Mental Causation and Ontology

This page intentionally left blank

Mental Causationand Ontology

edited byS. C. Gibb, E. J. Lowe,and R. D. Ingthorsson

1

3Great Clarendon Street, Oxford, ox2 6dp,United Kingdom

Oxford University Press is a department of the University of Oxford.It furthers the University’s objective of excellence in research, scholarship,and education by publishing worldwide. Oxford is a registered trade mark ofOxford University Press in the UK and in certain other countries

# the several contributors 2013

The moral rights of the authors have been asserted

First Edition published in 2013

Impression: 1

All rights reserved. No part of this publication may be reproduced, stored ina retrieval system, or transmitted, in any form or by any means, without theprior permission in writing of Oxford University Press, or as expressly permittedby law, by licence or under terms agreed with the appropriate reprographicsrights organization. Enquiries concerning reproduction outside the scope of theabove should be sent to the Rights Department, Oxford University Press, at theaddress above

You must not circulate this work in any other formand you must impose this same condition on any acquirer

British Library Cataloguing in Publication Data

Data available

ISBN 978–0–19–960377–0

Printed by the MPG Printgroup, UK

Links to third party websites are provided by Oxford in good faith andfor information only. Oxford disclaims any responsibility for the materialscontained in any third party website referenced in this work.

Contents

List of Contributors vii

Acknowledgements viii

Introduction 1

S. C. Gibb

Part 1: Levels of Being, Properties, and Mental Causation

1. Mental Causation 18

John Heil

2. Physical Realization without Preemption 35

Sydney Shoemaker

3. Mental Causation in the Physical World 58

Peter Menzies

4. Mental Causation: Ontology and Patterns of Variation 88

Paul Noordhof

5. Causation is Macroscopic but Not Irreducible 126

David Papineau

Part 2: Causal Relata, Substances, and Powers

6. Substance Causation, Powers, and Human Agency 153

E. J. Lowe

7. Agent Causation in a Neo-Aristotelian Metaphysics 173

Jonathan D. Jacobs and Timothy O’Connor

8. Mental Causation and Double Prevention 193

S. C. Gibb

9. The Identity Theory as a Solution to the Exclusion Problem 215

David Robb

10. Continuant Causation, Fundamentality, and Freedom 233

Peter Simons

11. There is No Exclusion Problem 248

Steinvor Tholl Arnadottir and Tim Crane

Index 267

vi CONTENTS

List of Contributors

Steinvor Tholl Arnadottir, University of Stirling

Tim Crane, University of Cambridge

S. C. Gibb University of Durham

John Heil, Washington University, St Louis

Jonathan D. Jacobs, St Louis University

E. J. Lowe, University of Durham

Peter Menzies, Macquarie University

Paul Noordhof, University of York

Timothy O’Connor, Indiana University

David Papineau, Kings College London

David Robb, Davidson College

Sydney Shoemaker, Cornell University

Peter Simons, Trinity College Dublin

Acknowledgements

We are very grateful indeed to the Arts and Humanities Research Council

(AHRC) for funding the research project on the new ontology of the

mental causation debate, of which this book is a consequence. (Arts and

Humanities Research Council’s Research Grant AH/F009615/1 ‘The New

Ontology of the Mental Causation Debate’.) We would also like to thank

the AHRC, the Mind Association and the Analysis Trust for helping to

fund the conference held at Durham University on the ontology of mental

causation, at which many of the contributors to this volume presented

their work.

Introduction

S. C. GIBB

The hypothesis of this book is that many of the central issues in the mental

causation debate are metaphysical, where metaphysics is to be understood as

the branch of philosophy that is concernedwith the fundamental structure of

reality as a whole. More specifically, we propose that it is ontology—the

domain of metaphysics which concerns what categories of being there are

and how they are related to one another—that is of particular concern to

the mental causation debate. It seems clear that, insofar as all branches of

philosophy deal with aspects of reality, they all incorporate—whether expli-

citly or implicitly—ontological claims. This is no less true of the philosophy

of mind and, in particular, the debate about mental causation. If this is the

case, then solutions to the problem of mental causation that attempt to

divorce themselves from ontology or which are based on ad hoc ontological

assumptions will inevitably prove to be inadequate. Let us begin by con-

sidering what the problem of mental causation is a problem about.

The thought that we are capable of performing intentional actions which

result in the movement of our limbs is central to our pretheoretical concep-

tion of human agency. It is my belief that it is going to rain together with my

desire not to get wet that causes me to run indoors. It is my desire to quench

my thirst in combinationwith my belief that there is a Coke in the fridge that

causes me to open the fridge door. However, despite the initial plausibility of

the claim that the mental is of causal relevance in the physical domain,

specifying a relation between the mental and the physical that is both

consistent with psychophysical causation and independently plausible has

proven to be one of the most enduring problems in the philosophy of mind.

Causal closure arguments appeal to the existence of psychophysical

causation as a premise, along with a further premise that the physical domain

is in some sense causally closed. All such arguments share the conclusion that

mental causes (that have physical effects) are identical with physical causes.

The structure of a causal closure argument will differ according to the

strength of causal closure principle that it appeals to. The most popular

form of causal closure argument, owing largely to the relative weakness of

the causal closure principle that it requires, is the argument from causal

overdetermination. It can be set out as follows:

(1) Some mental events are causally relevant in the physical domain.

(Call this ‘the principle of psychophysical causation’.)

(2) At every time at which a physical event has a cause it has a sufficient

physical cause. (Call this ‘the principle of causal closure’.)1

(3) There is no systematic causal overdetermination. (Call this ‘the

principle of causal non-overdetermination’.)

Mental events (that are causally relevant in the physical domain) are

identical with physical events.2

To explain the argument: In accordance with the principle of psycho-

physical causation, say thatM1 is a mental event existing at time t and thatM1

is a cause of physical event P2. The principle of causal closure states that at

every time at which a physical event has a cause it has a sufficient physical

cause, where for event e1 to be causally sufficient for event e2 is for e1 to

causally necessitate e2. Given the principle of causal closure, it follows that as

P2 has a cause at t1, it must have a sufficient physical cause at t1. Call this

sufficient physical cause P1. Now the principle of causal closure does not

entail that M1must be a physical event, despite the fact thatM1 has a physical

1 Note that if (2) were to be substituted for the weaker claim that ‘Every physical event that has a cause

has a sufficient physical cause’, then, provided that causation is transitive, the combination of the

premises of the argument from causal overdetermination would not be enough to establish its conclu-

sion. This is because, given the transitivity of causation, a physical event would have a sufficient physical

cause if it had a sufficient mental cause which in turn had a sufficient physical cause. If a principle of

causal closure allows mental events this causal role in the physical domain, then the three premises of the

argument from causal overdetermination are quite clearly compatible with various forms of dualism.

(2) avoids this problem because, according to it, at every time at which a physical event has a cause it has a

sufficient physical cause. However, it is arguable that an appeal to the possibility of simultaneous

causation demonstrates that even (2) is to too weak to establish the argument from causal overdetermi-

nation’s conclusion. For a detailed discussion of all of these points, see Lowe (2000a).2 For various formulations of this three-premise argument, see Hopkins (1978), Crane (1995), Lowe

(2000b), and Papineau (2001), among others. The formulation that I have presented here is Lowe’s (see

Lowe 2000b, 27).

2 S . C. GIBB

effect. Even if it is the case that at every time at which a physical event has a

cause it has a sufficient physical cause, some physical events might have non-

physical causes as well. It is only the principle of causal closure in combin-

ation with the principle of causal non-overdetermination that entails that

non-physical events are causally irrelevant in the physical domain. Given the

latter principle then arguably, as a general rule, if an event e has a sufficient

cause c at t, no event at t distinct from c can be a cause of e. It follows that if P1

is a sufficient cause of P2 at t1 andM1 is also a cause of P2 at t1, thenM1must

be identical with P1 or part of a cause that is identical with P1.

Both the principle of causal closure and the principle of causal non-

overdetermination enjoy general support in the mental causation debate.

The principle of causal closure is widely accepted, largely as a result of the

empirical support that it is thought to enjoy—that is, the great number of

cases in current physics that confirm this principle and the lack of any

disconfirming cases. Current physics, it is argued, has found sufficient

physical causes for many different kinds of physical events. Furthermore,

physics has never needed to appeal to a non-physical cause to provide a

sufficient cause for a physical event. This, it is thought, provides inductive

evidence for the claim that for every physical event, at every time at which it

has a cause, it has a sufficient physical cause.

Regarding the principle of causal non-overdetermination, depending on

one’s theory of causation, one may of course want to allow that there are

special cases in which events are causally overdetermined. For example, if

two shots are independently fired and each bullet reaches the victim’s heart

at exactly the same time, provided that each bullet striking was causally

sufficient for the victim’s death, the victim’s death was arguably causally

overdetermined by the shootings. However, even if there are isolated cases

of causal overdetermination, the claim that as a general rule some events are

causally overdetermined seems implausible. It is this claim that the principle

of causal non-overdetermination rules out—according to it, events are not

systematically causally overdetermined. And it is precisely this kind of sys-

tematic causal overdetermination that the combination of the principle of

psychophysical causation and the principle of causal closure seems to give

rise to, unless, of course, mental events and physical events are identified.

Despite the apparent plausibility of the argument from causal overdeter-

mination, the acceptance of its conclusion is problematical. Largely as a

result of the argument from multiple realizability—the argument that

INTRODUCTION 3

mental properties are multiply realized by, and hence cannot be identical

with, physical properties—most philosophers in the mental causation debate

would want to accept the following principle:

The principle of property dualism: Mental properties are not identical with physical

properties.

The acceptance of the principle of property dualism is, given certain accounts

of an event, directly inconsistent with the conclusion of the argument from

causal overdetermination. If, for example, events are property-instantiations

and the identity of property-instantiations requires the identity of their

properties, then the principle of property dualism entails that mental events

cannot, contrary to the argument from causal overdetermination, be physical

events. Thus, for example, take Kim’s account of an event.3 According to it,

the event (s1, p1, t1) exists just in case the substance s1 instantiates the property

p1 at time t1. A mental event is the instantiation of a mental property by a

substance at a time. A physical event is the instantiation of a physical property

by a substance at a time. Events (s1, p1, t1) and (s2, p2, t2) are identical if and

only if s1= s2, p1= p2, and t1= t2. Hence, the identity ofmental and physical

events requires the identity of mental and physical properties. Thus the

acceptance of the principle of property dualism entails the rejection of the

conclusion of the argument from causal overdetermination.

One might instead attempt to advance an account of events which allows

one to combine an event monism with a property dualism—events involve

properties, but the identity of events does not require the identity of these

properties. Given such an account, the principle of property dualism does

not entail the rejection of the conclusion of the argument from causal

overdetermination. Mental events may be physical, even though mental

properties are not. But even if it is combined with an event monism,

maintaining a property dualism still presents a problem given the argument

from causal overdetermination, as questions about the causal redundancy of

the mental simply arise at the level of properties, as opposed to the level of

causes. Events are presumably causes in virtue of the properties that they

involve—it is, in other words, the properties of an event that make the

causal difference. Furthermore, events are rarely causes in virtue of all of

their properties. Consequently, the question of whether a mental event is

3 See, for example, Kim (1993a).

4 S . C. GIBB

ever causally relevant within the physical domain in virtue of its mental (and

hence, non-physical) properties arises. If one denies mental properties such

causal efficacy, then one abandons any serious commitment to the principle

of psychophysical causation. Alternatively, if one allows them such causal

efficacy, then one seems forced to abandon either the principle of causal

closure or the principle of causal non-overdetermination. For these reasons,

it seems that one must either abandon the principle of property dualism or

provide a reason to reject the argument from causal overdetermination.

With the exception of the psychophysical reductionist, most of those in the

mental causation debate take the second option, and reject or disambiguate

one of the premises of the argument from causal overdetermination. Hence,

for example, eliminativists reject the existence of mental entities, and hence,

mental causes. They therefore reject the principle of psychophysical causation.

Interactive substance dualism and anti-physicalist forms of property dualism

typically deny the principle of causal closure. Most forms of non-reductive

physicalism deny or disambiguate the principle of causal non-overdetermin-

ation. Thus some argue that the combination of the principle of psychophys-

ical causation, the principle of causal closure, and the principle of property

dualism does not, given non-reductive physicalism, result in a systematic causal

overdetermination of the worrying type. For it to do so, it would have to be

the case that the effect would have occurred if only one of the causes was

present. On the non-reductive physicalist’s account, mental and physical

causes are not metaphysically independent—mental properties ontologically

depend on physical properties—and hence this condition cannot be satisfied.

But all of this is only to paint a partial picture of the possible responses to

the problem. An alternative kind of response is to argue that the acceptance

of the three premises of the argument from causal overdetermination does

not entail that mental causes are physical causes, and hence does not conflict

with the principle of property dualism. It is only given the implicit accept-

ance of further claims—claims that are themselves open to question—that

this would be so. Hence, for example, dualist attempts to respond to the

problem of mental causation by denying the homogeneity of the causal

relata are of this type, as are attempts to respond to the problem by denying

the homogeneity of the causal relation.4

4 See Lowe (1993), (2000a), and (2008) for the first kind of position. See Crane (1995), }9 for a

discussion of the second kind of approach.

INTRODUCTION 5

Despite the great amount of effort that has been devoted to responding to

the problem of mental causation, a number of central figures in the mental

causation debate have questioned whether we are any closer to a satisfactory

solution or, indeed, given the current form of the debate, whether we will

ever be. Hence, Kim concludes that those in the mental causation debate are

‘up against a dead end’ (Kim 1993b, 367), while McGinn suggests that a

solution to the problem of mental causation is forever beyond our reach

(McGinn 1989).

It is our hypothesis that the apparent insolubility of the problem of mental

causation stems, in part, from the fact that the debate has not always been

framed with sufficient ontological precision. Basing it in a plausible onto-

logical framework will constitute, we contend, a significant step towards its

resolution. Not only will this add clarity to the existing debate and serve to

identify those positions that are ontologically unsound, but it will also

potentially reveal new ways of responding to the problem of mental caus-

ation that have not been previously explored.

The metaphysical issues that are of relevance to the mental causation

debate fall into three interrelated groups. The first concerns the nature of

the causal relata. The second concerns the nature of a property. The third

concerns the nature of the causal relation. Let us consider each of these

issues in turn.

Assuming that causation is a relation, what is it a relation between? What

one takes the causal relata to be depends on the ontological system in which

one is basing one’s account of causation, and hence the ontological categories

to which one appeals. Making the plausible assumption that singular caus-

ation is a relation between particulars, there are several possible candidates for

the role. These include events, states of affairs, individual substances, and

tropes. In formulating the argument from causal overdetermination, we have

assumed that causation is ‘event causation’—that is, that causation is a

relation between events. Furthermore, in explicating what it is to maintain

an event dualism we have assumed that events involve properties. These

assumptions are open to question, and whether or not they are correct

impacts on the mental causation debate. Davidson’s theory of anomalous

monism and the discussion that surrounds it serves to provide an excellent

demonstration of this point (Davidson 1980b, 1980c, 1980d, 1993).

According to Davidson’s account of mental causation, mental events reduce

to physical events, but mental concepts do not reduce to physical concepts.

6 S . C. GIBB

Like the argument from causal overdetermination, Davidson’s argument for

the conclusion that at least some mental events are identical with physical

events has three premises and appeals to a premise that is similar to the first

premise of the argument from causal overdetermination:

(1) The Principle of Causal Interaction (CI): Some mental events and

physical events are related as cause and effect.

(2) The Principle of the Nomological Character of Causality (NCC):

Events related as cause and effect fall under strict laws.

(3) The Principle of the Anomalism of the Mental (AM): There are no

strict psychophysical laws.

To explain this argument, in accordance with CI, say that mental event, M,

and physical event, P, are related as cause and effect. Given NCC, it follows

that M and P must be characterizable in terms that allow them to fall under a

strict law. Given AM, if M is only characterizable in mental terms, M and

P cannot fall under a strict law. Consequently, M must also be characteriz-

able in physical terms. For this to be the case, M must be a physical event.

Furthermore, as there are no strict psychophysical laws that would support

the reduction of mental concepts to physical concepts, anomalous monism

leads to the rejection of any conceptual reduction.

Now anomalous monism is often accused of resulting in a property

epiphenomenalism.5 Davidson appears to be combining an event monism

with a property dualism. The worry, therefore, is how, according to

anomalous monism, the mental properties of an event can be causally

efficacious in the physical domain. More specifically, how can anomalous

properties of an event be causal properties given NCC?

If causal relations obtain between events because events involve certain

properties, this problem is arguably insoluble. However, unlike his critics,

Davidson does not consider events to involve properties. For Davidson,

properties are not objective aspects of things in the world. What makes an

event mental (or physical) is whether or not it has a mental (or physical)

description. Although a mental (or physical) description of an event is either

true or false of it, there is no ontological fact about or feature of an event that

makes its description as mental (or physical) true or false.6 Given this

5 See, for example, Honderich 1982, Kim 1993c, McLaughlin 1993, and Sosa 1993.6 See, for example, Davidson 1980b, 215.

INTRODUCTION 7

account of an event, it makes no sense to suggest that events are causes in

virtue of their properties. Therefore, anomalous monism cannot be accused

of property epiphenomenalism. It is arguably true that if one were to

combine, for example, a Kimean theory of events with anomalous monism,

then the resulting position would result in a property epiphenomenalism.

But Davidson’s account of mental causation should not be detached from

his account of an event. Of course, whether the ontological system in which

Davidson bases anomalous monism is plausible is a further issue. My point

here is simply that, given this ontological system, anomalous monism cannot

be accused of property epiphenomenalism.7

Even if one accepts, contra Davidson, that a cause is a cause in virtue of

the properties that it involves, there are still many further questions to be

raised regarding the nature of the causal relata. The first set of questions

concerns the nature of each of the categories which one’s account of the

causal relata invokes. Take, for example, the popular claim that events are

the causal relata, where events are property-instantiations—that is the

instantiation of a property by a substance at a time. Several questions arise

regarding this account: Are the properties that substances instantiate uni-

versals or particulars? What is the ontological status of substance—are

substances reducible to ‘bundles’ of properties or is this ontological category

fundamental? If the category of substance is fundamental, then what is

its nature? Is a substance to be understood as a ‘substratum’ or ‘bare

particular’—that is, as an entity which in itself has no properties, but

which plays the role of supporting and uniting an object’s properties? This

would suggest an ontology in which states of affairs are the basic building

blocks of reality. This is because, as Armstrong argues, given this notion of

substance, something is needed to weld substance and property together,

and it is states of affairs that most plausibly play this role. Where a substance a

instantiates property F, a’s being F is a state of affairs, and a and F are

constituents of the state of affairs which it holds together in a non-mereo-

logical form of composition. If so, then property-instantiations are states of

affairs.8 Alternatively, substances might be Aristotelian in nature.9

According to this account, a substance is not something that is wholly

separate from its properties in the way that a bare particular is (and for this

7 See further Crane (1995) and Gibb (2006). 8 See Armstrong (1997).9 See, for example, Lowe (2006) for such an account.

8 S . C. GIBB

reason, states of affairs are arguably not required to weld substance and

property together). But nor should it therefore be concluded that a sub-

stance just is a state of affairs, for unlike states of affairs, substances are not

complex entities. Rather, according to this account, substances comprise a

basic, irreducible category, which do not have properties as constituents,

but upon which properties ontologically depend. Property-instantiations

are therefore not to be identified with states of affairs, but with states of

substances.

A further set of questions concern the homogeneity of the causal relata. Is

all causation a relation between entities of the same kind? According to

several philosophers, it is not. Some advance an account of events and of

facts that allows them to distinguish between event causation and fact

causation. That is, causation is neither exclusively a relation between events

nor exclusively a relation between facts, but instead event causation and fact

causation are both basic and irreducible—either an event or a fact can cause

either an event or a fact.10 Alternatively, rather than drawing a distinction

between event causation and fact causation, others propose a distinction

between event (or fact) causation and agent causation. In the case of agent

causation, an agent causes an event (or fact).11 An ‘agent’ is to be understood

as an individual substance—whether the term is to be restricted to those

individual substances that are capable of performing intentional actions is a

matter for discussion.

These issues regarding the nature of the causal relata all have a bearing on

the mental causation debate. Indeed, the importance of some of them has

already been made clear by those in the debate. Robb’s ‘trope solution’

provides one such example (Robb 1997). Robb argues that if a cause is the

instantiation of a trope by a substance, as opposed to the instantiation of a

universal by a substance, then this resolves the problem of mental causation.

According to Robb, if properties are tropes then one can accept the

conclusion of the argument from causal overdetermination whilst avoiding

the problem of multiple realization—the denial of the principle of property

dualism is only problematical if properties are universals.

Lowe’s dualist response to the problem of mental causation provides

another example of the importance of the issue of the causal relata to the

10 See, for example, Bennett (1988).11 See, for example, O’Connor (2000) and Taylor (1966).

INTRODUCTION 9

mental causation debate.12 Lowe accepts the premises of the argument from

causal overdetermination, whilst rejecting its physicalist conclusion. He

does this by defending a psychophysical causal interactionism that rests on

a distinction between event causation and fact causation. According to

Lowe, mental events do not cause physical events. Rather mental events

cause physical facts. This, Lowe argues, is wholly consistent with each of the

premises of the argument from causal overdetermination.

Let us now turn to the second issue—that of property analysis. For those

in the mental causation debate who admit the ontological category of

property, the issue of how to analyse properties is a central one. What is it

for a property to exist? What is it for one property to be identical with

another? These questions matter to the mental causation debate at the most

general of levels. To determine whether mental properties exist, and, hence

whether mental causes exist, one must adopt some criterion of property

existence. To determine whether a mental property is identical with a

physical property, one must adopt some criterion of property identity.

A further closely related issue concerns the different kinds of relationship

that properties can stand in—both to each other and to items of any other

ontological category that one would want to admit. Of particular import-

ance to the mental causation debate, and more specifically to non-reductive

physicalism, is the question of whether properties can be said to depend

upon each other and, if so, in what sense.

The fundamental importance of property analysis to the mental causation

debate is made clear by Heil’s arguments regarding the mental causation

debate and levels of being.13 As Heil observes, a familiar point—but one that

is not always sufficiently respected in the mental causation debate—is the

importance of distinguishing property-types from predicates (see, for

example, Heil 2003, Ch. 3). A plausible analysis of the existence and identity

conditions for properties reveals that not every meaningful predicate picks

out a real property-type. This is the case regardless of whether properties are

universals or particulars and, hence, regardless of whether a property-type is

a universal or, for example, a set of resembling tropes. Whilst few in the

mental causation debate would admit to conflating predicates with proper-

ties, Heil considers that some of the central arguments and assumptions in

this debate are guilty of just such a conflation. In particular, Heil considers

12 See, for example, Lowe (1993). 13 See, for example, Heil (2003, 2004).

10 S . C. GIBB

non-reductive physicalism to be guilty of confusing the idea that mental and

physical predicates operate at different levels of description, with the idea

that mental and physical properties exist at different ontological levels.

Now, whether this criticism of non-reductive physicalism is correct is, of

course, a matter for debate. What seems to be clear, however, is that one

cannot take part in this debate without engaging in the analysis of a

property’s existence and identity conditions.

Finally, we would suggest that the mental causation debate is affected by

what the causal relation is. Some theories of causation say that c causes e

when e is counterfactually dependent on c (that is, if c had not existed, then

e would not have existed). Others maintain that for c and e to be causally

related they must exemplify types that are lawfully connected, but disagree

about what this lawful connection consists in. Still yet others maintain that

causation is the transference of some quantity from cause to effect, for

example, the transference of energy or momentum. Alternatively, others

consider that causation is to be accounted for in terms of irreducible causal

powers. Properties confer causal powers on the substances that they char-

acterize. Causation is the exercise of these causal powers. These are but a

few of the many different theories of causation that have been presented in

the literature on causation.

All theories of causation fall into one of two groups—those that consider

causal laws to be more basic than causal relations and those that consider

causal relations to be more basic than causal laws. In line with this, one may

distinguish between generalist and singularist accounts of causation.

A singularist denies that singular causation is grounded by type-level rela-

tions, while generalists maintain that for particular events to be causally

related at the token level, they must instantiate types of events that bear

suitable objective relations to one another. Most plausibly, laws furnish

causation’s generalist component. For example, the regularity theory of

causation is a generalist theory of causation. According to it causation is

grounded by laws, where laws are regularities. By contrast, the energy

transference theory of causation is a singularist account of causation—

according to it causation is not grounded by type-level relations.

There is also a narrower understanding of what a generalist theory of

causation is. According to this understanding, whether or not two particular

events are causally connected depends upon things that happen elsewhere or

elsewhen—causation is not a purely local matter. Contrary to this, others

INTRODUCTION 11

consider causation to be a purely local matter. The energy transference

theory of causation, for example, adopts the latter position—whether or not

two events are causally connected simply depends upon whether there is

transference of a quantity between them. Whether there is such a transfer

does not depend upon what is happening elsewhere or elsewhen. By

contrast, the regularity theory of causation adopts the former position.

According to it, a particular sequence is causal by virtue of being an instance

of a general pattern and the same particular sequence coupled with no

general pattern would not be causal. It is, however, quite clearly the case

that not all nomological theories of causation are generalist in this narrow

sense.

Here we have mentioned but a few of the more popular theories of the

causal relation. However, what should be obvious from this brief and partial

summary is just how much theories of causation can differ. It seems

reasonable to raise the question of whether any of these differences matter

to the mental causation debate.

Some of the main issues that might be raised regarding the mental

causation debate and theories of causation can be recognized by focusing

on the principle of causal closure. Very few theories of the causal relation

entail that causation must be a physical relation—one exception is an energy

transference theory of causation that identifies ‘energy’ with ‘physical

energy’. But, from the fact that few theories of the causal relation entail

that causation is physical causation, it would be far too hasty to conclude

that one’s theory of causation therefore does not affect the plausibility of the

principle of causal closure.

We have formulated the principle of causal closure as the claim that at

every time at which a physical event has a cause it has a sufficient physical

cause. What it is to be a ‘cause’ and what it is to be an ‘effect’, and thus what

it is to be a sufficient cause of an effect, depends on one’s theory of the causal

relation. Could it therefore be the case that viewed from the perspective of

one theory of causation, the empirical evidence suggests that at every time at

which a physical event has a cause it has a sufficient physical cause, whilst

viewed from the perspective of another theory of causation which entails a

broader understanding of what it is to be a cause, the very same empirical

evidence suggests no such thing?

A further issue concerns whether it is correct to assume, as this formula-

tion of the principle of causal closure does, that every physical event that has

12 S . C. GIBB

a cause has a sufficient cause. More specifically, is the principle’s assumption

that complete causes are always sufficient for their effects (where a cause is

sufficient for an effect if it causally necessitates it) correct according to all

theories of causation? Or, do some theories of causation allow cases in

which effects do not have sufficient causes? If so, does this leave a potential

gap in the argument from causal overdetermination?

Finally, even if one accepts the principle of causal closure, do certain

accounts of the causal relation still provide reasons to question other aspects

of the argument from causal overdetermination? Is the principle of causal

non-overdetermination equally plausible under various accounts of the

causal relation? Do certain accounts of the causal relation raise issues

regarding how this principle should be formulated? Is the argument from

causal overdetermination’s assumption that the causal relation is homo-

geneous correct? Or, do certain accounts of the causal relation allow one

room to question whether mental and physical causation differ, not merely

in what they relate, but also in the kind of causation that they involve?

Here, we have presented some of the issues that might be raised regarding

the ontology of the mental causation debate. We invited philosophers both

from within the mental causation debate and from within metaphysics to

think about the ontology of the mental causation debate. This book is a

product of what they think. The papers fall into two groups—those that

focus on issues relating to non-reductive physicalism and those that are

concerned with providing an alternative account of mental causation based

either on the acceptance of a powers ontology and/or a specific account of

the causal relata. All of the papers provide crucial insights into the relevance

of the causal relata or property analysis or the causal relation to the mental

causation debate, and, indeed, some of the papers touch upon all three of

these ontological areas.

John Heil’s paper ‘Mental Causation’ serves as an ideal point of entry into

the issues concerning the ontology of mental causation, highlighting some

of the questionable metaphysical assumptions that have arguably played a

role in the mental causation debate. One of Heil’s central aims in this paper

is to show that, given a proper understanding of properties, it is plausible

that the mental–physical distinction is not ontologically deep—that the

distinction is a distinction of conception only. This draws into question

appeals to levels of being—and, more specifically, non-reductive physicalist

attempts—to solve the problem of mental causation.

INTRODUCTION 13

Unlike Heil, the next three authors are united in considering that non-

reductive physicalism provides an attractive response to the problem of

mental causation. However, their accounts of non-reductive physicalism

differ greatly. According to Sydney Shoemaker’s account of non-reductive

physicalism, mental states are distinct from, but realized by, physical states,

where ‘realization’ is to be understood according to the subset account of

realization. In ‘Physical Realization without Preemption’, Shoemaker both

expands and corrects the account of the realization relation between mental

and physical states that he presented in his book Physical Realization.

In ‘Mental Causation in the Physical World’, Peter Menzies presents a

new version of Kim’s exclusion argument that targets all forms of physical-

ism, both reductive and non-reductive.14 Focusing on the principle of causal

exclusion to which this argument appeals, Menzies explores the extent to

which this principle is supported by different theories of the causal relation.

He argues that within the framework of a difference-making account of

causation, the principle is false. However, under special conditions, a

more plausible version of the principle of causal exclusion is supported

by this account of the causal relation. Menzies goes on to consider the

consequences—both surprising and encouraging—that this has for non-

reductive physicalism.

Paul Noordhof agrees with neither of these non-reductive physicalist

approaches. In ‘Mental Causation: Ontology and Patterns of Variation’ he

instead takes elements of those accounts which seek to understand the

efficacy of the mental non-ontologically in terms of patterns of variation

and elements of those accounts which attempt to resolve the problem of

mental causation by identifying mental property instances with physical

property instances, to motivate his own distinctive kind of non-reductive

physicalism.

David Papineau in ‘Causation is Macroscopic but Not Irreducible’ argues

that causation is not physically fundamental, but instead an essentially

macroscopic phenomenon. Causal processes are, according to Papineau,

analogous to thermodynamic processes, depending on probabilistic facts

about the ways in which macro-states are realized at the micro-level. This

is compatible with the popular non-reductive physicalist claim that causes

should be ‘proportional’ to their effects, and that the requirements of

14 For Kim’s version of this argument, see, for example, Kim (2005).

14 S . C. GIBB

proportionality can favour mental states as causes over their physical real-

izers. However, Papineau argues that these claims do not vindicate non-

reductive physicalism. He defends the thesis that mental properties are

instead reducible to some common physical feature of their realizers.

The first five papers in the second part of this book, unlike many of the

papers in the previous part, reject non-reductive physicalism. The different

accounts of mental causation that they advance hinge either on their

acceptance of a powers ontology and/or their understanding of the causal

relata. The first three papers place their discussion of the mental causation

debate in the context of recent developments in the ontology of powers.

In ‘Substance Causation, Powers, and Human Agency’, E. J. Lowe bases

his discussion within the general framework of a power-based theory of

substance causation. Lowe argues that the human will is a unique kind

of spontaneous power that is distinguished by the facts that it is a ‘two-way’

power and that it can be exercised rationally. Having such a power provides

human agents with all of the control that they could need over their

voluntary actions. Moreover, to deny that we have such a power is to

deny our own rationality.

Jonathan Jacobs and Timothy O’Connor’s paper ‘Agent Causation in a

Neo-Aristotelian Metaphysics’ explores how embracing a neo-Aristotelian

metaphysics of causal powers affects accounts of the nature of metaphysical

freedom. Jacobs and O’Connor argue that different versions of the neo-

Aristotelian metaphysics result in different metaphysical accounts of free and

responsible action. Consequently, action theory cannot properly be pursued

in isolation from general metaphysics.

In ‘Mental Causation and Double Prevention’, S. C. Gibb offers a new

solution to the problem of mental causation which has emerged from her

acceptance of a powers theory of causation. Gibb proposes that the causal

role of mental events in the physical domain is to serve as ‘double prevent-

ers’. She argues that if mental events are double preventers, and a powers

theory of causation is accepted, then the premises of the argument from

causal overdetermination can be reconciled with dualism.

David Robb’s response to the problem of mental causation depends on

his account of the causal relata, and, more specifically, his understanding of

the nature of a property. According to Robb, if properties are tropes then

this allows one to accept the conclusion of the argument from causal

overdetermination. In ‘The Identity Theory as a Solution to the Exclusion

INTRODUCTION 15

Problem’ he elaborates on this solution and responds to various objections

that have been raised against it.

Like Robb’s account, Peter Simons’ response to the problem of

mental causation arises from careful consideration of the causal relata. His

paper ‘Continuant Causation, Fundamentality, and Freedom’ distinguishes

between ‘continuant causation’ and ‘occurrent causation’. Simons maintains

that in the case of mental causation, the cause is not an event or other

occurrent, but instead a continuant. Hence, mental causation is continuant

causation. Simons goes on to argue that continuant causation is not funda-

mental, but is instead ontologically secondary to occurrent causation.

In the final paper, ‘There is no Exclusion Problem’, Steinvor Tholl

Arnadottir and Tim Crane argue that, contrary to many of the papers in

this book, to resolve the problem of mental causation one does not need to

take a stand on any controversial issues regarding either the nature of

causation or the ontology of the mental. A non-reductive physicalist or an

emergentist who accepts that mental and physical causes are not independ-

ent of one another does not face the problem of mental causation. They

would only face the problem if it were the case that an event cannot have

distinct but dependently sufficient causes. This is implausible. One does not

need to make any heavy-duty ontological claims to establish this.

References

Armstrong, D. (1997). A World of States of Affairs. Cambridge: Cambridge Univer-

sity Press.

Bennett, J. (1988). Events and their Names. Oxford: Clarendon Press.

Crane, T. (1995). ‘The Mental Causation Debate’. Proceedings of the Aristotelian

Society, Supp. Vol. 69: 211–36.

Davidson, D. (1980a). Essays on Actions and Events. Oxford: Clarendon Press.

—— (1980b). ‘Mental Events’. In Davidson (1980a): 207–28.

—— (1980c). ‘Psychology as Philosophy’. In Davidson (1980a): 229–44.

—— (1980d). ‘The Material Mind’. In Davidson (1980a): 245–60.

—— (1993). ‘Thinking Causes’. In Heil and Mele (1993): 3–18.

Gibb, S. C. (2006). ‘Why Davidson is not a Property Epiphenomenalist’. Inter-

national Journal of Philosophical Studies, 14(3): 407–22.

Heil, J. (2003). From an Ontological Point of View. Oxford: Clarendon Press.

16 S . C. GIBB

—— (2004). Philosophy of Mind. Cambridge: Cambridge University Press.

—— A. R. Mele (eds.) (1993). Mental Causation. Oxford: Clarendon Press.

Honderich, T. (1982). ‘The Argument for Anomalous Monism’. Analysis, 42: 59–64.

Hopkins, J. (1978). ‘Mental States, Natural Kinds and Psychophysical Laws’.

Proceedings of the Aristotelian Society, Supp. Vol. 52: 221–36.

Kim, J. (1993a). ‘Events as Property Exemplifications’. In Supervenience and Mind.

Cambridge: Cambridge University Press: 33–52.

—— (1993b). ‘Postscripts on Mental Causation’. In Supervenience and Mind. Cam-

bridge: Cambridge University Press: 358–68.

—— (1993c). ‘Can Supervenience and “Non-Strict Laws” Save Anomalous

Monism?’. In Heil and Mele (1993): 19–26.

—— (2005). Physicalism, or Something Near Enough. Princeton: Princeton Univer-

sity Press.

Lowe, E. J. (1993). ‘The Causal Autonomy of the Mental’. Mind, 102: 629–44.

—— (2000a). ‘Causal Closure Principles and Emergentism’. Philosophy, 75: 571–86.

—— (2000b). An Introduction to the Philosophy of Mind. Cambridge: Cambridge

University Press.

—— (2006). The Four-Category Ontology: A Metaphysical Foundation for Natural

Science. Oxford: Oxford University Press.

—— (2008). Personal Agency: The Metaphysics of Mind and Action. Oxford: Oxford

University Press.

McGinn, C. (1989). ‘Can We Solve the Mind–Body Problem?’Mind, 98: 349–66.

McLaughlin, B. (1993). ‘On Davidson’s Response to the Charge of Epipheno-

menalism’. In Heil and Mele (1993): 27–40.

O’Connor, T. (2000). Persons and Causes: The Metaphysics of Free Will. New York:

Oxford University Press.

Papineau, D. (2001). ‘The Rise of Physicalism’. In C. Gillett and B. M. Loewer

(eds.), Physicalism and its Discontents. Cambridge: Cambridge University Press:

3–36.

Robb, D. (1997). ‘The Properties of Mental Causation’. Philosophical Quarterly, 47:

178–94.

Sosa, E. (1993). ‘Davidson’s Thinking Causes’. In Heil and Mele (1993): 41–50.

Taylor, R. (1966). Action and Purpose. Englewood Cliffs, NJ: Prentice-Hall.

INTRODUCTION 17

1

Mental Causation*

JOHN HEIL

Howcouldmental events, states, or properties causally affect, or be affected by,

physical events, states, or properties? The question persists even though few

philosophers nowadays find themselves attracted to the kind of uncomprom-

ising dualism associated with Descartes, the most prominent source of the

modern mind–body problem. “Physicalism” rules. Mental items are, at the

very least, taken to be ‘dependent on’ physical goings-on.Cartesian sentiments

linger, however, in the widespread acceptance of a sharp-edged distinction

between mental properties and physical properties. “Non-reductive physical-

ists,” for instance, hold that, although mental properties are in some way

wholly dependent on physical properties, mental properties are nevertheless

distinct from physical properties. Physical properties, or their instances, ‘real-

ize’ mental properties, or their instances. A denial of distinctness is taken to

require either reduction (the mental is reducible to the physical; mental

properties are physical properties) or outright elimination (nothing thinks,

perceives, or feels: mentality is an illusion).

By my lights, almost everything expressed in the previous paragraph stems

from philosophical confusion that finds its source in the linguisticization of

metaphysics. I do not claim that the problem ofmental causation can be swept

away by a few deft philosophical flourishes. I do claim, however, that we have

managed to distract ourselves—and our students—from the real problems.

My procedure will be to offer up without much argument a handful of

observations that, taken together, are meant to afford a measure of hope.

* The paper has benefited from comments from a number of people, especially S. C. Gibb and

E. J. Lowe.

Causation

The problem of mental causation has many faces. You might wonder

whether mental states or events cause, or are apt to cause physical states or

events. You might wonder whether mental states or events cause whatever

they cause in virtue of beingmental. The question I shall address, however, is

not whether mental goings-on cause, or are “causally relevant” to, physical

goings-on, but how this could be possible, how it might work. In answering

this question, it is not to the point to call up examples of accepted causal

explanations in psychology or neuroscience as philosophers of science have

sometimes done (see, e.g., Woodward 2008). Princess Elizabeth does not

doubt that minds and bodies interact. Her challenge to Descartes is to

account for mind–body interaction given Descartes’ commitment to dual-

ism. Nowadays the difficulty is, more often than not, taken to be that of

accounting for mental causation, or the “causal relevance” of the mental,

given the truth of non-reductive physicalism.

I believe that the prospects of reconciling mental causation and non-

reductive physicalism are vanishingly small. This I find not at all troubling

because I have little enthusiasm for non-reductive physicalism. Indeed

I regard non-reductive physicalism as a kind of blight on the late twentieth-

century philosophical imagination. Abandoning non-reductive physicalism is

not something many philosophers seem prepared to do, however. Those

philosophers owe us all an account of how mental causation could work if the

non-reductivist picture is correct.

I have noted that one unsatisfying kind of answer would feature appeals to

empirical researchwidely taken to supportmental–physical causal claims. The

issue before us is notwhether the mental bears causally on the physical, but how

this could be so given the non-reductive physicalist’s metaphysical picture.

Another equally unsatisfying answer would incorporate a weakening of

the notion of causation so as to get the right answer. On the one hand, you

have mental–physical causation. On the other hand, you have mental

properties or states being distinct from, but dependent on physical proper-

ties and states. You need only massage the concept of causation in a way that

reconciles these two facts about the world.

Philosophers who solve the problem of mental causation by taking caus-

ation to be bare counterfactual dependence (or counterfactual dependence

MENTAL CAUSATION 19

suitably constrained), provide an illustration of what I have in mind. What,

you might wonder, grounds the pertinent counterfactuals? What is it about

the world in virtue of which these counterfactuals hold true? If you allow the

counterfactuals to be “barely true”, and if you embrace the idea that causality

is solely a matter of counterfactual dependence, the mind–body problem, the

problem of how the mental and physical could causally interact, evaporates.

Such an account would rescue a Cartesian, but that very fact ought to arouse

suspicion. To the extent that you regard the mind–body problem as a genuine

problem for Descartes, you are thinking of causation as something more than

bare counterfactual dependence.

A third ontologically barren category of response involves appeal to

psychological or psychophysical ceteris paribus laws figuring in psychological

explanation. Grant that accepting such laws amounts to accepting psycho-

physical causation. We are now back with the question how such causation

could be possible if non-reductive physicalism is true. In every case we are

led back to the non-reductivist’s ontology.

Non-Reductive Physicalism

Mainstream non-reductive physicalists embrace three theses.

(1) Distinctness: mental properties are distinct from physical properties.

(2) Dependence: mental properties depend on physical properties.

(3) Autonomy: the physical realm is causally self-contained.

Taken together, these three theses make it hard to see how mental properties

could “make a causal difference” to physical goings-on.Autonomy implies that

physical effects have purely physical causes. Distinctness ensures that mental

properties are removed from the physical causal mix. Dependence seems to

render mental properties inefficacious even as regards mental effects: a mental

effect depends on a physical effect; a mental property could be “causally

relevant” to the production of a mental effect only by playing a role in the

production of the physical effect on which the mental effect depends.

Figure 1.1, associated with the work of Jaegwon Kim, will be familiar to

most readers (Kim 1978, 1979, 1982, 1984, 1993a, 2005). HereM1 andM2 are

mental property instances, states, or events, P1 and P2, physical “realizers” of

these. Vertical double-arrows * represent ‘vertical’ dependence relations,

20 JOHN HEIL

and the horizontal arrow ! indicates a causal relation. P1 figures unprob-

lematically in the production of P2. It looks as though, owing to autonomy,

M1 could play no role in the production of P2, and thus, given dependence,

no role in the production of M2.

An industry dedicated to resolving this problemwhile preserving (1)–(3) has

yielded an assortment of ingenious philosophical theses. Kim himself thinks

that the only solution is to identify mental properties with their physical

realizers. To the extent that this appears unpromising (as in the case of qualia),

mental phenomena must be epiphenomenal. Sydney Shoemaker (2001, 2003)

and Dirk Pereboom and Hilary Kornblith (1991; see also Pereboom 2002)

defend an account of the “realizing” relation that builds instances of, or causal

powers “bestowed” by, mental properties into their realizers. John Gibbons

(2006), taking off from Fodor’s (1997) conception of properties as figuring

essentially in causal laws, argues that mental causes could have mental effects

without compromising autonomy. Karen Bennett (2003) suggests that mental

causes could “over-determine” physical effects in an unobjectionable sense

consistent with autonomy (see also Yablo 1992; Mills 1996; Thomasson 1998).

And Cynthia and Graham Macdonald (2006) echo others, including LePore

and Loewer (1987), andWoodward (2008) in construing causation as a species

of counterfactual dependence: becauseM2 (and very likelyP2, aswell) depends

counterfactually onM1, we are entitled to acceptM1 as a causal factor.

I don’t think it would be much of an exaggeration to say that none of

these views has attained anything approaching universal support. Their costs

exceed their benefits.

“Mental” and “Physical”

The tenets of non-reductive physicalism revolve around the distinctness of

the mental and the physical. The key breakthrough, it has been thought, lay

in recognizing that distinctness can be consistent with dependence: mental

M1 M2

P1 P2→

⇑ ⇑

Fig 1.1

MENTAL CAUSATION 21

properties, although distinct from physical properties, in some way depend

on physical properties. This idea was expressed in the 1970s and 1980s in

terms of supervenience: the mental supervenes on the physical, a doctrine

traceable to Davidson’s “Mental Events”: “there cannot be two events alike

in all physical respects but differing in some mental respect, or an object

cannot alter in some mental respect without altering in some physical

respect” (Davidson 1970, 214).

For a while supervenience seemed to provide an elegant, cost-free solu-

tion to the problem of mental causation. If mental properties supervened on

physical properties, and physical properties figured unproblematically in

causal transactions, thenmental properties, too, could be creditedwith causal

efficacy. When this happens, we have “supervenient causation” (Sosa 1984).

Initial euphoria over supervenient causation quickly faded, however.

Supervenience yields dependence, but dependence is trumped by distinct-

ness. Frederick Stoutland (1976) was (to my knowledge) the first of many to

identify the problem of “causal relevance.” Even if you concede the “token

identity” of mental and physical events, you can still ask whether a given

event had a particular effect in virtue of that event’s being mental or in

virtue of its being physical. Suppose your experiencing a painful sensation is

“token identical” with your C-fibers’ firing, and suppose your C-fibers’

firing causes your arm to move in a particular way. You can still ask whether

the event that caused your arm to move did so qua physical or qua mental,

whether the event caused your arm to move in virtue of its physical

properties or in virtue of its mental properties. We have token identity,

but type diversity—distinctness. And type diversity, coupled with depend-

ence and autonomy, threatens what Brian McLaughlin (1989) dubbed “type

epiphenomenalism” (nicely captured by Figure 1.1).

Most parties to these discussions regard causal relations as holding among

distinct token events. Thus, in a token identity thesis, tokens are meant to

be particular events; types are properties. Davidson, who is often credited as

the inspiration behind non-reductive physicalism, is taken to defend token-,

but not type-identity: mental events are physical events, but mental prop-

erties are distinct from physical properties. Indeed, the three principles with

which I led off are directly traceable to Davidson’s defense of “anomalous

monism.” But what, for Davidson, distinguishes the mental from the

physical? What distinguishes mental and physical types?

22 JOHN HEIL

Before answering this question, let me call your attention to a remarkable

feature of much of the discussion of mental causation over the past forty

years. Although philosophers insist on distinctness—mental properties are

distinct from physical properties—scant attention has been paid to what the

distinction is (Mellor 1973 and Crane and Mellor 1990 represent notable

exceptions). Philosophers involved in the debate seem to think that the

distinction is so obvious as not to be worth further discussion. This stems

in part from views on the autonomy of psychology (and other higher-level,

special sciences). If psychology is not reducible to neurobiology, psycho-

logical propertiesmust be distinct from neurological or biological properties.1

Philosophers sometimes characterize properties by reference to predicates

discoverable in formulations of laws of nature. We can see that predicates

deployed in psychology differ in their application conditions from predi-

cates occurring in various physical sciences: physics, chemistry, biology. We

have no reason to think that psychological predicates could be analyzed in

terms of, translated into, or otherwise defined in a non-psychological

vocabulary. Psychological descriptions and explanations appear autono-

mous with respect to the physical sciences. Physical properties, then, are

properties named by predicates figuring in laws of nature investigated by the

various physical sciences; mental properties are those designated by respect-

able psychological predicates.

On a view of this kind, there will be hosts of properties other than

those designated by predicates of the fundamental physical sciences

(biological properties, meteorological properties, geological properties,

sociological properties), mental properties being only one class of these.

Non-reductivists are fond of pointing out that philosophers who doubt the

causal efficacy of mental properties are obliged to doubt as well the causal

efficacy of biological, meteorological, geological properties, and socio-

logical properties. Scientific practice renders such a view comical, an

expression of the worst sort of philosophical hubris.

Returning to the mental–physical distinction, the move, as I have char-

acterized it, is a move from differences in conception to distinctions in

reality. Such a move is resistible. You can accept fundamental differences in

1 And of course neurological or biological properties will be distinct from properties figuring in laws

of still lower level sciences, thus spawning a hierarchical conception of reality, an updated version of the

Great Chain of Being.

MENTAL CAUSATION 23

conception, while doubting that these reflect real distinctions, fundamental

differences in reality.2 This is how a philosopher such as Spinoza might

think of the mental–physical dichotomy. It is also how Davidson (who cites

Spinoza explicitly: 1970, 212) is thinking of it. Events, for Davidson, are

mental if they answer to mental descriptions, physical if they are physically

describable. Every mental event, every event satisfying a mental predicate,

Davidson thinks, is a physical event, an event describable in a physical

vocabulary. On such a view, the mental–physical distinction is not onto-

logically deep, not a real distinction.

Many readers will be surprised to hear such a view attributed to David-

son. But only a philosopher who hears “predicate” as “property,” only a

philosopher who regards predicates, or predicates ineliminably deployed to

express truths about the world, as invariably naming real properties, could

possibly read Davidson any other way.3 My aim is not to engage in

exegetical debates about Davidson, however, but merely to place on the

table a position that seems rarely to have been considered by metaphysically

inclined philosophers of mind. If the mental–physical distinction is a

distinction of conception only, then there is no pressing need to nail

down an interesting, non-circular way of distinguishing mental and phys-

ical properties. There are properties, all right, properties of objects that can

serve as truthmakers for claims about the world framed in a mental or a

physical vocabulary. If Davidson is right, whenever you have a true mental

description of an event, a description framed in a psychological vocabulary,

a description of the same event couched in the language of physics could, at

least in principle, be formulated—although not necessarily by the wielder

of the mental description. This is what supervenience boils down to for

Davidson.

A view of this kind ought not be counted “physicalist.” In no sense does

it privilege the physical. Mental events can be picked out using a physical

2 Descartes was acutely aware that many of his critics were skeptical of his defense of the “real

distinction.” Modern interpreters have understood the critics as arguing that one and the same substance

could possess both mental and physical properties. A view of this kind, supplemented by Dependence,

amounts to non-reductive physicalism. I think it likely that at least some of Descartes’ critics regarded the

mental–physical distinction as one of conception only, not a real distinction—between kinds of

substance or kinds of property. See Strawson 2008.3 In the literature on mental causation, it has been customary to slide from talk of predicates to

property talk. Davidson, in contrast, regards talk of properties as an oblique way of talking about

predicates; see Davidson (1993).

24 JOHN HEIL

vocabulary, but physical events can be referred to via mental descriptions.

Application conditions for mental predicates are orthogonal to those for

physical predicates, so there is no possibility of reduction—at least not if

reduction requires translation of statements in one vocabulary to statements

in another, or the expression of application conditions of predicates in one

vocabulary in the vocabulary of the other.

I have put all this in terms of descriptions and predicates, but really all you

need is the thought that worldly goings-on can be regarded in very different

ways for different purposes. This could be so without its being the case that

there is just one correct way to regard the world. We turn to physics to

provide something like an exhaustive description of the fundamental things.

But biologists, meteorologists, and sportscasters also manage to give us true

descriptions that fasten onto broad similarities and differences that serve us

well in making sense of and predicting features of the world around us.

These similarities and differences are perfectly objective, mind-independent

in the pertinent sense.

You might be appalled at the thought that the mental–physical distinc-

tion is not ontologically deep. But bear with me, and consider its implica-

tions for “non-reductive physicalism.” If the physical—that is, the world

described via a physical vocabulary—lacks priority, reduction is out of the

picture. With the threat of reduction off the table, the motivation for

regarding mental items—items picked out by means of mental terms—as

“higher-level” entities dependent on, but distinct from, lower-level phys-

ical items evaporates. Mental states and events can have physical conse-

quences. And, as Davidson notes, the question whether these consequences

had mental or physical causes cannot arise.

Token Identity, Type Diversity

Non-reductive physicalism is regarded by its proponents as embracing

token identity and rejecting type identity. If you thought of causation as a

relation among events (and who doesn’t?), the tokens would be particular

events.4 What of types? Non-reductivism takes types to be properties,

4 Do not read this as an endorsement of “event causation.” I have severe reservations about the idea

that causal relations are asymmetrical relations among events.

MENTAL CAUSATION 25

properties to be universals.5 An event is mental if it includes a mental

property, physical if it includes a physical property. If you think that the

mental is in some fashion “determined” by the physical, then every mental

event will be a physical event: every event including a mental property

includes, as well, a physical property.

This is the picture that gives rise to the problem of causal relevance.

Given that mental events are physical events, mental events could be causes

if physical events could be causes. Difficulties arise, however, when you ask

whether a given event had the effect it had in virtue of being a mental event

or in virtue of being physical. Your forming the intention to move your

finger (thereby flipping the switch) is identical with some event in your

nervous system. When your forming the intention to move your finger

causes your finger to move, does it do so quamental, qua being an intention,

or qua physical, qua being a particular kind of neurological event? The

question needs only to be asked to raise doubts that the event’s being mental

could have played a role in the production of a physical sequence culminat-

ing in your finger’s moving.

Many philosophers have weighed in on this topic, offering ingenious

ways of insinuating mentality into the causal act, but if we are honest, we

shall have to admit that the results have not been encouraging. My own

belief is that a sensible answer to the qua problem requires a rejection of the

muddled ontology that gives rise to it in the first place. I have hinted at

how this might work in discussing Davidson. Before saying more, how-

ever, it is worth pausing to look more carefully at the underlying onto-

logical picture.

First, what exactly is an event? Philosophers sometimes divide the terri-

tory between “Davidsonian events” and “Kim-style events.” Davidsonian

events are thought of as “coarse-grained”; Kim’s are “fine-grained.”

According to Kim, an event is a substance’s having a property at a time:

a’s being F at t. (I shall omit the time reference in what follows.) Now

suppose thatM is a mental property and P is a physical property. In that case,

a’s being M and a’s being P are distinct events. There is token identity, a is

5 A variant substitutes collections of exactly resembling tropes for universals. Although ontologically

momentous, the distinction has no discernible effect on arguments I address here, so I shall ignore it in

what follows.

26 JOHN HEIL

identical with a, but the events are distinct—in the way a ball’s being red and

its being round are distinct. The events are not token identical.6

What of Davidson? I have argued that, for Davidson, the mental–physical

distinction is one of conception only. So for Davidson, a given token event

can answer to both a mental and a physical description. A single token event

answers to each description. Notice that this has nothing to do with the

event’s being “coarse-grained.” Were he ontologically less timid, Davidson

could accept Kim’s accounting of the ontology of events without modifying

his account of mental causation.

So? Well, it is hard to see how the qua problem could arise for either

Davidson or Kim. For Davidson, or my Davidson, it is crazy to ask whether

a given event caused what it did because it was described as mental or

because it was described as physical. And for Kim, the mental and physical

events are distinct if their constituent mental and physical properties are

distinct. You might wonder which event had a given effect, but not

whether a single event had the effect it had because it was mental or because

it was physical.

Properties of Events

In a paper published in 2006, Cynthia andGrahamMacdonald accept a Kim-

style conception of events, but proceed to argue that particular events can

cause what they do in virtue of their mental properties. When I introduced

this topic, I spoke of events “including” mental and physical properties.

I adopted this awkward form of speech because I wanted to leave open

whether the properties in question were constituents of the events. For the

Macdonalds, an event, a’s being F, can itself have a property. This property

can be, indeed must be, distinct from F, yet be “causally relevant” to the

event’s producing a given effect, provided various conditions are satisfied.

A view of this kind, might seem to allow that a mental event (an event

possessing a mental property) could be token identical with a physical event

(an event possessing a physical property) thereby leaving room for a qua

6 Does it matter that M supervenes on P? Not if supervenience is understood in the usual way: the

bearer of M is not P, M is not a second-order property, not a property of a property, but a property

possessed by a in virtue of a’s possession of P.

MENTAL CAUSATION 27

problem. The question is, why would you want to do this? Why go to a lot

of trouble to re-introduce a problem, then to a lot more trouble to devise a

convoluted solution to cope with it? It is hard not to think that the

Macdonalds are impressed by the usual way of describing Davidson’s view

as implying token, but not type, identity, and providing an explicit onto-

logical picture of events as, not only being partly constituted by properties but

as, in addition, possessing properties.

I find the Macdonalds’ position deeply uncongenial, but I am not con-

cerned here with its details. Rather I want to focus just on the idea of events

bearing properties. You can predicate—truly—many things of events, but it

needn’t follow from this that the truthmakers of such predications are

properties of those events. This is not a deep or technical point, merely an

expression of bewilderment over what it could be for an event—and here

we are thinking of “Kim-style” events, a’s being F—to bear a property.

Substances bear properties. I doubt that properties have properties, but even

if they did, it is very hard to see the Macdonalds’ events as being ontologic-

ally suited to be bearers of properties. Think of it. You have a’s being F

possessing, say G. The bearer of G here is not a; a’s being the bearer would

be fine, but then G would belong to a, not to the event, not to a’s being F.

Further,G’s belonging to a would in fact be a constituent of a new event: a’s

being G. Similarly, the bearer of G couldn’t be F. No, G must be possessed

by a’s being F, or, more particularly, by a’s being F at t.

You will not share my bewilderment at this thesis if you are content to

suppose that every true predication designates a property. Such a view turns

properties into what David Armstrong calls shadows of predicates. Under-

standing what it could mean for an event—a’s being F—to possess a

property, requires more than an appeal to the fact that we can and do

describe events using an assortment of predicates. Events can be brief or

drawn out, for instance, but what makes it true that an event is temporally

extended in a particular way need not be the possession by the event of a

temporal property, whatever that might mean.

The worry here is less an expression of a quirky ontological preference,

than a simple inability to comprehend what it would be for an event, as

characterized by the Macdonalds, to bear a property. There is certainly a

place for a relaxed conception of properties, according to which having a

property is simply a matter of answering to a predicate. When the question

is whether properties of a certain kind could figure in causal relations,

28 JOHN HEIL

however, you need an ontologically serious conception, you need to move

beyond the thought that, if you can predicate something truly of an event,

this must be because the event possesses a property corresponding to your

predicate (and shared by anything to which the predicate truly applies).

I do not consider these remarks on properties of events to amount to

anything approaching a knock-down argument. I will be satisfied if I have at

least pricked the conscience of those for whom talk of properties of events

trips off the tongue.

Monism

Moving on, then, let us suppose, at least for the sake of argument, that

causation is a relation among events and that events are substances’ possess-

ing properties at times. Now consider a particular causal sequence in which

your forming an intention to move your finger causes your finger to move.

If Davidson is right, there is a description of this sequence expressible in the

language of fundamental physics that is an instance of a fundamental,

“exceptionless” law. For Davidson, laws are linguistic items, statements.

Davidson is silent as to the truthmakers for law-statements, but we need not

be.7 If you are a Humean, the truthmakers will be uniformities, patterns of

events of particular types specifiable in the language of fundamental physics.

If you are like me, you will think the truthmakers are powers possessed by

the objects involved in virtue of those objects’ properties and relations

among these.

Whatever your view, the idea would be that what answers to a statement of

this fundamental law also, on this occasion, answers to a singular causal claim

couched in a psychological vocabulary. Because application conditions for

psychological predicates differ markedly from application conditions for

predicates deployed in fundamental physics, there is no prospect of framing

the former in terms of the latter. This, this conceptual or taxonomic mis-

match, is the anomalousness of the mental. The truthmaker for the original

psychological assertion is a state of the world that could be described in a fine-

grained physical vocabulary. The description would be, from the perspective

7 You might think that Davidson rejects the notion of truthmaking but I don’t think he has the

concept.

MENTAL CAUSATION 29

of physics, sprawling and ungainly. There is little reason to think—and if

arguments for “multiple realizability” are to be taken seriously, every reason

not to think—that there could be interesting connections between mental and

physical descriptions.

The emerging picture is of one world describable—truly describable—in

many different ways. The world, or rather ways the world is, serve as truth-

makers for these descriptions. If fundamental physics is in the business of

giving an exhaustive, maximally fine-grained description of the world, this in

no way excludes biological, anatomical, geological, meteorological, psycho-

logical, anthropological, or journalistic descriptions of worldly goings-on.

The mistake is to imagine that biology and the rest describe worlds distinct

from but dependent on the world described by physics. If physicalism is the

doctrine that every truth about the world could be expressed in the vocabu-

lary of fundamental physics, then this is not physicalism.

Powerful Qualities

All well and good, but doesn’t monism face the problem of finding a place

for the qualia in a universe of quarks and electrons? If you can move beyond

the thought that mental predicates designate a distinctive realm of proper-

ties, you can begin to think seriously about what truthmakers for psycho-

logical assertions, including assertions concerning qualities of conscious

experiences might be.

This is not the place to attempt a full-scale discussion of qualia, but I can at

least say why I think the non-reductive physicalist mind-set makes a difficult

problem much more difficult.

One facet of what might be thought of as the standard conception of the

physical world is the thought that qualities are exclusively mental. The

ancestor of this idea lies in the use to which the distinction between primary

and secondary qualities has been put by philosophers who have tended to

regard the physical sciences as the measure of all things, the source of all

truths. The primary qualities (shape, size, mass, charge, and the like) are

taken to be properties of physical objects. Secondary qualities (colors,

sounds, tastes, smells, etc.) are relegated to the minds of observers.

A conception of this kind bifurcates the world into minds and everything

else. Certain properties belong to minds that do not, indeed could not, belong

30 JOHN HEIL

to physical objects. If that is right, then there is no prospect of finding a place

for minds in the physical order.

You can see this conception at work in the thesis that physical properties

are powers: to have mass is to have the power to affect and be affected by

other massy things in particular ways. Most often, the thought that physical

properties are powers includes the thought that this exhausts their nature.

The result is a conception of the physical world as a world of “pure

powers,” a conception that apparently fits well with scientific practice.

Coupled with the idea that powers are individuated relationally, we are

led to the “structural realist” construal of the world as constituted by

relations (Ladyman 2007). Finding a place for qualities of conscious experi-

ences in such a world looks hopeless. At best such qualities are add-ons,

ontological fuzz, foam on the sea of being.

Suppose, however, you distinguished features of scientific accounts of the

world from features of the world in virtue of which those accounts are true.

A relational vocabulary, for instance, might suffice to represent the truths of

fundamental physics (Dipert 1997). From this, however, it does not follow

that truthmakers for assertions couched in a relational vocabulary are irredu-

cibly relational states of affairs (Heil 2009; see also Parsons 2009). Nor

from the fact that scientific explanation omits reference to qualities does it

follow that the world thus explained is qualitatively barren. To think

otherwise, to think that the assertion that the fundamental properties are

powers implies that they are not qualities, is to move from “the a’s are F” to

“the a’s are not G.” Properties could be powers, and powers could be

qualities; properties could be powerful qualities.

I believe there are excellent reasons to think this is so, to think that

properties are powerful qualities, for reasons having to do with the individu-

ation of powers.8You need not be fully convinced of such a view, however,

to grant that it is at least an option. The idea that properties are powerful

qualities narrows the gap between the qualia and everything else. If everything

has qualities, it is unsurprising that conscious states are qualitatively imbued.

States of mind differ qualitatively from states of an ordinary computing

machine. But qualitative differences are potentially explicable.

8 Berkeley argued the point but assumed that qualities must be immaterial. More recently, it has been

argued by Campbell (1976, 93), Martin (1997, 2008), Armstrong (1999), and Unger (2006). See Heil

(2008, 2010) for discussion.

MENTAL CAUSATION 31

Ontological Progress

I am not so foolish as to imagine that these brief remarks could persuade

skeptical readers that considering properties as powerful qualities affords an

immediate solution to the mind–body problem. Thinking of properties this

way does, however, nudge us in the right direction. Earlier, I described

Davidson as advancing the thesis that the mental–physical distinction is a

distinction of conception only, not a real distinction. This is all to the good,

but Davidson is concerned with just one class of psychological states, the

propositional attitudes. A natural reaction to Davidson’s argument would be

to concede that the argument might possibly provide a way of understand-

ing cases of mental causation involving beliefs, desires, intentions, and the

like, but to note, as Davidson himself notes, that the argument fails to

extend to conscious experiences.

This is where you need to move beyond Davidson’s hands-off approach

to ontology. Suppose I am right, suppose properties are powerful qualities.

This would mean that the distinction between qualities and powers is a

distinction of conception only, not a real distinction. Were that so, not only

would it be unsurprising that conscious experiences are qualitatively satur-

ated, but also that their qualities “make a causal difference.”

I believe we have excellent reasons to embrace an ontological picture that

regards properties as powerful qualities, reasons that have nothing in particular to

do with the mind–body problem. The mind–body problem as we have it

today—the problem of mental causation, the problem of causal relevance—

arises from acceptance of a very different ontological picture, one that has little to

recommend it. The influence of that picture stems, not from its being founded

on compelling arguments but from its permeating philosophers’ thoughts about

the mind, from its functioning as a lens through which we see problems and

evaluate competing solutions. Whether I am on the right track or not, the way

out is not to be found by making incremental ad hoc ocular adjustments—

adding epicycles to the prevailing,RubeGoldberg ontology—but by coming to

see that ontology as merely one among many possible ontologies.

References

Armstrong, D. M. (1999). ‘The Causal Theory of Properties: Properties According

to Ellis, Shoemaker, and Others’. Philosophical Topics, 26: 25–37.

32 JOHN HEIL

Bennett, K. (2003). ‘Why the Exclusion Problem Seems Intractable and How, Just

Maybe, to Tract It’. Nous, 37: 471–97.

Campbell, K. (1976).Metaphysics: An Introduction. Encino: Dickenson Publishing Co.

Crane, T. and D. H. Mellor (1990). ‘There is no Question of Physicalism’. Mind,

99: 185–206.

Davidson, D. (1970). ‘Mental Events’. In L. Foster and J. Swanson (eds.),Experience and

Theory. Amherst: University of Massachusetts Press: 79–101. Repr. in Davidson,

Essays on Actions and Events. Oxford: Clarendon Press. (1980, 2001): 207–25.

—— (1993). ‘Thinking Causes’. In Heil and Mele (1993): 3–18.

Dipert, R. R. (1997). ‘The Mathematical Structure of the World: The World as

Graph’. Journal of Philosophy, 94: 329–58.

Fodor, J. (1997). ‘Special Sciences: Still Autonomous after All These Years’.

Philosophical Perspectives, 11: 149–63. Repr. in Fodor (1998). In Critical Condition:

Polemical Essays on Cognitive Science and the Philosophy of Mind. Cambridge, MA:

MIT Press: 9–24.

Gibbons, J. (2006). ‘Mental Causation Without Downward Causation’. Philosoph-

ical Review, 115: 79–103.

Heil, J. (2008). ‘Critical Study of Peter Unger’s All the Power in the World’.Nous, 42:

336–48.

—— (2009). ‘Relations’. In R. Le Poidevin and R. Cameron (eds.), Routledge

Companion to Metaphysics. London: Routledge: 310–21.

—— (2010). ‘Powerful Qualities’. In A. Marmodoro (ed.), The Metaphysics of

Powers: Their Grounding and their Manifestations. London: Routledge: 58–72.

—— and A. R. Mele (eds.) (1993). Mental Causation. Oxford: Clarendon Press.

Hohwy, J. and J. Kallestrup (eds.) (2008). Being Reduced: New Essays on Reduction,

Explanation, and Causation. Oxford: Oxford University Press.

Kim, J. (1978). ‘Supervenience and Nomological Incommensurables’. American

Philosophical Quarterly, 15: 149–56.

—— (1979). ‘Causality, Identity, and Supervenience in the Mind–Body Problem’.

Midwest Studies in Philosophy, 4: 31–49.

—— (1982). ‘Psychophysical Supervenience’. Philosophical Studies, 41: 51–70.

—— (1984). ‘Epiphenomenal and Supervenient Causation’. Midwest Studies in

Philosophy, 9: 257–70.

—— (1993a). ‘The Non-Reductivist’s Troubles with Mental Causation’. In Heil

and Mele (1993): 189–210. Repr. in Kim (1993b): 336–57.

—— (1993b). Supervenience and Mind: Selected Philosophical Essays. Cambridge:

Cambridge University Press.

—— (2005). Physicalism, or Something Near Enough. Princeton: Princeton Univer-

sity Press.

MENTAL CAUSATION 33

Ladyman, J. (2007). ‘Structural Realism’. In E. N. Zalta (ed.), The Stanford Encyclo-

pedia of Philosophy, available online at: http://plato.stanford.edu/archives/

win2007/entries/structural-realism/.

Lepore, E. and B. Loewer (1987). ‘Mind Matters’. Journal of Philosophy, 84: 630–42.

Macdonald, C. and G. Macdonald (2006). ‘The Metaphysics of Mental Causation’.

Journal of Philosophy, 103: 539–76.

McLaughlin, B. P. (1989). ‘Type Epiphenomenalism, Type Dualism, and the

Causal Priority of the Physical’. Philosophical Perspectives, 3: 109–35.

Martin, C. B. (1997). ‘On the Need for Properties: The Road to Pythagoreanism

and Back’. Synthese, 112: 193–231.

—— (2008). The Mind in Nature. Oxford: Oxford University Press.

Mellor, D. H. (1973). ‘Materialism and Phenomenal Qualities’. Aristotelian Society

Proceedings, Supp. Vol. 47: 107–19.

Mills, E. (1996). ‘Interactionism and Overdetermination’. American Philosophical

Quarterly, 33: 105–17.

Parsons, J. (2009). ‘Are there Irreducibly Relational Facts?’ In E. J. Lowe and

A. Rami (eds.), Truth and Truth-Making. Stocksfield: Acumen Publishing: 217–26.

Pereboom, D. (2002). ‘Robust Nonreductive Materialism’. Journal of Philosophy,

99: 499–531.

Pereboom, D. and H. Kornblith (1991). ‘The Metaphysics of Irreducibility’.

Philosophical Studies, 63: 125–45.

Shoemaker, S. (2001). ‘Realization and Mental Causation’. In C. Gillett and

B. Loewer (eds.), Physicalism and Its Discontents. Cambridge: Cambridge Univer-

sity Press: 74–98.

—— (2003). ‘Realization, Microrealization, and Coincidence’. Philosophy and

Phenomenological Research, 67: 1–23.

Sosa, E. (1984). ‘Mind–Body Interaction and Supervenient Causation’. Midwest

Studies in Philosophy, 9: 271–81.

Stoutland, F. (1976). ‘The Causation of Behavior’. In J. Hintikka (ed.), Essays on

Wittgenstein in Honor of G. H. von Wright (Acta Philosophica Fennica, 28).

Amsterdam: North Holland Publishing: 286–325.

Strawson, G. (2008). Real Materialism and Other Essays. Oxford: Clarendon Press.

Thomasson, A. (1998). ‘A Non-Reductivist Solution to Mental Causation’. Philo-

sophical Studies, 89: 181–91.

Unger, P. (2006). All the Power in the World. New York: Oxford University Press.

Woodward, J. (2008). ‘Mental Causation and Neural Mechanisms’. In Hohwy and

Kallestrup (2008): 218–62.

Yablo, S. (1992). ‘Mental Causation’. Philosophical Review, 101: 245–80.

34 JOHN HEIL

2

Physical Realization withoutPreemption

SYDNEY SHOEMAKER

I

This paper expands, but also corrects, the account I gave in my book

Physical Realization of how we can hold that mental states are physically

realized without being forced to allow that the causal efficacy of mental

states is preempted by their physical realizers. In part the correction of the

account in my book involves returning to an account of realization I put

forward in an earlier paper, and retracting a concession I made nearly thirty

years ago to an objection to the original formulation of my causal theory of

properties. The expansion of the account involves clarifying and developing

the claim that all properties of macroscopic objects are realized in what I call

MSE properties, properties things have in virtue of their careers embedding

microphysical states of affairs of a certain sort.

II

I begin with the well-known multiple realization objection to type physic-

alism. The objection can be presented as follows. Ask yourself what would

be the best possible evidence that a particular mental state, say pain, is

identical with a particular physical state, say C-fiber firing. One might at

first think that the best evidence would be the discovery of a universal

correlation between the mental state and the physical state—the discovery

that a subject is in pain if and only if there is C-fiber firing occurring in her

brain. But this first thought is certainly wrong, for the existence of such a

correlation is perfectly compatible with a dualist view which holds that the

mental and physical states are distinct, although universally correlated,

perhaps because the mental state is always caused by the physical one.

A much better answer is that the best possible evidence would be the

discovery that the physical state has the same causes, and the same effects,

as the mental state—e.g., that C-fiber firing is caused by cuts and burns, and

causes grimaces, moans, the taking of aspirins, and calls to the doctor. This is

sometimes put by saying that the physical state plays the same causal role as

the mental state. Playing the causal role of a state can be equated with

sharing its causal profile, where the causal profile of a state is constituted by

the ‘forward-looking’ causal features of the state, its being apt to contribute

in certain ways to the causing of certain effects, and its “backward-looking”

causal features, its being such that its instantiation can be caused in certain

ways. It is hard to see what could be a better reason for identifying pain with

C-fiber firing than their having the same causal profile.

But now we face the objection that having discovered that a certain

physical state seems to play the causal role of pain in us, we might discover

another species in which some different physical state plays this causal role.

Thus David Lewis imagined that while C-fiber firing plays the causal role of

pain in Earthlings, in Martians this causal role is played by the inflation of

tiny cavities in the feet (Lewis 1980). And Hilary Putnam suggested that

what plays the causal role of pain in octipi is a physical state quite different

from the one that plays that role in us (Putnam 1967). We might even

discover that this causal role is played by different physical states in different

members of our own species, or even that it is played by different physical

states at different times in the history of the same person. Obviously the

same mental state cannot be identical to a number of different physical

states. But if states are identical, they have to be necessarily identical, which

means that there is no possible circumstance in which they are distinct. So if

it is so much as possible that different physical states can play the causal role

of a given mental state, playing that causal role cannot make a physical state

identical with the mental state.

The different physical states that play the causal role of a mental state can

also be said to play its functional role. The property of having some state or

other that plays this role is a functional property, which is realized in, or has

as possible realizers, those physical states. And now it seems that it is this

36 SYDNEY SHOEMAKER

functional property, not any of its physical realizers, that a physicalist should

identify with pain.

But now we face a problem. One problem which the psychophysical

identity theory did not face is that of explaining how, on a physicalist view,

mental states can have causal efficacy. For if a mental state is identical with a

physical state, it will obviously cause whatever that physical state causes. If

pain is identical with C-fiber firing, then pain will cause whatever C-fiber

firing causes—and as noted earlier, what would motivate the identification

of pain with C-fiber firing is precisely the fact (or supposed fact) that C-fiber

firing causes the things we take pain to cause, and is caused by the things we

take to cause pain. But if pain is instead identical with a functional property

of which C-fiber firing is one of the realizers, we do have a problem. What

would make C-fiber firing a realizer of pain would be the same thing that

initially made it a promising candidate for being identical with pain, namely

that it plays the causal role of pain. But if it plays the causal role of pain, but is

not identical with pain, how can pain itself play the causal role of pain?

More generally, if it is one or another of the realizers of a mental state that

plays its causal role when the state is instantiated, how can the mental state

itself be said to play the causal role? To suppose that a wince or groan

is caused both by the instantiation of some physical realizer of pain, say

C-fiber firing, and by the instantiation of pain, is to suppose that it is

overdetermined, And it seems implausible, at best, to suppose that every

case of mental causation involves overdetermination. It thus seems that if

the mental state is a multiply realizable functional state, its causal efficacy is

“preempted” by the realizers of that functional state. There is thus the threat

that the mental realm turns out to be causally inefficacious—that it is

epiphenomenal. But how can this be, given that we initially picked out

the physical realizers as states that play the causal role of the mental state they

realize? How can a mental state have a causal role if it doesn’t do any

causing? It may seem that we are led to the conclusion that our common-

sense psychological theory is really an error theory, and that instead of saying

that the physical realizers of a mental state are physical states that play the

causal role of that mental state we should say that they are physical states that

play the causal role that our commonsense theory mistakenly says is played

by that mental state.

One version of the preemption doctrine focuses on the relation between

instances of a functional property and instances of the physical properties

PHYSICAL REALIZATION WITHOUT PREEMPTION 37

that realize it. It seems prima facie plausible that if the property of being in

pain is realized on a given occasion by C-fiber firing, then the instance of

pain that occurs at that time is identical with an instance of C-fiber firing.

This goes with the idea that the causal powers of the instance of pain will be

identical with the causal powers of the instance of C-fiber firing, which is

what Jaegwon Kim’s Causal Inheritance Principle tells us. It will be in virtue

of being an instance of C-fiber firing that it will have these causal powers.

If the property of being in pain is not identical with the property of having

C-fiber firing occurring in one, there seems nothing for it to do—no way it

can contribute to bestowing causal powers on its instances. More generally,

the property instance that is an instance both of a realized property and one

of its physical realizers will have its causal powers in virtue of being an

instance of the physical realizer, and the fact that it is an instance of the

realized property will play no role in explaining why it has these causal

powers.

There have been various responses to this problem. One is eliminativism.

This is the view that there are no mental properties—there are only physical

properties that play the causal roles that we mistakenly take mental proper-

ties to play. This may seem to be forced on us if we hold the plausible view

that any genuine property must have a causal profile. Another response is

reductionism. This is the view that the mental is causally efficacious because

it somehow reduces to the physical. The identity theory we started with is

of course one version of reductionism, which we might call global reduc-

tionism. That theory we have seen reasons for rejecting. But some philoso-

phers, e.g., Jaegwon Kim, have suggested that we can have reductions that

are not global—that are species relative, or structure relative. We cannot

reduce pain, simpliciter, to something physical, but perhaps we can reduce

human pain, or pain of creatures with a certain physical structure, to

something physical. For each species, or for each structural type within a

species, we identify the pain of creatures in that species, or of creatures

having that structure, with a certain physical state. Viewed one way, this is a

version of eliminativism.We abandon pain as a state or event type, deny that

there is any property of being in pain, but we retain human pain, dog pain,

Martian pain, etc., about each of which we put forward an identity theory.

It will be allowed, on this view, that there is a single concept of pain that

applies to all of these states, but it is denied that there is a single property, or

state type, corresponding to this concept.

38 SYDNEY SHOEMAKER

But there is also a response which holds on to the view that mental states

are causally efficacious, without fragmenting them into different species-

relative or structure-relative states, and holds that they are not identical

with, although they are realized in, physical states. This is sometimes called

non-reductive physicalism. I will be presenting a version of this.

III

Such a view must hold that mental properties have causal profiles, and that

their instantiation involves their being realized by physical properties with

associated causal profiles, without there occurring causal overdetermination

of an objectionable sort. To see how this is possible we need to investigate

the notion of realization, and in particular the notion of physical realization.

I take the notion of a mental property being physically realized to be the

notion of there being a constitutive relation between the instantiation

of the property and something physical; the instantiation of the mental

property consists in the existence of its physical realizer, and the existence

of the realizer is metaphysically sufficient for the occurrence of the mental

property instantiation. As the term “realize” suggests, what realizes a

property instance is what makes it real. Of course, it is not only mental

properties and their instantiations that can be said to be realized in this

sense. Functional properties like being a braking system, or being a clock,

can be realized in mechanical systems of various sorts, and colors can

presumably be realized by such things as spectral reflectances. In fact, as

I will claim later on, every property of concrete things that we can refer to

is realized in other properties.

I distinguish two sorts of realization, which turn out to be closely related.

One sort, property realization, is what we have already met with—it is what

we have in the claim that a property like having C-fiber firing going on in

one can realize the property of being in pain but cannot be identical with it.

We can speak of this as a relation between properties, but at bottom it is a

relation between instantiations of properties, or what I will sometimes refer

to as property instances. One property is realized by another property if

instantiations of the one property can be realized by instantiations of the

other property—i.e., if the having of the one can be realized by the having

of the other. In the central case, the instantiation of a property and the

PHYSICAL REALIZATION WITHOUT PREEMPTION 39

instantiation of its realizer will occur in one and the same thing—although

we can define a derivative sense in which the instantiation of a property in

one thing is realized in the instantiation of a different property in a different

thing that is coincident with that thing (as it might be, a person’s body’s

having a certain property realizes the person’s having a certain other prop-

erty, although the person and her body are numerically different things). So,

as we imagined earlier, the instantiation of the property of being in pain

might be realized by the instantiation of the property of having C-fiber

firing occurring in one.

In the other sort of realization,microphysical realization, the instantiation of a

property will be realized in a microphysical state of affairs, one consisting of

micro-entities (atoms, electrons, quarks, or whatever) having certain proper-

ties and being related in certain ways—for short, it consists in micro-entities

being propertied and related in certain ways. Here the realizer of a property

instance will not be another property instance, as it will be in cases of property

realization, but will instead be such a state of affairs. On the physicalist view

that the microphysical facts about the world fix all of the facts about it, every

property instantiation will be microphysically realized. There will of course

be one massive microphysical state of affairs, involving all of the micro-

entities there are, that realizes every property instantiation. But there will be

less global states of affairs that realize some property instantiations and not

others, and it seems reasonable to assume that every property instantiationwill

have a microphysical realizer that is minimal in the sense that it does not

contain any proper part that is a realizer of that property instantiation.

These two sorts of realization are related, because for every type of

microphysical state of affairs whose members can realize a property instanti-

ation there will be a property something has just in case its career embeds a

state of affairs of this sort. One can speak of these as MSE properties, for

microphysical-state-of-affairs-embedding properties. But here I will reserve

the term “MSE property” for cases in which the embedded state of affairs is

maximally determinate. All microphysical states of affairs are realized by

maximally determinate ones, and every property instance will have a max-

imally determinate microphysical realizer. So every case of microphysical

realization, except for cases in which the realized property is itself an MSE

property, is a case of property realization in which the realizer is an MSE

property. Since all properties are microphysically realized, all properties that

are not themselves MSE properties are multiply realized by MSE properties.

40 SYDNEY SHOEMAKER

The usual characterization of property realization is what figured in my

introductory remarks. This says that the realized property is the second-order

property of having some property or other that plays a certain causal role, and

that its realizers are the properties that play that role. This is the formulation

that suggests that the realizers preempt whatever causal role we might think

belongs to the realized property. I favor an account that avoids that sugges-

tion. This is what is sometimes called the subset view, which I share with

Lenny Clapp andMichaelWatkins (Clapp 2001; Watkins 2002). My version

of it says, with certain qualifications that I will not go into, that property P has

property Q as a realizer if the forward-looking causal features of P—its

aptness to contribute when instantiated to the production of certain

effects—are a subset of the forward-looking causal features of Q. So, sticking

to the example of pain, the things we take pain to cause are included among

the things C-fiber firing causes. Initially I saw this view as a competitor to the

higher-order property account of property realization, but I have come to

see that it can be seen instead as a version of that account: let it be that P has

Q as a realizer if P is the property something has just in case it has some

property whose forward-looking causal features include those of P as a

subset. Notice that this view does not say that the realizers play exactly the

causal role that we take the realized property to play. There is a close relation

between the causal profile of the realized property and the causal profiles of

its realizers, which is given by the subset relation, but this relation is not

identity. This will be important later on.

Here I have reverted to the account of property realization I put forward

in my first published discussion of it (Shoemaker 2001), one in which it

consists in a subset relation between the forward-looking causal features of

the realized property and its realizers, and have abandoned the account

I gave in a later paper (2003) and in my book Physical Realization (2007),

which adds as a further requirement that the backward-looking causal

features of the realizer be a subset of the backward-looking causal features

of the realizer properties. This is not because I think that the holding of that

subset relation is not required, but because I think that it necessarily holds

whenever the subset relation between forward-looking causal features

holds. This would not be so if it were possible for two different properties

to share the same forward-looking causal features but differ in their back-

ward-looking causal features. Many years ago Richard Boyd offered a

putative case of this as an objection to my causal theory of properties. In

PHYSICAL REALIZATION WITHOUT PREEMPTION 41

his example substance X is what you get when you combine substances

A and B and substance Y is what you get when you combine the different

substances C and D, and the properties of being composed of X and being

composed of Y are exactly alike in their forward-looking causal features.

These properties are supposed to be different because of their different

compositional histories. I was not convinced by the example, but I saw

that the causal theory of properties could easily be modified so as to allow

for its possibility by having it individuate properties in terms of their

backward-looking causal features as well as their forward-looking causal

features (Shoemaker 1980, postscript). And so I thought that it would do no

harm to allow for the possibility of such cases in the account of realization by

adding the requirement about backward-looking causal features.

But I was wrong to think that doing this would do no harm. This was

shown by Brian McLaughlin (McLaughlin 2007). In a critique of my

account of realization he pointed out that, on the assumption that different

properties can have the same forward-looking causal features, there being an

instantiation of a property which my account says is a realizer of a given

property does not guarantee that there is an instantiation of that property. It

guarantees that there is an instantiation of a property having that property’s

forward-looking causal features, but it does not guarantee that there is an

instantiation of a property having that property’s backward-looking causal

features. The putative realizer’s backward-looking causal features will be a

subset of the backward-looking causal features of the supposedly realized

property, but its instantiation does not guarantee that there is instantiated a

property having precisely that set of backward-looking causal features.

The objection disappears, of course, if we reject the assumption that

properties can differ despite having the same forward-looking causal fea-

tures. And I now do reject this. Returning to Boyd’s example, the most it

could show is that compositional properties can be alike in forward-looking

causal features while differing in backward-looking causal features. But

I don’t think it shows even that. The fact that we get instances of the

property of being made of X by combining substances A and B and get

instances of the property of being made of Y by combining substances C and

D doesn’t show that being made of X is not the same property as being

made of Y—instances of the same property can be caused in different ways.

Of course, if being an X thing consists in being composed of A and B while

being a Y thing consists in being composed of C and D, then being an

42 SYDNEY SHOEMAKER

X thing and being a Y thing will be different properties. But given that

being an X thing and being a Y thing share all their forward-looking causal

features, there is no possibility of decomposing such things into their

supposedly different constituents. And that is a reason for denying that

they have different sets of constituents. I think that if we ran across a case

like this, the best thing for us to say would be that we have a single substance

that can be produced by combining either of two pairs of substances. If

someone insists that it is logically possible, although unverifiable, that in

such a case the substances would differ in composition and would for that

reason be different substances, and so the properties of being composed of

those substances would be different, then I will restrict my account of

property realization to properties that are not of this sort.

IV

I turn now to microphysical realization. Just as properties have causal

profiles, so too do types of microphysical states of affairs. A given type of

microphysical states of affairs will be such that members of that type are apt to

cause, or contribute to causing, microphysical states of affairs of certain other

types, andwill be caused bymicrophysical states of affairs of certain types.My

account of microphysical realization says that a microphysical state of affairs

realizes an instance of a particular property just in case the microphysical state

of affairs belongs to a type whose causal profile matches in a certain way

either the causal profile of the property whose instance it realizes or the causal

profile of a property realizer of that property, and the microphysical state of

affairs is embedded in the career of the subject of the property instance and is

simultaneous with it. Suppose again that pain is realized by C-fiber firing.

This formulation allows an instance of pain to have as a realizer a state of

affairs that is a microphysical realizer of an instance of C-fiber firing, and so is

of a type whose causal profile matches that of the property of C-fiber firing

rather than that of the property of being in pain. But it will also have a

microphysical realizer that is of a state of affairs type, a more abstract one,

whose causal profilematches that of the property of being in pain. Suchmore

abstract states of affairs will be realized by more concrete ones.

What is the relation of “matching” between causal profiles that I speak of

here? I would like to be able to say that it is identity, and I think that there is

PHYSICAL REALIZATION WITHOUT PREEMPTION 43

a good sense in which it is. But the causal features of the property instances

have to do with the causing of other property instances, while those of the

microphysical states of affairs have to do with the causing of other micro-

physical states of affairs. It is true that in contributing to the causing of other

microphysical states of affairs a microphysical state of affairs contributes to

the causing of the property instances realized by those states of affairs; and

that is a reason for saying that the states of affairs share the causal features

of the property instances they realize. But that invokes the notion of

microphysical realization, which is what I am trying to explain. So to say

this in explaining the notion of microphysical realization would involve a

kind of circularity. In Physical Realization I said that the causal profile of the

state of affairs type is isomorphic with the corresponding causal profile of the

instantiated property. But if that means only that there is a structural

similarity between the causal profiles involving a one–one correspondence

which pairs each causal feature in the one with a causal feature in the other,

such matching does not by itself make the state of affairs a realizer of the

property instance. It must further be the case that the property is instantiated

if there occurs at the same time and in the same career a state of affairs of the

corresponding type, and that this is true as a matter of necessity. And it must

be the case, again as a matter of necessity, that when an instance of the

property causes or contributes to causing an instance of another property,

the corresponding state of affairs causes or contributes to causing a state of

affairs of a type that is paired with that other property in the one-one

correspondence—likewise, when an instance of the property is caused by

an instance of another property, the corresponding state of affairs is caused

by a state of affairs of a type that is paired with that other property in the

one–one correspondence. A further requirement is that if we trace the

causal histories of the property instance and the microphysical state of affair,

both into their pasts and into their futures, they will converge—the very

same things will be involved in causing causal ancestors of both the property

instance and its microphysical realizer, and the very same things will be

caused by causal descendants of both. What all of this is intended to

guarantee is that the occurrence of the microphysical state of affairs consti-

tutes the instantiation of the realized property.

The microphysical states of affairs will be partly concrete, a matter of

particular micro-entities being propertied and related in certain ways, but

will consist in part in the truth of positive and negative existential propositions

44 SYDNEY SHOEMAKER

saying that there are, or are not, micro-entities related in certain ways to the

constituents of the concrete part of the state of affairs. The account involves

these complications because of the fact that what realizes a property instance

must guarantee the existence of the thing having the property instance, and

must guarantee the instantiation of whatever other properties a thing of that

sort must have in order to exist. For I take realizers to be sufficient for what

they realize—otherwise they would not be what make the realized things

real—and what is sufficient for the instantiation of a property must be

sufficient for the existence of something having it, and for that thing having

whatever properties it must have in order to exist. I deal with this by

suggesting a way of factoring the state of affairs that is a microphysical realizer

of a property instance. This will consist partly of what I call its core, which is

made up of states of affairs that contribute directly to implementing the causal

profile of the realized property. But it also consists in part of existential states

of affairs that exist in virtue of the concrete states of affairs involved in the

realizations of other properties of the thing that has the property. So, for

example, a microphysical realizer of an instance of the property of having a

certain height will have a core consisting of states of affairs that contribute

directly to instantiating the causal profile associated with that height, but will

also contain existential states of affairs that guarantee that it has some width or

other, some mass or other, and so on. And a microphysical realizer of an

instance of the mental property of having a certain belief will have a core

consisting of states of affairs that contribute directly to implementing the

causal profile of that belief, but will also contain existential states of affairs

that guarantee that the subject has whatever other mental properties a

subject of that belief must have. This is only a rough sketch of the account,

the details of which are too complex for me to go into here. (For a fuller

account see my 2007.)

V

As is apparent from this brief summary, the accounts of both sorts of

realization take it that realized properties, e.g., functional properties and

determinables of all sorts, have causal profiles, and so can be causally

efficacious. Certainly it should be the default view that mental properties,

and other physically realized properties, are causally efficacious. But how do

PHYSICAL REALIZATION WITHOUT PREEMPTION 45

I deal with the threat, mentioned in my opening remarks, that the causal

efficacy of realized properties is preempted by their realizers—and in par-

ticular, that the mental efficacy of mental properties is preempted by their

physical realizers, making mental properties causally inert?

I do this in part by focusing on JaegwonKim’s formulation of what he calls

the Causal Inheritance Principle.When Imentioned this earlier I put it as the

principle that the causal powers of an instance of a higher-order property are

identical with those of the instance of the lower-order property that realizes

it. This goes with the view that the instance of the higher-order property and

that of its lower-order realizer are one and the same—so, e.g., the instance of

pain might be identical with the instance of C-fiber firing. That is what Kim

thinks is true in the central case, and that is what seems to lead to the view

that higher-order properties have no causal role to play—for presumably the

instance of the physical property that is the lower-order realizer has its causal

powers in virtue of that physical property’s having the causal profile it has,

and that leaves the higher-order property, e.g., the mental property, with no

contribution to make to the causal powers of what is supposed to be an

instance of both it and its realizer. But the formulation of the Causal Inherit-

ance Principle that Kim gives in several places says that the causal powers of a

higher-order property are “identical with (or a subset of ) the causal powers of

[its] realizer” (Kim 1998, 116, my emphasis). His reason for including (in

parentheses) the phrase “or a subset of ” is to allow for cases in which the

realizer is a conjunctive property having as one of its conjuncts a property

that is itself a realizer of the higher-order property. Thus the conjunctive

property has C-fibers firing and is six feet tall will count as a realizer of pain if

having C-fibers firing is a realizer of it, and in this case the causal powers of an

instance of this conjunctive property will have as a proper subset those of an

instance of having C-fibers firing, where the latter are the causal powers of

the instance of pain. But the subset account of property realization suggests

that in general, and not just in the special case of such conjunctive realizers,

the causal powers of the realized property are a proper subset of those of the

realizer instance. This implies that the instance of the realized property is not

identical to the instance of the realizer, for the instances could be identical

only if their causal powers were the same. And this undermines the case for

the view that the higher-order property is causally idle.

But it is compatible with the claim that the instance of the higher-order

property and that of its realizer are not identical that the former is part of the

46 SYDNEY SHOEMAKER

latter. And that seems the right conclusion to draw from the fact that the

causal powers of the former are a proper subset of those of the latter. And

then it is open to us to say that while it is true that the instance of the realizer

property causes the various effects we attribute to the realized property, it

does so because it includes as a part the instance of the realized property. For

example, the instance of C-fiber firing causes, or contributes to causing, the

moaning, groaning, and calls to the doctor, but it does so because it includes

the instance of pain. It includes the instance of pain because its instantiation

guarantees, constitutively, the instantiation of a property having the causal

profile of pain—this because of the subset relations between the causal

profiles of the two properties. The part of the causal profile of C-fiber

firing that is exercised here is precisely the part it shares with the causal

profile of pain in virtue of having the forward-looking causal features of

pain as a subset. So while it is true that the instance of C-fiber firing “does

the causal work,” it does not do so in a way that leaves the instance of pain

with no work to do; on the contrary, it does the causal work because it

includes as a part the instance of pain.

Now let me turn to microphysical realization. As I mentioned earlier, a

property instance can have more than one microphysical realizer. If the

property instance is property-realized by another property instance, then it

will be realized by the microphysical state of affairs that realizes that property

instance, but will also be realized by a microphysical state of affairs that is

specific to it—one whose causal profile matches that of the property of

which it is a realizer. So, to stick with the rather tired example, the instance

of pain will be realized both by the microphysical state of affairs that realizes

the instance of C-fiber firing that property-realizes that pain instance, and

by a more abstract microphysical state of affairs that is peculiar to it.

A property that realizes another property may itself be realized by still

another property. So whenever there is property realization there will be

a hierarchy of property instances, having two or more members, each

member of which other than the one at the bottom is property realized

by those below it in the hierarchy. And there will be a corresponding

hierarchy of microphysical states of affairs, each member of which realizes

the corresponding property instance and all of the property instances above

it in that hierarchy. And each of these states of affairs can be said to realize

the states of affairs above it in the hierarchy of microphysical states of

affairs—there will be the same subset relations between the causal profiles

PHYSICAL REALIZATION WITHOUT PREEMPTION 47

of states of affairs at different levels in this hierarchy as there are between the

causal profiles of the properties instanced in the hierarchy of property

instances. As we go up the hierarchy of states of affair the states of affairs

will become more abstract. At the bottom of such a hierarchy will be a state

of affairs that is maximally determinate. The MSE properties referred to

earlier will be properties things have in virtue of their careers embedding

maximally determinate states of affairs of this sort.

VI

A few words of clarification about MSE properties. While these can be

regarded as a kind of microstructural, or micro-based, properties, not all of

them fit one natural characterization of microstructural properties. That

characterization says that a microstructural property is one that something

has just in case it contains micro-entities of certain sorts that are related to

one another in certain ways. So conceived, microstructural properties are

what I call thin properties. They are properties that can be shared by things

of different kinds. If, as I believe, there can be coincident entities—e.g., a

person and that person’s body—which are numerically different despite

being composed of the very same matter, then such coincident entities

share all of their thin properties and so share all of the same microstructural

properties of this sort. By contrast, thick properties are not shared by coinci-

dent entities, or by entities that are of different kinds and have different

persistence conditions. I take it that mental properties are thick properties.

Although I and my body are coincident entities, I have mental properties

and my body doesn’t. There aren’t two things here thinking my thoughts,

and feeling my sensations. Whatever properties are property realizers of a

thick property are themselves thick properties. And MSE properties that are

realizers of thick properties are themselves thick properties—these are the

ones that don’t fit the natural characterization of microstructural properties.

I said that an MSE property is one something has just in case its career

embeds a maximally determinate microphysical state of affairs that is a micro-

physical realizer of a property instance. But there are different things that can

be meant by ‘embed’ here. In one sense, for the career of a thing to embed a

state of affairs at a time it is sufficient that the state of affairs occurs at that time

in the career of the thing—e.g., in the case of microphysical states of affairs, it is

48 SYDNEY SHOEMAKER

sufficient that it is a state of affairs consisting in some of the micro-constituents

of the thing being propertied and related in certain ways. Call this weak

embedding. It is in this sense that microphysical realizers of thin property

instances are embedded in the careers of their possessors. But in the case of

micro-realization of thick properties we need a stronger sort of embedding,

which I will call strong embedding. We can think of the career of a thing of a

certain sort as consisting in a series of sets of property instances that are so

related, causally, that the persistence conditions for things of that sort, together

with the causal profiles of the properties instantiated, make that series the

career of a single thing of that sort. Assuming physicalism, this will be realized

in a series of microphysical states of affairs that realize the property instances.

Let’s say that a microphysical state of affairs that is a member of such a series at a

time is a momentary stage of that thing at that time. This stage will bemade up

of smaller states of affairs that are realizers of property instances occurring in the

thing at that time. What makes the stage made up of these states of affairs the

stage of a thing of a particular sort is the fact that the causal profiles of the states

of affairs that make it up are such that their occurrence contributes to the

implementation of the persistence conditions for things of the sort in question.

This will involve some of the states of affairs being realizers of thick properties

that can be instantiated only in careers of things of that sort. A microphysical

state of affairs is strongly embedded in the career of a thing of a given sort at a

time if it is such a part of themicrophysical state of affairs that is themomentary

stage of the thing at that time, i.e., if it has a causal profile that makes the

requisite contribution to the implementation of the thing’s persistence condi-

tions, qua thing of that sort. In the case of pairs of coincident entities, like me

and my body, there will be two different careers involving the same micro-

entities, and some of the microphysical states of affairs involving these entities

will be strongly embedded in one of these careers and some will be strongly

embedded in the other. Which is to say that one of these coincident entities

will have one set of thick MSE properties, and the other will have a different

set of such properties. A thick MSE property will be a property a thing has in

virtue of its career strongly embedding a microphysical state of affairs. Of

course, the things will share a number of thin properties, and the thin proper-

ties will also be realized by thinMSE properties. And here theMSE properties

will be realized by states of affairs that are weakly embedded in the careers of

their possessors. Some of these states of affairs will also be strongly embedded in

the careers of one or another of the coincident entities.

PHYSICAL REALIZATION WITHOUT PREEMPTION 49

Consider a microphysical state of affairs that realizes an instance in me of a

mental property. This state of affairs occurs in my career, and it also occurs

in the career of my body. But while it is strongly embedded in my career, in

the sense just explained, it is not strongly embedded in my body’s career,

although it is weakly embedded in it. And so my body, unlike me, does not

have the MSE property constituted by the strong embedding of that state of

affairs. The occurrence of that state of affairs in my body’s career does

guarantee that something has the mental property in question. But that

something is me, not my body. What its occurrence in my body entails is

that there is something coincident with my body that has the MSE property

of strongly embedding it and so has the mental property it realizes.

Every property instance that is not itself an instance of an MSE property

will be property realized by an instance of an MSE property. This will be

true of instances of properties that would not ordinarily be thought of as

higher-order properties—properties like having a certain shape, or a certain

mass. Such properties will have an infinite number of MSE properties as

possible realizers. The MSE properties will be epistemically inaccessible

to us, and will not be properties that figure in ordinary thought and

discourse—or, for that matter, in scientific thought and discourse. We can

easily know that two things differ in their MSE properties, but it will be

beyond our ability to know that two things share an MSE property, or that

something has remained unchanged with respect to MSE properties. And

sharing of MSE properties, or retention of them over time, will be exceed-

ingly rare, for the slightest difference in the location of a single electron

or quark will give us a difference in MSE properties. Such properties could

not figure in the taxonomy of any science, and could not figure in any laws

that we can formulate. Only an omniscient deity could have any knowledge

of them.

VII

Sometimes it is suggested that only first-order properties can be causally

efficacious and figure in causal laws. This goes with the view that the causal

efficacy ascribed to functional properties and other higher-order properties

is preempted by their lower order realizers. Those who hold this usually

assume that the first-order properties that “do the causal work” are ones that

50 SYDNEY SHOEMAKER

we can have knowledge of, and that the laws governing them are ones we

learn about from fundamental science. But if only MSE properties count as

first-order properties, this cannot be right. If the only causal laws are laws

about these, causal laws are unknowable by the likes of us. This is highly

implausible.

Consider the property of having a mass of one gram. One might have

thought that this is a paradigm first-order property. But it will have as

property realizers a vast—I think infinite—number of MSE properties.

These will differ from one another in the sorts of micro-entities that make

up the things having this mass, and in the ways these micro-entities are

configured. What they have in common is that the instantiation of each of

them bestows the causal powers bestowed by the property of having a mass

of one gram. When something has the mass of one gram in virtue of having

one of these MSE properties, it would be absurd to suggest that the effects

we attribute to its having this mass are really due instead to its having that

MSE property. For one thing, this would fly in the face of a plausible

proportionality constraint on the relation between causes and effects, for

the vast majority of the causal features of the MSE property will be irrele-

vant to the effects associated with having a mass of one gram (see Yablo

1992). It is by abstracting away from these causal features that we get to the

causal profile of the mass property. To be sure, it is in virtue of having that

MSE property that the thing has the mass of one gram, and so has the causal

powers that go with the possession of that mass. But it is the having of that

mass that gives it the relevant powers. In line with what I suggested earlier,

we can say that the instance of the MSE property contains as a part the

instance of the mass property, and that it is because of this that it bestows the

relevant powers.

What I have said about the property of having a mass of one gram, versus

the various MSE realizers of it, is similar to what Hilary Putnam said in his

famous discussion of why a cubical peg won’t go through one hole and will

go through another (Putnam 1975). The appropriate explanation of this, the

one having the generality we want from such an explanation, will be in

terms of the rigidity of the peg and board and the dimensions of the peg and

holes as described in macroscopic terms. The explanation that would be

perverse even if we were in a position to give it—Putnam says it is a “terrible

explanation” if it counts as an explanation at all—is one in terms of the

distribution of the micro-entities making up the objects. And that would

PHYSICAL REALIZATION WITHOUT PREEMPTION 51

amount to an explanation in terms of the MSE properties of the peg and of

the board that realize the shapes of the peg and the board and their rigidity.

We don’t, of course, think of properties like having a certain mass as

being property-realized by other properties that are not MSE properties—

these are usually thought of as ground level, first-order properties. On the

other hand, we do tend to think that the property of being in pain

has property realizers, perhaps C-fiber firing and the like, that are not

MSE properties. Let me observe in passing that it is not a consequence

of physicalism that this is so. Someone could be both a physicalist and a

functionalist while thinking that the property of being in pain has as good a

right as the property of having a mass of one gram to be regarded as a first-

order property, and that its only realizers are MSE properties. Whether it

has other realizers is an empirical question. But if it does, those realizers must

be realized, directly or indirectly, in MSE properties, and so also must the

property of being in pain. And what I just said about the property of having

a mass of one gram applies as well to the property of being in pain—and to

any other psychological property you care to mention. I said earlier that it is

only because the C-fiber firing instance contains the pain instance as a part

that it has the relevant effects. We can think of this as a consequence of the

relation between the microphysical realizers of these instances—it is because

the microphysical realizer of the C-fiber firing instance contains as a part the

microphysical realizer of the pain instance that it has the relevant effects.

And it is only because the maximally determinate microphysical state of

affairs that realizes an MSE property instance contains a state of affairs that is

a pain instance realizer (perhaps by way of containing a C-fiber firing

instance realizer) that it has the relevant effects.

How are we to understand this talk of states of affairs containing other states

of affairs as parts? As a first pass, state of affairs P contains state of affairs Q ifQ’s

existence is entailed by P’s existence. But the entailment can be seen as

holding in virtue of a relationship between the causal profiles of the states of

affairs that parallels the relationship between properties and their property

realizers. Where state of affairs P contains state of affairs Q, the forward-

looking causal features of P will contain as a subset the forward-looking causal

features of Q, and the backward-looking causal features of Q will contain as a

subset the backward-looking causal features of P. And this relationship, in

turn, will hold in part because the ways micro-entities are propertied and

related in Q are realized by the ways micro-entities are propertied and related

52 SYDNEY SHOEMAKER

in P. The contained states of affairs will be abstract relative to the containing

states of affairs, and can be said to be realized by them.

VIII

I said earlier that every case of microphysical realization, except for cases in

which the realized property is itself an MSE property, is a case of property

realization in which the realizer is an MSE property. This seems to be true

no matter what the world is like, and whether or not physicalism is true.

Will it also be true that every case of property realization will also be a case

of microphysical realization—i.e., that whenever a property instance is

realized by an instance of a different property, it is realized by a micro-

physical state of affairs? Assuming a version of physicalism on which the

microphysical facts fix all of the facts, this will be true. But there are

conceivable worlds, dualist ones, in which it is not true—in these many

property instances, in particular mental ones, do not have microphysical

realizers, but it may still be true that many property instances are realized by

other property instances. For example, the property of being a clock will be

realized by various mechanical and electronic properties. And there are

perhaps worlds in which the instantiation of physical properties in macro-

scopic entities is not realized by microphysical states of affairs. If so, in such

worlds there can be physical property realization unaccompanied by micro-

physical realization. My physicalist assumption is that the actual world is one

in which the microphysical facts fix all of the facts, and so one in which all

realization involves microphysical realization.

I began my discussion of realization by mentioning the multiple realiza-

tion argument against type psychophysical identity theory—the view that

mental properties are identical with physical properties. The objection was

that where a physical property might seem to be identical with a mental

property, because it plays its causal role, it is always possible for there to be

other properties that play the causal role of the mental property equally well

and so are equally good candidates for being identical with it, and that such

properties are at best realizers of the mental property rather than being

identical with it. But associated with any mental property there will be one

physical property that seems an especially good candidate for being identical

with it. If I had not restricted the application of the term “MSE property” to

PHYSICAL REALIZATION WITHOUT PREEMPTION 53

cases in which the embedded states of affairs are maximally determinate,

I could call these MSE properties. So let me call them MSE* properties.

Every property instance has a microphysical realizer of a type whose causal

profile exactly matches that of the property of which it is an instance, and

the thing having that property instance thereby has the property, an MSE*

property, something has just in case it embeds a microphysical state of affairs

of that type: (MSE properties are maximally determinate realizers of MSE*

properties). These MSE* properties should count as physical properties,

given that their instantiation just consists in microphysical entities being

propertied and related in certain ways. And they look to be excellent

candidates for being identical with the properties whose instances are realized

by the embedded microphysical states of affairs. So, for example, every

instance of the property of being in pain is directly realized by a microphysical

state of affairs of a certain type, and someone will have the property of being

in pain just in case he has the MSE* property something has in virtue of its

career embedding a microphysical state of affairs of that type.

But what I just said will be true only in worlds—of which I assume the

actual worlds is one—in which all of the facts are fixed by the microphysical

facts. In such worlds every property, including every mental property, will

be coextensive with an MSE* property. But if there are possible worlds in

which dualism is true, or worlds in which the microphysical facts do not fix

all of the facts, and if the properties instantiated in these worlds include some

of the properties, including the mental properties, instantiated in the actual

world, then these properties will not be necessarily coextensive with MSE*

properties and so cannot be identical with them.

Whether there are such possible worlds is a difficult question I cannot go

into here. I do not think that the conceivability or imaginability of such

worlds, or the fact that they are in some sense epistemologically possible,

shows that they are possible in the relevant sense—that they are metaphysic-

ally possible. And it is worth asking what the implications would be for the

philosophy of mind if it should turn out that they are not possible, and that

all properties of persisting things, including mental properties, are MSE*

properties, and so are physical properties.

Would this undermine the case for non-reductive physicalism? It would do

so, of course, if it is made definitive of non-reductive physicalism that it denies

any sort of type identity between mental properties and physical ones. But

I don’t think this should be made definitive of non-reductive physicalism.

54 SYDNEY SHOEMAKER

Notice that MSE* properties, though physical, are not for the most part

properties that figure under their physical descriptions in the laws of physics,

or are likely ever to so figure. And the ones that would be identical with mental

properties, on the supposition we are making, are not properties anyone has, or

ever will have, the ability to describe or define in canonical physical terms.

Accepting such an identification would not in the least support the claim that

psychology is reducible to physics, and would not put into question the

autonomy of psychological explanation, whether it be of the scientific sort or

the commonsense sort. Non-reductive physicalists should be happy to accept

psychophysical property identities involving MSE* properties.

Suppose, however, that dualist worlds aremetaphysically possible, and that

the coextensiveness of mental properties to MSE* properties is limited to a

subclass of possible worlds, those in which themicrophysical facts fix all of the

facts. What then is the relation between mental properties and coextensive

MSE* properties, given that it cannot be identity? It seems that it should be

realization or constitution of some sort. It could be property realization of the

sort defined earlier—the forward-looking causal features of the mental prop-

erty are a subset of those of the corresponding MSE* property and the

backward-looking causal features of the MSE* property are a subset of

those of the mental property. There is also a different subset relation between

them. The microphysical states of affairs that are realizers of instances of an

MSE* property would be a proper subset of the states of affairs that realize the

associated mental property. Just what the realizing states of affairs for mental

properties would be in a dualist world is hard to say—maybe, following

Putnam’s tongue-in-cheek suggestion, they would be ectoplasmic states of

affairs (Putnam 1967). Notice that we have this subset relation also in other

cases of property realization; if pain has C-fiber firing as a property realizer,

then given all of the states of affairs that obtain in all possible worlds, the states

of affairs that aremicrophysical realizers of instances of C-fiber stimulation are

a proper subset of the states of affairs that are realizers of pain. This gives us an

alternative account of property realization, one defined in terms of state of

affairs realization. Property P is a realizer of property Q if and only if the

possible states of affairs that are realizers of instances of P are a proper subset of

the possible states of affairs that are realizers of instances of Q—where a

possible state of affairs is a state of affairs obtaining in some possible world.

It seems to me that whether or not there are dualist worlds, mental

properties have as good a claim to be physical properties as automotive

PHYSICAL REALIZATION WITHOUT PREEMPTION 55

properties, architectural properties, computer properties, and botanical

properties. It is true of all these properties that all of their actual world

property realizers are physical, and that all of their actual world instances are

realized by microphysical states of affairs. If there are dualist worlds, mental

properties will in these have non-physical realizers—but so too, I think,

might automotive properties and botanical properties.

IX

To return from this digression, and conclude my paper, let me restate my

central claim. It is that a proper understanding of what it is for property

instances to be physically realized removes any threat that the causal efficacy

we ascribe to mental properties is preempted by their physical realizers. The

subset view of property realization suggests that the proper understanding of

“causal inheritance” is that the causal powers of an instance of a realized

property are always a proper subset of the causal powers of the instance of its

property realizer, and this suggests the view that the realized property

instance is included in the realizer property instance as a part. And the

same conclusion is suggested by a consideration of microphysical realization;

if microphysical realizers of higher-order properties are abstract states of

affairs that are realized by more concrete states of affairs that realize lower-

level properties that are property realizers of them, it seems appropriate to say

that the former states of affairs are included in the latter, and likewise that the

property instances realized by the former are included in those realized by

the latter. And then we can say that the lower-level property instances do the

relevant causal work because they have the higher-level property instances as

parts. In the case that primarily concerns us, physical property realizers of

mental property instances cause the effects we attribute to those mental

properties because they contain those mental property instances as parts.

References

Clapp, L. (2001). ‘Disjunctive Properties; Multiple Realizations’. The Journal of

Philosophy. 98(3): 111–36.

Kim, J. (1998). Mind in a Physical World. Cambridge, MA: MIT Press.

56 SYDNEY SHOEMAKER

Lewis, D. (1980). ‘Mad Pain and Martian Pain’, in N. Block (ed.), Readings in

Philosophy of Psychology, vol. 1. Cambridge, MA: Harvard University Press: 216–32.

McLaughlin, B. P. (2007). ‘Mental Causation and Shoemaker-Realization’.

Erkenntnis, 67: 141–72.

Putnam, H. (1967). ‘The Nature of Mental States’. First published as ‘Psychological

Predicates’ in W. H. Capitan and D. D. Merrill (eds.), Art, Mind and Religion.

Pittsburgh, PA: University of Pittsburgh Press. Repr. in Putnam,Mind, Language

and Reality: Philosophical Papers, Volume 2. Cambridge: Cambridge University

Press: 429–40.

—— (1975). ‘Philosophy and Our Mental Life’. In Putnam, Mind, Language and

Reality: Philosophical Papers, Volume 2. Cambridge: Cambridge University Press:

291–303.

Shoemaker, S. (1980). ‘Causality and Properties’. In P. van Inwagen (ed.), Time and

Cause. Dordrecht: D. Reidel: 109–35.

—— (2001). ‘Realization and Mental Causation’. In C. Gillett and B. Loewer

(eds.), Physicalism and Its Discontents. Cambridge: Cambridge University Press:

74–98.

—— (2003). ‘Realization, Micro-realization and Coincidence’, Philosophy and

Phenomenological Research, 67(1): 1–23.

—— (2007). Physical Realization. Oxford: Oxford University Press.

Watkins, M. (2002). Rediscovering Color. Dordrecht: Kluwer.

Yablo, S. (1992). ‘Mental Causation’. Philosophical Review, 101(2): 245–80.

PHYSICAL REALIZATION WITHOUT PREEMPTION 57

3

Mental Causation in thePhysical World

PETER MENZIES

1. Introduction

Mental causation is the phenomenon in which a mental state causes another

mental state or causes some behaviour. As Jaegwon Kim (1998, 2005)

reminds us, not much of commonsense psychology would make any sense

if mental causation were not real. Our conception of ourselves as conscious,

intentional agents capable of perception, memory, and reasoning is tied up

with the assumption of the reality of causal processes involving cognitive

phenomena.

However, philosophical questions about mental causation revolve

around, not so much whether it is important, but rather how it is possible

in the first place in the light of certain metaphysical assumptions and

principles. The classic instance of the philosophical problem of mental

causation is Descartes’ discussion of how mind–body interaction is possible

in his dualist metaphysics, according to which mind and body are two

radically different kinds of substances. Philosophers from Pierre Gassendi

onwards have pointed out such causal interaction is impossible within

Descartes’ metaphysics which accords primacy to causation by contact

forces and in which minds do not have any spatial location or extension.

The general consensus among philosophers is that Descartes was not suc-

cessful in solving this problem.

In contrast to Descartes’ problematic dualism, the monist metaphysics of

physicalism is thought to be more congenial to explaining how mental

causation is possible. For example, on the identity theory of the mind,

mental states are just neural states of the brain, so that mental causation is a

simple instance of neurophysiological causation. Unfortunately, this simple

solution to the problem doesn’t work if the identity theory isn’t tenable.

Many philosophers of mind now believe this to be the case in view of the

many serious objections faced by the theory, the most serious of which turns

on the multiple realizability of mental states (Fodor 1974; Putnam 1975).

However, physicalism as a more general framework has not lost its appeal

despite the waning popularity of the identity theory. Physicalists continue to

believe that the world and its contents are nothing over and above the

structures described by fundamental physics. The idea that physicalists try to

capture is that all the objects in the worlds are constituted out of physical

particles, and that all the properties and relations that these objects enjoy

depend, in some constitutive sense, on the properties and relations mapped

out in fundamental physics.

While there are still unresolved problems about the precise formulation

of this metaphysical view, most physicalists accept a formulation of physic-

alism in terms of a supervenience thesis. The following formulation by

Frank Jackson (1998) has become reasonably standard:

Physicalism about the mental: Any world that is a minimal physical duplicate of the

actual world is also a mental duplicate of it.

A minimal physical duplicate of the actual world is a world that contains the

same physical objects, physical properties and relations, and physical laws as

the actual world; and nothing else. It is important to restrict the set of worlds

used in the supervenience thesis to the set of minimal physical duplicate

worlds. For the physicalist should not accept that anyworld that is a physical

duplicate of the actual world is a mental duplicate. For there are worlds that

duplicate the physical entities of the actual world but include, in addition, a

number of non-physical entities such as Cartesian minds. Such worlds are

not relevant to the formulation of physicalism, which is supposed to be at

best a contingent truth about the actual world, not a necessary truth about

all worlds. Physicalists need not deny that there are such worlds with

Cartesian minds in the remote regions of logical space; they need insist

only that the actual world is not such a world.

The supervenience thesis above expresses a minimal commitment of

physicalism. It is a thesis endorsed by both reductive physicalists who

accept the identity of mental with physical properties and non-reductive

MENTAL CAUSATION IN THE PHYSICAL WORLD 59

physicalists who do not. In recent years, non-reductive physicalists have

tended to outnumber reductive physicalists, mostly because considerations

about multiple realizability have been regarded as biting into the plausibility

of the identity theory. Non-reductive physicalists express the hope that they

can explain how mental causation is possible within the austere metaphys-

ical framework of physicalism while avoiding the reductionism of the

identity theory. Indeed, they hope that it is possible to vindicate not only

the reality of mental causation, but also its independence and autonomy

from physical causation.

This paper will divide into two parts. In the first part I shall argue that

physicalism, whether of the reductive or the non-reductive variety, faces a

challenge just as serious as that faced by Cartesian dualism. I shall outline an

argument that proceeds from physicalist premises to the conclusion that

mental states are causally inert or epiphenomenal. The argument is related

to the well-known exclusion argument advanced by Jaegwon Kim (1998,

2005) that purports to show that non-reductive physicalism is an unstable

position that should be replaced by reductive physicalism. Like Kim’s

argument, the argument I advance appeals to an exclusion principle about

causation to the effect that a state that is causally sufficient for some effect

excludes any mental state that supervenes on it from being causally effica-

cious with respect to the effect. However, the exclusion principle I appeal

to is weaker than Kim’s principle. Also my argument is directed at all

versions of physicalism, reductive as well as non-reductive. Ultimately,

I shall conclude that the argument I describe is not sound: physicalism can

escape the conclusion about the causal inertness of the mental but only by

abandoning the exclusion principle about causation. Many physicalists will

find this conclusion hard to swallow, as the exclusion principle appears to be

very intuitive to them, with Kim (2005), for example, claiming that it is an

analytic truth. We shall see that the principle, when appropriately formu-

lated, is not a general truth of any kind, as there are straightforward counter-

examples to it.

The second half of the paper takes up the issue whether there is a better

formulation of the exclusion principle. Philosophical discussions of exclu-

sion principles seldom proceed in terms of a well-ground theory of

causation. I plan to remedy this defect by motivating a conception of

causation as difference-making and then using it to formulate an alterna-

tive, more satisfactory version of the exclusion principle not vulnerable to

60 PETER MENZIES

the counterexamples to the earlier version. Much of this discussion reports

on work done in collaboration with Christian List (List and Menzies 2009;

Menzies and List 2010). We have argued that the new principle is at best a

contingent truth about causal systems and have identified the conditions

that a causal system must satisfy in order for the principle to be true. It turns

out that the principle can apply in two non-trivial ways to a causal system.

The first—the case of upwards exclusion—is familiar from the argument

against physicalism: here a lower-level cause excludes a higher-level cause.

But the second—the case of downwards exclusion—is often overlooked:

here a higher-level cause excludes a lower-level one. These cases of

downwards exclusion are particularly interesting, as they support the causal

autonomy of higher-level properties. This is a surprising turn of events: far

from supporting reductionist thinking, the exclusion principle actually

turns out to be the linchpin of an argument that vindicates the causal

autonomy of mental properties. In the last section of the paper I turn to

consider the implications of this result for recently popular compatibilist

forms of non-reductive physicalism. Compatibilists attempt to answer the

exclusion argument against mental causation by claiming that mental and

physical states work in tandem to cause to behaviour in a form of non-

standard overdetermination. I concentrate on Sydney Shoemaker’s (2007)

version of compatibilism, arguing that the downwards exclusion result

demonstrates the untenability of his view that mental causation involves a

kind of non-standard overdetermination in which one kind of cause rides

piggyback on another.

2. A New Exclusion Argument

It may be best to illustrate the new exclusion argument against non-

reductive physicalism by way of an example, first introduced into philo-

sophical discussions by James Woodward (2008). The real-life example

concerns the research of Richard Andersen and colleagues at Caltech on

the neural encoding of intentions to act (Mussallam et al. 2004).1 Andersen

1 The ultimate goal of Andersen’s work is to develop neural prosthetics for paralysed subjects that

decode their intentions to reach for specific targets from neural signals and use these to control external

devices.

MENTAL CAUSATION IN THE PHYSICAL WORLD 61

and his colleagues made recordings from individual neurons in the parietal

reach region (PRR) of the motor cortex of monkeys. This region is known

to encode intentions or higher-order plans to reach for specific targets, say a

piece of fruit in a particular location. Andersen developed a programme that

correlated the monkey’s intentions to reach for specific goals, as revealed in

their movements, with certain patterns in the recorded firings of neurons in

their PRR. Using neural recordings, the programme was able to predict

with 67.5 per cent accuracy the reaching behaviour of the monkeys towards

eight targets.

The neural signals that encode themonkeys’ intentions to reach for certain

targets were recorded as averages of the firing rates (spikes per second) of

individual neurons. But clearly the same aggregate firing rate in a group of

neurons is consistent with a lot of variation in the behaviour of individual

neurons. For example, very different temporal sequences of neural firings

can give rise to the same firing rate. So an intention to reach for a certain

target can be realized in many different ways at the level of individual

neurons. Nonetheless, each intention is associated with a distinctive aggre-

gate pattern of firing rates. It is useful to introduce some simple notation.

Suppose themonkeys can have intentions to reach for certain targets, I1, I2, I3etc., and can perform the corresponding actionsA1,A2,A3, etc. Suppose that

intention Ii can be realized at the level of individual neurons in different

token patterns of neural firings, Ni1, Ni2, Ni3, etc. Suppose that on some

specific occasion a monkey forms the intention Ii to reach for a particular

object and performs the corresponding action Ai. Suppose further thatNi1 is

the particular token pattern of neural firing that realizes or encodes the

intention Ii on this occasion. What was the cause of the monkey’s action

Ai? Was it the intention Ii, or its particular neural realization Ni1? Let’s

assume that both the intention and its neural realization are causally sufficient

for the action. It is very tempting for physicalists to answer that it was the

highly specific, neural state Ni1 that caused the monkey to perform the

action. But if the neural state did all the causal work, it would appear that

the intention, which we are assuming is numerically distinct from its highly

specific neural realizer, has no causal role and so is epiphenomenal.

Let’s look at the argument in more detail. The argument relies on a

number of assumptions or principles, some of which were implicit in this

informal presentation of the argument. Let’s make these assumptions and

arguments explicit.

62 PETER MENZIES

(1) Supervenience and realization: mental properties supervene on distinct physical

properties; and so any given instance of a mental property will have an

instance of a distinct physical property as its supervenience base (alternatively,

any mental state will be realized by a distinct physical state).

This supervenience of mental properties on physical properties is a simple

consequence of the contingent supervenience thesis that we are taking to be

the minimal commitment of physicalism. Since we are discussing non-

reductive physicalism, the supervenience thesis is spelled out in terms of

mental properties supervening on distinct physical properties.

(2) Causation entails causal sufficiency: if the state S1 causes another state S2, then S1causally sufficient for S2.

What is meant by causal sufficiency here? I shall understand this as follows: a

state S1 is causally sufficient for S2 (in the actual world) if and only if all the

worlds among the set of minimal physical duplicates of the actual world in

which S1 holds are worlds in which S2 holds. Given that the minimal physical

duplicates of the actual world hold fixed the fundamental physical laws, this

means that these laws entail that an S1 state will lawfully evolve into an S2 state.

Of course, this is a questionable assumption that commits us to a deterministic

conception of causation.While I concede that causationmay involve probabil-

istic rather than deterministic processes, I make this assumptionmostly because

it simplifies our discussionwithout any significant loss of generality. It would be

misguided, I think, to imagine that the assumption of determinism is the source

of the difficulties affecting mental causation, which can be solved by repudiat-

ing this assumption. If there is a solution to the mental causation problem, it is

one that surely holds good even if we assume that causation is deterministic.

(3) The transmission of causal sufficiency across realization: if a mental state M is

causally sufficient for a behavioural state B and M is realized by a distinct

physical state P, then the physical state is causally sufficient for the behav-

ioural state B.

This principle should be no more controversial than the definition of causal

sufficiency given above, since it follows as an analytic consequence of this

definition. To see that the principlemust be true given the definition of causal

sufficiency, suppose, for reductio, that it is false; that is, suppose that the

mental state M is causally sufficient for B and that M is realized by a distinct

physical state P but P is not causally sufficient for B. Then it follows that

MENTAL CAUSATION IN THE PHYSICAL WORLD 63

among the minimal physical duplicate worlds there are someworlds in which

P holds but B does not. But by definition of supervenience, the P-worlds are

allM-worlds, so the worlds in question must be ones in whichM holds but B

does not hold. But this contradicts the assumption thatM is causally sufficient

for B, so demonstrating the falsity of our initial supposition.

(4) The new exclusion principle: if a mental state M is realized by a distinct physical

state P that is causally sufficient for B, then M does not cause B.

This principle is related to an exclusion principle that Jaegwon Kim (1998,

2005) formulates as follows:

(5) Kim’s exclusion principle: if a state S1 is causally sufficient for a state S2, then no

distinct state obtaining at the same time as S1 can cause S2.2

It’s easy to see that the new exclusion principle above follows from Kim’s

principle. Suppose that a stateM is realized by a physical state P that is causally

sufficient for B. ThenM and P obtain at the same time, and so it follows from

Kim’s principle thatM can’t cause B. On the other hand, it can be seen that

the new exclusion principle doesn’t imply Kim’s principle. Suppose that

mental stateM and physical state P obtain at the same time but are not related

by supervenience. Since the new exclusion principle only applies to pairs of

events related by supervenience, nothing follows from the principle con-

cerning whether M excludes P from causal efficacy. These considerations

show that the new principle is weaker than Kim’s. Kim (2005) says that his

exclusion principle is an analytic, a priori truth; and if this is correct it would

follow that the new exclusion principle is also such a truth. We shall see later

that neither principle is a truth of this kind. For now I simply rest content that

the new principle is no more implausible than Kim’s principle.

We are now in a position to formulate the new exclusion argument.

I present the argument schematically, but with a little effort it can be easily

2 Kim usually formulates his exclusion principle with the qualifying clause ‘unless it is a case of

genuine overdetermination’. However, I omit this qualification in order to simplify my discussion. By

the term ‘a case of genuine overdetermination’, Kim means the kind of situation in which the multiple

causes operate independently of each other in much the manner of two assassins who, completely

independently of each other, shoot the same victim at the same time. Since there is general agreement

that mental causation doesn’t involve this kind of overdetermination, it’s reasonable to assume that the

condition specified in the clause is not satisfied in the case of mental causation and so can be safely

ignored. Nonetheless, the whole question of whether mental causation involves overdetermination is

discussed further in section 6 below.

64 PETER MENZIES

translated into a concrete argument using the example about the monkeys’

neurally encoded intentions.

Suppose, for reductio, that a particular state M causes some behaviour B.

By (1) the state M is realized by a distinct physical state P.

By (2) the state M is causally sufficient for B

By (3) the state P is causally sufficient for B.

By (4) the state P excludes M as a cause of B.

Hence, a contradiction.

This argument is extremely simple. But a physicalist who accepts the

assumptions and principles listed above must accept the conclusion that

the causal efficacy of any mental state is excluded by that of its underlying

physical realizer.

This argument is related to Jaegwon Kim’s famous exclusion argument

(1998, 2005), which has a slightly broader target. Kim’s exclusion argument

attempts to show that any kind of property dualism that implies the mental

properties are distinct from physical properties is committed to epipheno-

menalism about the mental. Property dualists include both non-reductive

physicalists who accept the supervenience of mental properties on physical

and non-physicalists who deny this. Kim’s argument proceeds from slightly

different premises to the same conclusion about epiphenomenalism about the

mental. The differences between the arguments result from the fact that they

have different targets. Kim’s argument establishes a stronger conclusion that all

forms of property dualism are committed to epiphenomenalism about the

mental, but it proceeds from stronger premises, in particular relying on a

contingent principle to the effect that the physical world is causally closed.

3. Descartes’ Revenge

I have argued that the new exclusion argument is a problem for non-

reductive physicalists. However, it might be doubted whether the argument

is effective against reductive physicalists who assert the identity of mental

properties with physical properties. There is some reason for this scepticism,

as the premises of the argument all concern the relationship between mental

properties and distinct physical properties that form their supervenience

bases. For example, the supervenience premise states that a mental property

MENTAL CAUSATION IN THE PHYSICAL WORLD 65

supervenes on a distinct physical property, or alternatively, that the mental

stateM is realized by a distinct physical state P. In any case, even without the

distinctness qualification, the argument would be powerless against the

identity theory. If M is identical to P, then the argument generates no

contradiction at all: the fact that P is causally sufficient for B doesn’t

undermine the causal efficacy of M with respect to B given that P and M

are the same state.

In this connection, it is worth noting there was something unrealistic in

the way I presented the example about Andersen’s work on the neural

encoding of monkeys’ intentions. I ran an informal version of the argument

by supposing that the monkey’s intention Ii is realized by a distinct neural

state Ni1; and then appealing to exclusion reasoning to show that the causal

powers of the intention are excluded by those of the underlying neural state.

But it would be more reasonable to think in this example that the monkey’s

specific intention is actually identical with a neural state: namely, the

aggregate neural state of neurons in the cluster having a collective average

firing rate. This abstract neural state is plausibly represented as a disjunction

of all the more specific realizers Ni =Ni1 vNi2 vNi3 v . . . vNin, where each

disjunct represents a highly specific neural state consisting of a temporally

ordered sequence of the individual neural firings such that together these

states have a specified average firing rate. It is indeed reasonable to think that

Andersen and his colleagues thought of the monkeys’ intentions as being

identical with such aggregate neural states, as what they were directly trying

to manipulate were the aggregate patterns of firings. So these experimenters

are best seen as reductive physicalists who could plausibly claim that the new

exclusion argument is not effective against their view.

However, let’s subject this thought to more scrutiny. Does the reductive

physicalist really emerge unscathed by the new exclusion argument? Let’s

suppose that every mental state is identical to a physical neural state, perhaps

a fairly abstract one. Let’s suppose that each of the monkey’s intentions can

be identified with an abstract aggregate pattern of neural firing, as suggested

above. It would still seem that the argument should go through to show that

this abstract physical state will have its causally efficacy pre-empted by the

more specific neural state which realizes it. The only assumption required

to kick-start the argument is the assumption that the target state is realized

by a distinct underlying physical state. And this is true for the neural state

Ni, which on a given occasion will be realized by one state from the set

66 PETER MENZIES

{Ni1,Ni2,Ni3, . . . ,Nin}. These are more highly specific versions ofNi, each

being a complex state consisting in a temporally ordered sequence of

individual neural firings that satisfies the aggregate average associated with

Ni. These highly specific states are distinct fromNi and so the new exclusion

argument goes through to show that the causal efficacy of Ii =Ni is excluded

or pre-empted by the causal sufficiency of one of the highly specific realizer

states. It is easy to see that the argument can be reproduced for any physical

property that one cares to identify with a mental property: by taking its

realizers to be more fine-grained specifications of this physical property, one

can show that the physical property has its causal powers pre-empted by its

more fine-grained realizers.

It doesn’t require much effort to see that the physicalist world picture

itself will lead to a natural generalization of the argument. As defined above,

physicalism is committed to a multilayered model of reality, stratified into

different levels and bottoming out in fundamental physical level.3 Entities

belonging to a given level have an exhaustive decomposition without

remainder into entities belonging to entities belonging to the next level

down. So living organisms can be decomposed into cells, which can be

decomposed into molecules, then atoms and so on to the basic fundamental

physical particles—perhaps the quarks, leptons, and bosons of the standard

model. Physicalism, as I have defined it, accepts that the distribution of

fundamental particles with their properties and relations, together with the

way this distribution evolves in conformity with the fundamental physical

laws, will fix everything else in reality.

If this picture is correct, then any non-fundamental state, whether it be a

mental state, a neurophysiological state, a biochemical state, molecular state,

or atomic state, will have fine-grained realizers at the next level down that

are distinct from it. These realizers will consist in lower-level states involv-

ing constituents of the objects of the higher-level states configured into

3 It might be argued that reductive physicalism is better understood as committed to a flat, one-layer

view of reality rather than a multilayered view like non-reductive physicalism. However, contrary to this

view, I maintain that while reductive physicalists believe that the fundamental physical level is the basis

for all real entities (objects and properties), they must nevertheless accept the derivative reality of entities

at higher levels. Reductive physicalists can, for example, accept the reality of biological organisms and

biological properties because they are constituted out of fundamental physical objects and properties.

Accordingly, even the austere metaphysics of reductive physicalism makes room for a multilayered

conception of reality: the different levels of reality are ordered in terms of the complexity of the

constructions of the higher-order objects and in terms of the supervenience relations between properties.

MENTAL CAUSATION IN THE PHYSICAL WORLD 67

complex arrangements. So for example, if we take one of the fine-grained

neural states Nij above, which consists in a specific temporally ordered

sequence of individual neuron firings, then this state can be decomposed

into a highly complex state involving molecular arrangements and processes

that actually realize the sequence of neural firings. This complex molecular

state will be distinct from the given neural state because there are many

other arrangements of molecules and molecular processes that could realize

the same neural state. This process of decomposition can reiterate until one

finally reaches a decomposition of the neural state in terms of a state at the

level of fundamental physics. An application of the new exclusion principle

at each stage of the decomposition process will show that the causal efficacy

of each state resulting from the decomposition is rendered void by the causal

powers of its realizer, until we finally reach the last stage where only the

fundamental physical realizer state has any causal powers.

In this way, I would argue, the new exclusion argument can be general-

ized so that it applies even in the situation in which the mental state is

identified with a neural state. The generalization of the argument depends

only on the assumption guaranteed by physicalism that any non-fundamental

state, whether it be a mental, neural, chemical, or atomic, will be realized by a

distinct finer-grained physical state, and indeed ultimately by the finest-

grained state of all, the realizing state specified by fundamental physics. The

upshot of the generalization of the argument is that the only causal powers are

those possessed by fundamental physical properties and states. Ned Block

(2003) aptly calls this the causal drainage problem because the causal powers of

properties and states at all but the lowest level drain away to the lowest level.

This is an indeed serious problem for any physicalist, whether reductive or

non-reductive, who accepts the five assumptions and principles required to

generate the exclusion argument.

Let’s take stock of where we have arrived in our reasoning. At the outset,

we saw that Descartes’ substance dualism was jettisoned because of its

inability to explain the causal interactions between mind and body. It’s

impossible, so it is said, to vindicate mental causation within the metaphys-

ical model of substance dualism. At first sight, physicalists seem to be in a

better position to carry out the project of vindication. After all, what could

be clearer than the fact that mental states at the very least supervene on

physical neural states that can evidently produce or give rise to behaviour?

But now it would seem that the new simple exclusion argument threatens to

68 PETER MENZIES

muddy this clarity. By generalizing the argument in an apparently unprob-

lematic way, one can show that not only mental causation is unreal, but

biological, chemical, atomic causation are all unreal too. Moreover, the

premises from which the new exclusion argument proceeds are principles

that are definitive of physicalism (such as the supervenience thesis), or are

analytic or close to being so (like the principle of transmission of causal

sufficiency across realization), or are assumptions that many physicalists take

to be unexceptionable (like the new exclusion principle). If we should

repudiate substance dualism because of its failure to vindicate mental caus-

ation, what should we say about physicalism, both of the reductive and non-

reductive variety, that evidently fails to vindicate any kind of upper-level

causation? I entitled this section “Descartes’ Revenge”, because it would

seem that, from the point of view of commonsense plausibility, physicalism

is in a much worse position than substance dualism.

I have argued that the new exclusion argument applies more generally to

show that all non-fundamental properties and states are epiphenomenal.

This surely gives us reason to suspect that the argument is unsound. I will

eventually point the finger at the new exclusion principle as the false

premise that physicalists should reject. However, before moving on to

this, I wish to consider whether some philosophical manoeuvre will stop

the drainage of causal powers to the fundamental levels of physics. In this

connection, a number of philosophers have claimed that Kim’s superve-

nience argument, which is closely related to his exclusion argument, also

generates a causal drainage problem and must be defective for this reason.

However, Kim has responded that his supervenience argument doesn’t

generalize in this untoward way. So let’s examine Kim’s defence to see

whether it can be used to invalidate the generalization of the new exclusion

argument mooted above.

Kim’s defence (1998, 2005) involves his distinctive conception of the

hierarchy of levels. According to Kim, the hierarchy of levels applies in the

first instance to objects: objects are ordered into the hierarchy of levels on

the basis of the part–whole relation so that entities at one level are composed

out of smaller constituent entities at the next level down. The crucial point

that Kim makes is that his supervenience argument posits a supervenience

relation between mental and physical properties, which must, by virtue of

the definition of property supervenience, belong to the same object. In

order for the supervenience argument to give rise to causal drainage from

MENTAL CAUSATION IN THE PHYSICAL WORLD 69

the macro to micro-levels, the supervenience relation would have to apply

to properties belonging to objects at different levels. But this is not possible.

So, the worries raised by his supervenience argument are intra-level con-

cerns that do not cut across levels, in particular micro–macro boundaries.

Kim writes:

In general, supervenient properties and their base properties are instantiated by

the same objects and hence are on the same level . . . So the microphysical, or

mereological, supervenience does not track the micro–macro hierarchy: the

series of supervenient properties, one mereologically supervenient on the next,

when we go deeper and deeper into the micro, remains at the same level in the

micro–macro hierarchy . . . This means that the supervenience argument, which

exploits the supervenience relation, does not have the effect of emptying macro-

levels of causal powers and rendering familiar macro-objects and their properties

causally impotent. (1998, 86)

What is to be said in response to this? First, whatever is true of Kim’s

supervenience argument, it is not true of his own exclusion argument that

the properties that compete for causal efficacy belong to the same object.

His exclusion argument starts from the supposition that an organism’s

having a mental property causes its physical behaviour; and then, invoking

the physical causal closure principle, posits a simultaneous physical state that

is causally sufficient for this behaviour. The principle doesn’t require that

this physical state should be an instantiation of a physical property by the

very organism with the mental property: the physical state may consist in a

configuration of more basic physical objects having certain properties and

bearing certain relations to each other. So, whatever may be true of Kim’s

supervenience argument, his exclusion argument can cut across levels to

render higher-level properties and states causally redundant.

Secondly, this defence turns on the fact that the standard definition of

property supervenience requires that supervening and base properties

belong to the same object. But this is simply a peculiarity of the definition

of this type of supervenience. What’s actually crucial to the new exclusion

argument is the assumption that a higher-level state supervenes on, or is

realized by, a lower-level state. It is possible to define the supervenience of

one kind of state on another in a way that doesn’t require that the super-

vening and subvening states should be instantiations of properties by the

same object. For example, here is one such definition: a state S1 is realized

70 PETER MENZIES

by a state S2 in the actual world if and only if in the set of minimal physical

duplicates of the actual world, all worlds in which S2 holds are worlds in

which S1 holds. This definition doesn’t require that S1 and S2 should consist

in the same object instantiating different properties.

Thirdly, even assuming supervening and subvening properties must belong

to the same object doesn’t stop the causal drainage from non-fundamental to

fundamental properties. For Kim’s own notion of a micro-based property

(Kim 1998, 114) can be invoked to allow property supervenience to cut across

micro–macro boundaries. For an object to have a micro-based property is just

for the object to be decomposable into non-overlapping proper parts, each of

which has certain properties and all of which bear a certain relation to each

other. In other words, an object’s having a micro-based property is simply

constituted by the complex state of its proper parts being configured in a

certain way. So the property of instantiating a temporally ordered sequence

of neural firings may be a micro-based property, since a person has this

property just when the neurons of his brain enter into a certain complex

spatiotemporal relations with each other and have distinctive properties. So

a person may have a mental property and also this micro-based property,

meaning that the supervenience and realization relations can hold between

these properties. Clearly then the generalization of the new exclusion

argument can appeal to this kind of micro-based property to generate its

unacceptable conclusion that the causal powers of all macro-properties

drain away to the fundamental physical level.

So, I conclude that nothing that Kim says against the generalization

objection can assuage the worries about causal drainage.

4. Causal Relevance4

Where does the new exclusion argument go awry? I suggest that the error of

the new exclusion principle lies in its claim that one state’s causal sufficiency

for an effect excludes the causal efficacy of any state supervening on the first

state. The fundamental error of this principle is that it mistakes causal

sufficiency for causation. The fact that causal sufficiency doesn’t amount

4 This section and the next report on work done in collaboration with Christian List in List and

Menzies (2009).

MENTAL CAUSATION IN THE PHYSICAL WORLD 71

to causation has been known for some time. Wesley Salmon (1971) pointed

out that causal explanation requires information of the right kind and in the

right amount, or in other words, causal explanations must cite causally

relevant factors. However, while a man’s taking a contraceptive pill is

causally sufficient for his not getting pregnant, there is no causal relevance

here, as the man’s taking a contraceptive pill makes no difference to his not

getting pregnant. Even if had not taken the pill, he wouldn’t have got

pregnant.

As several philosophers have noted, the man’s taking the contraceptive

pill does not satisfactorily fit the role of a cause because it is overly specific

and involves extraneous detail. To be sure, the man’s taking the contracep-

tive pill is causally sufficient for the effect, but causal sufficiency is not the

same thing as causation, which requires causal relevance. To illustrate the

difference, consider an example of Stephen Yablo’s (1992) concerning a

pigeon that has been trained to peck at all and only red objects. The pigeon

is presented with a red target and she pecks at it. As it happens, the target is a

specific shade of crimson. What caused the pigeon to peck? Was it the fact

that the target was red or the fact that it was crimson? The exclusion

principle would say that since being red is realized by being crimson and

being crimson is causally sufficient for the pigeon’s pecking, the redness of

the target is not the cause. But this seems wrong, as Yablo points out: the

target’s being red is of the right degree of specificity to count as a cause of

the pigeon’s action. In contrast, the target’s being crimson is too specific to

count as the cause: citing it as the cause of the pecking might give the

erroneous impression that the pigeon would not peck at anything non-

crimson.

Many philosophers have sought to capture the idea of causal relevance in

terms of the dictum that causes make a difference to their effects.5 How are

we to cash out this dictum in more precise terms? I agree with those

philosophers (Pearl 2000; Spirtes et al. 2000; Hitchcock 2001; Woodward

5 Only the simplest form of causal relevance can be captured in terms of difference-making, as

explained below. The full complexity of the concept of causal relevance requires an account that goes

well beyond the simple outline I sketch below. For example, Woodward’s (2003) interventionist

account provides detailed explanations of a range of type-level causal concepts including direct, total,

and contributing causes that involve sophisticated elaborations of the basic notion of difference-making.

It is also important to note that the account below is not intended to handle the complications involved

in pre-emption and overdetermination examples. Such examples require a more sophisticated treatment.

72 PETER MENZIES

2003) who interpret causal claims as claims about relationships between

variables, and so interpret the dictum, quite literally, as requiring that

changing the value of the cause variable changes the value of the effect

variable. Applied to binary variables representing the presence or absence of

some state, the dictum says that changing the causal state from being absent

to being present (or vice versa) changes the effect state from being absent to

being present (or vice versa). Formally, I suggest that one state makes a

difference to another just when the following conditions are satisfied:

Truth conditions for causal relevance (or making a difference): The state S1 makes a

difference to the state S2 in the actual world just in case (i) if in any relevantly

similar possible situation S1 holds, S2 also holds; and (ii) if in any relevantly

similar situation world S1 does not hold, S2 does not hold.

For example, the target’s being red makes a difference to the pigeon’s

pecking because in any relevantly similar situation in which the pigeon is

presented with a red target it pecks, and in any relevantly similar situation in

which it is not presented with a red target it does not. The relevantly similar

situations in this example are ones in which the pigeon has received the

same training, the targets are presented to the pigeon in the same experi-

mental setting, there are no confounding influences on the pigeon and so

on. But under this construal of the relevantly similar situations, the target’s

being crimson does not make a difference to the pigeon’s pecking. Condi-

tion (ii) is not met: in a relevantly similar situation in which the pigeon is

presented with a non-crimson but red target, it still pecks. These observa-

tions confirm the conjecture that the requirement that causes make a

difference to their effects captures the crucial notion of causal relevance.

Further confirmation of this conjecture comes from examining how the

suggested truth conditions constrain the specificity of causes: satisfaction of

these conditions ensures that causes are specific enough for their effects, but

no more specific than needed. This is revealed most clearly in the case of

many-valued causal variables. Suppose, for example, there is a drug that

causes patients to recover from an illness. The effect variable is a binary

variable whose values are recovery or non-recovery. But the cause variable

is many-valued, with possible values 0mg, 50 mg, 100mg, 150 mg, and 200

mg. Suppose that any regular dose at or above 150 mg cures a patient, but

any lower dose does not. Suppose a patient has taken a regular dose of 150

mg and has recovered from the illness. What made the difference to the

MENTAL CAUSATION IN THE PHYSICAL WORLD 73

patient’s recovery? According to the truth conditions above, the answer is

‘Giving the patient a dose of at least 150 mg’. It satisfies both conditions (i)

and (ii): all relevantly similar patients who take a regular dose at or above 150

mg recover and all those who take a lower dose don’t. Other answers are

either too specific, or not specific enough. For example, the cause cannot be

‘Giving the patient a dose above 50 mg’ because that does not meet

condition (i): some relevantly similar patients who are given a dose above

50 mg, say 100 mg, do not recover. Similarly, it cannot be ‘Giving the

patient a dose of exactly 150 mg’ because that does not meet condition

(ii): some relevantly similar patients who are not given a dose of exactly

150 mg—say they are given 200 mg—nonetheless recover. In this way,

condition (i) rules out causes that are not specific enough to account for the

change in the effect variable, while condition (ii) rules out causes that are

too specific to account for it.

The truth conditions for making a difference can be expressed more

formally using counterfactuals, as understood in a possible-world semantics.

Specifically, let’s replace the notion of a relevantly similar situation with that

of a relevantly similar possible world, and thus rewrite the conditionals in

the truth conditions above as counterfactuals:

Truth conditions for causal relevance (making a difference): The S1makes a difference to

S2 in the actual world if and only if it is true in the actual world that (i) S1 holds

&? S2 holds; and (ii) S1 doesn’t hold &? S2 doesn’t hold.

I interpret the counterfactuals in accordance with the standard possible-

worlds semantics of David Lewis (1973), which provides truth conditions

for counterfactuals in terms of a similarity relation between possible worlds.

The similarity relation, which may vary with context, is represented by an

assignment to each possible world w of a system of spheres of worlds centred

on w. The system of spheres conveys information about the similarity of

worlds to the world w at the centre. The smaller a sphere, the more similar

to w are the worlds in it. So whenever one world lies in some sphere around

w and another lies outside it, the first world is more similar to w than the

second. In terms of this system of spheres, I now state the truth conditions

for counterfactuals as follows: P &? Q is true in world w if and only if Q is

true in all the closest P-worlds to w.

By adopting this semantic framework, I follow Lewis rather than Stalna-

ker, in allowing that there may be more than one closest P-world to w.

74 PETER MENZIES

Although there may sometimes be just one such world, this is not the

general rule. However, I diverge from Lewis in imposing only a weak

centring requirement on the systems of spheres. I allow the smallest sphere

around w to contain more than one world. Lewis imposes the stronger

requirement that the smallest sphere around w contains only w. This corres-

ponds to a constraint on the similarity relation whereby no world is as

similar to w as w itself. It also corresponds to the inference rule from the

premise P & Q to the conclusion P &? Q. In other words, if P and Q are

true in some world so is P &? Q. Lewis’ strong centring requirement, the

corresponding constraint on similarity and the corresponding inference rule

may appear plausible. But I cannot accept them. If the counterfactual

formulation of the truth conditions for causal relevance is to match the

earlier formulation, clause (i) of the counterfactual formulation must capture

the idea that every relevantly similar situation in which S1 holds S2 also

holds. In the original formulation, this condition is non-trivial: it rules out

insufficiently specific causes, provided the set of relevantly similar situations

in which S1 holds includes more than one such situation. To match this

condition, the counterfactual formulation must require that even if S1 and

S2 hold in the actual world, the smallest sphere around it also contains some

other worlds in which S1 holds.

Before I apply the difference-making account of causation to the exclu-

sion principle, I note an implication of the account. Several philosophers

(Hitchcock 1993; Woodward 2003; Schaffer 2006) have observed that causal

statements are contrastive in character. They have pointed out that descrip-

tions of both cause and effect seem to involve reference to a contrast

situation, or set of contrast situations. Sometimes the contrasts are made

obvious by the use of contrastive focus. For example, asserting a sentence

such as ‘Giving the patient a 150 mg dose of the drug caused his recovery’

highlights the fact that the 150 mg dose was one in a range of doses and not

all doses within this range cause recovery. But often the contrast situations

are left implicit. The rule for reconstructing the contrast situations is

straightforward in the case of causal claims involving binary variables.

Here the contrast situation is simply the opposite value to the actual one.

So the causal claim ‘The state S1 caused S2’ is to be understood as ‘S1’s

holding rather than not holding caused S2 to hold rather than not hold.’ All

these observations are predictable based on the account of causation as

difference making. If causal statements convey information about how

MENTAL CAUSATION IN THE PHYSICAL WORLD 75

variation in one variable is associated with variation in another, as explicated

by a pair of counterfactuals, it is no surprise that they can be expressed

contrastively.

The various examples discussed above—Andersen’s monkey and Yablo’s

pigeon—can be seen as counterexamples to the exclusion principle. In each

case, the exclusion principle leads us to identify the intuitively wrong

property as the cause of the given effect. In Yablo’s example, the causally

relevant cause of the pigeon’s pecking is not the crimson, but the redness of

the target, contrary to what the exclusion principle implies. This is sup-

ported by the truth of the counterfactuals:

Target is red &? pigeon pecks.

Target is not red &? pigeon does not peck.

In contrast, the following counterfactuals are not both true:

Target is crimson &? pigeon pecks.

Target is not crimson &? pigeon does not peck.

It is natural to interpret these counterfactuals in terms of a similarity relation

that makes the closest worlds in which the target is not crimson ones where

it is some other shade of red.6 Given this assumption, the second counter-

factual is false: in the closest worlds in which the target is not crimson it is

some other shade of red, in which case the pigeon will still peck.

A similar treatment can be given for the example of the monkey. The

causally relevant cause of the monkey’s reaching actionAi is not its particular

neural state Ni1, but its intention Ii. The following counterfactuals are true:

Monkey has intention Ii &? monkey performs Ai.

Monkey doesn’t have intention Ii &? monkey doesn’t perform Ai.

Whereas the following counterfactuals are not both true:

Monkey has neural property Ni1 &? monkey performs Ai.

Monkey doesn’t have neural property Ni1 &? monkey doesn’t perform Ai.

Assuming that the closest worlds in which the monkey doesn’t have neural

property Ni1 are ones in which it has another neural property realizing the

intention Ii, one can see that the second counterfactual is false: in any such

6 I discuss this assumption further in the next section.

76 PETER MENZIES

world, the monkey has another neural property that realizes Ii and so

performs Ai.

In summary, requiring causes to be causally relevant, one can see that the

exclusion principle is false. Even when some state S1 is causally sufficient for

another state S2, a state that supervenes on S1 can nonetheless be a cause of

S2. The monkey’s intention Ii to reach for a specific target is the cause of its

reaching action Ai even though it is realized by the neural property Ni1,

which is causally sufficient for the action.

5. Revised Exclusion Principle

We have seen that within the framework of a difference-making account of

causation there are some persuasive counterexamples to the exclusion

principle. A central feature of this principle is that it is couched in terms

of causal sufficiency: it states that a property that is causally sufficient for some

effect excludes certain other properties from being causes of that effect. But

one might ask: “Why talk of causal sufficiency rather than causation?”

Naturally, this raises the question of what happens if we reformulate the

exclusion principle, replacing the reference to causal sufficiency with one to

causation in a more adequate sense, understood as difference-making. So

let’s consider the following revised principle:

Revised exclusion principle: For all distinct states S and S* such that S* is realized by

S, S and S* do not both cause state T.

Here the truth conditions for causation are those for difference-making

introduced above. The principle can also be formulated in two different,

but logically equivalent ways. The first is the counterpart of the original

principle, whereas the second is seldom explored in the debate about the

exclusion problem:

Revised exclusion principle (upwards formulation): If a state S causes a state T, then no

distinct state S* that supervenes on S causes T.

Revised exclusion principle (downwards formulation): If a state S causes a property T,

then no distinct state S* that realizes S causes T.

Although logically equivalent, the two formulations draw our attention to

two different ways in which the exclusion principle can apply. An instance

MENTAL CAUSATION IN THE PHYSICAL WORLD 77

of upwards exclusion occurs when there exists a subvenient difference-making

cause that excludes a supervenient one; and an instance of downwards

exclusion, usually overlooked, occurs when there exists a supervenient

difference-making cause that excludes a subvenient one.

Is the revised exclusion principle true or false? Let’s focus on the instance

of the principle that concerns the causal relationships between a mental state

M, a neural state N, and a behavioural state B. Throughout the discussion,

I assume that N realizes M in the actual world. We are interested in the

logical relationship between the following two propositions:

(1) M is a difference-making cause of B.

(2) N is a difference-making cause of B.

Using the truth conditions introduced above, each of these propositions is

equivalent to a conjunction of counterfactuals:

(1a) M holds &? B holds.

(1b) M doesn’t hold &? B doesn’t hold.

(2a) N holds &? B holds.

(2b) N doesn’t hold &? B doesn’t hold.

The revised exclusion principle dictates that propositions (1) and (2), or

equivalently (1a), (1b), (2a) and (2b), are never simultaneously true. But is

this claim actually correct? One benefit of formulating the difference-

making conception of causation in terms of counterfactuals is that it

makes this question logically tractable. One can prove that these four

counterfactuals hold only under very special conditions. To state this result,

call a causal relation betweenM and B realization-sensitive if B fails to hold in

all thoseM-worlds that are closest �N-worlds (i.e., whereM has a different

realizer from the actual one). The result is the following:

Compatibility Result (List andMenzies 2009): IfM causes B, thenN causes B if and

only if the causal relation between M and B is realization-sensitive.

Rather than prove this result here, it is more instructive to describe a

situation that exemplifies the result. So consider the situation represented

in Figure 3.1. The concentric spheres represent sets of more andmore similar

worlds to the actual world; the innermost sphere contains the actual world,

labelled w, and the other worlds deemed maximally similar to it. The set of

N-worlds is represented by the convex region with light shading, and the set

78 PETER MENZIES

of M-worlds by the larger convex region that includes the set of N-worlds.

The regionwith darker shading represents the set ofB-worlds and this region

coincides with the innermost sphere. In this situation, it is easy to see thatM

causes B. First, since M holds throughout the innermost sphere, that sphere

picks out the closest M-worlds, and since B also holds in it, counterfactual

(1a) is true. Second, since B does not hold in any�M-worlds, it fails to hold

in all the closest�M-worlds and thus counterfactual (1b) is true. Further, the

causal relation between M and B is realization-sensitive: since B does not

hold in any�N-worlds, it follows a fortiori that it does not hold in any of the

closest �N-worlds that are M-worlds. And finally, N does indeed cause B:

counterfactuals (2a) and (2b) can easily be verified to be true.

It is important to note, however, that the conditions under which the

counterfactual pair (1a)–(1b) implies the pair (2a)–(2b) are very special.

Figure 3.1 illustrates this point nicely. Although both M and its actual

realizing state N are difference-making causes of B here, the realization-

sensitivity of the causal relation betweenM and Bmeans that small perturb-

ations in the way in whichM is realized would result in the absence of B. In

other words, if M were realized by any neural state other than N then B

would cease to hold. When might we expect the conditions for realization-

sensitivity to obtain? If the mental property M were identical to the neural

property N, then we would certainly expect instances of M to stand in

realization-sensitive causal relations with respect to instances of N. The

fact that M-instances had certain effects when and only when N-instances

are present would simply reflect the identity of the properties. However,

M

B

N

.w

Fig. 3.1

MENTAL CAUSATION IN THE PHYSICAL WORLD 79

I do not rule out the possibility of other explanations of the existence of

realization-sensitive causal relations.

At this point it is useful to consider a logically equivalent formulation

of the Compatibility Result that is relevant to Downwards Exclusion. In

analogy with the earlier definition, call a causal relation between M and B

realization-insensitive if B holds in someM-worlds that are closest�N-worlds

(i.e., where M has a different realizer from the actual one). The following

proposition is an immediate corollary of the Compatibility Result:

Downwards Exclusion Result (List and Menzies 2009): If M causes B, then N does

not cause B if and only if the causal relation between M and B is realization-

insensitive.

Again let’s consider a schematic example that exemplifies this proposition,

focusing on the situation represented in Figure 3.2. As before, the system of

spheres represents sets of worlds with greater or lesser degrees of similarity to

the actual world, labelled w. The set ofN-worlds is represented by the convex

region with lighter shading, and the set ofM-worlds by the larger vex region

that includes the set of N-worlds. The region with darker shading, which

coincides with the innermost sphere, represents the set of B-worlds. This

figure shows thatM causesB, sinceB holds in all the closestM-worlds and fails

to hold in all the closest�M-worlds, i.e., counterfactuals (1a) and (1b) are both

true. It is also easy to see that this causal relation is realization-insensitive:

B continues to hold in some, indeed all, of the M-worlds that are closest

�N-worlds. Finally, it is easy to see thatN does not causeB: the counterfactual

(2b) is false, since B holds in all the closest �N-worlds.

B

M

N

.w

Fig. 3.2

80 PETER MENZIES

6. Some Implications for Compatibilism

To highlight the significance of these results I want to examine some of

their implications for a non-reductive physicalist solution to the mental

causation problem that has come to be called compatibilism. This has

recently become a popular position among philosophers. (See Thomasson

1998; Crisp and Warfield 2001; Pereboom 2002; Bennett 2003, 2008.)

This solution says that any piece of intentional behaviour has two causes: a

mental state and the neural state that realizes it. These causes are not partial

causes in the way that a short circuit and the presence of oxygen are each

partial causes of a fire since the mental state and its realizing neural state are

each causally sufficient by themselves for the physical behaviour. Nor are they

standard overdetermining causes in the way that two assassins who fatally

shoot their victim at the same time are each causes of the victim’s death. For

themental state and its realizing neural state aremetaphysically connected in a

way that the overdetermining causes usually are not. In short, the compatibi-

list solution says, very roughly, that a mental state and its realizing neural state

are non-standard overdetermining causes of a piece of intentional behaviour.

Sydney Shoemaker (2007) has provided the most fully developed version

of the compatibilist solution. Shoemaker argues that in any given world a

property’s identity is determined by its causal profile—the set of its forward-

looking causal powers to cause other properties and its backward-looking

causal powers to be caused by other properties. In other words, whether

properties count as the same or different depends on whether they have the

same causes and effects.7 In his discussion Shoemaker gives equal weight to a

property’s forward- and backward-looking causal powers. But it will serve

my exposition to focus just on properties’ forward-looking causal powers.

Further, talking about a property’s forward-looking causal powers is short-

hand for talking about the causal powers of instances of the property. An

instance of a property F has the causal power to produce an instance of the

property G just in case the first instance can in suitable conditions cause the

second. The causal powers of properties are to be understood in terms of

generalizations about the causal powers of their instances.

7 While Shoemaker actually believes that a property’s causal profile defines its inter-world identity, his

account of the realization relation depends only on causal profiles providing an intra-world criterion of

identity.

MENTAL CAUSATION IN THE PHYSICAL WORLD 81

Shoemaker (2007, Chap. 2) explains realization in terms of the inclusion

of causal powers. He states that one property G realizes another property F

just in case the forward-looking causal powers of F are a subset of the causal

powers of G. So a neural state N realizes a mental state M just in case the

forward-looking causal powers of the mental property are a subset of the

causal powers of the neural property. This definition is supposed to avoid

the problem that the causal role of the realized property is pre-empted by

the causal role of the realizer property. He remarks that, on the contrary, his

account starts with the supposition that the realized property has a causal

profile, and nothing in the account takes this assumption back. Moreover,

his account avoids appealing to a problematic kind of overdetermination.

While it is true that a mental state and a neural state are both causes of

behaviour, the neural state has its causal role in virtue of the causal role of the

mental state. He explains this by analogy with an example. Suppose a firing

squad fires a salvo of shots at Smith but the only shot that hits Smith is one

fired by Jones. In this case the salvo killed Smith, but it did so in virtue of a

particular shot, Jones’ shot, that killed Smith. As a further illustration, he

cites Yablo’s example about Sophie, the pigeon:

Now we take scarlet as a realizer of red. The forward-looking causal features of a

red are a subset of the forward-looking features of scarlet . . . This instantiation of

red was realized in an instantiation of scarlet, and the instantiation of scarlet was

of course causally sufficient (in the circumstances) for the occurrence of Sophie’s

pecking. But it seems right to say that it was the instantiation of red, not the

instantiation of scarlet, that caused Sophie’s pecking. (2007, 14)

The instance of scarlet caused Sophie’s pecking but it did so in virtue of the

realizing the instance of red.

Crucial to the feasibility of the compatibilist solution is acceptance of two

principles that are implicit in Shoemaker’s discussion. The first principle is that

causal sufficiency is the same thing as causation. Shoemaker is tacitly committed

to this principle. In the quoted passage above, for example, Shoemaker moves

back and forwards from the claim that an instance of a colour is causally sufficient

for the pigeon’s pecking to the claim that it is a cause. The second principle is

what I earlier called the transmission of causal sufficiency across the realization

relation: if a property is causally sufficient for another property then any

property that realizes the first property is also causally sufficient for the second.

This principle holds as an analytic consequence of Shoemaker’s definition of

82 PETER MENZIES

realization in terms of inclusion of causal powers and his assumption that causal

sufficiency is the same thing as causation.8 Acceptance of these principles

allows one to argue that if a mental property is causally sufficient for some

piece of intentional behaviour, then both the mental property and its realizing

neural property are causes of the behaviour and that the neural property is a

cause in virtue of the fact that the mental property is. Though there are other

routes to compatibilism, many philosophers who offer compatibilist solutions

to the mental causation problem seem to be committed to these principles.

Despite its initial plausibility, the compatibilist solution does not, I think,

stand up to critical scrutiny. The solution entails that in Yablo’s example the

red and the scarlet properties are both causes of the pigeon’s pecking; and

that a mental property and its realizing neural property are both causes of

intentional behaviour. If this looks like one too many causes, the multipli-

cation of causes does not end with two causes. In section 3, I argued that the

new exclusion argument generated a causal drainage problem in that causal

powers of higher-level properties drain away to the fundamental physical

level. The principles used in this argument generate a descending sequence

of properties, one for each level in the hierarchy of levels, each of which is

causally sufficient for some effect. To generate such a sequence all that is

required is the assumption that some higher-level property, say a mental

property, is causally sufficient for some effect, say some behaviour. Then by

successive applications of the supervenience principle and the principle of

the transmission of causal sufficiency across realization, one can show there

is a descending sequence of properties, each of which (bar the first) is a

realizer of the one above and each of which is causally sufficient for the

effect. The upshot of this argument is that an effect of a higher-level causal

property such as a mental property has many causes: besides its mental cause,

it has a neural cause, a biochemical cause, a molecular cause, an atomic

cause and so on, one for each level in the hierarchy of levels. Positing

two piggybacking causes is somewhat implausible, but positing a whole

sequence of piggybacking causes for any effect strains credibility.

In any case, a central assumption of compatibilist solutions to the mental

causation problem, namely, that causal sufficiency is the same thing as

8 The principle of the transmission of causal sufficiency across the realization relation does not require

the specific doctrines of Shoemaker’s framework. A compatibilist who held that properties are sets of

possibilia and that realization is a matter of the inclusion of one set of possibilia in another would also be

committed to the principle.

MENTAL CAUSATION IN THE PHYSICAL WORLD 83

causation, is seriously mistaken, as I have been at pains to argue. Replacing the

causal sufficiency conception of causation with a more sensible one like the

difference-making conception does not increase the plausibility of the com-

patibilist solution. Indeed, the downwards-exclusion result, described in the

last section, implies that when a mental state causes some behaviour in a

realization-insensitive way, then the neural state that realizes it cannot also

cause the same behaviour. So when the realization-insensitivity condition is

met, the downwards-exclusion result rules out as impossible the kind of dual

cause solution entailed by compatibilism. There is good reason to think the

realization-insensitivity condition is met in the kinds of cases that compatibilist

take as their standard models. For instance, it is reasonable to think about the

Yablo example that the causal relation between the target’s being red and the

pigeon’s pecking is realization-insensitive in that the causal relation doesn’t

depend on the particularway inwhich the redness of the target is realized by its

being scarlet. More precisely, it is plausible to think that the similarity ordering

of possible worlds relevant to this example is such that in some of the closest

worlds in which the target is not scarlet but still red the pigeon pecks. Given

the reasonableness of this assumption, it follows from the downwards-exclu-

sion result that the target’s being scarlet can’t also be a cause of the pigeon’s

pecking along with the target’s being red. The easiest way to see this is to

consider the counterfactual ‘If the target had not been scarlet, the pigeon

would not have pecked.’ The realization-insensitivity condition implies that

in some of the closest worlds inwhich the target is not scarlet but is nonetheless

red the pigeon pecks, which then implies that this counterfactual must be false.

More generally, it follows from the downwards-exclusion result that if the

causal relation between a mental state and some piece of physical behaviour is

realization-insensitive in theway specified above, then it’s logically impossible for

any state that realizes themental state to be a cause of the same behaviour. So any

kind of compatibilist solution to themental causation is ruled outwhere there is a

reasonable assumption that the mental causation is realization-insensitive.

7. Conclusion

I began this paper by formulating a new exclusion argument that is similar to

Kim’s exclusion argument in attempting to show that a physicalist who

accepts certain principles including a weak causal exclusion principle is

84 PETER MENZIES

committed to the view that mental states and other higher-level states

are causally inefficacious. The argument differed from Kim’s argument,

however, in being directed at all forms of physicalism, both reductive and

non-reductive. My aim was not to endorse the new exclusion argument,

but rather to highlight the absurd implications of the exclusion principle it

employs. I subsequently showed that, within the framework of a difference-

making conception of causation, there are straightforward counterexamples

to this exclusion principle. My intermediate conclusion was that both the

new exclusion argument and Kim’s exclusion argument are unsound to the

extent they rely on an exclusion principle that implies that a mental state is

causally inefficacious with respect to some behaviour if it supervenes on a

physical state that is causally sufficient for the behaviour.

In the second half of the paper I explored a version of the exclusion

principle that is formulated in terms of difference-making rather than causal

sufficiency. The principle states that if two states are related by super-

venience, it cannot be that both states are difference-making causes of

another state. It turns out that while this principle is not a logical truth,

there are nonetheless certain contingent conditions concerning the realiza-

tion-insensitivity of difference-making counterfactuals under which the

principle is true. Moreover, the principle can be applied in an upwards

direction to support exclusion reasoning about upper-level states and, less

familiarly, in a downwards direction to support exclusion reasoning about

lower-level causes. The application of the principle in a downwards direc-

tion has several significant implications, one of which I explored in con-

nection with Shoemaker’s compatibilist solution to the mental causation

problem. The downwards exclusion result implies that the compatibilist

view that a mental state and its underlying realizer are both causes of some

piece of behaviour can’t be correct when causation is understood in a

plausible way in terms of difference-making and when the relevant assump-

tion of realization-insensitivity holds.

References

Bennett, K. (2003). ‘Why the Exclusion Problem Seems Intractable and How, Just

Maybe, to Tract It’. Nous, 37: 471–97.

MENTAL CAUSATION IN THE PHYSICAL WORLD 85

Bennett, K. (2008). ‘Exclusion Again’. In J. Hohwy and J. Kallestrup (eds.), Being

Reduced: New Essays on Reduction, Explanation, and Causation. Oxford: Oxford

University Press: 280–305.

Block, N. (2003). ‘Do Causal Powers Drain Away?’ Philosophy and Phenomenological

Research, 67: 133–50.

Crisp, T. and T. Warfield (2001). ‘Kim’s Master Argument’. Nous, 35: 305–16.

Fodor, J. (1974). ‘Special Sciences, or the Disunity of Sciences as a Working

Hypothesis’, Synthese, 28: 97–115.

Hitchcock, C. (1993). ‘The Role of Contrast in Causal and Explanatory Claims’.

Synthese, 107: 395–419.

—— (2001). ‘The Intransitivity of Causation Revealed in Equations and Graphs’,

Journal of Philosophy, 98: 273–99.

Jackson, F. (1998). From Metaphysics to Ethics: A Defence of Conceptual Analysis.

Oxford: Oxford University Press.

Kim, J. (1998). Mind in a Physical World: An Essay on the Mind–Body Problem and

Mental Causation. Cambridge, MA: MIT Press.

—— (2005). Physicalism, or Something Near Enough. Princeton: Princeton Univer-

sity Press.

Lewis, D. (1973). Counterfactuals. Oxford: Basil Blackwell.

List, C. and P. Menzies (2009). ‘Nonreductive Physicalism and the Limits of the

Exclusion Principle’. Journal of Philosophy, 1006: 475–502.

Menzies, P. and C. List (2010). ‘The Causal Autonomy of the Special Sciences’. In

G. and C. Macdonald (eds.), Emergence in Mind. Oxford: Oxford University

Press: 108–28.

Mussallam, S., B. Corneil, B. Greger, H. Scherberger, and R. Andersen (2004).

‘Cognitive Control Signals for Neural Prosthetics’. Science, 305: 258–62.

Pearl, J. (2000). Causality: Models, Reasoning, and Inference. New York: Cambridge

University Press.

Pereboom, D. (2002). ‘Robust Nonreductive Materialism’. Journal of Philosophy,

99: 499–531.

Putnam, H. (1975). ‘The Nature of Mental States’. In Mind, Language, and Reality:

Philosophical Papers, Volume 2. Cambridge: Cambridge University Press: 429–40.

Salmon, W. (1971). ‘Statistical Explanation’. In Salmon (ed.), Statistical Explanation

and Statistical Relevance. Pittsburgh, PA: University of Pittsburgh Press: 29–87.

Schaffer, J. (2006). ‘Contrastive Causation’. Philosophical Review, 114: 297–328.

Shoemaker, S. (2007). Physical Realization. Oxford: Oxford University Press.

Spirtes, P., C. Glymour, and R. Scheines (2000). Causation, Prediction, and Search.

Cambridge, MA: MIT Press.

Thomasson, A. (1998). ‘A Nonreductive Solution to Mental Causation’. Philosoph-

ical Studies, 89: 181–95.

86 PETER MENZIES

Woodward, J. (2003). Making Things Happen: A Theory of Causal Explanation.

New York: Oxford University Press.

—— (2008). ‘Mental Causation and Neural Mechanisms’. In J. Hohwy and

J. Kallestrup (eds.), Being Reduced: New Essays on Reduction, Explanation, and

Causation. Oxford: Oxford University Press: 218–62.

Yablo, S. (1992). ‘Mental Causation’. Philosophical Review, 101: 245–80.

MENTAL CAUSATION IN THE PHYSICAL WORLD 87

4

Mental Causation: Ontologyand Patterns of Variation

PAUL NOORDHOF

Physicalism was initially motivated by its ability to deal with the problems of

mental–physical interaction. The most attractive version of physicalism,

though, is one which allows the mental some degree of autonomy with regard

to the physical. Few physicalists feel driven to defend the claim that mental

properties are identical with those which are identified by some suitably refined

version of current physics. Unfortunately, as is only too familiar, non-reductive

physicalism—that which denies such an identification—seems to have signifi-

cant problems withmental causation of its own. In this paper, I begin by setting

out the challenge to its efficacy due to Jaegwon Kim. I shall do this briefly

because I am sure the reader has, by now, tired of seeing this argument stated.

I just want to make a couple of comments upon it for the discussion ahead.

I then discuss two over-reactions to it—one which seeks to understand mental

efficacy non-ontologically in terms of patterns of variation, the other of which

uses the problem to motivate a particular ontology—trope metaphysics.

I explain why I consider these over-reactions, identify what is unsatisfactory

about them, and then take elements of each to motivate my own approach

which, you may not be surprised to learn, captures what is best in both. It also

dealswith an issue aboutmental causation untouched byKim’s initial challenge.

1. Kim’s Exclusion Argument

I’m going to set out Kim’s argument making certain assumptions to fix ideas.

These assumptions don’t change the import of the argument, nor affect the

responses to it I am going to consider. They just simplify presentation.

Specifically, I will assume that the non-reductive physicalist is committed

to holding that there is more than one arrangement (A1, A2, A3 . . . ) of

narrowly physical properties (P1, P2, P3 . . . )—those properties identified

by current physics or a future development of it which suitably resembles

it—such that, for each of them, it is metaphysically necessary that if they are

instantiated, then a certain broadly physical property (BP) is instantiated

(henceforth I will use upper case letters to designate type-properties and

lower case letters to designate specific instances). One subclass of broadly

physical properties is that of mental properties and behavioural properties. It

is because all properties instantiated in the world are either narrowly physical

or broadly physical, that non-reductive physicalism is true.

The appeal to metaphysical necessity is required to capture the fact that

non-reductive physicalism is committed to a tighter connection than mere

nomological necessity between arrangements of physical properties and

mental properties. The latter type of connection would be acceptable to

the emergent dualist. Debate has raged over whether appeal to metaphysical

necessity is sufficient to capture what is required. I have defended this

conclusion (Noordhof 2003, 2010). Nevertheless, all that matters is that it

is stronger than the relation allowed by emergent dualists (bracketing an

issue I touch on in section 5 about a powers ontology). If it is not, not only

do we not have a version of physicalism but the issue set aside in the

comment below about other events in the causal chain or causal circum-

stances becomes salient.

Appeal to metaphysical necessitation may appear too strong (e.g., Kim

2005, 49). Consider the relationship between O, a property occupying a

certain causal role R, and the property of having role R. On some accounts

of the connection between properties and laws, the relationship between

O and R is one of merely nomological necessity. Laws independent of O,

but governing its causal relations, give O the R-role. Nevertheless, the

thought runs, the instantiation of the property of having role R is explained

by the presence of O and the laws which hold relating to O. Instead of

metaphysical necessity, we have nomological necessity plus explanation.

This issue can be set aside by allowing that narrowly physical laws—those

identified by physics—can be part of the metaphysical necessitation-base for

a property. Thus we do have metaphysical necessitation still in play between

Os and laws on the one hand, and R on the other.

MENTAL CAUSATION: ONTOLOGY AND PATTERNS OF VARIATION 89

With these assumptions in place, the argument against the efficacy of

those broadly physical properties recognized by non-reductive physicalism

runs as follows.

(1) A1(p1, p2, p3 . . . ) is causally sufficient for, or fixes the probability of,

A2(p100, p101, p102 . . . ) (necessitation-bases for, but not identical to,

bp1, bp2, respectively).

(2) bp1 is a cause of bp2 (Assumption).

(3) bp1 causes bp2 either directly or by causing A2(p100, p101, p102 . . . ).

(4) If bp1 causes bp2 directly, then either A1(p1, p2, p3 . . . ) is insufficient

for bp2 by causing A2(p100, p101, p102 . . . ) or bp1 is an overdetermin-

ing cause.

(5) If bp1 causes bp2 by causing A2(p100, p101, p102 . . . ), then the same

choice holds regarding A2(p100, p101, p102 . . . ).

(6) There is no systematic overdetermination in this way.

Therefore,

(7) bp1 is inefficacious (see, e.g., Kim 1998, 41–7; Kim 2005, 39–52).

The argument does not claim that if A1(p1, p2, p3 . . . ) is causally sufficient

for A2(p100, p101, p102 . . . ) then there can be no other sufficient cause

without overdetermination. There may be other sufficient causes which

are part of the causal circumstances, or further up or down the causal chain.

The focus is just on the efficacy of A1(p1, p2, p3 . . . ) and bp1 for target effects

A2(p100, p101, p102 . . . ) and bp2 standing in the same relationship. The

question is whether, at that point in the causal network, there is any

contribution for bp1 to make given A1(p1, p2, p3 . . . )’s presence.

Talk of position in a causal network may raise alarm bells because of

putative difficulties in fitting conditions, in particular negative conditions,

into the framework (Steward 1997, 135–40). It should not. The argument

does not require an exhaustive causal network. All that is required is that we

can make sense of the idea that token events, or property instances, stand in

a causal network against a backdrop of assumed causal conditions and that, as

a result of this, we can see two or more events, or property instances, as in

potential competition for efficacy at a certain position in this network.

Although I have dubbed this argument Kim’s exclusion argument, the

appeal to the causal exclusion principle is implicit. The principle holds that

90 PAUL NOORDHOF

No single event can have more than one sufficient cause occurring at any given

time—unless it is a genuine case of causal overdetermination (Kim 2005, 42).

(4) and (5) each claim that the choices are insufficiency of one of the

putative causes or overdetermination. This is what the causal exclusion

principle claims. I do not appeal to the causal exclusion principle explicitly

because it is inadequately formulated given the first point I made about what

the argument does not claim. At a given time, there may be two or more

sufficient causes each of which is sufficient, given causal circumstances that

include the other of the causes.

The argument involves a simplification relating to Jaegwon Kim’s dis-

tinction between supervening and micro-based properties. My appeal to

metaphysical necessitation does not distinguish between these two cases.

Nevertheless, Kim holds that the kind of argument I rehearse works against

the former and not the latter. Since the argument only seems to need to

appeal, at the crucial point, to the idea that A1(p1, p2, p3 . . . ) is sufficient for,

or fixes the probability of, bp2 by being sufficient for, or fixing the prob-

ability of, something which is sufficient for bp2, it is hard to see how to

justify the distinction between the cases (for more detailed discussion, see

Kim 1999; Noordhof 1999b, 2010).

The argument also works at a certain level of abstraction that may seem to

reduce its threat or make its application uncertain. Candidate BPs will include

those we attribute by attributing the belief that . . . where ‘ . . . ’ is filled in by

some specification of content, sensation of . . . where ‘ . . . ’ might be filled in by

‘burning feeling in the foot’, and so on. Thosewho put forward the argument,

and those who discuss it, oftenworkwith the standard picture that Ai(Pj, Pj + 1,

Pj + 2 . . . ) refers to some arrangement of narrowly physical properties in a

subject’s brain. It may well be plausible that the following is true:

If S has BP1 and BP2, and bp1 causes bp2 (where these are the particular instances

of the properties attributed to S), then there are some arrangements of narrowly

physical properties in S’s brain, say A1(p1, p2, p3 . . . ) and A2(p100, p101, p102 . . . ),

which are part of the metaphysical necessitation-bases of BP1 and BP2, and A1(p1,

p2, p3 . . . ) causes A2(p100, p101, p102 . . . ).

That is, corresponding to mental efficacy, there is related efficacy at the

narrowly physical level in the brain. However, it is no part of the argument

that this assumption is written in. All it needs is the idea that, however

MENTAL CAUSATION: ONTOLOGY AND PATTERNS OF VARIATION 91

extensive the metaphysical necessitation-base for mental properties needs to

be, putative causal relations between them imply corresponding causal

relations between these bases (where a metaphysical necessitation-base for

a property is one whose instantiation metaphysically necessitates the instan-

tiation of the property in question).

Two reactions to Kim’s argument are popular but I will argue are over-

reactions. The first says that, in fact, bp1 is efficacious because it stands in

different patterns of variation to bp2 than A1(p1, p2, p3 . . . ) as a result of

which it plays a distinctive causal explanatory role. I say that this relies on an

understanding of causation that is, at once, too strong and too weak. We

don’t have to adopt an account of causation with such counterintuitive

consequences (as we shall see) to have an answer to the exclusion argument.

We should not take distinct patterns of variation, and the inferential conse-

quences which flow from this, as fully capturing the reality of causation.

The second says that bp1 is efficacious because it is identical to A1(p1, p2,

p3 . . . ) even though BP1 is not identical to A1(P1, P2, P3 . . . )—we have a

property instance identity without an identity of properties. One version of

the latter proposal—attractive because it provides a prima facie answer to an

immediate objection—is formulated in terms of a trope metaphysics. Here

I will argue that instance identification is, in itself, questionable, inadequate

to support the whole weight of the response and leads one to a dubious

metaphysics. Most importantly, it conflates property causation with prop-

erty instance causation in its attempt to provide a defensible position.

I consider these responses in turn in the next two sections as preliminaries

to my own preferred approach, with, of course, nary a hint of over-reaction

to be found.

2. Different Patterns of Variation

The first line of response to the argument appeals to, in the limiting case,

different patterns of absence. For example, it is noted that the following

are true.

(PA1) If bp1 had not occurred, then no necessitation-base of bp1 would have

occurred.

Hence there would be no bp2.

92 PAUL NOORDHOF

(PA2) If A1(p1, p2, p3 . . . ) had not occurred, then another necessitation-base of

bp1 might have.

Hence, it is not the case that there would be no bp2 (e.g. List and Menzies

2009, 487–9; Menzies 2008, 210).

Truths such as this have been used in various contexts. Sometimes it is

said that causation is a contrastive matter. The basic form of causation is that

c rather than c’ causes e rather than e’. Different ways of describing what

might be thought to be one property instance (e.g., A1(p1, p2, p3 . . . )/bp1),

or identifications of distinct property instances, set up different comparison

classes. Contrastive accounts of causation typically take property instances

(or events) to be coarsely individuated—so that a property instance involves

the co-instantiation of multiple properties—because the case for more finely

individuated property instances (or events) can be answered if causation is

contrastive (e.g., Schaffer 2005, 347). Nevertheless, this is not mandatory.

Taking causation to be contrastive is often accompanied by the claim that

causal statements are context-sensitive. Context-sensitive statements

convey different propositions in different contexts of use. In the case of

causation, the context-sensitivity concerns what is the foil to the target

property instance or event. If there is variation in this—because, in some

contexts, the foil is absence of the target event, in others a specific alternative

event—then causal statements would be context-sensitive in this respect.

(PA1) and (PA2) take the foil to be the absence of a property instance

satisfying a particular description. Thus, describing a property instance as

bp1 determines the comparison class to be the absence of any property

instance correctly described as an instance of BP1. This will include lots of

other necessitation-bases of BP1. Whereas, describing a property instance as

A1(p1, p2, p3 . . . ) determines the comparison class to be the absence of any

property instance correctly described as an instance of A1(p1, p2, p3 . . . ) (see,

e.g., Menzies 2008, 206–8). By focusing simply on the case of absence, as

(PA1) and (PA2 do), we bracket the question of context-sensitivity. Never-

theless, the considerations offered below with regard to the case of absence

may be generalized.

Counterfactual theories of causation promise an immediate explanation

of the relevance of (PA1) and (PA2) though, as we shall see, this promise is

not kept. They are generally formulated as contrastive theories in which the

contrast is always with the absence of the target cause. Non-counterfactual

MENTAL CAUSATION: ONTOLOGY AND PATTERNS OF VARIATION 93

theories have to generate (PA1) and (PA2) either from taking their truth as a

constraint—their approach is geared to make such counterfactuals true—or

by writing in the contrastive component as an additional element, for

example, by holding that in a layered world of natural kinds ‘same level

causation is the norm’ (Gibbons 2006, 88, where the talk is of systematic

difference-making rather than explicitly of counterfactuals such as these

which express difference-making).

In any event, the claim is that the mental is shown to be efficacious, by

identifying the right difference-making as plausibly revealed in such coun-

terfactuals. The reasoning runs as follows. ‘If c were not the case, then

e would not be the case’ is a plausible sufficient condition for causation.

A1(p1, p2, p3 . . . ) fails to satisfy this condition if (PA2) is true. That is, if there

might be some other necessitation-base of bp1 so that bp2 may still be the

case. By itself, this doesn’t show that A1(p1, p2, p3 . . . ) fails to be a cause if

the counterfactual dependence of e upon c were merely a sufficient condi-

tion. So it seems that it is being taken as a necessary condition too, in the

circumstances.

The first thing to note is that we don’t allow that the possible occurrence

of replacements to a cause to discredit that cause from being efficacious, on

pain of making the world’s causal processes very gappy affairs. For example,

suppose my head of department comes to me and points out that I have done

very little administration for the department recently and other folks have

done lots of stuff. I don’t undermine what they have done by saying that,

since, if they hadn’t done it, I would have done it in their place, they cannot

be credited with having done anything. Yet, the situation seems analogous.

There were two, or doubtless more, possible undertakers of these adminis-

trative tasks. Undertaking these tasks was just realized in them rather than

me. Causes are those things which are actually involved in the process which

led to a certain target effect. Otherwise, at every point in the process at which

there might have been a replacement, we would have a causal gap. Of

course, you could decide to call the gap ‘a gap of causation’—a gap which

is filled by the occurrence of actual determination—but the decision to talk

this way has no particular utility and, as we shall shortly see, would not

alleviate the worries about mental causation in any case.

A second, and related point is that the counterfactual reasoning which is

meant to support bp1’s efficacy over A1(p1, p2, p3 . . . )’s claims is the same

reasoning that is judged inappropriate in all cases of redundant causation,

94 PAUL NOORDHOF

especially pre-emption. Recognizing the existence of pre-emptive caus-

ation precisely turns on supposing that the possible occurrence of replace-

ments does not undermine the pre-empting cause’s entitlement to be called

a cause. It is a significant cost to appeal to a pattern of reasoning which

would discredit all pre-emptive causes in order to discredit the causal claims

of A1(p1, p2, p3 . . . ), and yet this is what is required to answer the exclusion

argument without rejecting a version of the exclusion principle. Perhaps

Peter Menzies will say that it is appropriate to appeal to this pattern of

reasoning given that redundant causation has been tacitly ruled out in this

kind of case. However, it is hard to see this move as legitimate bearing in

mind that no characterization of the difference between this type of case and

redundant causation has been provided and the latter is very much an option

which is under consideration in discussions of this issue.

An unfortunate consequence of the patterns of variation approach is that it

makes the resolution of Kim’s argument turn upon brain plasticity, in our

terminology, the plasticity of arrangements of narrowly physical properties

supporting the causal relationship between mental properties. If it is the case

that no replacement arrangements of narrowly physical properties would

subserve the relationship, if the actual arrangement of physical properties

were absent, then we would be back with causal competition once more

with mental properties the potential losers. This may not, in fact, be an issue

because neuroscientists have observed that, as a result of damage, different

parts of the brain can be used to play the same function.Nevertheless, it would

be surprising if the efficacy of mental properties turned onwhether or not this

held on a case by case basis. Furthermore, since brain plasticity reduces with

age, this proposed response seems stuck with the potential consequence that

subjects’ mental properties may lose efficacy during the course of their lives.

The counterfactual reasoning with which I began this section has been

taken to express another feature of causes which, thereby, provides a

motivation for taking the previous points I’ve made to be inconclusive.

This is the idea that causes should be proportional to their effects and not

contain lots of redundant elements (Yablo 1992; Menzies 2008; List and

Menzies 2009, 488–9). This is alleged to be the difference between bp1 and

any of the A( . . . )s.

As things stand, this last claim is susceptible to a deflationary response. The

objector to the efficacy of bp1 can concede that talk of bp1 has a causal

implication that talk of A1(p1, p2, p3 . . . ) does not: bp1’s absence ensures the

MENTAL CAUSATION: ONTOLOGY AND PATTERNS OF VARIATION 95

absence of sufficient causes/chance-fixers, that is, any of the A( . . . )s; talk of a

particular A( . . . ) does not.Nevertheless, it can be argued, it is not that, by these

means, bp1 is revealed to be the cause itself. bp1 is not, by anybody’s lights, a

cause of any of its necessitation-bases, rather its absence entails the absence of

any of them. In brief, we have causal explanatory impact without causation.

The claim of proportionality is, plausibly, overstated in any case and,

thus, doesn’t get past the difficulty raised by the exclusion argument.

Considerations of proportionality entitle something to be counted a cause

in the following sense.

If c and c’ are putative competitor causes of e at the same point in the causal

network, and c is more proportional than c’ for e, then if c’ is a cause, c is a cause.

In brief, the reason for this is that more proportional causes are specified in

terms of properties which enable us to capture a generality that less propor-

tional causes miss. So if the latter is a cause, the former will be too. This will

become clearer on the development of my own approach in section 5. For

the moment I observe, first, that the attempts to discredit the efficacy of

A1(p1, p2, p3 . . . ) have not been successful and, second, there is no motiv-

ation for adopting a distinction between causing on the one hand, and causal

sufficiency for, or determination of, the probability of the target effect on

the other. Causing something may involve additional features than simply

being causally sufficient for, or a determinant of, the probability of the target

effect. Nevertheless, the latter is plausibly a necessary condition for the

former and, as a result, an exclusion argument run in terms of causal

sufficiency or determination would appear almost as damaging, if not as

damaging (for more discussion, see Noordhof 1999c, 374–5).

Thus, we are left with an apparent causal explanatory difference that we

must evaluate to see whether we have a corresponding difference in causal

reality. To conclude that difference in causal reality just falls out of the causal

explanatory difference identified is the first of the two over-reactions

I promised to identify.

3. Identity of Property Instances

An alternative fashionable approach to Kim’s argument is to argue that bp1 is

identical to A1(p1, p2, p3 . . . ). Although the properties, BP1 and A1(P1, P2,

96 PAUL NOORDHOF

P3 . . . ) are not identical, when it comes the instances—bp1 and A1(p1, p2,

p3 . . . )—they are. It cannot be denied that, if these property instances are

identical, then this particular problem is resolved. bp1 is a cause given

agreement that A1(p1, p2, p3 . . . ) is, and so on for all other instances of

broadly physical properties so identical. Evaluation of this approach doesn’t

focus on whether it works so much as whether, and indeed how, the

identification can be justified. On the positive side, there is the satisfactory

result for the problem of mental causation the exclusion argument raises.

But are there things to be said on the negative side? Identifications need to

be justified by more than the fact that they offer a convenient simplification

of our problems. They need to be independently plausible or, at least, not

implausible.

Some will take this challenge as unfair. They will remark that identity is a

primitive relation so one cannot expect any justification of it. To the extent

that we need a reason for recognizing the identity, resolution of the problem

of mental causation in this vicinity supplies us with one. However, this is

mistake. First, there is the slide from metaphysics to epistemology. Identity

may be a primitive relation but that doesn’t mean that justification for

supposing it to hold must be taken to be primitive. We may have complex

reasons to believe simple things. Second, the combination of views pro-

posed is that two properties may be distinct yet have identical instances. We

need an account of why this combination is coherent. Third, within a

metaphysical framework which makes this combination of views possible,

the considerations in favour of taking instances of mental properties to be

identical with instances of arrangements of physical properties must have

general application. We can’t have mental property instances as a special

case. That would be unmotivated.

It is no surprise, then, that sophisticated proponents of this strategy

address these issues. It is convenient to divide the approaches into those

which take properties as universals to be the fundamental element and those

which take property instances or tropes to be the fundamental element.

I shall consider these in turn.

The apparent problem for the first approach—which takes properties as

universals to be the fundamental element—is how one instance could involve

the instantiation of two distinct fundamental elements. Instance identity and

distinctness, it would seem, must follow universal identity and distinctness

(Ehring 1997, 462–3). The following makes the connection explicit.

MENTAL CAUSATION: ONTOLOGY AND PATTERNS OF VARIATION 97

An instance of F is identical to an instance of G only if F = G (where ‘F’, ‘G’ are

universals).

A sufficient condition for instance identity will draw upon additional factors

that serve to distinguish between instances, e.g. spatiotemporal location.

This problem seems overstated. The, by now, standard, subset, approach

to property co-instantiation seems available to those who take properties to

be universals. The subset view of property instance identity holds that bp1 is

identical with A1(p1, p2, p3 . . . ) if the causal powers of BP1 are a subset of the

causal powers of A1(P1, P2, P3 . . . ). A property instance with a set of causal

powers {CP1, CP2, CP3, CP4 . . . } will count as a property instance of a

property with that set, and also of a property with, say, {CP1, CP2 . . . }

alone (Whittle 2007, 68–9, who doubts that the subset view can be used in

the straightforward fashion recommended here). The apparent distinctness

of the instantiation of F and the instantiation of G is shown to be mistaken

because the causal powers of the former are a subset of the causal powers of

the latter. Co-instantiation as partial coincidence in causal powers is not

ruled out to those with universals in their ontology.

Some prefer to say that bp1 is realized by, but is not identical to, A1(p1, p2,

p3 . . . ) when the causal powers of the former are a subset of the causal

powers of the latter (Shoemaker 2007, 17; for other grounds for resisting

identity, see ibid. 48–9). They rightly point out that instances cannot be

identical if the causal powers of one stands in the subset relation to the causal

powers of the other (or, with qualifications, more precisely, causal profile

which includes the ways in which an instantiation may be caused too, as

well as its causal powers, see Shoemaker 2007, 11–12, 16–17). Their strategy

is not strictly speaking an example of the identity of instance strategy;

however, it resembles it in important respects. They take it that there is a

state of affairs A1(p1, p2, p3 . . . ) which realizes bp1, by having its causal

powers as a subset, and bp1 is efficacious when the subset of causal powers

relating to it are in play. Talk of states of affairs allows them to resist

characterizing A1(p1, p2, p3 . . . ) itself as a property instance—as opposed

to an arrangement of property instances—while retaining the subset picture.

They can, of course, allow that there is a property of being a certain kind of

state of affairs if they wish (e.g., Shoemaker 2007, 32–4). The difference

between instance-identity and realization just mentioned brings out the

slipperiness of the term ‘co-instantiation’. It can either mean identity of

98 PAUL NOORDHOF

instances or coincidence of instances in a particular instantiation. My

remarks below apply equally to both characterizations.

The basic problem is that the subset view is not true if you take the

existence of BP properties seriously. To fix ideas, consider a case of pain.

A part of me hurts and my experience of it—my pain experience—is of a

type that causes writhing and general unhappiness in humans and, we may

presume, similar writhing and unhappiness in sentient robots. The hallmark

of this activity is that the writhing is because of heat damage done to an arm.

So the writhing stems from there and is, in some way, directed towards

alleviating what is going on there.

In humans, this property is necessitated by particular kind of A@-fibre or

C-fibre firing (let us suppose), which kind depending upon the type of

hurting involved. Let’s focus on the famous C-fibre firing and take the

particular hurting to be necessitated by C-fibre firing in way W. In robots,

the pain will be necessitated by something different, let’s call that C-circuit

activity in way V. Now the question is whether the causal powers of pain

are a subset of C-fibre firings’ causal powers and of C-circuit activity’s causal

powers. The answer seems to be no. Pain experiences have the capacity to

cause pain behaviour in humans and robots whereas C-fibre firing in way

W can only cause such pain-behaviour in humans, and C-circuit activity in

way V can only cause such pain-behaviour in robots. So, the subset view

would deny that instances of pain experience of the kind specified are

identical with, or for that matter realized by, either.

One move would be to claim that pain-experiences-in-humans and pain-

experiences-in-robots, rather than simply pain experiences, are to be identi-

fied with C-fibre firings and C-circuit activations. This was Kim’s proposed

response to the existence of variable realization in defence of reductive

physicalism (i.e., type-type identity theory) (Kim 1992, 330–5). It rested

upon the claim that the nomic relationships, and hence causal powers, of

instances of mental properties or states are to be explained in terms of the

nomic relationships, and causal powers, of narrowly physical properties (Kim

1992, 322). As we shall see later, this claim is susceptible to a number of

different interpretations. Its use in this context, though, is questionable.

Denying that there are pain experiences, as opposed to species-specific

pain experiences, is, quite obviously, refusing to take them seriously.

The existence of a cross-species psychology reflecting general truths

about the causal implications of having pain experiences requires more

MENTAL CAUSATION: ONTOLOGY AND PATTERNS OF VARIATION 99

than merely species-specific pain states. Information about one type of

species-specific pain experience would, if pain experiences failed to exist,

imply nothing about the role of pain experience in other creatures. Yet,

when we reflect upon how a creature would respond if something hurt like

this—thinking about a particular kind of pain experience—we think that

there are general psychological commonalities between how they would

respond, and how we would respond, independent of variation of physical

constitution. There may be differences too, as a result of our differences in

physical constitution, but recognizing that there may be psychological

differences is quite compatible with also recognizing commonalities. This

is a phenomenon with which we are familiar for individuals too. Recogniz-

ing individual differences does not imply that all that is possible are individ-

ual psychologies and individual-relative states.

Let me state a bit more precisely how this might work in the face of

Jaegwon Kim’s scepticism (e.g., Kim 1992, 323–5). Suppose that BP1 is

metaphysically necessitated by each of the following arrangements of nar-

rowly physical properties A1(P1, P2, P3 . . . ), A1(P11, P21, P31 . . . ), A1(P12,

P22, P32 . . . ) . . . Then, BP1 can be related to radically disjunctive narrowly

physical conditions—A1(P1, P2, P3 . . . ), A1(P11, P21, P31 . . . ), A1(P12, P22,

P32 . . . ) . . .—in that the corresponding causal powers of these conditions

are {CP1, CP2, CP3, CP4, CP5, CP6 . . . }, {CP1, CP2, CP31, CP41, CP51,CP61 . . . }, {CP1, CP2, CP32, CP42, CP52, CP62 . . . }. The vast majority of

the causal powers are disparate.

According to the subset approach, instances of A1(P1, P2, P3 . . . ), A1(P11,

P21, P31 . . . ), A1(P12, P22, P32 . . . ) . . . are each instances of BP1 because a

subset of their causal powers are {CP1, CP2}, the causal powers of BP1.

Some of the other causal powers listed may be responsible for psychological

differences but there is a significant psychological commonality. The prob-

lem I’m raising for this approach with regard to the case of c-fibre and

c-circuit activity is that the arrangements of narrowly physical properties

that we find it plausible to count as instances, or realizations, of mental

properties do not seem to have the causal powers that would constitute the

commonality. That is, {CP1, CP2} are not part of the sets of powers,

contrary to how they have been represented above.

A second move in response to this case may seem to help with the

difficulty just identified: the distinction between core realization and total

realization. The core realization of a BP is that arrangement of narrowly

100 PAUL NOORDHOF

physical properties which play the causal role associated with BP’s instanti-

ation. The total realization is the arrangement of narrowly physical proper-

ties together with the context in which they occur which, taken together,

necessitate that an instance of BP is presently typically playing the role in

question. As Sydney Shoemaker puts it, to motivate the distinction, firing

C-fibres in a Petri dish is not a case of pain (Shoemaker 2007, 21; the

distinction goes back to Shoemaker 1981). For mental properties to be

instantiated, it seems plausible that not only must properties with a certain

causal profile be instantiated but, in addition, key elements of the profile

must be typically manifested.

With this distinction in place, it might be argued that, if C-fibre firing in

way W occurs in a robot, then it fails to cause writhing because the total

realization of the relevant experience of pain is not present. So an instance of

BP does not display a causal power that an instance of C-fibre firing in way

W fails to have. This response is mistaken. For BP to have a causal power

A1(P1, P2, P3 . . . ) fails to have, BP does not have to display that causal power

in circumstances in which A1(P1, P2, P3 . . . ) is present and does not display

it. BP can display the causal power when it is realized in a different way and

causes something that A1(P1, P2, P3 . . . ) would not in those circumstances.

Appeal to the idea of total realization explains how individual differences

are compatible with cross-species psychological laws. The total realizations

of mental properties ensure that certain causal relations typically hold across

differences of constitution. Compatible with this, other differences in

constitution may result in differences of psychology which disrupt these

typical causal relations. Psychological laws may have written into them

conditions under the typical causal relations don’t hold and these conditions

may imply that the regularities do not hold at all in some other species. This

is no more exceptionable than our appreciation that, for example, if we

were more secure, a rejection would have less of a significant effect than it

does in our case.

The distinction between core and total realization raises a question mark

over the efficacy of instances of BP. If the conditions under which BP is

necessitated include the circumstances, then can it be attributed causal

powers with regard to those circumstances? An answer to this will come

in the development of my own proposal in section 5. However, in brief, the

response is that BP is efficacious in virtue of the fact that a part of its

realization is efficacious.

MENTAL CAUSATION: ONTOLOGY AND PATTERNS OF VARIATION 101

A third move in support of the subset approach to mental causation is to

claim that the causal powers of an instance of pain experience are no more

than that of the property with which it is co-instantiated even though the

property of being a pain experience has causal powers which exceed those

properties with which it is co-instantiated. Two ways in which this might

be achieved are either indirectly from a claim about the individuation of

instances or directly from a claim about what is required for instantiation.

The indirect method would be to say that, suppose that a particular pain

experience is co-instantiated with an instance of C-fibres firing in way W,

then that instance of pain experience could not be co-instantiated with

something else. There is no possible world in which, say, my instance of

pain experience could be co-instantiated with C-circuit activity in way

V. So, this particular instance of pain experience cannot have causal powers

which exceed those of that with which it is co-instantiated.

Obviously, if the instance of pain experience is identical with an instance

of C-fibre firing in way W, then it is plausible that they share modal

properties. However, it would be illegitimate to appeal to instance identity

by itself to establish that the causal powers of one of the instances don’t

outstrip those of the other. That’s supposed to be the conclusion of the

subset approach. Instance identity is not meant to establish the correctness of

the subset approach. Furthermore, an actual future case, rather than appeal

to a possible case, might put this under pressure. Suppose that I have a

throbbing pain and I receive prosthetic neural fibre replacement without

anaesthetic in the hope of stopping the pain and the pain persists. It is very

plausible to say that that instance of pain experience—and not just a pain

experience of the same type—is continuing although realized by different

neural fibres.

It could be argued that the instances of prosthetic C-fibres have the same

causal powers as the instance of pain experience in the case described. So

they present no problem. By contrast, the instance of my pain experience

could not be co-instantiated with C-circuit activity because I could not be a

robot. The object in which a property is instantiated is, the claim would

run, one of the essential features of an instantiation.

The success of this response partly turns on whether it is plausible to insist

that I could not have been constituted from the same material as a robot. If

I am essentially a particular human animal, then the response receives

support. Those friendly to a psychological characterization of personal

102 PAUL NOORDHOF

identity will resist the claim that I could not be made from the same material

as a robot and even those who insist the psychological view is false, and that

we are animals, don’t have to conclude that we are animals essentially

(Olson 1997, 125, for the claim I could not become a robot; for the claim

that I am an animal is compatible with my becoming a robot, see Olson

2007, 27). If I might be a robot, then it is possible that my life could have run

a different course so that my pain experience now, in fact co-instantiated

with C-fibre firing in way W, could have been co-instantiated with

C-circuit activity firing in way V. In which case, the causal powers of the

instance of pain experience threaten to outstrip that which is co-instantiated

with it. So the instance of my pain experience is not identical with instances

of either C-fibre firing or C-circuit activity.

Nor is this the extent of the problems the response faces. Certain cases of

causation seem to involve the transfer of a property instance from one object

to another. For example, there is a difference between an object being sticky

on one side, it ceasing to be, and another object being sticky on one side,

and the stickiness of the first object wholly transferring to the second object

by contact. In the latter case, it is plausible to say that the same instance of

the property of being sticky has moved from one object to the other. In

which case, the object in which the property is instantiated cannot be

essential to the identity of the instance (Ehring 1997, 123–4). It is usually

thought that only a trope metaphysics can account for this kind of fact but

this does not seem correct.

The reasoning goes like this. Exemplifications are essentially momentary.

For suppose otherwise, then we should allow the same in the case of spatial

location. If an exemplification of squareness on one side of my office is not

distinct from an exemplification of squareness on the other side of my office,

then we have an exemplification which is wholly present in two places. We

have lost the distinction between exemplifications and universals. We

would do the same if we recognized non-momentary exemplifications

(Ehring 1997, 87–9).

This line of reasoning can be resisted. Exemplifications do not have to be

momentary. They can have duration. Thus if an object remains square over

a period of two hours, 11 a.m. to 1 p.m., there is a single exemplification of

the universal of squareness. What should we say about the object’s square-

ness at noon? Rather than recognize momentary exemplifications of square-

ness which, then, have to constitute the exemplification of squareness over

MENTAL CAUSATION: ONTOLOGY AND PATTERNS OF VARIATION 103

two hours, we should deny that there is an exemplification of squareness at 12

noon. There is an exemplification of squareness between 12 and 2 in virtue

of which it is true that the object is square at 12 noon. Corresponding to

every timely predicate, there does not have to be an exemplification.

The same applies to spatial extent. An expanse of red involves a single

exemplification of red. We do not have to suppose that exemplifications of

red at much smaller regions make up this single exemplification otherwise,

amongst other difficulties, we would need to consider the smallest extent in

which red could be realized and question whether that smallest extent could

be described as red at various points within it. If it is allowed that it can be so

described without there being an exemplification of red at the points with

the smallest extent, then we might as well accept that there is a single

exemplification of red across the extent with it still being true that portions

of that single exemplification are also described as red.

The difference between what it is plausible to say in this case, and what

I said previously concerning two spatially separated exemplifications of

squareness, derives from the fact that the latter property has defined bound-

aries. So an exemplification of squareness falling outside the boundaries

cannot constitute the same exemplification of squareness. Whereas, in the

case of an instance of redness, the extent is not fixed. So we don’t have to

concede that a lesser extent must also count as an exemplification of redness.

The claim that that lesser extent is red can be true simply in virtue of the

larger extent which is an exemplification of redness.

Since exemplifications of properties are not individuated by the objects

which possess them, there is nothing to rule out an instantiation of pain in

me being instantiated in another creature with a different constitution. Even

if they were so individuated, it is an additional step to hold that such

individuation of exemplifications requires that the means of individuation

is in terms of essential properties. We individuate objects by their spatio-

temporal position. From that, it does not follow that their actual spatio-

temporal position is essential to them.

The second version of the third move I identified in support of the subset

approach was to tinker with what is required for instantiation. So it may be

suggested that A1(p1, p2, p3 . . . ) is identical to bp1 by having truncated causal

powers of BP1. This will still not take the existence of BP1 seriously unless

you also allow that the same would hold for A1(P1, P2, P3 . . . ). Thus, an

instance of the latter may be identical to A1(p1, p3, p4 . . . ) because, whatever

104 PAUL NOORDHOF

causal powers it loses from not having p2 as part of the arrangement are not

sufficient to undermine the instance identity. You can’t resist this by

claiming that p2 must be part of the instantiation of A1(P1, P2, P3 . . . )

because instance identity is meant to be determined by the subset of causal

powers relation—or now the truncated subset relation—and not what

properties constitute the property to be instantiated. After all, if the latter

were in play, we would have grounds for denying that instances of pain

were identical with instances of arrangements of narrowly physical proper-

ties and the argument would centre around providing a justification for

thinking that narrowly physical properties were constituents of pain. How-

ever, since it is unacceptable to hold that A1(p1, p3, p4 . . . ) is an instance of

A1(P1, P2, P3 . . . )—otherwise how would be distinguish between instances

of A1(P1, P2, P3 . . . ) and instances of A1(P1, P3, P4 . . . )—this must be

because having a truncated set of causal powers is not enough.

The second problem with the truncated powers proposal is that our

evidence about bp1 in us would provide no grounds for supposing that pain

in robots would make silicon creatures writhe. The only causal powers that

my pain would reveal to me would concern how it affected me. I would not

be able reasonably to assert that that pain—in me—is so bad that, if it were

instantiated in a robot, they would be writhing about too unless they had

much greater powers of pain control to me and could focus their attention

away from it. I’d have no idea at all—by the truncated powers view—what

powers pain would have in robots. But this not correct. The reason why we

know how others would behave if they had pain—even if they had a different

constitution—is that we know the causal powers that pain would have in

them from our own case (Gibbons 2006, 95–7, also emphasizes this).

Our starting position was that there were two distinct properties under-

stood to be universals—being a pain experience of a certain type and being

c-fibre firing in way W—which, therefore, would naturally be thought to

have distinct instances. We were looking for a justification for concluding

that some of their instances are identical. The subset proposal fails to provide

it. This may not be altogether surprising. Proponents of trope metaphysics

take the situation to be different within their framework. In fact, many of

the points travel across and trope metaphysics has problems of its own with

providing what is needed.

Proponents of a trope metaphysics take property instances as fundamental

and construct physical and mental properties from these elements. This

MENTAL CAUSATION: ONTOLOGY AND PATTERNS OF VARIATION 105

gives them a motivated way of avoiding the question of how distinct

properties could be identical in instances because this will be built into the

construction. In addition, by insisting that property instances are fundamen-

tal, they have a way of resisting the claim that they have a structure which

will raise, once more, the question of whether it is the mental or physical

component of the instance which is causally relevant. Thus, a response to

Kim’s argument would then be a full response to the question of causal

relevance of properties. A further question cannot be raised about in virtue

of what features of an instance, is that instance efficacious. Unfortunately, a

substantial motivation in favour of their metaphysics tells against this solu-

tion to the problem of mental causation.

Trope metaphysicians construct properties from exact resemblance classes

of tropes. This only works if two properties don’t share the same instance. If

two properties do share the same instance, and are not coextensive, then we

cannot appeal to exact resemblance. We must appeal to rough resemblance

and, indeed, that is what those who hold that mental properties share

property instances with physical properties do.

Just in case, the issue isn’t obvious, let me explain why they need to move

to rough resemblance. Let m1, m2, m3 . . . mn be a particular class of mental

tropes (e.g., each of which is a pain experience) and p1, p2, p3 . . . pn be a class

of physical tropes. Remember that the physical class and the mental class

cannot be coextensive because, according to non-reductive physicalism,

mental properties are not identical to physical properties. Let p3 = m3. Then

if the classes of M and P were constructed from exact resemblance, then m3

would exactly resemble all the ms and exactly resemble all the ps too. But in

that case, each of the ps would exactly resemble each of the ms (because

exact resemblance is transitive) and we would just have one class after all

(Gibb 2004, 471–2). Or, put it another way, m1 and m2 must exactly

resemble each other. But if they are identical to different physical properties

(that is, if they are variable realized), then those physical properties cannot

exactly resemble each other. Again, to be able to construct a mental

property class, trope metaphysicians need to appeal to rough resemblance.

Members of a class roughly resemble each other to a certain degree which is

greater than any non-member.

The appeal to rough, rather than exact, resemblance undermines the

motivation for trope metaphysics in the first place. A principal reason for

adopting a trope metaphysics, rather than resemblance nominalism, is

106 PAUL NOORDHOF

because of the problem of imperfect community (Campbell 1990, 32–4, 72–3).

The difficulty identified for resemblance nominalism under this heading is

that if you try to construct universals from resemblance classes of objects,

then you will be committed to surrogates for universals which are not

united in a resemblance but rather may be united by different resemblances

between objects in the specified class. Moreover, it is unclear that even

some of these classes will be surrogates for universals because a class bound

together by a resemblance may be legitimately made a more extensive class

in virtue of other resemblances if all that is required is rough resemblance

(Manley 2002, 77–9, who raises the latter difficulty for worlds with limited

members but it is not clear that more populous worlds avoid the difficulty).

Trope metaphysicians claimed to avoid the problem of imperfect com-

munity by appealing to the notion of exact resemblance as opposed, simply,

to resemblance. Appeal to exact resemblance only works, though, for the

trope metaphysician, if each property instance is a property instance of only

one property (Campbell 1981, 134–5; Campbell 1990, 66, 72–3; abandoning

his Campbell 1981 position on p. 137). Once you allow that a property

instance may be a property instance of two or more properties the problem

reasserts itself.

The result is that proponents to the trope metaphysics solution to mental

causation face a trilemma depending upon whether they appeal to exact

resemblance, rough resemblance, or resemblance in a certain respect. If they

appeal to exact resemblance to construct classes of properties, then they

must either concede that there are no mental properties (in which case, non-

reductive physicalism is false) or that mental property instances are distinct

from physical property instances (in which case, they have no solution to the

problem of mental causation). That is, either they have no solution to the

problem or the doctrine for which they sought to provide a solution cannot

be formulated.

On the other hand, if they appeal to rough resemblance, or resemblance in

a certain respect, they face one of two difficulties. If the appeal is to rough

resemblance, then they face the problem of imperfect community and under-

mine the motivation for adoption of a trope metaphysics. That is not to

presume that the problem of imperfect community cannot be resolved.

Perhaps it can. Indeed, Resemblance Nominalists, or friends on their behalf,

have suggested solutions which we do not have to (and don’t have the space

to) evaluate here (e.g., Hirsch 1993, 58–9; Rodriguez-Pereyra 2002, Ch. 9).

MENTAL CAUSATION: ONTOLOGY AND PATTERNS OF VARIATION 107

The point is that unless it turns out these solutions work only for tropes—

which is unlikely—there is no particular reason for adopting a trope meta-

physics rather than resemblance nominalism.

I guess it can be argued that, if a solution were available, then the appeal

of resolving the problem of mental causation may make trope metaphysics

independently attractive. A lot would turn on whether causation by

events—say—rather than causation by property instances—preserves less

of our intuitions concerning whether the mental is efficacious. I have

argued elsewhere that proponents of the trope solution have to make

versions of the same moves that they criticize in those who claim that

mental causation just involves events and not properties (or their instances).

They have to suggest that apparently intuitive claims about efficacy reveal

something, instead, about the pragmatics of explanation (Robb 1997;

Noordhof 1998, 225–6). In section 5, I will explain how the claims have a

plausible ontological basis instead.

If trope metaphysicians appeal to resemblance in a certain respect—

perhaps even exact resemblance in a certain respect—then they needn’t

face the problem of imperfect community but the respects (be they mental

or physical) allow the problem of mental causation to be raised once more.

It can be legitimately asked, are mental property instances efficacious in

virtue of their mental respect or their physical respect? The trope metaphys-

ician cannot deny this structure because they have appealed to it to resolve

the problem of imperfect community (Gibb 2004, 473–5).

We must conclude that proponents of the trope solution fail to establish

that a successful response to Kim’s argument constitutes a successful defence

of the claim that mental properties are causally relevant. Nor does their

proposal sidestep the problem I raised with regard to the application of the

subset approach. It indicates that we need to recognize the existence of

mental tropes in addition to narrowly physical tropes (and their arrange-

ment) to capture the additional causal powers that mental properties possess.

In this section, I have examined how a certain ontological response to

Kim’s argument—identity of property instances—leads to distortion of

what we should say about the causal powers of mental property instances,

implausible theses concerning instance identity, or abandonment of the

advantages of trope metaphysics. This is the second of the two over-

reactions I identified. Before I turn to my own approach, let me briefly

discuss an approach which self-consciously does not take BP seriously.

108 PAUL NOORDHOF

4. Challenge from the Unilevellers

Unilevellers deny that the world is layered, the idea that there are different

levels of properties. One unileveller position would be to adopt a trope

metaphysics which takes properties to be fully determinate tropes in exact

resemblance classes and concludes that, for example, while there are con-

cepts of mental properties, and their instances, there are, in fact, no such

properties or instances of them. Instead, ourmental concepts capture families

of similar tropes (see, e.g., Heil 2003, 140–3, 153; Heil’s modes differ from

tropes in that the objects they characterize are essential to their identity).

If there are no mental properties, then they cannot be causally relevant.

Nevertheless, unilevellers suppose that our mental discourse picks out

something which is causally relevant. Take the case of pain experience of

a particular type. The picture is captured by a negative and positive claim.

First, the negative:

There is no single respect R (exact resemblance in some way) in virtue of which

{A1(p1, p2, p3 . . . ), A1 (p11, p12, p13 . . . ) . . . } . . . } are all members of the class of

property instances that fall under my concept of a particular type of pain

experience (hereafter, pain experience instances) (cf. Heil 2003, 153).

Second, the positive claim:

‘A particular instance of pain experience caused a particular instance of writhing’

is true in virtue of there being some member of the set of pain experience

instances, say a particular instance of C-fibre firing in way W, which caused

some member of the set of writhing instances.

Our talk of pain experience of a certain type is causally relevant because,

although there are no such properties, our concept of it has conditions of

application which pick out a class of property instances one of which was

efficacious in the circumstances.

I have two related objections to this position. The first is that it turns

inferences about how creatures behave as a result of being in pain—or being

in other mental states—into relatively shaky inferences. We are inclined to

assert that if a pain experience of a certain type were instantiated in a silicon

creature (a sentient robot), it would writhe. But exactly how it would

behave is open to question if the chain of resemblances that bundle all

pain experiences together allow for significant differences. Perhaps my pain

MENTAL CAUSATION: ONTOLOGY AND PATTERNS OF VARIATION 109

experience and a silicon creature’s pain experience resemble each other in

being distracting, but not in giving rise to certain bodily responses. Identical

causal powers are not guaranteed across subjects.

Unilevellers may dissipate this initial worry by insisting that the required

amount of similarity will include the robot version of writhing in circum-

stances in which c-circuit activity occurs. This is not guaranteed because,

once you go approximate, there is always the possibility that there will be

sufficient similarity without this element. However, let me for the sake of

argument concede it.

That will still not deal with the particular case. Suppose I am currently

having a pain experience. Then I can reasonably think, if my friend Robbie

the Robot was experiencing this, he would be writhing about in agony.

However, in thinking about this pain experience, I am thinking about

(according to the unileveller picture) C-fibre firing in way W. This

C-fibre firing would not cause Robbie to writhe about in agony. Unile-

vellers deny that there is any mental property apart from this upon which

my belief may be grounded. So they are committed to holding that I have

no grounds for the belief in question.

It might be argued that this upshot is intuitive. Humans do respond to

pain in different ways. Different creatures are likely to do so even more. But

these observations are compatible with shared causal profile (as I noted

before). The causal profile of a property will, in different contexts, manifest

itself in different ways. The unileveller position is more radical than this.

According to them, there is no shared causal profile—at best, just an

approximate similarity in causal profiles of different property instances.

This brings me tomy second objection. As we shall see in the next section,

what is required for the causal relevance of mental properties is not simply

the efficacy of an instance of a mental property (nor for that matter the

efficacy of an instance of something picked out by a mental concept) but a

condition-relative general relationship between the instantiation of mental

properties and their target effects. By denying that such properties exist,

unilevellers give up on this requirement. As such, this is a point against them.

The unileveller position derives much of its motivation from an appeal to

truthmaking. The basic idea is that if A1(p1, p2, p3 . . . ) putatively metaphys-

ically necessitates bp1, then it counts as the relevant part of a truthmaker of

sentences with terms putatively referring to bp1. There is no need for bp1 to

exist. There is, however, another dimension which A1(p1, p2, p3 . . . ) seems

110 PAUL NOORDHOF

less well suited to provide. That is, the generality that we associate with bp1expressed in the inferences we are inclined to make. The precise nature of

those will occupy us in the next section of this paper. But, to summarize the

concern in a phrase, unilevellers mistakenly emphasize truthmaking at the

expense of inference-basing.

5. Property Causation

Kim’s original challenge to non-reductive physicalismwas that itmademental

property instances inefficacious. An answer to his argument is a necessary

condition for a defence of non-reductive physicalism against the charge of

epiphenomenalism but it is not sufficient. In addition, we need an explanation

of how mental property instances are efficacious in virtue of being mental

property instances. The appeal to different patterns of variation seemed to

draw on material that might be helpful in this regard but at the expense of

neglecting the detail needed to defend the claim that mental property

instances were efficacious. The instance identity strategy was more focused

on the latter but the problems with the subset view started raising issues about

the efficacy of the mental instances qua being mental, which were revealed in

cross-subject judgements about the effects of a particular pain instance.

My proposal is an attempt to satisfy both requirements. It focuses both on

what is required for a particular instance of a mental property, or indeed any

broadly physical property to be efficacious, and also on the element of

generality that shows that the instance is efficacious in virtue of being a

mental property. It runs as follows.

F is a property cause of G if and only if

Particularity: part of the (minimal) necessitation-base for the instance of F causes

part of the (minimal) necessitation-base for the instance of G.

Generality: (part of ) each (minimal) necessitation-base of F is such that all its

instantiations would cause (or in the case of indeterminism, raise the probability

of ) an instantiation of one of the (minimal) necessitation-bases of G if they were

in some causal circumstances C—where C may vary for each kind of necessita-

tion-base.

Let me comment on various elements of this proposal.

First, the appeal to necessitation-base is meant to be understood in terms of

metaphysical necessity—just as with the characterization of non-reductive

MENTAL CAUSATION: ONTOLOGY AND PATTERNS OF VARIATION 111

physicalism. The insistence on a minimal necessitation-base is addressed to

the following difficulty. If T is the necessitation-base for F, then so is T plus

the whole world apart from T. But we wouldn’t want to conclude that F is

efficacious because of the efficacy of some feature of the world unrelated to

T (or, indeed, some feature of T unrelated to the minimal necessitation-base

for F if T is also not the minimal necessitation-base for F). I have character-

ized the minimal supervenience-base in a previous publication; talk of

minimal necessitation-base just focuses on the key element (see Noordhof

1999a, 307).

The intuitive idea is straightforward. The minimal necessitation-base for

F is all that needs to be instantiated for a particular way of instantiating F. It is

meant to capture, in some sense, the thought that broadly physical proper-

ties are constituted from arrangements of narrowly physical ones, although

these ways may vary. We know that a certain understanding of property

constitution cannot be right. Complex universals (if they exist) cannot be

composed from other universals. The classic example to illustrate this is

Lewis’ case of methane (Lewis 1986). It is not composed of four hydrogen

universals and one carbon universal because there are not four hydrogen

universals. We can’t understand property constitution simply in terms

of property instance constitution either. Instances of methane may be

composed from one instance of the property of carbon and four instances

of the property of hydrogen but even this does not work for variably

realized properties. A single universal cannot be constituted in various

ways, even if its instances can be. So variably realized universals can’t be

said to have other properties as constituents.

For properties, rather than property instances, it is better to appeal straight

to the idea of minimal metaphysical necessitation—if it can be defended

against counterexamples as I urge. When it holds, it seems to follow that

there is nothing over and above arrangements of instances of properties

needed to constitute an instance of the target broadly physical property.

If there were something over and above arrangements of instances of

these properties, then there would always be the possibility that the add-

itional element could fail to occur, even if it wouldn’t (given the physico-

psychological laws).

Of course, particular analyses of minimal necessitation-base—and the

background idea of property instance constitution—may fail. But since

the idea is natural and, more importantly, does not implicitly draw on

112 PAUL NOORDHOF

claims concerning the efficacy of the target broadly physical properties, we

could safely take it as a primitive without concern that it vitiates the

substance of the account of property causation.

Second, the appeal to ‘part of ’ is to allow for the possibility that some-

thing may count as efficacious only in virtue of an element of it being

efficacious. The fire burned because of the presence of air in virtue of the

fact that oxygen is part of air. I shall discuss this no further here but it is

relevant to the issue of the efficacy of mental properties for which extern-

alism is true (see Segal and Sober 1991; Noordhof 1999a).

So much for a preliminary understanding of the first element. Let me

now turn to the second element, that of generality: the every minimal

necessitation-base clause. Of course, I am not the first to recognize the

implicit generality. Anybody who has offered an account of causal relevance

in terms of law has also done so (e.g., Fodor 1989; Segal and Sober 1991, 15).

However, first, my proposal does not appeal to law because it is question-

able whether there is a law if the pattern I have identified holds, and, second,

those who offered such an account often failed to appeal to the idea of

minimal necessitation too. Yet an appeal to the latter is also required.

Appeals to law by themselves struggle to explain whether correlation

between broadly physical properties reveals that their instances are standing

in a causal relationship. Broadly physical properties which are nomically,

but not metaphysically, necessitated by arrangements of narrowly physical

properties will have a true general statement concerning their co-

occurrence even if the broadly physical properties are intuitively ineffi-

cacious (Segal and Sober 1991, 4–5). So something extra is needed. Either

this can be part of the conditions under which the generality would count as

a causal law, or it can be characterized independently. That these conditions

are needed is not in dispute.

Turning to the first point, my condition bears most resemblance to an

account which appeals to a ceteris paribus law relating F and G to capture

the generality involved in causal relevance. A preliminary analysis of ceteris

paribus laws is that there is a ceteris paribus law relating F and G, if and only

if, for all R, where R realizes F, there are some conditions C, such that,

whenever R & C, then G and it is nomologically possible that R without

C (Fodor 1991, 23–4; Schiffer 1991, 6–7). If the second condition were not

met, then the law would be strict. The possibility of R without C provides

conditions in which the correlation between F and G fails.

MENTAL CAUSATION: ONTOLOGY AND PATTERNS OF VARIATION 113

An objection to this analysis is that ceteris paribus laws have what Fodor

has dubbed ‘absolute exceptions’: realizations for F for which there are no

circumstances C which, together with the realization, are sufficient for an

instance of G. One way in which conditions may be unequal is if F is

realized by a dud. Fodor accommodates this by allowing that F can figure in

a ceteris paribus law if most of the time, it is not realized by duds for G, and

for other properties, say H, with which it also stands in a ceteris paribus law,

the dud realization does have circumstances in which it yields an instance of

G (Fodor 1991, 27–8). Others respond to this objection by denying the

existence of ceteris paribus laws (Schiffer 1991).

Whichever way one goes, the characterization of my generality condition

does not, then, involve an appeal to laws. However, its motivation remains

intact. If two properties are co-instantiated, then the effects of this instanti-

ation may be due to one or the other of the properties. One famous

illustration is the soprano’s singing of ‘my love’, at a certain pitch and

loudness, causing the glass to crack. It is plausible that the soprano’s singing

is an instance of that pitch, that loudness, and those words. Yet we would

not conclude that the glass cracking occurred in virtue of those words. So

how should we differentiate?

According to the generality condition, the property of involving the

words ‘my love’ does not serve to explain the pattern of causal relations

concerning glass crackings, taking into account different ways in which the

property of involving the words ‘my love’ may be realized. If the generality

condition holds for a certain property for a target effect, then we have such

an explanation. The causal relevance of a property, and not just one of its

instances, is hard to deny if, for every type of minimal necessitation-base of a

property, there are circumstances in which an instance of that property

always causes the target effect.

Consider the property of being rickety. If something is rickety, it is likely

to collapse. However, whether or not something will collapse depends

upon the precise circumstances in which it is located and the precise way

in which the property of being rickety is realized. If for every way of being

rickety, there are circumstances in which collapse follows, then we can say

that a collapse followed in virtue of being rickety. Otherwise, talk of

ricketiness, at best, figures in a ceteris paribus law. If it does not meet the

generality condition, then, while various ways of being rickety may be

causally relevant, ricketiness is not.

114 PAUL NOORDHOF

Closer to our interests to begin with, consider the case of pain. If pain is

realized by C-fibres firing in me, together with, perhaps, the laws which

govern their causal role, and these firing fibres are transplanted into our

silicon friend, Robbie the Robert, we would expect no writhing to ensue.

Now it might be urged that, in that case, what is shown to be efficacious is

the way in which the pain is realized and not pain itself. Nevertheless, there

is a fact unaddressed by this suggestion. It is that all of the realizations of pain

(e.g., in the case of Robbie the Robert C-circuit activity) have conducive

circumstances in which they give rise to writhing behaviour. This is not the

case for other properties with regard to that kind of behaviour. So it is

plausible that there is something about pain, and not just pain in such and

such a type of creature, which is responsible for the link between pain and

writhing. It is this which grounds the claim that pain is causally relevant for

this behaviour.

Suppose that there is a lone species in which instances of pain do

not cause writhing. Would that imply that human pain does not cause

writhing?1 It would not because, by limiting the question to human pain,

the generality condition would only apply to necessitation-bases of human

pain. Nor would it even imply that pain is not a property cause of writhing.

That would depend on whether the way in which pain is realized in this

lone species might also be realized in other creatures in which it did cause

writhings, or whether there were conditions in the lone species in which

writhing might be so caused.

The generality condition is also related to, but importantly distinct from,

a distinction drawn recently between sensitive and insensitive causation.

Often, the latter distinction is made within the context of taking causation

to be difference making, something I discussed in the second section of this

paper (e.g., Woodward 2006, 7). A causal relation is relatively insensitive—

between particulars, or types of things—if the counterfactual dependence

between the causal relata holds in a variety of different background condi-

tions. It is sensitive if this dependence is easily disrupted. Christian List and

Peter Menzies extend this idea to include sensitivity, or otherwise, to the

way in which the properties standing in the putative causal relation are

realized. Cases of sensitive causation in this sense are taken to be counter-

examples to the exclusion principle I mentioned earlier. List and Menzies

1 This question was asked by an anonymous referee.

MENTAL CAUSATION: ONTOLOGY AND PATTERNS OF VARIATION 115

hold that both the way in which a property is realized, and the instance of

the property, are to be counted as causes in such cases (List and Menzies

2009, 491–2, 497–9; in Menzies 2008, he seems committed to an exclusion

principle to which this kind of case is a counterexample).

To illustrate, suppose that a certain kind of pain, Pa, has four necessitation-

bases N1, N2, N3, and N4 and let Bg be the utterance ‘That hurts!’ Suppose

that, further, the following counterfactuals were true.

If Pa were not instantiated in S, then S would not utter Bg.

If N1 were not instantiated in S, then S would not utter Bg.

For the latter to be true, the closest worlds in which N1 is not present are

ones in which Bg wouldn’t occur even though there is a replacement, N2,

and Pa is, thus, present. In those circumstances, List and Menzies conclude

that both Pa and N1 are causes of Bg.

I can see why it is plausible to suppose that N1 is a property cause in that

situation. It is far less clear why it is plausible to suppose that Pa is. Givenwhat

has been previously been argued, we are allowed the question: Does N1

cause Bg partly in virtue of necessitating Pa? Evidence that it is not in virtue of

Pa is that, when a substitute realization, N2, is present, Bg does not occur. List

and Menzies suggest that the relationship between Pa and B is sensitive,

depending upon the precise way in which Pa is realized. Instead, the sensi-

tivity supplies evidence that it is N1 rather than Pa that is the causally relevant

property. If the sensitivity were just the result of a failure of the right causal

circumstances, then the case List and Menzies cite would not be a problem.

The verdicts of the two approaches would coincide. The difference stems

from the decision to count as one source of sensitivity the way in which Pa is

realized. It is here that I think their account yields counterintuitive verdicts.

Sensitivity is not compatible with causal relevance.

A consequence of my favoured account is that it delivers the verdict that

there are causal relations between broadly physical properties. Kim’s argu-

ment may be viewed as questioning this on the grounds that all the work is

being done by the arrangements of narrowly physical properties. Since my

proposal does not make instances of broadly physical properties identical to

instances of arrangements of narrowly physical properties, I don’t have an

immediate response to this worry. True my proposal may get the right

verdict in the sense of what we want to believe but the charge is that it

shouldn’t.

116 PAUL NOORDHOF

Part of my response to this objection is contained in my reply to the

challenge of the unilevellers. At this point, I emphasize another issue. The

debate in this area begins by conceding that there are broadly physical

properties but then challenges their existence by arguing that they are

inefficacious. However, the initial concession undermines the challenge.

Either you don’t think broadly physical properties exist, in which case

I draw your attention to the way in which they back inferences we want

to make about how things will behave in different circumstances. Or you

accept that they do, barring an argument to the contrary. Appeal to causal

considerations will not provide such an argument because, in allowing that

broadly physical properties exist, you must also allow that broadly physical

causal relations exist. They are just one more species of property whose

existence we have allowed as a result of their necessitation by arrangements

of, in this case, causal relations between narrowly physical properties. There

seems an entirely unmotivated asymmetry in the debate whereby causal

relations are treated differently to any other kind of property. One illustra-

tion of this last point is that, just as other properties seem to stand in relations

of determinable to determinate, so do various types of causal relations, for

example, 6 inch diameter ball depression, ball depression, depression, spe-

cify causal relations at different degrees of generality. These are determin-

able causal relations in which determinable properties may stand.

There might be other reasons to resist the claim that broadly physical

causal relations exist. My point is simply that these considerations had better

not take the same form as considerations, independent of causation, for

rejecting the existence of broadly physical properties in general. We were

supposed to be provided with a consideration from causation against the

latter, not just a blanket favouring of the narrowly physical. My account of

property causation is an attempt to identify when these broadly physical

property causal relations are present, and how they capture something in

addition to particular arrangements of narrowly physical properties, through

the generality condition.2

Another objection to the proposal discussed recently derives from the

possible truth of a powers ontology. A powers ontology takes the causal

profile of a property to be internal to it. By that I mean that the causal profile

of the property does not depend upon laws which hold, in addition, but

2 This paragraph was written in response to an objection by an anonymous reviewer.

MENTAL CAUSATION: ONTOLOGY AND PATTERNS OF VARIATION 117

rather given that the property is instantiated, certain laws hold. Suppose that

emergent dualism is true. Then one element of the causal profile of an

arrangement of narrowly physical properties is that they cause the presence

of an emergent non-physical property. If arrangements of narrowly physical

properties in such an ontology could not fail to have their causal profiles,

then it follows of metaphysical necessity that, if the arrangement of narrowly

physical properties is present, then the emergent dualist property is instanti-

ated. Nevertheless, it could still be the case that it is not part of the causal

profile of the emergent dualist property that it cause some target effect

which is part of the causal profile of the arrangement of narrowly physical

properties. Indeed, that is what epiphenomenal emergent dualists assert.

The objection to my proposal is that it suggests a certain account of how

broadly physical properties can inherit the efficacy of arrangements of

narrowly physical properties that cannot allow for this possibility (O’Con-

nor 1994, 97; Wilson 2005, 436–47).

One response to the specifics of the objection is to say that if a powers

ontology were true, there would be no basis for being an epiphenomenal

emergent dualist. The grounds for being a dualist are usually the intrinsic

features of phenomenal states. If a powers ontology were true, there would

either be no intrinsic features, or the intrinsic features in question would not

be different for narrowly physical properties. I mention this last possibility to

take into account C. B. Martin’s position that every property has both a

qualitative and dispositional aspect (e.g., Martin 1997).

Nevertheless, this does not deal with the general structure of the objec-

tion. Suppose that there is a property C1 which has a causal role CR1 which

includes, if C1 is instantiated in S, then E1 and F1 is instantiated. Then C1 and

S metaphysically necessitate E1 and they also metaphysically necessitate F1.

Doesn’t my position have as the upshot that E1’s causal role ER1 must

include the instantiation of F1 when, intuitively, it need have nothing to do

with the instantiation of F1?

Here are two more general lines of response. First, my talk of metaphys-

ical necessitation was meant to capture the important characteristic of

previous talk of constitution, namely that if that which was necessitated

by the necessitation-base involved nothing more than what was in the base,

then given the base, the necessitated must also be instantiated. It might be

argued that, if a powers ontology is true, metaphysical necessitation cannot

suffice to capture our notion of constitution even given the assumption that

118 PAUL NOORDHOF

the entities it associates are contingent. In which case, we might take

property instance constitution as a primitive and note that it supports

metaphysical necessitation claims but is not the only possible support. The

proposal would be reformulated in terms of constitution. This is not a

particularly damaging adjustment because there is no reason to think that

the proper understanding of property instance constitution must appeal to

causation or kindred notions that I am seeking to illuminate by my proposal.

Second, we can deny that a powers ontology implies dispositional essen-

tialism, the view that the causal profile of a property is essential to it. In

which case, there is no reason to accept that C1 and S metaphysically

necessitate E1 and hence no grounds for supposing that my proposal must

accept the verdict that E1 has the instantiation of F1 as part of its causal

profile. The point is especially plausible with regard to the fundamental laws

relating arrangements of narrowly physical properties and the properties put

forward by epiphenomenal emergent dualists as mental properties. They

envisage that these stand in isolation from other narrowly physical proper-

ties and so it is perfectly conceivable that the same narrowly physical

properties may be instantiated without this part of their causal profile.

However, more generally, any particular aspect of the causal profile of a

property could be plausibly supposed to be absent with the rest still present.

One way to think of these possibilities is in terms of counterpart theory.

We can suppose that, in other possible worlds, there are properties with a

strong similarity to the causal profiles instantiated in our world—structurally

speaking—and yet some differences. The question arises whether it is

plausible to consider these properties counterparts of the property in our

world. It is hard to see why not. Properties with different causal profiles may

be counterparts and yet nothing that has been said rules out the possibility

that the causal profile is internal to the property. Indeed, counterpart theory

was introduced to, amongst other things, deal with the problem of acciden-

tal intrinsic properties of particulars. The suggested strategy just applies this

to the case of properties (for further discussion, see Noordhof 2010).

A second objection to the proposal follows from something I said earlier.

I remarked that the causal powers of many broadly physical properties

exceeded those of the arrangement of narrowly physical properties that

metaphysically necessitated them. This can seem wrong on one of two

counts. First, how can it be that novel causal powers are metaphysically

necessitated by arrangements of properties which, it is alleged, individually

MENTAL CAUSATION: ONTOLOGY AND PATTERNS OF VARIATION 119

or together don’t possess them? Second, if it is allowed that they do

metaphysically necessitate novel causal powers for the sake of argument,

then why doesn’t just admitting this make the causal powers accrue to the

arrangements of narrowly physical properties which do the necessitating?

The answer to the second question is that the powers of BP1 don’t

transmit to A1(p1, p2, p3 . . . ) because, first, the instance of BP1 is not

identical to, nor caused by, the instance of A1(p1, p2, p3 . . . ) and, second,

downward transmission of causal powers does not apply because A1(p1, p2,

p3 . . . ); the latter stops some of the powers of BP1 from being manifestable,

namely those associated with other physical realizations of BP1. Of course,

part of the causal profile of a property F need not be manifested in order for

the property to have that causal profile. However, F cannot have, as part of

its causal profile, the potential for causal relations it could not stand in while

remaining the property it is, given the laws which hold. I mention this

second point in case it is thought that a version of my proposal should

explain how efficacy of broadly physical properties transmits downwards

even if the attribution of the causal powers is not immediate in virtue of the

first point.

This response to the second question makes it harder to see how one

could provide an answer to the first. How can a particular arrangement of

narrowly physical properties necessitate a property which has causal powers

more extensive than it? An incomplete answer would be that, although the

powers of BP1 exceed particular minimal necessitation-bases of it, if we

consider all the various minimal necessitation-bases, then the complete set

of causal powers that these minimal necessitation-bases have is possessed by

BP1. There are two problems with this response. The first is that its

plausibility partially rests upon the assumption that all the possible minimal

necessitation-bases of BP1, which give it distinct causal powers, are instanti-

ated in a particular world. In the absence of this, upon what basis could we

conclude that the other elements of the causal power were present? This is

not merely a notional objection. Many candidate BPs actually allow for

physical and non-physical minimal necessitation-bases with the presump-

tion that there are none of the latter if physicalism is true. Second, even if we

have some explanation of why we may allow that all the powers associated

with BP1 are instantiated, it is unclear why we should conclude that they are

instantiated with regard to a particular instance of BP1 when necessitated by

a minimal necessitation-base that cannot have at least some of the powers.

120 PAUL NOORDHOF

The proper response is to distinguish constitution as co-ordination of the small

in making up the bigger from constitution as involving grounding, in which the

constituents are viewed as fundamental. These are clearly distinct notions

otherwise we would have a fast argument from something being a constitu-

ent to monism (the priority of the whole) being false. We should reject the

idea that arrangements of narrowly physical properties constitute broadly

physical properties and, more specifically, that the causal relations of the

narrowly physical properties so arranged constitute the causal relations of the

broadly physical properties in a metaphysically fundamental sense in which

the constituents are taken to be primary. Instead, the proper relationship

between narrowly physical and broadly physical properties is one of har-

monization (see Noordhof 2003, 105–6). The right metaphor is not of an

economical God who, if only he were to fix the arrangements of the

physical, he would have the broadly physical properties fixed, but rather

of a God subject to constraints. He is not allowed to instantiate some of the

first lot without instantiating some of the second lot too. Broadly physical

properties, and their causal relations, are no less fundamental than the

arrangements of narrowly physical properties with which they are closely

related.

From this alternative perspective, the relations of metaphysical necessita-

tion between the arrangements of narrowly physical properties and broadly

physical properties capture the constraints upon instantiation, and co-in-

stantiation, between these properties. If the constraints are not observed,

then the causal relations of the properties would literally be incompatible

with each other. We would have an impossible world. Arrangements of

narrowly physical properties only appear to be ontologically fundamental

because the causal relations identified at that level are more detailed than

those identified between broadly physical properties. Since there are various

ways more general causal relations may be realized by more detailed causal

relations, we have an asymmetry. Arrangements of narrowly physical prop-

erties fix what broadly physical properties there are but the latter only imply

that one or other of various arrangements of narrowly physical properties

are present. However, interpreting this asymmetry as implying that arrange-

ments of narrowly physical properties are fundamental is not mandatory if

constitution is just co-ordination of the small.

Of course it is true that instances of narrowly physical properties may be

present, and stand in causal relations, whether or not broadly physical

MENTAL CAUSATION: ONTOLOGY AND PATTERNS OF VARIATION 121

properties are present. That might suggest that they have some priority.

However, once the alternative picture is in play, this fact needs to be set in

the context of other observations. First, since broadly physical properties

may be related to different arrangements of narrowly physical properties and,

indeed, in some cases, to arrangements of non-physical properties, there is no

reason to take broadly physical properties to be dependent on their instances’

actual constituents. Second, arrangements of narrowly physical properties

are subject to constraints on co-instantiation stemming from their constitut-

ing, in the co-ordination of the small sense, broadly physical properties. If a

broadly physical property is to be instantiated with certain causal powers, and

certain constituent instantiations of narrowly physical properties are to be

instantiated, then certain other constituent instantiations of narrowly phys-

ical properties must be instantiated too, namely those implied by the instanti-

ation of the broadly physical property with those other narrowly physical

properties as constituents.

Recognition of novel causal powers, in the way that I have sketched,

does not constitute a rejection of a weak causal closure principle like ‘every

event with a cause has a narrowly physical cause’ for, at least, two reasons.

First, that principle is compatible with there being non-physical causes too.

But, second, and more important in the present context, denying that

arrangements of narrowly physical properties constitute the causal relations

of broadly physical properties, does not mean that there are events with

broadly physical properties as causes without arrangements of narrowly

physical properties as causes. Allowing that there are cases in which the

broadly physical properties are no less fundamental than narrowly physical

ones does not imply that they have causal consequences without arrange-

ments of narrowly physical properties being present.3

6. Concluding Remarks

The causal relevance of properties, or property causation as opposed to

property instance causation, turns on two issues: first, causal facts about their

instances; second, the causal significance of a generality captured in terms of

the properties in question. Focus on difference making, or patterns of

3 This paragraph was written in response to a question by an anonymous referee.

122 PAUL NOORDHOF

variation, are better at capturing the second element but they ignore the first

element at their peril. Even with regard to the required display of generality,

they fail to observe the conditions which should be met for a property to be

said to be causally relevant (rather than just its instances).My own proposal—

involving a particularity condition plus a generality condition—has themerit

of indicating how the two elements should be integrated. It suggests that the

concern about efficacy, within the context of non-reductive physicalism,

partly stems from an inadequate understanding of how this integration

should be undertaken given that there are different minimal necessitation-

bases for broadly physical properties together with an unmotivated asym-

metry in the treatment of causation itself, as just one kind of property

amongst others. The other root of the trouble is the focus on developing a

picture of the world which places the emphasis on truthmaking rather than

inference-basing. The recognition that non-reductive physicalists should

allow broadly physical properties to have causal powers which outstrip

their bases in a circumscribed sense and, as a result, resist the constitution-

as-grounding assumption, suggests a different understanding of how we

should see the relationship between broadly and narrowly physical proper-

ties. If this proposal is along the right lines, it provides support for the view

that the apparent problem of mental causation is only properly resolved if the

metaphysical picture, in which it might figure, is made much clearer: the

theme of the AHRC funded project to which this paper was a contribution.4

References

Campbell, K. (1981). ‘The Metaphysic of Abstract Particulars’. Midwest Studies in

Philosophy, 6: 477–86.

—— (1990). Abstract Particulars. Oxford: Basil Blackwell.

Ehring, D. (1997). Causation and Persistence. Oxford: Oxford University Press.

4 Many thanks to the audience of the Metaphysics of Mind conference in Durham to which an earlier

version of this was presented, S. C. Gibb, for inviting me to participate and, more recently, to the

members of the York Mind and Reason group who subjected my ideas to scrutiny: Keith Allen, Will

McNeill, Ema Sullivan-Bissett, and Rachael Wiseman. I would also like to thank the Leverhulme Trust

for supporting a period of research, of which this was an element, by a Major Leverhulme Research

Fellowship.

MENTAL CAUSATION: ONTOLOGY AND PATTERNS OF VARIATION 123

Fodor, J. A. (1989). ‘Making Mind Matter More’. Philosophical Topics, 17(1): 59–79.

Repr. in Fodor (1990), A Theory of Content and Other Essays. Cambridge, MA:

MIT Press: 137–59.

—— (1991). ‘You Can Fool Some of the People All of the Time, Everything Else

Being Equal; Hedged Laws and Psychological Explanation’.Mind, 100(1):19–34.

Gibb, S. (2004). ‘The Problem of Mental Causation and the Nature of Properties’.

Australasian Journal of Philosophy, 83(3): 464–76.

Gibbons, J. (2006). ‘Mental Causation without Downward Causation’. Philosophical

Review, 115(1): 79–103.

Heil, J. (2003). From an Ontological Point of View. Oxford: Oxford University Press.

Hirsch, E. (1993). Dividing Reality. Oxford: Oxford University Press.

Kim, J. (1992). ‘Multiple Realization and the Metaphysics of Reduction’. Philoso-

phy and Phenomenological Research, 52: 1–26. Repr. in Kim (1993), Supervenience

and Mind. Cambridge: Cambridge University Press: 309–35 [page references in

text to latter].

—— (1998). Mind in a Physical World. Cambridge, MA: MIT Press.

—— (1999). ‘Supervenience Properties and Micro-Based Properties: A Reply to

Noordhof ’. Proceedings of the Aristotelian Society, 99: 115–18.

—— (2005). Physicalism, or Something Near Enough. Princeton: Princeton Univer-

sity Press.

Lewis, D. (1986). ‘Against Structural Universals’. Australasian Journal of Philosophy,

64: 25–46.

List, C. and P. Menzies (2009). ‘Nonreductive Physicalism and the Limits of the

Exclusion Principle’. Journal of Philosophy, 106(9): 475–502.

Manley, D. (2002). ‘Properties and Resemblance Classes’. Nous, 36(1): 75–96.

Martin, C. B. (1997). ‘On the Need for Properties: The Road to Pythagoreanism

and Back’. Synthese, 112(2): 193–231.

Menzies, P. (2008). ‘The Exclusion Problem, the Determination Relation, and

Contrastive Causation’. In J. Hohwy and J. Kallestrup (eds.), Being Reduced: New

Essays on Reductive Explanation and Special Science Causation. Oxford: Oxford

University Press: 196–217.

Noordhof, P. (1998). ‘Do Tropes Resolve the Problem of Mental Causation’.

Philosophical Quarterly, 48 (191): 221–6.

—— (1999a). ‘Causation by Content?’. Mind and Language, 14(3): 291–320.

—— (1999b). ‘Micro-Based Properties and the Supervenience Argument:

A Response to Kim’. Proceedings of the Aristotelian Society, 99: 109–14.

—— (1999c). ‘The Overdetermination Argument versus the Cause-and-Essence

Principle—No Contest’. Mind, 108(430): 367–75.

124 PAUL NOORDHOF

—— (2003). ‘Not old . . . but not that new either: Explicability, Emergence and the

Characterisation of Materialism’. In S. Walter and H.-D. Heckman (eds.),

Physicalism and Mental Causation: The Metaphysics of Mind and Action. Charlottes-

ville: Imprint Academic: 85–108.

—— (2010). ‘Emergent Causation and Property Causation’. In C. Macdonald and

G.Macdonald (eds.),Emergence inMind. Oxford: OxfordUniversity Press: 69–99.

O’Connor, T. (1994). ‘Emergent Properties’. American Philosophical Quarterly, 31(2):

91–104.

Olson, E. (1997). The Human Animal. Oxford: Oxford University Press.

—— (2007). What Are We? Oxford: Oxford University Press.

Robb, D. (1997). ‘The Properties of Mental Causation’. Philosophical Quarterly, 47:

178–94.

Rodriguez-Pereyra, G. (2002). Resemblance Nominalism. Oxford: Oxford Univer-

sity Press.

Schaffer, J. (2005). ‘Contrastive Causation’. Philosophical Review, 114(3): 327–58.

Schiffer, S. (1991). ‘Ceterus Paribus Laws’. Mind, 100(1): 1–17.

Segal, G. and E. Sober (1991). ‘The Causal Efficacy of Content’. Philosophical

Studies, 63: 1–30.

Shoemaker, S. (1981). ‘Some Varieties of Functionalism’. Philosophical Topics, 12(1):

83–118. Repr. in Shoemaker (1984, 2003), Identity, Cause and Mind. Oxford:

Oxford University Press: 261–86.

—— (2007). Physical Realization. Oxford: Oxford University Press.

Steward, H. (1997). ‘On the Notion of a Cause “Philosophically Speaking”’.

Proceedings of the Aristotelian Society, 97: 125–40.

Whittle, A. (2007). ‘The Co-Instantiation Thesis’. Australasian Journal of Philosophy,

85(1): 61–79.

Wilson, J. (2005). ‘Supervenience-based Formulations of Physicalism’. Nous, 39(3):

426–59.

Woodward, J. (2006). ‘Sensitive and Insensitive Causation’. Philosophical Review,

115(1): 1–50.

Yablo, S. (1992). ‘Mental Causation’. Philosophical Review, 101(2): 245–80.

MENTAL CAUSATION: ONTOLOGY AND PATTERNS OF VARIATION 125

5

Causation is Macroscopicbut Not Irreducible

DAVID PAPINEAU

1. Introduction

In this paper I argue that causation is an essentially macroscopic phenom-

enon, and that mental causes are therefore capable of out-competing their

more specific physical realizers as causes of physical effects. But I also argue

that any causes must be type-identical with physical properties, on pain of

positing inexplicable physical conspiracies. I therefore allow macroscopic

mental causation, but only when it is physically reducible.

2. Causation is Macroscopic and Not Physically

Fundamental

It is widely supposed, at least among philosophers, that causation is grounded

in basic dynamical processes. At bottom, according to this supposition, causal

relations consist in the way that basic dynamical laws govern the temporal

evolution of precise arrangements of fundamental physical particles and fields.

This supposition is often on display in contemporary debates about mental

and other ‘higher-level’ causation. Suppose you think that some mental state

M—your wanting to hail a taxi, say—supervenes on physical state P without

being reducible to it. (P is more specific than M: it metaphysically necessitates

M, but M can be realized by physical states other than P.) Now consider some

further physical effect P*—your arm moving—that apparently results from

M. Can M really be the cause of P*? Some philosophers deny that it can, on

the grounds thatMwill always be ‘out-competed’ as a cause by P.Others assert

that M can be such a cause, arguing that there is nothing wrong with both

M and P ‘overdetermining’ the effect P* in such cases. But scarcely anyone

queries whether the realizing P will itself qualify as a cause of P*.1

Thus Jaegwon Kim, in discussing just this kind of case, insists that ‘The

question is not whether P should be considered a cause of P*; on anyone’s

account, it should be’ (Kim 1993, 207, my italics). (Kim then continues with

his familiar query: ‘What causal work is left over for M, or any other mental

property, to do?’) However, I myself am very doubtful about the claim that

Kim takes to be agreed on all sides.

Why does Kim take it to be obvious that P is a cause of P*? Presumably

he is reasoning from the above supposition that causation is constituted by

the way maximally precise physical arrangements evolve in accord with

basic dynamical laws. If P is a full specification of physical initial conditions,

and these evolve in line with basic dynamic laws into P*, then of course—or

so Kim assumes—P must cause P*.

However, there is strong reason to doubt that causation is constituted by

basic dynamical processes. The objection is simple. Causation is asymmetric

in time, but basic dynamics is not. So it seems that causation must involve

something more than basic dynamics.

The basic laws of dynamics determine no direction in time. Take a

specification of what happens at each point of spacetime in some closed

physical system. Then you can view the ‘initial’ conditions as evolving into

the ‘final’ conditions in accord with the basic dynamical laws. But you could

equally well think of time as ‘flowing’ in the opposite direction, with the

‘final’ conditions evolving into the ‘initial’ ones, again in accord with the

basic dynamical laws. In this sense, the basic laws of dynamics don’t care

which direction is ‘earlier’ and which ‘later’. Accordingly, if you are given

a basic dynamical description of a physical system, but not told which

temporal direction is which, you won’t be able to read this off from the

description.

But causation is different. If you can discern the causes and effects within

a physical process, then this alone will tell you which way time is directed.

1 I take causes and effects to be facts or states of affairs, paradigmatically consisting of some particular

possessing some property. Given this, it will sometimes be natural to talk as if properties themselves are

causes or effects; but strictly what is meant by such usages are the facts involving those properties.

CAUSATION IS MACROSCOPIC BUT NOT IRREDUCIBLE 127

Causes always come earlier than their effects, and so a specification of causal

structure will tell you which temporal direction is which.

Of course, this would be trivial if the difference between ‘causes’ and

‘effects’ simply consisted in the former occurring earlier than the latter. Thus

suppose that the relationship between ‘causes’ and ‘effects’ were just like

that between the sets of conditions at the two temporal ends of a basic

dynamic process, save that the direction of time was independently given,

and it was specified that the ‘cause’ was the earlier set of conditions, and the

‘effect’ the later set. Then it would scarcely be surprising that we could read

off temporal order from information about ‘causes’ and ‘effects’—for such

talk would simply add this temporal information explicitly to the temporally

neutral dynamic facts.

But it is arguable that the difference between causes and effects lies deeper

than this, and can be discerned prior to any independently given infor-

mation about temporal order. Causal relations have a characteristic prob-

abilistic signature which is asymmetric in time. It is this signature that the

recent tradition of ‘Bayesian nets’ exploits in order to infer causal structure

from probabilistic correlations. It is noteworthy that the techniques

exploited by the Bayesian net tradition do not need to assume temporal

order in order to distinguish causes from effects. Sufficiently rich correla-

tional information on its own will always determine a causal order among

related variables. (This is not the place to go into details. But, to get a

flavour, note that the correlation between the joint effects of a common

cause will disappear when we ‘control’ for the common cause—that is,

consider separately cases where the cause is absent and where it is present.

By contrast, any correlations between the joint causes of a common effectwill

not disappear when we ‘control’ for that effect. For more on the asymmetric

probabilistic dimension of causation see Hausman 1998; Spirtes et al. 2000;

Papineau 2001.)

There is room here to debate the precise metaphysical relationship

between the underlying causal structure and the temporally asymmetric

correlational structure which manifests it. But the very possibility of epi-

stemologically distinguishing causes from effects without assuming temporal

order suggests that there must be something in the nature of causation that

orientates it in time. If so, this means that there must be more to causation

than the temporally symmetric structures of basic dynamics.

128 DAVID PAPINEAU

3. Thermodynamics and Causation

It is illuminating to compare causation with thermodynamics in the above

respects. As is well known, the second law of thermodynamics is also inexplic-

able in terms of basic dynamics alone, precisely because it refers to a specific

direction in time: later entropy is always greater than earlier entropy within a

closed physical system. So an explanation of the second law needs to invoke

assumptions that go beyond basic dynamics. In particular, such an explanation

needs to posit, in addition to basic dynamics, first, that entropy was low in the

past, and, second, that there is a certain probability distribution over all the

precise physical microstates consistent with given ‘macrostates’ of tempera-

ture, energy, entropy, and so on (see Albert 2000).

I take causal asymmetry to have an analogous basis. There is no estab-

lished way of relating causation to thermodynamics. But the asymmetric

correlational structures displayed by causal relationships suggest that causal

processes are akin to thermodynamic processes. In particular, it suggests that

causation is also is an essentially macroscopic phenomenon, constituted by

the nature of past facts together with probability distributions over the

maximally specific microstates that can realize given macrostates.

Some readers might be puzzled by the suggestion that thermodynamic

processes, and causal relationships alongwith them, are essentiallymacroscopic

phenomena. I alluded above to the way that thermodynamic processes like

entropy increase can be explained in terms of particle physics, together with

past facts and probability distributions over microstates. But if such explan-

ations are possible, then don’t they show that themacroscopic thermodynamic

phenomena can all be reduced to microscopic processes, and so aren’t really

macroscopic after all?

But it does not work like that. The explanation of thermodynamic

phenomena by particle physics does not eliminate macroscopic features,

but makes essential use of them. Take a volume of gas that is hot in one half

and cold in another. Thermodynamics tells us that in a while the tempera-

ture will almost certainly be uniform throughout. Now, you could in

principle have analysed this particular system by applying basic dynamics

to the precise initial conditions of all the particles involved, and this would

no doubt have told you that the later temperature would be uniform. But

this microanalysis would owe nothing to the general principle that almost

CAUSATION IS MACROSCOPIC BUT NOT IRREDUCIBLE 129

any system in that initial macrostate would end up at a uniform temperature.

(After all, you could have applied an entirely analogous microanalysis to

predict the evolution of one of the very unlikely ‘rogue’ microstates that

would not end up with a uniform temperature.) To bring out the general

principle, you need to ‘throw away’ the information about the precise

microstate, and note instead that the system is in a macrostate which is

overwhelmingly likely to (be realized by a microstate that will) end up

with a uniform temperature. This is why thermodynamics is essentially

macroscopic. Without probabilistic information about the way in which

given macrostates get realized by microstates, you cannot infer any thermo-

dynamic patterns from microphysics.

Similarly, I suggest, causation is an essentially macroscopic phenomenon.

If you focus on the precise microstate of some physical process, you will lose

sight of causation. The causal structure of the world depends on probabil-

istic facts about the ways in which given macrostates are realized at the

micro-level, rather than on the actual micro-realizations themselves.

4. Intuitions are Irrelevant

Of course, this is not howwe think about causation intuitively. The intuitive

paradigm of a causal interaction is of one physical object bumping into

another and the latter’s motion changing. We humans are naturally prone

to judge without further ado that in such cases the impact of the former

caused the new motion of the latter (see Michotte 1946/1963). In line with

this, our intuitive conception of causation contains no mention of probabil-

istic distributions over the microstates that realize different macrostates, and

correspondingly our concept of causation sees no contradiction in the idea of

causation existing even when such probabilistic distributions do not.

In this connection, consider Ernest Sosa and Michael Tooley’s objection

to the way David Lewis analyses causation in terms of the ‘asymmetry of

overdetermination’. Lewis’s analysis is in the spirit of theories that account

for causal asymmetry in terms of probabilistic asymmetry. True, Lewis does

not put this analysis explicitly in probabilistic terms, but the appeal to

probabilistic facts is not far beneath the surface. (Thus note how Lewis

appeals to the way that causes typically issue in many independent chains of

effects, yet typically only stem from one chain of causes.)

130 DAVID PAPINEAU

Sosa and Tooley object to Lewis’s theory on the grounds that

it is not a necessary truth that any world containing causally related events is one

where events typically have more effects than causes. The world . . . could have

been a very simple one, where there were no causal forks . . . Lewis’s analysis cannot

be sound, therefore, since there are logically possible causal worlds for which it

yields the wrong results . . . (Sosa and Tooley 1993, 27)

Well, no doubt there are conceivable scenarios which contain causal relations

but lack the asymmetric probabilistic structure to which Lewis appeals in

explaining causation. And if Lewis’s theory were put forward as a piece of

conceptual analysis, then such scenarios would suffice to refute it.

But there is no reason to read theories like Lewis’s in this conceptual way.

Rather, they are better understood as synthetic metaphysical theories,

which aim to uncover the nature of causation, not via analysis of our

concepts, but through a posteriori investigation of the world we live in.

You can’t argue against theories of this kind by appealing to merely conceiv-

able scenarios, any more than you can argue against orthodox chemistry by

appealing to the conceivability of a world with water but no H2O. Of

course, if it could be established that worlds with causation but no probabil-

istic structure were metaphysically possible, then this would indeed defeat

probabilistic accounts of the nature of causation. But the mere conceiv-

ability of such worlds does not show that they are metaphysically possible. If

causation is indeed constituted by temporally asymmetric probabilistic

structure, then there is no metaphysical possibility of the one without the

other, however much this may be conceivable.

5. Autonomous Mental Causes

So far I have argued that it is a mistake to think of causal relationships as

being determined by some maximally specific level of physical facts. Rather

causation depends on general patterns essentially involving macroscopic

properties, where these macroscopic properties will be realized by different

arrangements of fundamental physical facts on different occasions.

Over the last couple of decades a number of philosophers have argued

that mental facts M are no less causes of subsequent physical results P* than

their physical realizers P. On this view, the result P* can be attributed to

CAUSATION IS MACROSCOPIC BUT NOT IRREDUCIBLE 131

both of the ‘parallel causes’ M and P. Of course, nobody wants to view all

mental causation as overdetermination by two ontologically distinct causes,

like the death of the man who is shot and struck by lightning at the same

time. But defenders of the ‘parallel causes view’ can observe that M and

P are not so ontologically distinct, in that M metaphysically supervenes on

P. True, M is not identical to P, and so in a sense a kind of ‘benign

overdetermination’ is being posited. But precisely because M supervenes

on P, and is not ontologically independent, it is not obvious that there is

anything wrong with such benign overdetermination. (See Shoemaker

2001; Pereboom 2002; Bennett 2003.)

This ‘parallel causes view’ has the virtue of recognizing macroscopic

mental facts as causes in their own right. But the points made in the last

section open the way to a more radical position. Why shouldn’t the mental

state M be the cause of P* rather than the physical state P? If causation derives

from patterns essentially involving macrostates, then perhaps it is the mental

M that figures in these patterns, not the physical P.

Just this possibility has been explored by some recent writers. (See

Menzies 2008; List and Menzies 2009; Menzies and List 2010. Also relevant

are LePore and Loewer 1987; Yablo 1992.) Their standard form of argument

appeals to plausible counterfactual requirements on causation. Let us sup-

pose that, if C causes E, then

(1) E wouldn’t have occurred if C hadn’t occurred, and

(2) E would still have occurred if C had occurred differently.

Take the case where you are waving for a taxi. Let the physical effect P* be the

movement of your arm. M is your wanting to hail a taxi. P is the definite

neuronal arrangement which realizes this mental state. Now, both M and

P satisfy clause (2)—your arm would still have moved as long as either M or

P occurred, even if they had occurred in a different way. But only M satisfies

clause (1)—your arm wouldn’t have moved if you hadn’t wanted a taxi—

where P does not—the absence of just that precise neuronal arrangement

wouldn’t have stopped your arm moving, for you would still have wanted to

wave even if your desire had been realized by a slightly different neuronal state.2

2 Laurie Paul has queried (in correspondence) whether M would have been different realized, rather

than simply being absent, if P had not occurred. My reading does sound natural to my ear, but in any case

let me simply stipulate that the suggested counterfactual requirement on causation be understood this way.

132 DAVID PAPINEAU

So here M is the cause and not P. P is too specific. This analysis is in line

with Stephen Yablo’s thesis (1992) that causes should be proportional to their

effects. Causes must be specific enough for their effects, but no more

specific than this requires. In the above example, the neuronal arrangement

P is too specific for the effect, but the wanting M is just right.

Note how the example comes out differently if the effect P* is not your

arm moving as such, but your arm moving in the precise way that it did on

this occasion. Now both M and P satisfy clause (1)—if you hadn’t wanted to,

you wouldn’t have waved at all, and so a fortiori not just as you did; and if

you hadn’t had just that precise neuronal set-up, you also wouldn’t have

waved just as you did. But, with this precise effect P*, only P satisfies clause

(2)—while you would still have moved just like that if P had occurred a bit

differently, you wouldn’t have moved just like that if you’d still wanted to

move your arm but this desire had been realized with some different

neuronal arrangement.

So now P is the cause and not M. The wanting M is not specific enough

to account for your moving just like that, but the neuronal arrangement P is

just right. Does this last example not run against the points made in my first

section? There I said that causation is an essentially macroscopic phenom-

enon, and disappears at the level of maximally specific physical processes.

Now I am saying that your neuronal arrangement P can be the cause of your

particular movements, rather than your mental state M. However, these

claims are not inconsistent. The reason is that the neuronal arrangement

P needn’t constitute a maximally specific physical state. Just as a given mental

state M can be realized by different neuronal arrangements, so too can a

given neuronal arrangement be different realized at the maximally specific

level of precise fundamental particles and fields. So in both cases there is

room for the kinds of probabilistic facts which I say are essential to asym-

metrical causal relationships. It is only at the level of fully specific physical

arrangements that causation disappears.

Let us return to the analogy between causation and thermodynamics.

Suppose we have a volume of gas in a container with a safety valve. If the gas

is heated, there is a temperature T at which the valve will open. This

temperature can be realized by the many different sets of particle move-

ments which would yield the requisite mean kinetic energy. Which causes

the valve’s opening on some given occasion, the temperature T or the

specific particle movements which there realize that temperature? Intuition

CAUSATION IS MACROSCOPIC BUT NOT IRREDUCIBLE 133

might suggest that it is the particle movements. But if the effect at issue is the

opening as such, the approach I am defending argues that it is the tempera-

ture and not the particle movements that cause this effect. The particle

movements are too specific. We would still have had the opening even if

the temperature T had been realized by different particle movements.

Does thismean that the specific particlemovements cannot cause anything?

That would be an undesirable conclusion. There are more fine-grained

effects, such as the precise trajectory of the valve’s opening, which will surely

be the results of the specific particle movements. Perhaps the valve’s opening

in that specific manner is due to precise sequence of high-energy particles that

impact its inner surface. With the more fine-grained effect, it is the more

specific particle movements and not the generic temperature T which is

proportional to the cause: if the temperature T had been differently realized,

then the valve would not have opened in just that manner.

Note how the causal efficacy of the particle movements does not under-

mine my claim that causation is essentially a macroscopic phenomenon. Even

after we have focused on the definite particle movements, there will be yet

further features of the set-up—such as the bonding properties of the gas’s

molecules, the molecular structure of the valve’s inner surface, and so on—

that will still be variably realized at the level of fully specific physical arrange-

ments. And it is still probabilistic facts about the distribution of such further

realizers that underpin the asymmetric causal relationship between the par-

ticle movements and the manner of opening. If we descend to a level where

all physical facts are fully determinate, then I say that we lose sight of any

asymmetric causal relationships. But this leaves plenty of room for relatively

definite physical facts like given particle movements to function as genuine

causes of relatively fine-grained effects. We can descend to particle move-

ments without descending to the level of fully specific physical states.

I alluded earlier to the idea that a mental cause M and a physical realizer

P might both be the cause of some physical effect P*, via a sort of benign

overdetermination. The points made so far in this section argue that this is

not a possibility—not on the grounds that such overdetermination per se

would generate any unacceptable consequences, but simply because the

requirements of proportionality rule out two such causes. Once we have

fixed on a specific effect P*, then it can’t be that some M and some more

specific realizer P are both causes of P*. If M is proportionally ‘just right’, then

the more specific P will violate requirement (1), in that we would still have

134 DAVID PAPINEAU

had P* without P. And if P is proportionally ‘just right’, then the less specific

Mwill violate requirement (2), in that we wouldn’t still have had P* if M had

been realized differently. It can’t be the case that a more and less specific

state are both causally proportional to a given effect. (Cf. LePore and

Loewer 1987; Menzies and List 2010.)

Suppose that some mental state M out-competes its more specific phys-

ical realizer P as the cause of some physical effect P*. What then is the

relation of the realizer P to the effect P*? Many of those who defend the

causal status of M nevertheless retain the idea that the physical realizer P is

‘causally sufficient’ for the physical result P*. Their thought is that P still

causally determines P* even though it is too specific to count as ‘the cause’

of that result. But from the perspective being defended here, even this seems

to concede too much to the intuition that causation is grounded in basic

physical processes. Of course we might wish to allow that such specific

physical antecedents are nomologically sufficient for the subsequent physical

results. But there is no reason to think of this sufficiency as a causal matter,

in cases where the precise physical detail omits any mention of the macro-

scopic pattern that constitutes the causal relationship.

6. Proportionality and Reduction

Does the fact mental states can eclipse their physical realizers as causes of

certain effects vindicate the possibility of non-reductive physicalism in the

philosophy of mind? This conclusion is typically drawn by those philoso-

phers who stress that proportionality requirements can favourmental states as

causes over their physical realizers. But it is by no means clear that it follows.

Reductive physicalism requires type identity of mental properties with

physical properties. Non-reductive physicalists maintain that no such type

identities are available. It is important to realize that, in order to establish

non-reductive physicalism, it is not enough to show that there are some

physical differences present on the different occasions where M is realized.

Rather we need to show that there is no distinctive physical commonality

present on all those occasions.

When Putnam and Fodor introduced the idea that mental and other

special science properties might fail to reduce to physical properties, they

weren’t just making the weak claim that different instances of these properties

CAUSATION IS MACROSCOPIC BUT NOT IRREDUCIBLE 135

will display some physical differences. Rather their idea was that there would

be no common physical feature of different instances. There would be

nothing physically in common between the different computers that can

run a given program, or between the different organisms across the universe

that can think a given thought.

So far in this paper we have been dealing with cases where some mental

M is realized by different more specific physical Ps on different occasions.

That is, we have been dealing with cases where distinguishable physical Ps

can metaphysically determine the same mental M. This by itself fails to

establish the anti-reductive thesis that there is no further physical feature

Q which is type identical to M.

To see that this stronger anti-reductive thesis does not follow from the

fact that M is determined on different occasions by distinguishable Ps, we

need only consider a thermodynamic example once more. Take the case

where the gas reaching temperature T opens the valve. Temperature T can

be realized by many distinct arrangements of specific particle movements.

But it does not follow that there is no further physical property which

characterizes all the instances of T. And of course in this case there is. All the

instances of T involve arrangement of particles with the same mean kinetic

energy. And it is precisely this common physical feature which allows the

possibility of a uniform thermodynamic explanation of why the valve will

open at that temperature. The probability distribution over the possible

microstates that realize that mean kinetic energy implies that the valve is

overwhelmingly likely to open at that temperature.

This is surely the paradigm of a type–type reduction. We identity tem-

perature with some common physical feature specifiable in terms of particle

movements, namely a given mean kinetic energy, and thereby explain

patterns involving temperature in terms of particle physics.

But as well as being the paradigm of a type–type reduction, this is also a

case where proportionality considerations point to the macroscopic tem-

perature as the cause of the valve opening, rather than the more specific

particle movements which realize it on given occasions. I infer that there is

nothing in the idea of macroscopic facts being proportionate causes to rule

out fully reductive physicalism.

Recall Putnam’s famous example of the square peg and the round hole

(1975). Putnam argued that the properties of squareness and roundness will

be a much better explanation for why the peg does not fit in the hole than

136 DAVID PAPINEAU

any detailed specification of the quantum mechanical arrangement and

properties of relevant bodies’ molecules. Quite so. It is the squareness and

roundness that are proportional to the peg’s failure to fit, not the very

specific molecular arrangements that realize these properties.

But this does not mean that squareness and roundness are not physically

reducible. We can still specify features of their molecular arrangements that

will be common to all square pegs and round holes, and we can appeal to the

so-specified features to explain at that level why square pegs don’t go into

round holes. Here again we see that it does not follow from the causal

dominance of macroscopic facts over their more precise realizers that those

macroscopic facts must be physically irreducible.

These examples manifest a typical set-up in physics. Some macroscopic

property common to many microscopically distinguishable states can be

identified with some common feature of those microscopic realizers, and

this common feature then accounts for the way that the macroscopic

property features essentially in some general pattern.

7. Against Unreduced Causes

The last section showed that macroscopic causes are one thing,

non-reducibility another. There are plenty of cases where macroscopic

properties can feature as proportionate causes of certain physical effects,

and thereby causally eclipse their more specific microphysical realizations,

and yet these macroscopic properties are fully reducible to some common

physical feature of their microscopic realizations.

I now want to argue that macroscopic causation is not just consistent with

physical reducibility, but that it positively requires this.

I have argued in this paper that asymmetric causal relations derive from

probabilistic facts about the way in which macrostates are realized at the

micro-level. This picture assumes that each macro-cause corresponds to

some constraint specifiable at the micro-level, in the way that temperature

corresponds to mean kinetic energy. The probabilistic facts about the

different ways this condition can be satisfied by precise microstates then

accounts for the asymmetric causal patterns involving macrostates.

If this is the right general story about causation, then it is hard to see how

macroscopic causes can fail to be physically reducible. Their very nature as

CAUSATION IS MACROSCOPIC BUT NOT IRREDUCIBLE 137

causes will derive from their type-identity to some physically specifiable

constraint, for it will only be in virtue of this identity that they systematically

generate their effects.

What options are open to non-reductive physicalists here? There seem to

be two ways they might go. First, they might argue that, when a given

macroscopic cause is variably realized, it generates its effects via different

causal processes at the physical level. Alternatively, they might argue that

there is no need to invoke any casual processes at the physical level to

explain how a variably realized macro-cause generates its effect. However,

neither of these options seems at all attractive.

To bring out the difficulties here, note that proportionate causation

involves an element of generality. Recall that our two requirements for

C to cause E were that

(1) E wouldn’t have occurred if C hadn’t occurred, and

(2) E would still have occurred if C had occurred differently.

Clause (2) here tells us that in other similar circumstances where C occurs,

E will occur too. On other similar occasions where I want to hail a taxi, my

arm still moves. And clause (1) tells us that when C doesn’t occur in similar

circumstances, E will fail to occur too. On other similar occasions where

I don’t want to hail a taxi, my arm doesn’t move. In short, there is a general

co-variation of C and E in similar circumstances.

Now, the problem facing non-reductive physicalists is to explain why

C and E should so co-vary if there is no uniform physical condition

corresponding to C which can account for this. The answer is obvious if

C can be identified with some physical condition which systematically

generates the result E. But in the absence of any such identification, non-

reductive physicalists seem to face a challenge.

The first non-reductive response to this challenge would be to hold that

the different realizations of C give rise to E via different causal processes. This

is probably how non-reductive physicalism is normally understood. On

different occasions when people want to hail a taxi, their desire is realized

by different physiological arrangements—but each of these different physio-

logical arrangements has the causal power to produce an arm movement.

The trouble facing this option is that we have been given no account of

why the different causal processes that realize C should all alike be ones that

give rise to E. For all that has been said so far, this looks like a mystery. If the

138 DAVID PAPINEAU

processes at the physiological level are all so different on different occasions

of desiring to hail a taxi, why ever should they all be followed by E?

To focus this issue, it will be helpful to consider an inorganic example, as

there are features of mental phenomena that can obscure the difficulty at

hand, in various ways to be considered below. Let us imagine that the water

from a certain lake seems to have a distinctive power to destroy rubber. But

when we look into the mechanism, we find no common causal process. In

one case, the water contains rubber-eating bacteria. In another, the water

turns out to be highly acidic. In yet another, there are high levels of ozone in

the water and this produces a rubber-destructive agent. And so on. In each

of the cases that we examine, we find a physical explanation for the rubber’s

deterioration, but the explanation is different in each case.

I take it that this story does not hang together. If we really came across a

case like this, and discovered a different mechanism in each case, we would

surely conclude that it wasn’t a genuine causal relationship after all, and that

the observed pattern was just a coincidental feature of the cases so far

observed. I think that we should have the same reaction to the suggestion

that some mental C can produce a physical effect E via different causal

mechanisms on different occasions. In the absence of any further infor-

mation, it seems incredible that nature should work like this.

Let me now consider the alternative non-reductive response to the

challenge of explaining why some C and E should co-vary if there is no

uniform physical condition corresponding to C which can account for this.

The alternative non-reductive option would be to deny that we need any

causal accounts at the physical level for macroscopic causal processes. Now

the idea is not that different causal processes account for the C–E link in

different cases, but that there are no further causal stories to be told at the

physical level at all.

But this too looks like mystery-mongering. Remember that we are

exploring the possibility of non-reductive physicalism. It is not as if we are

positing some ontologically independent realm of mental causes with brute

powers to produce physical effects. Rather, we are taking it that mental and

other macroscopic causes metaphysically supervene on the physical facts—

nothing more is required for their presence than those physical facts. But

then causation without a physical explanation looks like a conspiracy. On

different occasions C is realized by different physical microstates, and

somehow these all evolve into later microstates that determine E. But

CAUSATION IS MACROSCOPIC BUT NOT IRREDUCIBLE 139

there are no conditions satisfied by the initial microstates that might account

for their all evolving into states that determine E. I see no reason to accept

that there are macroscopic causal patterns which correspond to no causal

patterns at the physical level in this way. Imagine that temperatures super-

vened on molecular motions, and that certain temperatures produced

certain regular effects, but that there was no uniform story available at the

molecular level of why this should be so. This doesn’t seem the way that

things work in our world.

Note that I am not accusing non-reductive physicalism of any outright

inconsistency. There is nothing contradictory in the idea that the physical

realizers of some C should all just happen to eventuate in some E, either via

different causal paths at the physical level, or via no such casual paths at all.

But I take it that our experience shows us that the world just doesn’t work

like that. Macroscopic causal patterns do not depend on massive coinci-

dences at the physical level. Rather any macroscopic cause corresponds to a

common physical condition satisfied by its realizers, and there is a physical

story to be told about why this condition gives rise to the relevant effect.

8. Causes not Laws

In a number of previous papers (Papineau 1985, 1992, 2010) I have offered a

similar argument against non-reductive physicalism. However those earlier

papers focused on laws, not causes. My earlier arguments owed nothing to

the way that asymmetric causation depends on probability distributions over

the microstates consistent with a given macro-condition. Rather I simply

appealed to our supposed knowledge of the way general laws of any kind

depend on uniform physical processes, arguing that it would be incredible

that there should be a law involving physically supervenient properties, yet

no uniform physical account of the way that the physical realizations of the

initial condition evolve into physical realizations of the final condition.

However, this line of argument can be criticized for its appeal to the

unexplained notion of a ‘uniform’ physical account. Take the full range of

nomologically possible physical microstates that can realize the relevant

initial macrostate. Now suppose that those initial conditions evolve

according to the basic dynamical law (Newton’s second law in a classical

context, or Schrodinger’s equation in a quantum context). The upshot will

140 DAVID PAPINEAU

be that this range of initial microstates will be shown to end up in a range of

later microstates that determine the relevant final macro-condition. On

what grounds do I say that this is not a ‘uniform’ account of the original

macroscopic law? It can’t just be that the basic dynamic law is being applied

to a range of distinguishable microstates. That would rule out pretty much

anything as a uniformly explained law, including any laws of thermodynam-

ics, chemistry, or planetary motion.

Yet in my earlier papers I offered no other account of what might render

a physical account non-uniform. Given this, it is unclear what force there is

to my insistence that it is ‘incredible’ that there should be macroscopic laws

that lack a uniform physical account. In the absence of some further

explanation of what counts as uniform, what exactly is it that I say I find

incredible?3

Of course, this is not an objection that can happily be made by those who

want to define themselves as non-reductive physicalists, since they too will

need to appeal to a notion of ‘uniform’. This is because they want to insist

that the application of the basic dynamic law to the collection of initial

conditions in not a uniform physical reduction, but a derivation that covers a

physically heterogeneous range of cases. Still, the point remains that, with-

out some further account of what counts as a ‘uniform’ physical process, it is

unclear what substance there is to my dispute with non-reductive physical-

ism. It seems as if there may be nothing at issue when I claim that macro-

scopic laws must be physically reducible, and they deny this.

I take this paper to add substance to my position by focusing on causal

processes. I no longer wish to argue that all laws are physically reducible—I

concede that there may be no good sense in which all macroscopic laws

must have a uniform physical reduction. Rather my focus is now specifically

on asymmetric causal patterns, and my claim is that for any such causal

pattern, there will be a constraint specifiable at the physical level common to

all realizations of the cause, and that a probability distribution over the

microstates satisfying this constraint will play a part in explaining why the

effect follows. This is what I mean by a uniform physical explanation for a

causal pattern. So my present thesis is that there are no causal patterns in our

world that lack uniform explanations of this kind.

3 Barry Loewer has pressed me on this point in conversation on a number of occasions.

CAUSATION IS MACROSCOPIC BUT NOT IRREDUCIBLE 141

9. Does Functionalism Help?

It might have occurred to some readers to wonder whether a functionalist

account of mental states might not help to explain how a given mental cause

may produce its results via different physical-level causal processes on

different occasions. If mental states are defined functionally as states which

produce certain effects, then won’t such variable causal mechanisms be just

what we would expect? Suppose the mental state of wanting to hail a taxi is

defined as a state that will produce arm movements or similar signals. There

may well be lots of different physiological states that satisfy this requirement.

But it will scarcely be ‘incredible’ that they should all alike give rise to arm

movements—for it is just this tendency that qualifies them as realizations of

wanting to hail a taxi in the first place.

However, the appeal to functionalism does not help. As is well known,

there are two very different versions of the functionalist thought that mental

states can be ‘defined’ as states which produce certain effects. Once they are

clearly distinguished, we can see that neither of them helps non-reductive

physicalism to explain how one state can cause another via different causal

processes. The impression that functionalism helps with this problem only

arises if the two versions are run together.

The first version—realizer functionalism—is a thesis about how the

reference of mental terms is fixed. On this view, mental states are physical

states that are identified via their connection to certain causes and effects.

For example, the term ‘desire to hail a taxi’, applied to some person, is to be

understood as referring to that physical state which causes appropriate arm

movements in that person. This term might thus refer to different physical

states in different people, just as the term ‘your watch’ might refer to

different devices when different people are being addressed.

This view does nothing to explain variably realized causes, for the simple

reason that it does not trade in variably realized states of any kinds. The only

states it countenances are ordinary physical states—such as the physical state

which causes my arm to move, say—and these physical states cause their

effects in an ordinary uniform manner. True, different physical states may

well be picked out by the same mental word, in virtue of producing some

common effect—but there is nothing here to suggest that any given such

state produces its effect via different routes on different occasions.

142 DAVID PAPINEAU

Then there is role functionalism. This does recognize states which are

variably realized at the physical level. Role functionalism takes mental terms

to refer, not to the first-order physical states that have certain specified

causes and effects, but to the second-order states of having some first-order

state that plays that causal role. On this view, mental terms will have the

same referent even when applied to differently constituted beings: they refer

to the second-order state shared by all beings who instantiate the relevant

causal role.

We can usefully bring out the difference between realizer and role

functionalism by thinking of a term like ‘dormitive virtue’ as applied to

sleeping pills. The realizer option would take this term to refer to the

narcotic chemical constituent present in whichever sleeping pill is under

discussion. The role option, by contrast, would take the term to refer to the

property common to all sleeping pills, namely, their tendency to produce

sleep by whatever means.

Now, role functionalism does arguably give us variably realized states.

Just as a tendency to produce sleep can be realized by different chemical

processes, so can a tendency to move one’s arm be realized by different

physiological processes. The trouble is that, if mental states are like tenden-

cies to produce sleep, then surely they are disqualified as causes of the effects

that constitute them. A tendency to produce sleep isn’t sufficiently distinct

from the sleep itself to qualify as its cause. Similarly, if a desire to hail a taxi

constitutively requires appropriate arm movements, it isn’t distinct enough

from the arm movements to cause them.

So whichever way we turn functionalism, it doesn’t give us causes which

produce their effects via non-uniform physical processes. Realizer function-

alism gives us causes all right, but they operate in a physically uniform

manner. Role functionalism gives us variably realized states all right, but

they aren’t causes of the relevant effects.

10. Selectional Properties

Perhaps a different kind of functionalism can account for variably realized

causes. Rather than considering states that are defined or constituted by a

causal role, let us instead consider states that are functional in the sense that

they have been selected to play some causal role.

CAUSATION IS MACROSCOPIC BUT NOT IRREDUCIBLE 143

The puzzle I have been pressing so far is how some physically superven-

ient putative cause C can regularly co-vary with some putative effect E if

there is no common feature at the level of its physical realizations to account

for this. In the absence of any such commonality at the physical level, it

seems mysterious that E should generally follow.

But now suppose that the instances of C have been selected because they

produce result E. That is, they occur as the result of some selection process

that favours items that produce E. Then the puzzle would be explained.

There would be an explanation for why C generally leads to E even though

there is no uniform explanation at the physical level. E generally follows

because different instances of C have been selected to produce precisely

that result.

To illustrate, consider the simple example of thermostats in electrically

controlled domestic hot-water heating systems. Any such system contains a

thermostat which stops the heating once the water reaches some set tem-

perature. But these thermostats involve a range of different mechanisms at

the physical level, including bi-metallic strips, expansion gases, mercury

bulbs, and thermocouples. Yet there is clearly no puzzle here as to why

these different kinds of thermostat always produce the same effect of

stopping the heating. Their mechanisms have been selected by the heating

designers precisely in order to produce this effect.

So maybe this is a model for unreduced causes. Take the property, in a

heating system, of containing a thermostat. Let us suppose that this property

does not constitutively involve the effect of stopping the heating, and so is a

candidate for causing that effect. Won’t this now amount to a case where

this physical effect is caused by a variably realized property, namely the

property of containing a thermostat?4

Certainly the counterfactuals seem to vindicate this claim. It is the generic

presence of a thermostat per se, rather than the specific mechanism that

realizes it in a given case, that comes out as proportional to the effect of

4 The most natural way of construing selectional properties like being a thermostat is as constitutively

involving some past history of selection. So understood, selectional properties arguably won’t consti-

tutively involve their effects as role properties do—something can be selected to do F and yet have no

tendency to do F in the future. Even so, selectional properties might be held to be ineligible as causes on

the different grounds that historical provenance cannot matter to causal significance. I shall not press this

particular worry, however. Perhaps it can be avoided by construing selectional properties in some way

that disconnects them from their history. But even if it can be so avoided, the causal status of selectional

properties would still be open to the more fundamental objection made below.

144 DAVID PAPINEAU

stopping the heating. If (1) there hadn’t been a thermostat, the heating

wouldn’t have stopped. And if (2) the thermostat had been realized differ-

ently, the heating would still have stopped. By contrast, the specific mech-

anism does not seem proportional. While it is true (2) that a differently

realized bi-metallic strip would still have stopped the heating, it isn’t true (1)

that if there hadn’t been a bi-metallic strip, the heating wouldn’t have

stopped—because in that case a different design of thermostat would no

doubt have done the job instead.

Of course mental systems are not designed by intelligent engineers in the

way that heating systems are. But, to the extent that they are designed by

phylogenetic and ontogenetic selection processes, the same moral will

apply. These selection processes will ensure that there is some mental

component available to produce a given effect, but the precise mechanism

that does this may vary from case to case.

Thus consider pain across different species. Intergenerational genetic

selection will have ensured that all organisms have some mechanism that

responds to bodily damage by seeking to avoid the source of the damage.

But it may well have lit on different things to do this job in different species.

Nor is the point restricted to the way that the products of intergenera-

tional genetic selection can vary across species. Humans and other complex

animals are sophisticated learning machines that embody a hierarchy of

processes that operate to preserve items that produce such-and-such effects.

The items selected may well be physically different in different individuals,

or even in the same individual at different times, but this won’t matter to the

selection mechanisms, provided they produce the reinforcing effects. The

state which leads me to hail a taxi when one is needed may be quite

differently realized in me and in you, but we are both likely to possess

some such state.

Just as with the thermostats, proportionality considerations again suggest

that such selectional mental states can qualify as variably realized causes

of physical effects in their own right. Consider again the state of wanting

to hail a taxi, and the effect of my arm moving. If (1) I hadn’t wanted to hail

a taxi, my arm wouldn’t have moved. And (2) if this desire had been realized

differently, my arm would still have moved. But now consider the specific

brain state that realizes the desire in me. While it is true (2) that if this brain

state had been realized differently my arm would still have moved, it

isn’t true (1) that if I hadn’t had that brain state, my arm wouldn’t have

CAUSATION IS MACROSCOPIC BUT NOT IRREDUCIBLE 145

moved—because in that case a different brain state would have been

selected to move my arm instead.

11. Too Many Causes

This might all now look like good news for unreduced mental causes.

However, I think that appearances are deceptive. Despite the points made

above, there is a strong reason to doubt that selectedmental items, and indeed

selected items generally, can feature as non-reduced causes in their own right.

To the extent that the proportionately counterfactuals argue differently, I say

that these counterfactuals are misleading as to causal structure.

Let us ask why certain physical states are selected to play a certain role in a

cognitive structure or other designed system. The answer is that these states

are apt to cause some specific effect, and the relevant selection mechanism

favours items with this feature. However, if this is the reason why these

physical states are selected, it rules out the more generic variably realized

selectional state from also causing that effect.

Take the thermostat example again. If a heating engineer chooses to put a

bi-metallic strip into the electric circuit, this is because this itemwill cause the

circuit to break when the temperature rises. It is precisely the causal status of

this item that renders it suitable for the engineer’s purpose. But this then

undermines the thought that having a thermostat per se causes the circuit

breaking. This generic property is common to different kinds of circuits, and

in each of these the breaking is caused by a different mechanism. Having a

thermostat itself does not cause the result, for having a thermostat depends on

being in some more specific state which does cause the result.

The same point applies to selectedmental causes.Whyhave phylogenetic and

ontogenetic selection processes picked certain brain states for the wanting-to-

hail-a-taxi-role? Because those brain states get activated when a taxi is needed

and they then cause arm waving or similar movements. The relevant selection

mechanisms will favour just those brain states that have this causal profile. But

this again argues that the selectional property of wanting-to-hail-a-taxi cannot

itself cause anything. Wanting-to-hail-a-taxi involves being in a brain state

which itself causes arm waving and so on. It was because this brain state already

caused this result, so to speak, that it was selected. Given this, it makes little sense

to think of the generic selectional state as also causing the result.

146 DAVID PAPINEAU

Note that this analysis does not appeal to some unthinking intuition that

more specific physical states always casually out-compete any more generic

states that supervene on them. As the earlier sections of this paper will have

made clear, I regard this intuition as fundamentally misguided. Rather

I have a more particular objection to viewing generic selectional states as

causes. This objection derives from the structure of the selection processes

that account for such selectional states, and in particular which explain how

they can produce uniform effects despite being variably realized. Selection

processes operate on causal facts. Their workings hinge essentially on the

causal properties of the items selected. They preserve items that cause certain

effects (see Papineau 2003). This is why we are forced to accept that it is

these realizing items that cause those effects, and not the generic selectional

states that supervene on them.

What about the counterfactuals? As we saw, they do seem to indicate the

generic selectional states as causes, in preference to the more specific mech-

anisms that realized them. If we hadn’t had the generic state, we wouldn’t

have had the result. By contrast, it’s not true that the result wouldn’t have

occurred if we hadn’t had the specific realizing mechanism—since in that case

some other item would no doubt have been selected to produce the result

instead. So it looks as if the generic state is proportional to the effect, rather

than the specific realizer.

However, I take this to be analogous to the many familiar cases where the

counterfactuals fail match causal structure because of back-up arrangements.

When I make an assassination plan with a contingency arrangement (for

example, a back-up assassin lest the first one fail), it is the whole plan that is

proportional to the death of the prisoner, not the shooting by the first

assassin. Yet it is that first assassin that caused the death, and the back-up

assassin played no causal part. Similarly in the cases at hand: it is the selec-

tional state of being designed for some end that is proportional to the effect,

but the specific mechanism that fulfils the design that actually causes it.

12. Explanation is Different

Variably realized selectional states may not cause physical effects, but this

does not mean that they cannot be used to explain them. It will be worth

CAUSATION IS MACROSCOPIC BUT NOT IRREDUCIBLE 147

clarifying this issue, in order to forestall any inference from the explanatory

significance of variably realized states to their causal efficacy.

In this connection, note first that we can often refer to genuine causes

indirectly, by citing variably realized selectional states like wanting to hail a

taxi. This is possible because we can use descriptions involving the selec-

tional states to construct variable names for the genuinely causal physical

states. The way this works has already been discussed under the heading of

‘realizer functionalism’. We saw there howwe can read ‘dormitive virtue’ as

referring to the specific chemical property present in whichever sleeping pill

is under discussion. Similarly, we can understand ‘wanting to hail a taxi’ as

referring to the specific brain state that makes the relevant subject’s arm

move. So understood, claims like ‘he fell asleep because he took a pill with

dormitive virtue’, or ‘his arm waved because he wanted to hail a taxi’, will

state causal truths.

Not only will such claims state causal truths, but they can also be

explanatory. Explanations of particular facts need to name causes. But

they can do so indirectly, using descriptions involving the selectional status

of those causes. As long as this mode of reference shows us how those causes

fit into patterns that can be used to anticipate and control, the attribution of

causes will be explanatory.

Thus it can certainly be explanatory to say that someone fell asleep

because he took a pill with a dormitive virtue (as opposed to having had a

very tiring day, say). Similarly, it can be genuinely explanatory to say that

someone’s arm waved because they wanted to hail a taxi. (Not all indirect

references to causes are explanatory. It is not explanatory to say I fell asleep

because I was caused to fall asleep. We need to cite the cause in a way that

fits it into a practically significant pattern.)

Selectional states of all kinds are very commonly cited in explanations.

I might explain the high temperature in the room by the setting on the

thermostat, or the improved performance of my car by its new carburettor.

I may have no idea of the actual mechanisms in either case. But knowledge

of design properties tells me how the relevant items will work and so suffices

for explanatory purposes. Similarly with mental explanations. The states we

cite may be variably realized selectional states which are not themselves

causes, but they can be genuinely explanatory for all that.

148 DAVID PAPINEAU

13. Causal Closure

The principle of the ‘causal closure of the physical’ has played a significant

role in recent philosophy of mind. According to this principle, every

physical effect must have a physical cause. It is this principle that lies behind

the widespread modern acceptance of physicalism (Papineau, 2002). It

allows us to argue that any non-physical realm can only be epiphenomenal,

since it would generate an unacceptable overdetermination of physical

effects to attribute them to non-physical causes in addition to the physical

ones already guaranteed by closure.5

It is tempting to infer the falsity of the closure principle from the

possibility of macroscopic causes. If, as proportionality considerations

argue, macroscopic causes can out-compete the more specific realizers as

causes of certain physical effects, does this not show that those physical

effects at least will have macroscopic causes rather than physical ones, and

therefore that the physical realm is not causally closed? (Cf. Menzies and List

2010.)

Rejecting causal closure would have a cost. Without a principle of causal

closure, we would be left with no argument against interactive Cartesianism

and other strong forms of dualism. Fortunately, the possibility of macro-

scopic causes does not refute causal closure. This would only follow if

macroscopic implied non-physical. I have argued that it does not. To repeat

my standard example, temperature is a macroscopic property, but it can be

type-identified with the physical property of mean kinetic energy.

Moreover I have argued that, not only is macroscopicity consistent with

physicality, but that macroscopic causation positively requires physicality. If a

macroscopic cause cannot be type-identified with a physical property, we

can’t give a uniform explanation of why the same physical effect always

follows from its different realizations. Nor does it help to appeal to selection

processes to explain this, for it is built into the nature of selection that the

5 The literature displays different uses of this argument. Some use it only to rule out forms of dualism

on which the mental realm does not even supervene on the physical. But others, most prominently

Jaegwon Kim, also use it to argue against ‘non-reductive physicalisms’ that respect supervenience but

deny type identity. The latter form of argument assumes that any kind of overdetermination is unaccept-

able, even when one cause supervenes on the other. The former can allow such supervenient overde-

termination, and need assume only that overdetermination by metaphysically distinct causes is

unacceptable.

CAUSATION IS MACROSCOPIC BUT NOT IRREDUCIBLE 149

relevant effects are caused by realizing mechanisms, rather than by the

generic selectional states that these mechanisms determine.

So my overall analysis reinforces the causal closure of the physical. Cer-

tainly many physical effects should be attributed tomacroscopic causes rather

than their more specific realizers. But these macroscopic causes will still

always be physical, thus upholding the principle that every physical effect

must have a physical cause, and leaving the argument against dualism intact.

14. Mental Causes

One last point. In the latter half of this paper I have been arguing that

variably realized mental states cannot be causes. But this does not of course

mean that mental states as such can never be causes. For there remains the

possibility that some mental states can be type-identified with physical

states, in the way that temperature is type-identified with mean kinetic

energy.

I have paid little attention to this possibility so far, given that my main

concern has been to establish that variably realized states cannot be causes.

But the physical type identity of at least some mental causes is a serious

option. Remember that type identity does not require that there can be no

physical differences between the bearers of a given mental state, just that

there should be some physical commonality which might explain why the

state regularly produces certain effects.

It seems very likely that a wide range of mental states are so uniformly

realized within humans, and indeed across many of the other taxa to which

we belong. For example, there is every reason to suppose that the pain

mechanism is uniformly realized across humans and similar mammals.

Again, many sensory mechanisms can be expected to be physically uniform

in this way. Perhaps the basic mechanisms of learning and reasoning will also

be uniformly realized in all humans, even if not in other species. Provided

that we understand our mental terms for these categories as indexed to the

appropriate range of species, we can read them as referring to physically

reducible types, and hence to fully causal states.

On the other hand, I accept that many other mental states will be variably

realized across humans. These will be states which derive from ontogenetic

selection processes. For example, I would expect wanting to hail a taxi to be

150 DAVID PAPINEAU

variably realized within humans, and even perhaps within individuals. States

like these will thus not be causally efficacious, even though they can be

explanatory significant in the way explained above.

It is an intriguing question which states are which. For everyday explana-

tory purposes the difference may not matter much, given that both kinds

can equally be invoked in explanation. But the contrast will be significant

for cognitive science. Investigation of the physical nature of physically

reducible states could bring important scientific benefits, but a similar

investigation of variably realized states would inevitably be fruitless. Cogni-

tive science thus needs to know which mental states are causal in their own

right, and which play only an explanatory role.6

References

Albert, D. (2000). Time and Chance. Cambridge, MA: Harvard University Press.

Bennett, K. (2003). ‘Why the Exclusion Problem Seems Intractable and How, Just

Maybe, to Tract It’. Nous, 37: 471–97.

Hausman, D. (1998). Causal Asymmetry. Cambridge: Cambridge University Press.

Kim, J. (1993). ‘The Non-reductivist’s Troubles with Mental Causation’. In J. Heil

and A. Mele (eds.), Mental Causation. Oxford: Clarendon Press: 336–57.

LePore, E. and B. Loewer (1987). ‘MindMatters’. Journal of Philosophy, 84: 630–42.

List, C. and P. Menzies (2009). ‘Non-Reductive Physicalism and the Limits of the

Exclusion Principle’. Journal of Philosophy, 106(9): 475–502.

Menzies, P. (2008). ‘Causal Exclusion, the Determination Relation, and Contrast-

ive Causation’. In J. Kallestrup and J. Hohwy (eds.), Being Reduced: New

Essays on Reductive Explanation and Special Science Causation. Oxford:

Oxford University Press: 196–217.

—— and C. List (2010). ‘The Causal Autonomy of the Special Sciences’. In

G. and C. Macdonald (eds.), Emergence in Mind. Oxford: Oxford University

Press: 108–28.

Michotte, A. (1946/1963). The Perception of Causality. English translation 1963 by

E. and T. Miles. New York: Basic Books.

Papineau, D. (1985). ‘Social Facts and Psychological Facts’. In G. Currie and

A. Musgrave (eds.), Popper and the Human Sciences. Dordrecht: Nijhoff: 57–71.

6 I would like to thank Eleanor Knox, Barry Loewer, Laurie Paul and David Yates for valuable

comments on earlier drafts.

CAUSATION IS MACROSCOPIC BUT NOT IRREDUCIBLE 151

Papineau, D. (1992). ‘Irreducibility and Teleology’. In D. Charles and K. Lennon

(eds.), Reduction, Explanation and Realism. Oxford: Oxford University Press:

45–68.

—— (2001). ‘Metaphysics over Methodology—or, Why Infidelity Provides no

Grounds to Divorce Causes from Probabilities’. In M.-C. Galavotti, P. Suppes,

and D. Costantini (eds.), Stochastic Causality. Stanford: CSLI Publications: 15–38.

—— (2002). Thinking about Consciousness. Oxford: Oxford University Press.

—— (2003). ‘Causation as a Guide to Life’. In The Roots of Reason: Philosophical

Essays on Rationality, Evolution, and Probability.Oxford: Oxford University Press:

167–211.

—— (2010). ‘Can Any Sciences be Special?’ In C. and G. Macdonald (eds.),

Emergence in Mind. Oxford: Oxford University Press: 179–97.

Pereboom, D. (2002). ‘Robust Nonreductive Physicalism’. Journal of Philosophy,

99: 499–531.

Putnam, H. (1975). ‘Philosophy and Our Mental Life’. In Mind, Language and

Reality: Philosophical Papers, Volume 2. Cambridge: Cambridge University Press:

291–303.

Shoemaker, S. (2001). ‘Realization and Mental Causation’. In C. Gillett and

B. Loewer (eds.), Physicalism and Its Discontents. Cambridge: Cambridge University

Press: 74–98.

Sosa, E. andM. Tooley (1993). ‘Introduction’. In Sosa and Tooley (eds.),Causation.

Oxford: Oxford University Press: 1–32.

Spirtes, P., Glymour, C., and Scheines, R. (2000). Causation, Prediction, and Search.

Cambridge, MA: MIT Press.

Yablo, S. (1992). ‘Mental Causation’. Philosophical Review, 101(2): 245–80.

152 DAVID PAPINEAU

6

Substance Causation, Powers,and Human Agency

E. J. LOWE

The powers of individual substances may, I believe, be distinguished in at

least the following two ways. First, some of these powers are causal powers,

while others are non-causal powers. Second, some of these powers are active

powers, while others are passive powers. But all powers, as we shall see, are

individuated by their manifestation types, that is, by the characteristic types of

activity that constitute their exercise. A causal power is one whose exercise

consists in the bearer of the power acting on one or more substances to bring

about a change in them. A passive power is one whose exercise is always

caused by one or more substances acting upon the bearer of the power. This

classification of powers leaves open the possibility of there being a type

of power that is at once active and non-causal: a power whose exercise is

(1) not caused by any substance acting upon its bearer and (2) does not

consist in its bearer bringing about a change in any substance. Such a power

may be called a spontaneous power. It seems clear that such powers do exist in

nature, the power of a radium atom to undergo radioactive decay being an

example. This fact shows that there need be nothing anti-naturalistic, or

incompatible with current physical science, in supposing that the human

will, as it is exercised in episodes of voluntary action, is another such power.

In saying that the will is a non-causal power, it is not being implied that the

will is causally inefficacious, only that its exercise does not consist in the

agent’s bringing about any sort of effect. Agent causalists who suppose that

agents cause their own volitions by exercising agent-causal powers are,

I believe, mistaken in this regard and mistaken too if they think that their

view explains the special sense in which free agents have control over their

voluntary actions. What, in my view, distinguishes the will from any other

kind of spontaneous power is (1) that it is a two-way power—a power either to

will or not to will a particular course of action—and (2) that it can be exercised

rationally, that is, ‘in the light of reason’. The possession of such a power

would, I believe, give human agents all the control that they could need or

want over their voluntary actions. And very arguably, as we shall see, we

cannot—on pain of undermining our entitlement to regard ourselves as

rational beings—deny that we have such a power.

Individual substances

Individual substances—as I propose to construe this term—are persisting,

concrete bearers of properties, including powers of various kinds, at least

some of which are causal in character. There cannot, in my view, be any

such thing as a causally inert individual substance. Furthermore, individual

substances are ontologically independent entities, in a sense of ‘independent’

that needs to be carefully defined. The sense in question is this: individual

substances do not depend for their identity on other entities of any kind.1 This

is quite consistent with saying that they may depend for their existence on

various other entities, such as their proper parts (if they have any). I shall

illustrate this point by means of examples in a moment.

Now, ostensibly at least, individual substances include both material objects

of certain kinds (‘bodies’, in a broad sense of that term) and also human

persons, such as ourselves. Items of these two types constitute what

P. F. Strawson called ‘basic particulars’—‘basic’ in the sense that it is ultim-

ately only by reference to items of these two types that we can in general

succeed in identifying concrete particulars of any other types, such as

particular events and particular states, including mental events and mental

states.2 Different kinds of individual substance are primarily distinguished,

moreover, by their distinctive identity conditions and their distinctive powers—

just as John Locke maintained.3 Thus, amongst material objects, mountains

have very different identity conditions and powers from, say, mice. The sorts

of changes through which a mountain can survive identically are mostly very

1 I defend this claim more fully in Lowe (1998), Chapter 6. 2 See Strawson (1959).3 See Locke (1975) [1690], Book II, Chapter 23.

154 E. J . LOWE

different from those through which a mouse can survive identically. And

what a mountain can characteristically cause is very different from what a

mouse can characteristically cause. (By a ‘concrete’ particular, incidentally,

I simply mean a particular that exists in space and time, as opposed to an

abstract particular, such as the number 7.4)

Returning, however, to the point that individual substances do not

depend for their identity on other entities of any kind, consider, for example,

a particular mouse, Mortimer. Mortimer certainly depends for his existence

on other entities, such as the atoms and molecules that compose him at any

given time, in the absence of which he could not exist at all. But, since

Mortimer can survive a change of these atoms andmolecules, his identity does

not depend on their identity: that is to say, whichmouseMortimer is does not

depend on which atoms and molecules compose him at any given time. After

all, Mortimer’s identity is essential to him, but it is not essential to him that he

is composed at any given time of the particular atoms and molecules that do

happen to compose him then, for the latter is a purely contingent circum-

stance. Nor does Mortimer depend for his identity on any event in which he

‘participates’, such as his birth or death—since, on the contrary, any such

event depends at least in part for its identity on him, and identity dependence

is asymmetrical.5 Which death is the death of Mortimer is at least partly

determined by which mouse Mortimer is, not vice versa.

Powers: token and type

When we speak of powers it is important to distinguish carefully between

token powers and power types. By token powers, I mean the particular powers

of individual substances, such as a particular copper wire’s power to conduct

electricity, as opposed to the type of power of which this particular power is a

token, namely, electrical conductivity. Different copper wires share this

same power type, inasmuch as they are all electrically conductive, but each

has a different token power of that type—its own particular power to conduct

electricity. A token power belongs essentially to the individual substance that

is its bearer and cannot be ‘transferred’ to another individual substance.

4 I explain my conception of the abstract/concrete distinction more fully in Lowe (1998), Chapter 10.5 See again Lowe (1998), Chapter 6.

SUBSTANCE CAUSATION, POWERS, AND HUMAN AGENCY 155

As I mentioned earlier, powers—both token powers and power types—

may be distinguished in two ways. Some are causal, some non-causal. And

some are active, some passive. Moreover, these two distinctions are mutually

independent, generating a fourfold division of powers to which I shall

return shortly. All powers, however, are individuated at least partly by

their manifestation types.6 This again applies both to token powers and to

power types. Every power, P, is essentially a power to ç, for some distinct-

ive type of manifestation, ç-ing. For example, electrical conductivity is a

power to conduct an electrical current. Again, water-solubility is a power to

dissolve in water. Now, it is important to recognize that there are higher-order

powers—powers to acquire lower-order powers. For instance, magnetiz-

ability is a power to acquire the power of being magnetic, which in turn is a power

to attract ferrous metals. However, in what follows I shall concentrate on first-

order powers, these being powers to act in a certain way.

Now, where power types are concerned, I am happy to say that they are

individuated entirely by their manifestation types—and I consider moreover that

each power type has only onemanifestation type.7 Inmy view, it makes no sense

to say that the very same powermight bemanifested in two quite different ways. If

we are sometimes tempted to think this, it may be because we are apt to confuse

what is properly speaking the manifestation type of a given power with certain

further effects that such a manifestation may give rise to. For example, we may be

inclined to say that the very same power—magnetism—ismanifested both by the

attraction of ferrous metals and, say, by the navigational behaviour of homing

pigeons, which relies on a sensitivity to the earth’s magnetic field. But, in fact, it

appears that homing pigeons can only navigate in this way because they contain

particles of ferrous metals in their sensory systems which are attracted by the

earth in virtue of its magnetism. Their homing behaviour is not amanifestation of

the earth’s magnetism.Rather, the earth’s magnetismmanifests itself by attracting

the particles of ferrous metals in the pigeons’ sensory systems and these systems have

evolved so as to adjust the flight behaviour of the pigeons in response to the

movements of those particles.

As for token powers, however, we need to say that they are individuated

not just by their manifestation types but in addition by their bearers and by

6 I defend this claim more fully in Lowe (2010).7 Here, then, I agree with George Molnar: see Molnar (2003), 195. For further discussion, see Lowe

(2010).

156 E. J . LOWE

their time of existence. This is because token powers, unlike power types,

are concrete particulars. Thus, this particular copper wire’s token power of

electrical conductivity is individuated by its manifestation type—conducting

an electrical current—together with its bearer, the copper wire in question,

and the time (which may be of long or short duration) at which that wire

possesses the token power. In the case of at least some token powers,

including this one, we have to be able to say that they can be manifested

more than once, but also that they need not be manifested even once. That is

why we cannot say that a token power is individuated by its manifestation

tokens, only by its manifestation type in conjunction with its bearer and time

of existence. (Other token powers, such as the token power of a stick of

dynamite to explode, can be manifested only once, since any manifestation

of the token power destroys its bearer. But even in this case there is no

necessity for the token power actually to be manifested at all.) Incidentally,

when I say that an item, such as a token power, is individuated by certain

other items—such as its manifestation type, bearer, and time of existence—I

mean to imply that it thereby depends for its identity on those other items.

Causal and non-causal powers

At this point it is appropriate to return to the topic of the fourfold division of

powers alluded to earlier. A causal power, as I shall construe this term, is one

whose manifestation or ‘exercise’ consists in its bearer’s acting on one or

more other individual substances (or sometimes on itself) so as to bring about

a certain kind of change in them (or it). Take, for instance, water’s power to

dissolve salt. Dissolving something is a matter of causing that thing to become

dissolved. And here, most importantly, we have a case of what I would call

substance causation. More particularly, we have a case of one substance, water,

exercising one of its distinctive causal powers by bringing about a certain

kind of change in another substance, salt.

My own view is that, fundamentally speaking, all causation is substance

causation, because only substances strictly and literally possess causal powers.8

If events and properties, for example, are sometimes said to ‘have’ causal powers,

this can in my view only be understood to be the case in a derivative or

8 For earlier defence of this claim, see Lowe (2008), Chapters 6–8.

SUBSTANCE CAUSATION, POWERS, AND HUMAN AGENCY 157

secondary sense. For causing is a kind of action—abringing about of change—

and events and properties cannot literally act: only substances can do that.

Events may be the effects of action—what is brought about when an agent

acts—since they just are changes. But they are never literally agents themselves

and so can never literally cause anything. If we are often inclined to speak as if

they do, this should be understood as no more than a facon de parler.

We might say, for instance, that the explosion of the stick of dynamite

caused the collapse of the building. But really, in my view, this is just an

elaborate way of saying that the stick of dynamite, by exploding, caused the

building to collapse. It is the dynamite that literally possesses the destructive

power, not the explosion. To treat the latter as a powerful particular is

indulge in an illicit hypostatization: the treatment of a non-substance as if it

were substance. This is not, of course, to deny the relevance of the explosion

to the kind of change that was brought about. If the detonator had failed, the

building would have remained intact. Substances can bring about effects

only by acting in appropriate ways: but it is nonetheless the substances that

bring those effects about, not their actions.

Now, in contrast with a causal power, such as water’s power to dissolve salt,

we have also non-causal powers. An example would be the power of a spherical

object to roll down an inclined plane. The manifestation of this power—the action

of rolling down an inclined plane—does not consist in its bearer (the spherical

object) bringing about any distinctive kind of change in anything (not even in

itself). It simply consists in a certain kind of translational motion—a movement

from one location to another. Of course, by so moving the spherical object

might cause some change to occur in another object: for instance, itmight cause

an object in its path to be crushed. But the important point is that any such effect

is not part of themanifestation of the power in question, anymore than a pigeon’s

homing behaviour is part of a manifestation of the earth’s power of magnetism.

The proof of this is that the spherical object could roll without anything’s

actually being crushed by it. By contrast, water obviously could not manifest

its power to dissolve salt without some salt’s actually being dissolved by it.

Active and passive powers

Having explained the distinction between causal and non-causal powers,

I come next to the distinction between active and passive powers. A passive

158 E. J . LOWE

power, as I propose to use this term, is one whose manifestation or exercise

always needs to be caused by one or more substances acting on its bearer (and

thereby exercising their causal powers). For example, salt’s water-solubility

is a passive power—we might alternatively call it a ‘liability’—because its

manifestation type, which is dissolving in water, has to be brought about by

some water exercising its power to dissolve the salt in question, by causing it

to dissolve.

An active power, by contrast, is one whose characteristic manifestation

never needs to be ‘triggered’ in this way. In the current jargon, such a power

has a manifestation type, but no stimulus type. An example would be the

power of a radium atom to undergo spontaneous radioactive decay.

Radium has a characteristic half-life, implying that there is a certain object-

ive probability or chance of any given radium atom decaying within a

specified interval of time—a probability that is the same for all radium

atoms, no matter how long such an atom may have been in existence. No

external circumstances or conditions can affect this probability. When such

an atom decays, then, this isn’t a matter of probabilistic causation. That is to

say, it isn’t a case of anything’s raising or fixing the chances of the atom’s

decaying, since those chances are already fixed by the nature of the atom

itself, independently of any external conditions that it may happen to find

itself in. Rather, in such a case, there is no causation at all. This means, of

course, that causal determinism—the doctrine that every event is either

causally necessitated, or at least has its chances of occurrence fixed, by

antecedent events—is false. As we shall shortly see, this has important

implications in the domain of mental causation and voluntary action.

The fourfold classification of powers

It may be helpful if I present here in diagrammatic form the fourfold

classification of natural powers that I have just been describing, providing

in the case of each basic type of power a paradigmatic example. This is set

out in Figure 6.1.

A few words of further explanation are perhaps called for. Matter’s power

of gravitational attraction is clearly a causal power, since its manifestation

consists in the causal activity of attracting other matter. But it is also an active

power, in the sense defined earlier, since its manifestation does not need to

SUBSTANCE CAUSATION, POWERS, AND HUMAN AGENCY 159

be ‘triggered’ or ‘stimulated’ by anything acting on the bearer of the power.

This is because all matter, by its very nature, is always exercising or mani-

festing this power.Water’s power to dissolve salt, by contrast—although it is

likewise a causal power—is a passive power, precisely because its manifest-

ation needs to be ‘triggered’ by the introduction of some salt into any body

of water that possesses the power.

The will as a spontaneous power

As we have just seen, there can be and in fact are powers in nature that are at

once active and non-causal—the radioactivity of a radium atom being an

example. This is a non-causal power because its manifestation just consists in

a change in the properties of the atom, not in the atom’s causing a change in

anything else (or even in itself). Now, I want to say that the human will is

another such active, non-causal power—in short, what I propose to call a

spontaneous power.9 Locke, of course, likewise held the will to be a power,

with volitions or acts of will as its manifestations or exercises.10 He took

volitions to be a species of ‘thoughts’, in the broadest sense of that term—a

kind of ‘inner command’ of the mind to itself. But thinking is not a causal

activity. It is not a matter of bringing about some effect, not even in one’s

own mind. Of course, this is not to deny that thoughts may have effects,

or (as I would prefer to put it) that by thinking we can sometimes cause

Causal Non-causal

E.g., matter’sgravitational power

of attractionActive

Passive

E.g., radium’spower of spontaneous

radioactive decay

E.g., water’spower to dissolve

salt

E.g., a sphere’spower to roll downan inclined plane

Fig. 6.1

9 I propose this in Lowe (2008), Chapter 8. 10 See Locke (1975) [1690], Book II, Chapter 21.

160 E. J . LOWE

something to happen. It is just to say that thinking does not consist inmaking

something happen. (Clearly, it would be absurd to say that thinking is a

matter of causing thoughts to happen or occur in our minds. A thought just is

an episode of thinking and hence to say that thinking consists in causing

thinking to occur would result in either circularity or an infinite regress.)

The same applies, more specifically, to willing. I may, by willing, cause my

arm to rise. But my willing in this instance does not consist inmy causing this

(or indeed anything else) to happen. That is demonstrated by the fact that

I may will to raise my arm and yet my arm may still fail to rise. If my willing

to raise it consisted in my causing it to rise, my willing could not occur

without my arm’s rising.

Here, however, it may be asked how it can be the case that, by willing, an

agent can cause something to happen, even though the will is not a causal

power. But we have already seen that something similar occurs in other

cases of substance causation. Thus, a spherical object’s power to roll down

an inclined plane is a non-causal power and yet, by so rolling, the object can

cause another object in its path to be crushed. This implies, indeed, that the

spherical object does also have a certain causal power, namely, the power to

crush objects in its path, for crushing something is causing it to be crushed.

However, the power to roll and the power to crush are distinct powers,

since they have distinct manifestation types. What is true, nonetheless, is

that sometimes an object exercises or manifests its power to crush by means of

manifesting its power to roll, when another object with a suitable passive

power lies in its path. I take something similar to be case where the will is

concerned. By willing an agent may exercise or manifest a causal power to

raise his or her arm, but the power to will is nonetheless distinct from the

power to cause one’s arm to rise. Whether by willing the agent succeeds in

raising his or her arm will depend on whether the arm possesses a suitable

passive power to rise, which can be ‘triggered’ by the agent’s act of will.

As for the will’s being an active power—one whose manifestation or

exercise is never caused by something acting upon the agent whose will it

is—this is a claim that I believe to be supported on both phenomenological

and metaphysical grounds. But I shall return to this matter later. At present

I merely wish to observe once more that one cannot object to such a claim

on the empirical grounds that no such powers exist in nature, since modern

atomic physics tells us otherwise, the power of spontaneous radioactive

decay being a case in point.

SUBSTANCE CAUSATION, POWERS, AND HUMAN AGENCY 161

Agent causation

What I have just been saying about the will sets me apart from most self-

styled agent causalists, who typically say that human agents possess a special

power to cause their own volitions or intentions.11By this they mean a power

to cause such volitions or intentions not by doing anything—not, that is, by

acting in any manner—but simply in virtue of being a cause of them. Since

agents are, by almost anyone’s account, not events but substances, this means

that such agent causalists are committed to the notion of substance causation:

the causation of something by a substance. Where they differ from me,

however, is in supposing that a human agent can be a cause simpliciter of

something, namely, of a volition or intention. By contrast, I consider that

substances, including human agents, can only cause anything by acting in some

way. I can make no clear sense of the idea of an agent’s being a cause of

somethingwithout doing anything to cause it. Apart from anything else, there

is the familiar difficulty here of explaining why the effect should occur when

it does, if its cause is merely the agent as such.12 For the same agent may cause

many different effects which occur at different times, so that the agent’s time

of existence, which extends over the times of existence of all these different

effects, does not serve to explain why each of them occurs when it does.

The human power of will, as I understand it, is not at all like the sort of

agent-causal power that typical agent causalists subscribe to. As I see it,

when I exercise or manifest my power of will, the exercise consists in my

willing to do something, such as raise my arm. It doesn’t consist in my causing

anything, let alone causing myself to have a volition or intention to raise my

arm. As I say, I regard the will as a non-causal power, although I do want to

say that by exercising this power an agent may, in suitable circumstances,

succeed in causing something, such as the rising of his or her arm. Success or

failure will depend on whether some other thing has a suitable passive

power which can be ‘triggered’ by the act of will in question. If I have an

arm which has a passive power to respond to my volitions to raise it, then

I may succeed in causing it to rise by willing to raise it. This is the only

model of a powers-based conception of voluntary human agency that I can

11 Perhaps the best recent account and defence of agent causalism is O’Connor (2000).12 Perhaps the earliest formulation of this objection is to be found in Broad (1952). I acknowledge that

present-day agent causalists have replies to this objection, but I do not have space to discuss them here.

162 E. J . LOWE

understand. It has the advantage over standard agent causalism that it does

not invoke a special, sui generis type of substance causation which is confined

to the domain of human agency. Rather, it appeals only to types of

causation and powers that are found ubiquitously in the natural world.

Freedom of action

Standard agent causalists typically claim that their account of human agency

can alone provide a satisfactory sense in which voluntary human action can

be free, because by their account agents are originating causes of their own

voluntary actions, making those actions causally determined and yet at the

same time self-determined—that is, determined by the self or agent whose

actions they are, rather than by other agents or events. However, I believe

that my own account of human agency is no less able to explain the sense in

which human agents can freely determine their own voluntary actions. By

my account, they do this precisely by exercising their power of will, whose

manifestations—in the form of volitions or acts of will—help to causally

determine, for example, their bodily movements. To say that agents ‘deter-

mine’ their own volitions by causing them in a sui generis way is, in my view,

no advance on saying that they ‘determine’ them simply by exercising their

power of will, whose manifestations these volitions are. And saying the former

has all the disadvantages mentioned earlier. My account still allows us to say

that agents have an ‘originating’ role in their own actions, inasmuch as

causal chains leading to their voluntary bodily movements begin, by my

account, with an agent’s uncaused exercise of his or her will.

Of course, many philosophers opposed to ‘libertarian’ theories of free will

may be expected to reject both my account and that of standard agent

causalism, on the grounds that these accounts render our voluntary actions

mere ‘chance’ events. The idea is that if nothing determines the agent to

exercise his or her supposed agent-causal power or power of will in a

particular way, then any action determined by the agent in such a fashion

will be indistinguishable from a merely chance happening, such as the fall of

a die. Indeed, there is a notorious argument—the ‘roll-back argument’—

designed to demonstrate precisely this.13 If we hypothetically ‘re-run’ any

13 See van Inwagen (2002).

SUBSTANCE CAUSATION, POWERS, AND HUMAN AGENCY 163

supposedly free course of action, as libertarians would characterize it, a

sufficient number of times, then we may expect a certain proportion of

these ‘re-runs’ to replicate what actually happened in the original case, but

also expect a proportion of them to turn out differently—for, after all, if

they all turned out the same way, that would suggest that the agent was not

really free, in the libertarian sense, to determine the outcome. But now it

seems that we may regard the ratio of same to different outcomes in a

sufficiently large number of hypothetical ‘re-runs’ precisely as a measure of

the probability or chance of the original action turning out as it actually did.

And this makes it a purely chance event, like the fall of a die or indeed the

spontaneous decay of a radium atom.

Now, since I have already likened the power of will to the spontaneous

power of a radium atom to undergo radioactive decay, it might seem that

I am particularly vulnerable to this objection. However, although I do

indeed classify the will as a spontaneous power—meaning thereby merely

that it is an active, non-causal power—I by no means want to say that it is in

every other respect just like the power to undergo radioactive decay. I shall

now try to explain in what ways it is crucially different from a power such as

the latter.

The will as a rational two-way power

The first way in which the will differs crucially from the spontaneous power

of radioactive decay is in being a two-way power. As Locke recognized, the

will is a power to will or to refrain from willing any particular course of action

that presents itself to the mind. Presented with the possibility of raising my

arm on a given occasion, I can either will to raise it or alternatively refrain

from so willing by willing not to raise it. In contrast, a radium atom cannot

in any coherent sense refrain from decaying on any given occasion: at most it

can simply fail to decay, because it happens not to manifest its power to

decay on this particular occasion. Its failing to decay is not a manifestation of

a power it has not to decay, since it has no such power. As a free human

agent, however, I do have a power not to will to raise my arm because I have

a power to will not to raise it. And this is the very same power that I can

alternatively exercise by instead willing to raise it. (This, incidentally, is not

to contradict my own doctrine that each power has only one manifestation

164 E. J . LOWE

type, since both in willing to raise my arm and in willing not to raise it my

will manifests itself in the form of acts of will or volitions, albeit volitions with

different intentional contents.) This, then, is why I say that the will is a two-

way power—and in this respect it appears to be utterly unlike any other

power to be found in the natural world.

The second crucial way in which the will differs from other spontaneous

powers is in being a rational power, as indeed Aristotle maintained in Book Ł

of the Metaphysics. By this I mean that it is a power whose exercises are

responsive to reasons, or which is exercised ‘in the light of ’ reasons. A reason for

action I take to be any consideration which speaks in favour of the agent’s

acting in a certain way in certain circumstances.14 When deliberating about

how to act, an agent reflects on such reasons and then exercises his or her

will in a manner that, typically, corresponds to his or her judgement as to

where the weight of reasons for or against any particular course of action

falls. This is not to imply that the agent’s judgement causes him or her to will

or not to will a particular course of action, since that would obviously be

incompatible with saying that the will is a spontaneous power whose

exercises are, accordingly, uncaused. Even so, by this account, we may say

that an agent’s voluntary action, as determined by the exercise of his or her

will, typically has a rational explanation, in the shape of the particular reasons

that the agent judged to merit him or her acting in that particular way.

Obviously, there is a great deal more that needs to be said about such

matters,15 but enough has been said already to set the will radically apart

from a power like that of a radium atom to undergo spontaneous radioactive

decay. When an atom so decays, there is evidently no rational explanation, in

the foregoing sense, for this occurrence. A radium atom can have no

‘reason’ to decay, since it makes no sense to say that there could be

considerations which ‘speak in favour’ of its decaying at any given time,

or indeed ever. Of course, a human agentmight prefer some radium atom to

decay or not decay at some particular time, but the atom itself can have no

such ‘preference’, nor can the inanimate natural world at large. Inanimate

things simply do not belong to ‘the space of reasons’.

These two crucial features of the will serve, I believe, to nullify the ‘roll-

back’ argument. Even if an agent might have willed to act differently in

14 Compare Dancy (2000) and see further Lowe (2008), Chapter 9.15 I say a good deal more myself in Lowe (2008), Chapter 9.

SUBSTANCE CAUSATION, POWERS, AND HUMAN AGENCY 165

exactly the same circumstances, this by no means implies that his or her

willing in the particular way that he or she did is just a pure ‘chance’ affair.

For in either case the agent will, typically, have exercised his or her will in

the light of the reasons that were presented to his or her mind and since, in

many cases, it is simply a matter of judgement as to where the weight of

reasons falls—not a matter of incontestable fact—the agent will typically

have acted rationally either way, not just arbitrarily. Moreover, precisely

because this is largely a matter of judgement rather than of fact, there can be

no incontestable way of estimating the ‘probability’ of the agent’s exercising

his or her will in one way rather than another when presented with the same

reasons in the same circumstances. That being so, it is incoherent to suppose

that there is some measurable objective chance of his or her voluntarily acting

in this way or that in these circumstances, in the way that there is a certain

measurable objective chance of a radium atom’s decaying during any given

interval of time.

Agential control

Some may object that the account I have just given leaves us with no real

control over our actions, since it leaves us with no control over our will. The

thought here is that it is not enough for proper agential control over our

actions that we can determine them by exercising our will, unless we can

also determine the exercises of our will. But this then seems to lead us into

a vicious infinite regress of higher-order volitions, with second-order voli-

tions determining first-order ones, third-order ones determining second-

order ones and so on ad infinitum. It might seem, indeed, that standard

agent causalism gets around this difficulty by having the agent as such being a

cause of his or her own volitions and thereby determining them. But I, for

my part, certainly cannot allow that our volitions are ‘determined’, in sense

of being caused, by anything whatever, since I regard the will as a spontan-

eous power.

However, I have already implicitly answered this objection earlier, when

I said that an agent ‘determines’ his or her volitions or acts of will simply by

exercising his or her will. A volition is itself a kind of action—indeed, it is the

most primitive or basic kind of action that any agent can perform. To have a

power of will is to be able to will to do various things, such as raise one’s arm.

166 E. J . LOWE

One exercises this ability simply by willing. Because willing is a basic action,

however, one does not and cannot do it by doing anything else, and so a

fortiori not by willing to will. The idea that one should be able to ‘control’

one’s will, in the sense of doing something which will determine its oper-

ations, simply misrepresents the essential nature of the will, alienating it

from the agent whose will it is. The will is not some mechanism inside

me that I need to have power over. It is my power, for me to exercise as

I see fit, in the light of the reasons for action that present themselves to me.

When I do exercise it, I am demonstrating par excellence my control over

my actions: it is the very source of that control, not some means that I use

whereby to control my actions. For the same reason, it makes no sense to

speak of another agent controlling my will, even if we sometimes mislead-

ingly talk in these terms. Certainly, another agent can offer me strong

inducements to exercise my will in a certain way. But what another agent

cannot do is to determine my will for me. Saying this would be as absurd as

saying that someone else can literally make my mind up for me—indeed, it

really amounts to the same absurdity. Only I can make my own mind up

about some matter, however forceful the inducements that others may offer

me to make it up in a particular way. And it is in this fact that our true

freedom ultimately lies. (Of course, another agent may well be able to render

my will ineffective on a particular occasion, for instance by forcing my arm to

rise when I will not to raise it—but that is entirely different from making me

will to raise it by determining my will for me, which I hold to be impossible

because unintelligible.)

Incidentally, none of what I have just said is to deny that we can school

ourselves to be less impetuous in our decision-making. In that sense, we can

‘rein in’ our will and learn from unwise or hasty decisions taken in the past.

But this is not the same as having to ‘control’ our will by constantly having

to monitor and determine its operations, as though it were some homun-

culus inside us. That is just absurd and no consequence of the theory of the

will that I am advocating.

Causal closure and physicalism

Many contemporary physicalists, who almost always conceive of physical

causality in terms of event causation rather than substance causation,

SUBSTANCE CAUSATION, POWERS, AND HUMAN AGENCY 167

maintain that a principle of causal closure reigns in the physical domain.

Although such a principle has been variously formulated by different philo-

sophers, a commonly favoured formulation would be this: at any time at

which a physical event has a cause, it has a sufficient wholly physical cause—

where, by a ‘sufficient cause’, is meant an event or conjunction of events

which causally necessitates the physical effect in question.16 It might appear

that this principle is logically consistent with everything that I have said so

far about voluntary human action. For one thing, nothing that I have said so

far is inconsistent with the view that volitions just are physical events of a

certain kind. For another, the principle as just stated allows that some

physical events may lack causes altogether, just as I say volitions do.

However, further reflection shows that the principle as just stated is in fact

false, quite independently of anything that I want to say about volitions. The

earlier example of spontaneous radioactive decay shows why, recalling that

any such decay is an event that has no cause whatever. Suppose a radium atom,

R, decays at a certain time t1, and that this event—call it e1—is part of a

sufficient physical cause, at t1, of a later physical event, e2, occurring at the

later time t2. To make matters more concrete, suppose that e2 is the

registering of e1 by a Geiger counter, in the form of a ‘bleep’ emitted by

the machine. Now consider an earlier time, t0, preceding that of the decay

e1. At t0 there will surely be some physical events that are causal antecedents

of the bleep e2. After all, amongst the physical events at t1 which jointly

constitute a sufficient physical cause of the bleep e2 there will be some that

have antecedent physical causes, even though the decay e1 does not.

(Clearly, it would be ridiculous to suppose that the decay e1 was, all by

itself, a sufficient cause of the bleep e2.) Hence, by the transitivity of

causation, the bleep e2 has some physical causes which occur earlier than t1,

such as at t0. However, we know that all of e2’s causes at t0 cannot jointly

constitute a sufficient cause of e2, because they cannot jointly causally neces-

sitate e2. This is quite simply because at t0 it is still causally possible for the

decay e1 not to occur and hence for the bleep e2 not to occur. Thus, t0 is a

time at which the bleep e2 has a cause, but not a sufficient wholly physical

cause. Consequently, the principle of causal closure as just stated is false.

16 Compare Kim (1993), 280. I discuss various forms of causal closure principle, including this one, in

Lowe (2008), Chapter 2.

168 E. J . LOWE

The significance of this finding, as far as I am concerned, is the following.

Physicalists often appeal to the causal closure principle, usually in conjunc-

tion with a principle ruling out systematic causal overdetermination, in

order to argue against dualists who claim that mental events and states are

not identical with, nor even ‘realized by’, physical events and states. And,

although I have not so far proclaimed myself to be in favour of dualism in

the present essay, I am in fact firmly of the opinion that it makes no sense to

think of volitions as being either identical with or ‘realized by’ physical

events, such as neural events in the brains of human agents. I do not think

that the human will is a physical power at all, even if human agents need to

have certain physical powers in order to have an efficacious will. Since,

however, we have seen that the principle of causal closure, at least in the

relatively strong form just described, is simply false, it does not present a

threat to the kind of dualism concerning volitions to which I have now

committed myself. I fully acknowledge that this kind of dualism is inconsist-

ent with the causal closure principle as just stated. After all, according to my

account, a volition may be part of a sufficient cause of a subsequent bodily

movement, so that at the time of the volition the bodily movement has a

cause, but typically not a sufficient wholly physical cause, since the volition

itself, by my account, is not a physical event. But since, as we have seen, the

causal closure principle in this form is false in any case, it need be of no

concern to me that my account is inconsistent with it. I strongly suspect that

the same may be said with regard to other variants of the principle which do

not simply beg the question against dualism by ruling it out in the very way

they are formulated. (Incidentally, although I have couched all these

remarks in terms of event causation, this is only because physicalists typically

speak in these terms. The same points could be rephrased in the language of

substance causation that I regard as preferable.)

The rational undeniability of free will

It is pertinent to ask at this point what reason we have for believing

ourselves to possess a rational, spontaneous, two-way power of will of

the sort described in this essay. I have tried to show that our possession of

such a will is metaphysically consistent with current physical science, since

current physical science already admits the existence of spontaneous powers.

SUBSTANCE CAUSATION, POWERS, AND HUMAN AGENCY 169

Moreover, physical science has nothing to say, one way or another, about

matters of rationality, since these are normative matters, whereas physical

science is entirely factual or descriptive in its purview and content. I also

consider that my kind of account is phenomenologically plausible. We seem to

possess a rational will that we can freely exercise in deliberative action—

why else would such a doctrine have proved so popular for so many

centuries? Physical determinists will contend, of course, that this is just an

illusion. But we know that current physical science is not on the side of

physical determinism. Can we, though, say anything even more compelling

in favour of the current proposal? I believe we can. I believe we can show

that it would just be irrational to deny that we possess a will of the kind that

I have described. This is not to say that it is metaphysically impossible for us to

lack such a will, only that it is logically impossible for us to believe that we

lack it and still qualify as genuinely rational beings.

The reason for this is as follows. Our voluntary, deliberative actions are

not confined only to our bodily movements, but include also our processes

of thought and reasoning. Now, in any process of reasoning, the mind

draws conclusions from certain premises that it entertains. But for the mind

to draw a conclusion rationally from certain premises, it must draw that

conclusion in virtue of apprehending the support that the premises confer

upon the conclusion. That is to say, the mind—or, more properly speaking,

the personwhose mind it is—must perform the mental action of drawing the

conclusion ‘in the light of ’ the reasons that the premises supply in favour of

that conclusion. The mind must respond to those reasons precisely as reasons

favouring the conclusion. However, if the movement of the mind from

premises to conclusion were purely causally determined, then the mind

could not be said to embrace the conclusion in virtue of apprehending the

support that the premises confer upon the conclusion. For the notion of

‘support’ in this sense is a thoroughly normative one and hence not one that

can be captured by the content of any universal causal law supposedly

governing movements of the mind. Causal laws are merely descriptive of

the domain of non-normative facts. No such law can intelligibly be said to

regulate movements of the mind in accordance with the demands of

rationality, because what those demands are is itself an essentially contestable

matter for rational debate between rational beings. To put it another way: no

cause brings about its effects in recognition of some rational consideration’s

170 E. J . LOWE

favouring the obtaining of those effects.17 (Of course, standard agent caus-

alists will presumably deny this, but can do so only because they invoke a sui

generis notion of agent causation which has no connection with the notion

of a universal causal law.) Only if we are free to draw conclusions because

we apprehend the cogency of the arguments in which they figure can we

truly be said to be rational thinkers—as opposed to mere computing

machines, which may mechanically replicate certain patterns of reasoning

deemed valid by their designers, but without ever actually conducting such

reasoning by and for themselves.

Those who deny that we have a rational free will of the kind I propose are

therefore faced with the following dilemma.18 If what they say is true, then

the movements of their minds that have led them to say it are simply

consequences of certain causal laws governing those movements. Hence,

these movements of their minds may at most replicate valid reasoning but do

not and cannot constitute it. Consequently, their belief in the conclusion—

that we have no rational free will—is not a rationally held belief. On the

other hand, if what they say is false, and unsupported for the reasons that

I have advanced, then again their belief is not a rationally held one, because

they hold it in defiance of the reasons that count against it and only because

they are party to a deterministic dogma which has no foundation in

empirical science. Either way, then, their belief that we lack a rational free

will cannot be a rationally held one. But if it is not rational to believe that p,

then one should not believe that p. Consequently, we should not believe

that we lack a rational free will: we should, on the contrary, believe that we

do possess one.

I have no doubt, of course, that my opponents will want to challenge this

conclusion vigorously and seek every means available to question the

cogency of my argument for it. But, ironically enough, that merely serves

to emphasize its cogency. For in reacting in this fashion my opponents will

precisely be evincing their implicit conviction that they really are rational

thinkers, capable of responding to an argument on its rational merits, not

merely as a result of the playing out of causal laws governing the movements

of their minds.

17 As I have elsewhere expressed it, causation is ‘blind to reason’. See Lowe (2008), 9.18 Compare Malcolm (1968).

SUBSTANCE CAUSATION, POWERS, AND HUMAN AGENCY 171

References

Broad, C. D. (1952). ‘Determinism, Indeterminism, and Libertarianism’. In Ethics

and the History of Philosophy: Selected Essays. London: Routledge & Kegan Paul:

195–217.

Dancy, J. (2000). Practical Reality. Oxford: Oxford University Press.

Kim, J. (1993). Supervenience and Mind: Selected Philosophical Essays. Cambridge:

Cambridge University Press.

Locke, J. (1975 [1690]).AnEssay Concerning HumanUnderstanding, ed. P. H.Nidditch.

Oxford: Clarendon Press.

Lowe, E. J. (1998). The Possibility of Metaphysics: Substance, Identity, and Time.

Oxford: Clarendon Press.

—— (2008). Personal Agency: The Metaphysics of Mind and Action. Oxford: Oxford

University Press.

—— (2010). ‘On the Individuation of Powers’. In A. Marmodoro (ed.), The

Metaphysics of Powers: Their Grounding and Their Manifestations. London and

New York: Routledge: 8–26.

Malcolm, N. (1968). ‘The Conceivability of Mechanism’. Philosophical Review, 77:

45–72.

Molnar, G. (2003). Powers: A Study in Metaphysics. Oxford: Oxford University

Press.

O’Connor, T. (2000). Persons and Causes: The Metaphysics of Free Will. New York:

Oxford University Press.

Strawson, P. F. (1959). Individuals: An Essay in Descriptive Metaphysics. London:

Methuen.

van Inwagen, P. (2002). ‘Free Will Remains a Mystery’. In R. Kane (ed.), The

Oxford Handbook of Free Will. Oxford: Oxford University Press: 158–79.

172 E. J . LOWE

7

Agent Causation in aNeo-Aristotelian Metaphysics

JONATHAN D. JACOBS AND

TIMOTHY O’CONNOR

Freedom and moral responsibility have one foot in the practical realm

of human affairs and the other in the esoteric realm of fundamental

metaphysics—or so we believe. This has been denied, especially in the

metaphysics-bashing era occupying the first two-thirds or so of the twenti-

eth century, traces of which linger in the present day. But the reasons for

this denial seem to us quite implausible. Certainly, the argument for the

general bankruptcy of metaphysics has been soundly discredited. Arguments

from Strawson and others that our moral practices are too deeply embedded

in human life to rest on anything as tenuous as a metaphysical doctrine far

from the thoughts of ordinary people would seem to prove too much: we

can easily imagine fantastic scenarios far from the thoughts of ordinary

people—involving, say, alien manipulation or massive deception—that, if

true, would clearly undermine claims to freedom and responsibility. For still

other philosophers, the separation of the moral life from (some) metaphys-

ical issues is prescriptive, not descriptive: it is a recommendation that we

revise ordinary moral thought by severing its allegedly problematic links to

metaphysics. (Some philosophers appear to hover undecided between such

a prescriptive project and a Strawsonian descriptive claim.) We suspect that

the prospects of retaining the binding force of ordinary moral thought, were

such a reconceived moral practice widely embraced, are bleak. A transition

to something closer to moral nihilism seems at least as likely. In any case, our

interest here is in descriptive metaphysics, not revisionary.

To say as we do that freedom and moral responsibility have a partly

metaphysical character is not to suggest that they can be had only if some

highly specific version of a particular metaphysical framework is correct.

Instead, we suggest in what follows, it is a broadly neo-Humean metaphysics

that is not hospitable to freedom (for reasons distinctive to the metaphysics),

while a broadly neo-Aristotelian metaphysics is. But we also think (and it is

the main aim of our paper to show) that different versions of the neo-

Aristotelian metaphysics lead to rather different metaphysical accounts of

free and responsible action. Specifically, we will argue that (1) the most

satisfactory account of human freedom within the broadly neo-Aristotelian

metaphysics is agent-causal, but that (2) two different versions of the general

metaphysics will lead to important differences in the agent-causal account of

freedom. Adjust the details of your general metaphysics, and the details of

your account of freedom are transformed in significant ways. Action theory

cannot properly be pursued in isolation from general metaphysics.

1. Freedom and neo-Humeanism

David Lewis popularized a certain form of neo-Humean metaphysics,

according to which causal facts and the laws of nature are reducible to

facts concerning the global spatiotemporal arrangement of fundamental

natural properties (which we allegedly may conceive in non-dispositional

terms). Roughly, the laws are the best system of generalizations over such

natural facts, where bestness is determined by the optimal balance of

simplicity and “strength” (explanatory power). Causation in turn consists

in a restricted kind of counterfactual dependence of one event on another,

where the counterfactuals are grounded in cross-world similarities.1 There

are well-known problems with counterfactual accounts of causation, but we

will not render any pessimistic verdict here.2 Furthermore, the problem that

1 The locus classicus is Lewis’s article “Causation,” reprinted in 1986. (We note that Lewis allows for

temporally remote causation by defining causal chains in terms of stepwise counterfactual dependencies,

but it is unnecessary to fuss about such details here.)

2 For discussion, see the essays in Collins et al. (2004), which includes ‘Causation as Influence’, in

which Lewis proposed a revision of his theory. Hitchcock (2001) and Woodward (2003) have defendedrather different counterfactual accounts that employ the structural equations framework that was given a

major articulation and development by Pearl (2000). For discussion of these developments, see Menzies

(2008).

174 JONATHAN D. JACOBS AND TIMOTHY O’CONNOR

we do press against a neo-Humean account of free action is not dependent on

a counterfactual theory of causation. It is a problem for any reductive account

of causation, and we discuss Lewis’s picture simply for the sake of concreteness.

Within the neo-Humean framework, intentional agency is naturally under-

stood in terms of the counterfactual dependence of behavior or behavior-

guiding intentions on appropriate beliefs, desires, or intentions the agent had

immediately before and as the behavior occurs. That human beings act is

(nearly!) uncontroversial. That we act freely can more plausibly be questioned.

We assume here that both metaphysical freedom and moral responsibility are

incompatible with causal determinism. Necessary conditions on free actions

include plausible compatibilist constraints (e.g., the absence of strong internal

or external compulsion) and that they are not determined to occur over some

interval terminating in the initiation of the action.

The inclusion of a non-negligible degree of indeterminism in one’s

account of the proximate genesis of free actions is thought by many to give

rise to problems of explanation and control. But questions of explanation and

control are better posed within particular metaphysical frameworks. It seems

to us that if the neo-Humean framework is accepted, indeterminism need

not present a special problem of control. Causation is just counterfactual

co-variation of a certain kind, and the neo-Humean can readily describe a

form of co-variation of motivational factors and behavior that applies to the

indeterministic case.3 (Indeed, this fact has been insufficiently recognized by

compatibilists who have held that something approximating determinism is

necessary for freedom.) We should require only that the objective chance of

the behavior’s occurring would have been much less in the absence of those

factors. Furthermore, the counterfactual dependence of the chance of

behavior on psychological facts with which the agent identifies is all that it

could be for a person freely to form a choice. (Irreducible agent causation, for

example, makes no sense in this metaphysics, so its omission can hardly be

judged a deficiency.4) Hence, a suitably textured, causally indeterministic

theory of free action gives everything that a neo-Humean could sensibly

3 For an excellent discussion of this issue, see Clarke (1995).4 Obviously, we are further assuming, though less contentiously, that whatever broad metaphysical

account of contingent reality is correct for our world will hold for all worlds involving contingent concrete

particulars. It is not the case that some worlds are neo-Aristotelian while others are neo-Humean. Without

this assumption, the neo-Human account might well be deficient on grounds that there is a kind of direct

control of action had by some possible agents though by no agents in a neo-Humean world.

AGENT CAUSATION IN A NEO-ARISTOTELIAN METAPHYSICS 175

want for an account of metaphysical freedom (a fact that is insufficiently

recognized by some agent causationists).5

We defer consideration of what is or is not explainable with respect to

undetermined action, treating it in the context of our preferred libertarian

account of freedom. It is true that, given a position held by some neo-

Humeans (and others as well) that there are special explanatory limits in

indeterministic worlds, there will indeed be a serious problem of explan-

ation facing any indeterministic account of agency. But that position is not

compulsory, and we will suggest below that it is implausible.

In our view, the above neo-Humean account of free agency is founded

on a deeply problematic general thesis of causal reductionism. By taking the

fact of A’s being a cause of B to be a reducible, massively extrinsic relation—

grounded in what occurs elsewhere and elsewhen—we empty the funda-

mental idea that causes “produce” or “bring about” their effects of any clear

content.6 Since agency is a causal notion, this problematic consequence

carries over: on a neo-Humean analysis, the sense in which my beliefs and

desires here and now bring about my present action is at best very weak tea.

A fortiori, extrinsic analyses, on which whether or not psychological factors

are causes of behavior is metaphysically determined in large measure by

what happens in the distant reaches of spacetime, provide a bizarre account

of a free action’s being, as we commonly say, “directly controlled by” the

agent, such that it was “up to her” what she would do in the particular

circumstances.7 Our ordinary sense of control with respect to freedom of

5 Indeed, the extrinsic grounding of particular causal facts in the neo-Humean framework might lead

one to doubt the necessity of indeterminism for freedom. See Beebee and Mele (2002). Unfortunately

for the neo-Humean, this same extrinsicality renders it doubtful as an account of causation generally and

of agency in particular, as we argue immediately below.

6 We should acknowledge that “causation” in folk usage probably cannot be neatly lined up with a

fundamental relation in the world, on any likely metaphysical account. The folk, for example, often

speak of causation by absences, as when one says that Susan’s failing to water her neighbor’s plant caused it

to die. On any plausible metaphysical account, there simply are no absences available to stand in a

fundamental relation. In our view, it is most plausible to suppose that ordinary causal talk only roughly

tracks an important fundamental relation in the world, which, to avoid contentious semantic disputes,

we may call “M-causation.” The folk speak truly (often enough) even when speaking of causation by

absences. But such truths are grounded in facts concerning “positive” circumstances that stand in the M-

causal relation (whose nature we sketch in the next section). Conversation with Gunnar Bjornsson has

helped clarify our own thinking here. There is also a nice discussion of this matter in Ted Sider’sWriting

the Book of the World (2011), 15–16 and 75–6.7 See O’Connor (2009) for a development of this point. Gunnar Bjornsson has pointed out in

discussion that it is open to the neo-Humean to modify her account as follows: our concept of natural

law require there to be some minimal score on the balance of simplicity and strength. In neo-Humean

176 JONATHAN D. JACOBS AND TIMOTHY O’CONNOR

action manifestly points to something that supervenes on the local circum-

stances in which we act—or, at any rate, circumstances much more local

than those thousands of years in the past or future.

2. A Neo-Aristotelian Metaphysics

and Event-Causal Libertarianism

There is more than one path away from the neo-Humean’s causal reduction-

ism. Here we will consider only the path that we favor: a neo-Aristotelian

metaphysics that assigns a central role to primitive causal powers. On this

view, natural properties are, or of necessity confer, causal powers on their

bearers.8While the neo-Humean’s properties are intrinsically inert, the neo-

Aristotelian’s are intrinsically powerful.9 If the neo-Humean world is

ungoverned, since laws are merely descriptions of contingent regularities,

the neo-Aristotelian world is self-governed, since laws are necessary descrip-

tions of the powerful natures of properties.10Neo-Humean causation is a sort

of counterfactual co-variance, but neo-Aristotelian causation is the exercise

of an irreducible causal power.11 The details of the broadly neo-Aristotelian

metaphysics need not concern us here. What are important are the general

ideas, first, that properties are (or confer) primitive causal powers and,

second, that causation is the exercise of such powers.

In order to understand the nature of indeterministic causation within the

neo-Aristotelian metaphysics, it is helpful to contrast it with another sort of

picture that some contemporary philosophers endorse. On the latter, causal

worlds where the patterns in one cosmic neighborhood or compact world-segment (such as the one we

currently occupy) sharply differ from those in others, we should say that the laws themselves vary from

one world segment to another. In this way, we needn’t say that what contingently occurs in very remote

regions of spacetime are needed to fix what actions I bring about (or whether I ever so much as act at all).

We grant that reducing the extent of extrinsicality serves to improve the view. But since it is the very

nature of the view to give an extrinsic account of causation, this move cannot make the implausibility go

away. There are ever so many neo-Humean worlds where memories and seeming historical traces are

radically misleading beyond a short threshold into the past and where the patterns will abruptly change or

simply cease in the very short future.Whether or not these things are in fact so just seems beside the point

when we ask whether a present bodily motion is something that I freely bring about. (Note that the point

concerns metaphysical determination, not epistemic justification.)

8 See, e.g., Shoemaker (1980, 1998), Heil (2003), Mumford (2004), Bird (2007), and Martin (2008).9 For full discussion, see Jacobs (2011).10 See, e.g., Mumford (2004) and Bird (2007).11 See, e.g., Martin (2008), Mumford (2009), and Bird (2010) for discussion.

AGENT CAUSATION IN A NEO-ARISTOTELIAN METAPHYSICS 177

indeterminism is thought of as causation of probability. Indeterministic causes,

no less than deterministic ones, are always and uniformly efficacious. They

do not cause the undetermined outcome of an indeterministic process, but

instead cause the outcome’s objective chance of occurring (generally by

raising the probability, though in certain kinds of cases the probability may

actually be lowered). Beyond helping to fix the prior chances of an event,

there is nothing more that a cause does. Where the chance is 1, the cause

suffices for the effect, and so we naturally, if misleadingly, say that it brings

about the effect. But strictly speaking, nothing brings about the effect,

whether the chance is 1 or less than 1. Only the prior chance is brought

about.

This causation of probability view is perhaps assumed (consciously or not)

in objections to the intelligibility of indeterministic agency. If an outcome is

not brought about by anything, it’s hard to see how it can be something that

the agent controls and that we may fully explain in terms of her reasons for

acting. But we should reject the causation of probability interpretation in favor

of a probability of causation alternative precisely because the former makes the

occurrences of events in indeterministic worlds utterly mysterious. There is

no reason within a causal powers metaphysics to suppose that causes must

always produce their characteristic effects, so that in indeterministic scen-

arios we have to resort to the fiction of regular causings of objective

chances. We should suppose instead that indeterministic causes produce

their effects though they need not have done so: they are propensities towards

a plurality of possible effects. They are sufficient for each of them only in the

sense that they are all that is needed, not in the sense that they are a causally

sufficient condition.12 Every indeterministic event is produced, though

none is necessitated. Causation, whether deterministic or indeterministic,

is a singular relation—the very same relation. The prior probability of one

event’s causing another (with limit case of 1) is simply a measure of the

strength of its (single-case) propensity to do so, which helps to fix applicable

laws of nature.

Let us apply this understanding to an indeterministic account of human

free action that, like the neo-Humean account above, is rooted in a causal

theory of action generally. According to it, when an agent freely acts, her

web of motivational states is jointly disposed towards two or more choices,

12 Anscombe (1971) famously develops this point.

178 JONATHAN D. JACOBS AND TIMOTHY O’CONNOR

to varying degrees. Whichever choice is made, it will have been caused by

some relevant motivation of the agent, a motivation with which she

identifies.13 The exercise of agent-control consists in the causal efficacy of

one’s motivations, and freedom further requires the openness of the future

to (or consistency of the past and laws with) a plurality of specific outcomes.

Such is a plain vanilla version of event-causal libertarianism.14

While agents, on this account, do not have any less control over what

they do than agents in a corresponding deterministic scenario, they also do

not have more. Indeterminism in the causal link between motivations and

choice opens up a plurality of alternatives unavailable on determinism, but

the agent does not seem to settle which of the options is taken in a sense

robust enough for the agent to be morally responsible. Autonomous

control seems to require more than compatibilist control plus plural alter-

natives. Consider two event-causal libertarian universes, whose histories

have been precisely the same until a time at which two intrinsically

identical agents (including psychological propensities towards the same

possible choices with the same degrees of strength) make diverging

choices. It does not seem correct to say that it was up to the respective

agents, something that they were individually responsible for, that one

chose the path of insult and the other that of gracious forbearance. It’s not

that choices in these worlds would be “freakish,” the “result of pure

chance,” and so not something that the agents in any sense did. It’s merely

that the control that is exercised is of an insufficient variety to ground

robust freedom and responsibility.

But this objection presupposes the intelligibility of a stronger, more

robust variety of control. Unlike on the neo-Humean metaphysics, there

does seem to be space on the neo-Aristotelian account for such an alterna-

tive, as we will now show.

13 The condition that the causing motivation be one with which the agent “identifies” is intended to

handle possible cases where an agent might be subject to a powerful and perhaps momentary “alien”

desire. We needn’t concern ourselves here with different accounts of this notion of “identification.”

14 The foremost recent defender of this theory, Kane (1996), augments the account with further

conditions on the process by which reasons result in choices. Ekstrom (2000) locates the requisite

indeterminism in a special subset of actions—those in which an agent critically evaluates her own

conception of the good and comes thereby to have certain preferences that regulate ordinary actions.

These proposals are interesting and it is worth considering the issues that they raise in their own rights.

But they do not, in our judgment, suffice to answer the fundamental concern with event-causal

libertarianism that we raise immediately below.

AGENT CAUSATION IN A NEO-ARISTOTELIAN METAPHYSICS 179

3. The Standard Agent-Causal Alternative

Agent causalists maintain that freedom requires a distinct, enhanced kind

of control from the causal efficacy of internal states with which one

identifies. Responsibility-grounding control resides in an indeterministic,

ontologically fundamental causation of a choice or action-guiding inten-

tion by the agent. Taking a feature as a metaphysical primitive is a reliable

way to ensure that one’s overall theory really does allow for the feature,

instead of offering a pale substitute in the manner of various implausible

reductionisms. And the longstanding difficulty of giving a plausible analy-

sis of our pre-theoretical notion of autonomous control suggests that the

gambit of primitive posit is not simply absurd.

We should be careful to distinguish the agent causationist’s position from

that of non-causalists (e.g., Ginet 1990, Goetz 1988, McCann 1998, and

Pink 2004). Both positions agree that autonomous control rests on a

primitive capacity to form intentions (or volitions, according to the theo-

rist’s preference). But the agent causationist insists that this capacity is—and

can only be—causal in nature. The non-causalist, by contrast, ascribes

‘active power’ or ‘the power of choice’ to the agent while insisting that

these terms are to be understood non-causally. However, it is unclear to us

what this means. It seems to us that the term “power” is being misappropri-

ated for rhetorical purposes. Better that these theorists simply say that

nothing causes free choices or volitions but that, notwithstanding, which

choice is made is controlled by the agent, in virtue of the fact that the choice

is his.15 Such a statement is clearer—though clearly false, in our estimation.

In assessing the agent-causal account, we need to consider the role of the

agent’s motivational states in the production of their undetermined choices.

Randolph Clarke (2003) proposes an “integrationist” account on which free

actions are caused both by the agent (qua substance cause) and by certain of

the agent’s motivational states (qua event cause). Clarke proposes that, in the

presence of a “live” agent-causal capacity, it is a law of nature that:

15 It is worth noting that defenders of libertarian accounts of freedom that are ostensibly event causal

sometimes respond to the problem of control by emphasizing not that the choice is caused by the agent’s

reasons but simply that it is “his”—it occurs within the agent. We think that the tendency of event

causalists when pressed to shift between a causal and a non-causal, “ownership” account of control is

revealing.

180 JONATHAN D. JACOBS AND TIMOTHY O’CONNOR

(a) whatever action is performed will be caused by the agent,

(b) a particular reason will cause an action only if the agent causes it, and

(c) the agent will cause an action only if some corresponding reason also

causes it.

It seems to us that, absent further explanation, a lawful and symmetrical

causal yoking of this sort is mysterious. Surely one or the other causal factor

will be in the driver’s seat (given, as Clarke says, that they are not each partial

causes). And we want there to be at least one sort of explanatory asymmetry:

it is because the agent had those reasons that he (qua agent cause) caused the

action that he did, not the other way around. Yet Clarke can’t say that the

state of having those reasons indeterministically brought about the agent-causal

event on pain of making indeterministic causation by reasons more funda-

mental than (because prior to) agent causation in the production of an action.

The resulting account would seem to offer no improvement over a simple

event-causal account that dispenses with primitive agent causation.16

O’Connor (2008) suggests a different account of the way that reasons

influence agent-causal actions. He suggests that while agent-causal events

are unproduced by other events, they are probabilistically structured by myriad

factors, especially the agent’s own motivational states. As a result, agents

have a continuously evolving, objective propensity to cause intentions to

act in ways they take to be suited to their ends. More carefully, the idea is

that motivational states act causally on the persisting capacity of an agent

freely to form an intention to act, altering the objective strength of (or

generating) the dispositions the agent has to form specific intentions within

certain intervals. The influence of reasons so conceived is not unlike how

things go according to the causation of probability interpretation of indeter-

ministic causation. However, O’Connor’s account of the influence of

reasons on agent-causal choices is not offered as an account of the nature

of causation itself, and it does not have the absurd consequence that nothing

brings about the specific outcome of an indeterministic process. This

account has the advantage (over Clarke’s integrationist account) of offering

a unified picture of the flow of causal influence, and it does so without

sacrificing the core agent-causal commitment to the exercise of a power that

is not itself in turn produced by previous events.

16 For further discussion, see O’Connor and Churchill (2006).

AGENT CAUSATION IN A NEO-ARISTOTELIAN METAPHYSICS 181

Note that on this neo-Aristotelian framework, having reasons (under-

stood as motivational states) is having certain kinds of causal powers.17 So, as

an agent first comes to have reasons for a course of action that was

previously (subjectively) unmotivated, she comes to have new powers of

choice and action. In a case where she comes to have additional reasons for

an already motivated action type, the power so to act is not altered.

However, insofar as the different reasons also motivate different actions of

however fine-grained a type, new powers are thereby acquired. And

distinct reasons must have the potential to motivate somewhat different

action types under at least some possible circumstances, for otherwise they

could not be individuated within the causal powers framework.

On O’Connor’s account of agent-causal power, there is one persistent

agent-causal power, a power to form an intention to act. And an agent with

that power can have differing specific propensities so to act, depending on

what reasons the agent has. The reasons are, in part, powers to act on the

persistent agent-causal power, to alter its strength. It is worth noting,

however, that the causal powers metaphysics does not, by itself, entail

O’Connor’s view of agent-causal power. Indeed, on some versions of the

causal powers metaphysics, specific propensities are essential to causal

powers.18 On this way of thinking about it, when a substance has a power

to do some action, A, the power is a power to do A with some specific

probability in specific circumstances. Were the probabilities different, even

slightly, it would be a different power. Hence, on this view of powers,

reasons are powers to bring about various agent-causal powers, each confer-

ring specific probabilistic tendencies towards specific outcomes in specific

circumstances. Speaking loosely, one has “agent-causal power.” In strict

truth, however, there are a family of related agent-causal powers; all such

powers are similar in being powers of the agent to bring about an intention

with some specific content. This issue, however, is orthogonal to our

central topic, as either view—O’Connor’s view of a persistent agent-causal

17 “Reasons” can refer to normative reasons, or the conditions (generally external to the agent’s

psychological states) that rationally or morally justify a particular course of action for an agent in a

given circumstance, whether or not the course of action is taken or the agent even acknowledges the

existence of the reason. “Reasons” can also refer to motivational reasons, the agent’s own reasons for doing

what he does, wise or foolish as may be. In this latter sense, having a reason is a psychological state or set

of states (such as beliefs, desires, and intentions) that motivates the agent towards and potentially explains

certain courses of action. It is this latter, motivational sense of “reasons” that is in view here.

18 See, e.g., Jacobs (2011).

182 JONATHAN D. JACOBS AND TIMOTHY O’CONNOR

power, or the alternative view of a family of related agent-causal powers—

can be accepted in both the above causal powers metaphysics and the

modified causal powers metaphysics to which we turn now.

4. A Modified Causal Powers Metaphysics

On the above causal powers metaphysics, the causes of events that do not

involve agents, and indeed of many events that do involve agents, are

events. The addition of agent causation to such a picture therefore involves

the addition of a new kind of causation. But there is an alternative account

of causation that fits well within a causal powers metaphysics, on which the

causes of all events are substances. All causation, on this view, is substance

causation. E. J. Lowe (2008) argues for this view roughly in the following

way. Causation is the exercise or manifestation of a power. The cause is the

thing that has the power. But only substances have powers. Therefore, only

substances are causes. Events are the having of a power by substances, and

those powers are exercised or manifested by substances.

On Lowe’s version of the substance causation view, substances cause

effects by manifesting a power. But whenever a substance causes an effect, it

does so ultimately by manifesting a non-causal power. The rock caused the

tree’s breaking, by rolling into it. And the rolling of the rock is a manifest-

ation of the rock’s non-causal power to roll. In other words, the rolling of

the rock does not consist in the rock’s causing anything. It consists in, say,

the rock’s changing position.

In many cases, such manifestations of non-causal powers may themselves

be effects. Not so with what Lowe calls spontaneous powers: when a sub-

stance manifests its spontaneous power, it is not caused to do so by anything,

and its manifestation of the spontaneous power does not consist in its causing

anything. Such is Lowe’s account of both radioactive decay and free action. In

both cases, the fundamental source of action is the exercise of a non-causal,

spontaneous power. That is to say, the fundamental action, by which the

agent or atom does cause something, when it does, is an event that is neither a

causing nor caused. In the case of free action, it is a willing. A willing is not a

causing—that is, it does not consist in the substance causing an effect. And it is

not caused by anything. Still, the agent does cause something, say, raising her

hand, by willing to raise her hand, when her willing is effective.

AGENT CAUSATION IN A NEO-ARISTOTELIAN METAPHYSICS 183

In our judgment, Lowe’s analysis unhelpfully complicates the substance-

causal powers metaphysics. By introducing a fundamental distinction

between the manifesting of a power and that by which a substance manifests

the power, he invites a question concerning the nature of the “by” relation

that has no satisfactory answer that we can see. And it is unclear how this

posit offers any improvement over a simpler analysis on which a substance’s

causing an effect simply is its exercising a causal power. What’s more,

Lowe’s view leads to an even clearer problem in the account of freedom.

For central to the account is a non-causal sort of power, something that we

encountered above in discussing non-causalism. Lowe’s uncaused volitions

appear to be no different, intrinsically, from the non-causalist’s volitions or

choices. As there, so too here: it just seems misleading to call the spontan-

eous occurrence of such events “the exercise of non-causal powers,” given

that the events have no causes and no internal causal structure. We think it

doubtful that control can be understood in non-causal terms.

For these reasons, we find the following analysis to be preferable: a

substance’s having a property is its having a causal power of a specific sort.

A substance’s causing an effect is its manifesting such a power or its

co-manifesting a power with other substances. (Note that on this view,

causation is non-transitive, since causes are substances and effects are

events.) In some possible cases, given the totality of properties had by an

object and its situation, the effect is causally determined to occur. The

conjunction of interacting powers yields a probability of 1 that the substance

or substances will cause that very type of effect. In other possible cases, the

effect is causally undetermined. Here, the exercise of more than one power

is possible and presumably each is probable to some specific degree.

The modified neo-Aristotelian ontology, then, is one on which substances

have powers, and all and only substances are causes of effects.When a substance

causes an effect, it exercises its power to do so—that is, a substance’s causing of

an effect is identical to its exercising its power to bring the effect about.

5. Agent-Causal Libertarianism and the NewOntology

Some of the basic claims of the agent causationist concerning freedom carry

over from the event-causal to the substance-causal powers metaphysics.

First, agents can be literal causes of actions. (Better: actions are agent-causings

184 JONATHAN D. JACOBS AND TIMOTHY O’CONNOR

of intentions.) Second, nothing produces an agent’s causing of an intention.

(Since on this metaphysics, in general, nothing produces any causing of an

event by a substance, whether its activity is determined or not.) Finally, the

motivation for incompatibilism about freedom, such as it is, remains.19

However, the substance-causal powers metaphysics forces some changes

to an agent-causal account of freedom. On the event-causal powers meta-

physics, reasons either structure the agent-causal power, in O’Connor’s

sense, or they cause the agent to have the specific agent-causal power she

has. What do reasons do, according to the modified framework now under

discussion? Strictly speaking, nothing. The agent (and the particles that

compose her—see below) do things. Some of those doings are the exercise

of a fundamental agent-causal power to form intentions to act which the

agent has in part because she has the reasons she does. So, reasons are

causally relevant. She wouldn’t have done what she did, and wouldn’t have

been capable of doing what she did, were she not to have those (or other)

reasons so to act. But, strictly speaking, reasons are not causes, since reasons

are not substances, and only substances are causes.

As noted in the previous section, O’Connor (2000, 2008) proposes from

within an event-causal powers framework that what reasons do is to confer

“carried propensities” or “tendencies” on a generic and persisting agent-

causal capacity. This causation of probability has to be taken on board as a

kind of influence that differs from the probability of causation that charac-

terizes non-intentional indeterministic causes. An advantage of the frame-

work now under consideration is thus ideological simplification on this point.

We also have the further ideological simplification resulting from the fact

that agent causation is not a fundamentally distinct kind of causation. What

is distinctive about agent causation among other varieties of substance

causation on this view is merely that the cause is conscious, intentional,

and freely chooses the ends for which it will act.20 As we see it, each of these

19 We note the interesting fact that the substance-causal metaphysics enables one to make good sense

of Markosian’s (1999) claim that agent causation is consistent with compatibilism. (Whereas it is not clear

that it is compatible with the more common event-causal powers metaphysics.) We disagree with

Markosian, however, that compatibilism becomes more attractive once one endorses an agent-causal

theory of free action. Once conceptual space is opened up for determined agent causation, the question

of whether there is a substantive distinction between being produced by me and being freely produced by

me is on the table. As we see it, a suitably formulated version of the Consequence Argument for

incompatibilism is compelling (see O’Connor 2000, Ch. 1).20 Ruth Groff suggested a similar view in conversation.

AGENT CAUSATION IN A NEO-ARISTOTELIAN METAPHYSICS 185

distinctive sorts of capacity is ontologically fundamental. (They are intercon-

nected in certain ways, with the last, in particular, presupposing the first two.)

“Fundamental” here means that having specific instances of these types of

capacity does not consist in (and is not “constituted” or “realized” by) the

agent’s having some set of other capacities, or in the agent’s parts’ having

certain capacities and standing in certain relations. Given (what we take to be)

the fact that human agents are composed systems, we must think of human

persons as ontologically emergent substances. Among the powers of our

fundamental parts are powers collectively to cause system-level properties/

powers, powers that are sustained as long as the system (the person) retains the

requisite form of organized complexity. As the bearers of fundamental powers,

the person is an ontologically fundamental, albeit composed substance.21

Agents freely act because they (literally) cause their effects with the conscious

aim of attaining certain ends, and their doing so is not settled by features of

their situation (whether external or internal) up to the time of the action.

We have, then, two versions of agent causalism, one embedded with an

event-causal, neo-Aristotelian metaphysics, the other within a substance-

causal, neo-Aristotelian metaphysics. On the first view, agent-causal power

is a power to cause things in a fundamentally distinct sort of way, involving

agent causation rather than the typical event causation. The reasons that an

agent has are powers to structure her agent-causal power, either by causally

influencing the strength of a persisting agent-causal power or by bringing

about a new, slightly different agent-causal power. On the second view,

when an agent causes an event, it does not involve a unique sort of

causation. All causation is substance causation. Rather, the uniqueness

comes from the sort of substance, a conscious substance influenced by

reasons, and the sort of effect, an intention to act for a certain reason.

6. Revisiting the Alleged Problems of Controland Explanation

Some philosophers contend that agent causation, even if coherent, cannot

solve the causal indeterminist’s problems of explanation and control. We

21 See O’Connor and Jacobs (2003) for a detailed account of this picture of ontologically emergent

substances within an event-causal powers metaphysics.

186 JONATHAN D. JACOBS AND TIMOTHY O’CONNOR

consider a typical way that each of these problems is presented against agent

causation and argue that they fail.

Agent Causation and the Problem of Explanation

Consider first the problem of explanation.Where an event is undetermined,

there was some objective chance that an alternative type of event might

have occurred—the alternative’s obtaining had a non-zero causal probabil-

ity in the total set of circumstances. This implies that any cause or causes that

one might cite in a putative explanation of the actual event was consistent

with the occurrence of the alternative. It appears to follow that the cause

cannot explain why the actual event obtained rather than the possible

alternative. From this it might seem to follow that the cause cannot, after

all, ‘fully’ explain the actually occurring event itself, since to fully explain

why an event occurred is inter alia to explain why it occurred rather than any

alternative.

Against this argument, we note that not all causal explanations of events

must be contrastive or imply the availability of contrastive explanations, for

every possible contrast. As Peter Lipton (1990) made clear, a request for a

contrastive explanation (“Why P rather than Q?”) presumes that there is an

explanatory relationship between fact (P) and “foil” (not-Q); it presumes

that the occurrence of P and the non-occurrence of Q can be given a

unifying explanation. But this assumption plainly will not hold for every

such pairing even in a deterministic world—as when the occurrence of

P and the absence of Q are completely unrelated matters. In an indetermin-

istic world, contrastive explanation will also fail (plausibly) wherever P and

Q are mutually exclusive; each had a substantial chance of occurring, and

P was not significantly more probable than Q. But it does not follow that there

can be no explanation of P, or that whatever non-contrastive explanation there may

be of P will be somehow deficient—of a lesser variety of explanation than contrastive

explanation. We explain—really explain—an indeterministic outcome P by

citing and describing the causal factor or factors that brought it about,

including the cause’s objective set of objective propensities and the most

salient, proximate causes of its having such propensities.

The point is a familiar one in scientific explanations of indeterministic

phenomena unrelated to free action. If there are a plurality of possible

outcomes of the interaction of a pair of particles, the particular outcome

AGENT CAUSATION IN A NEO-ARISTOTELIAN METAPHYSICS 187

that obtains has an explanation in terms of propensities of the two particles

which actually were manifested, bringing about that particular result. Once

one understands the indeterministic nature of those propensities and others

that were not, but might have been, manifested on that occasion, one

realizes that there is nothing further to explain about the situation.

We can even explain why there can be no true contrastive explanation of

the fact that P occurred rather than Q by underscoring the indeterministic

nature of the causal source in question. Philosophers in the grip of the

Principle of Sufficient Reason profess mystification at this scenario, but

don’t give any kind of argument. If we grant that there can be indetermin-

istic causal mechanisms (or agents), then deterministic causes are just the

limit case of a continuum of probabilistic causes, and which sorts of explan-

ation it is appropriate to seek depends on which sort of world we occupy.

The application of this general point to our account of human free action

is as follows: as we come through various causes to have motivations to act

in various ways, the interplay of these motivations and other influences

result in an array of propensities to choose and act of varying strengths.

Suppose that while deliberating on what to do on a Saturday afternoon, I am

disposed with a strength of 0.3 to help a friend repair her deck, owing to my

awareness that she wants to get it done soon, could do so more easily if she is

helped, has helped me in similar ways in the past, together with my desires

to be and to be perceived as helpful to her. And suppose that I am disposed

with a strength of 0.7 to watch a football game instead, for the obvious

reasons. I choose to help my friend. Question: Why did I so choose?

Answer: I so chose because I was motivated by my awareness that my friend

wanted to get the job done, etc. That is to say, those beliefs and desires were

the predominant factors determining my propensity so to choose. And that

is a perfectly good explanation of why I chose as I did, even though there

will not be an explanation of why I chose to help rather than to watch the

football game.22

22 It is worth making explicit here how we would respond to Davidson’s (1963) famous challenge to

theories of (free) action according to which reasons are not causes. (As we noted earlier in the text, on

our view, motivational reasons are not causes, strictly speaking.) Davidson asks how, if reasons are not

causes, we will distinguish cases in which an agent has distinct reasons A and B for choice C but does so

for reason A from similar cases in which she does so for B, and from still other cases in which she does so

for both A and B. But our view supplies a ready answer: the agent acts for all and only those reasons she

had which made a non-negligible contribution to her propensity to choose as she did.

188 JONATHAN D. JACOBS AND TIMOTHY O’CONNOR

Agent Causation and the Problem of Control

Recall the argument directed against the causal indeterminist that was based

on a comparison of indeterministic worlds identical up to a certain time at

which intrinsically identical agents make diverging choices. As the causal

indeterminist conceives things, agents don’t cause events—only events do.

Where choices are causally undetermined, some motivational state of the

agent brings about the choice though it was possible that another state have

brought about a different choice. Hence, one set of motivations cause the

choice to insult in world W1, while a distinct set of motivations cause

the choice to refrain in W2. We judged that in this scenario, it is not up to

the respective agents, something that they were individually responsible for,

to make the choices they do. Indeterminism of this sort confers opportunity

but not an enhanced form of control that enables the agent responsibly to

exploit the opportunity.

Some argue that the agent causationist can do no better (Haji 2004; see

also the related, intra-world “rollback” argument in van Inwagen 2000). For

here, too, there is nothing whatsoever about the one agent, right up to the

moment of the choice, that distinguished her from the other, and so nothing

about her that made the difference to what she did. Each had the same

propensity to insult or refrain. Once again, it seems that neither agent

controlled the way their respective cases unfolded in such a way that it

was up to her that she spoke the insult or refrained.

How one replies to this objection depends, in part, on whether one

accepts the event-causal or substance-causal powers metaphysics. On the

former, there is available the very simple reply that the objection fails to take

seriously the concept of agent causation, which is conceived on this meta-

physics as a primitive form of control over undetermined, single-case

outcomes. The agent’s control is exercised not through the efficacy of

prior states of the agent (as on causal theories of action), but in the action

itself. Susan’s causing her intention to publicly insult her opponent is itself

an exercise of control. And since, ex hypothesi, it is quite literally the agent

herself generating the outcome, it is hard to see how the posited form of

control could possibly be improved upon.

But on the substance-causal powers metaphysics, agent causation is not a

special sort of causation, and so cannot be a special, agential form of control

simply by dint of the basic sort of causation manifested. Return, then, to the

AGENT CAUSATION IN A NEO-ARISTOTELIAN METAPHYSICS 189

two indeterministic worlds, W1 and W2, identical up to a certain time at

which intrinsically identical agents make diverging choices. And let us

consider three such pairings: a neo-Humean pair of ‘worlds,’ an event-

causal libertarian pair of ‘worlds’, and a purely substance-causal pair of

‘worlds.’ (We put ‘worlds’ in single quotes to mark the fact that it will be

disputed whether any of these descriptions match genuinely possible

worlds.) As on the event-causal powers metaphysics, the proponent of

the substance-causal powers metaphysics can insist on the significance of

the fact that only in the latter pair of worlds are agents literally the causes of

their choices. This fact, construed narrowly, is not sufficient to ground the

claim that agents in these substance-causal worlds but not the others are

responsible for their choices, since it is possible (and indeed common) for

substances to cause their effects without purpose, consciously or otherwise.

But it does mark a relevant and important difference that, when combined

with the fact that the causation is consciously goal-directed, grounds a more

robust variety of control than is possible in either neo-Humean or purely

event-causal neo-Aristotelian worlds. The agent, herself, consciously and

intentionally brought about the effect in both worlds. That, we submit, is a

sufficiently robust form of control to ground freedom and responsibility.

7. Conclusion

We have not here endorsed, let alone argued for, the substance causation

metaphysics. Our aim was only to explore how embracing it would alter the

way that we conceive of metaphysical freedom. There is a more general

moral that we hope will become more widely embraced in action theory

and philosophy of mind: metaphysics matters.23 Debates over reductionist,

epiphenomenalist, or emergentist alternatives concerning both intentional-

ity and consciousness, and over freedom and determinism, turn more on

general metaphysical positions than is commonly acknowledged.24

23 We are not alone in drawing that conclusion. See Beebee and Mele (2002) and the evolving debatebetween Jaegwon Kim and his recent critics over non-reductive physicalist accounts of mental causation.

24 Jonathan Jacobs worked on this project during an NEH summer seminar, “Metaphysics and

Mind,” led by John Heil in 2009. He would like to thank the NEH for its support, and John Heil,

Jonathan Lowe, and the participants in the seminar for their feedback. He would also like to thank the

John Templeton Foundation, for a grant supporting work on this project.

190 JONATHAN D. JACOBS AND TIMOTHY O’CONNOR

References

Anscombe, E. (1971). Causality and Determination: An Inaugural Lecture. Cambridge:

Cambridge University Press.

Beebee,H. andA.R.Mele (2002). ‘HumeanCompatibilism’.Mind, 111(442): 201–23.

Bird, A. (2007). Nature’s Metaphysics: Laws and Properties. Oxford: Oxford Univer-

sity Press.

—— (2010). ‘Causation and the Manifestation of Powers’. In A. Marmodoro (ed.),

Powers: Their Grounding and their Manifestations. Abingdon: Routledge: 160–8.

Clarke, R. (1995). ‘Indeterminism and Control’. American Philosophical Quarterly,

32: 125–38.

—— (2003). Libertarian Accounts of Free Will. New York: Oxford University Press.

Collins, J., N. Hall, and L. Paul (eds.) (2004). Causation and Counterfactuals. Cam-

bridge, MA: MIT Press.

Davidson, D. (1963). ‘Actions, Reasons, andCauses’. Journal of Philosophy,60: 685–700.

Ekstrom, L. W. (2000). Free Will: A Philosophical Study. Boulder, CO: Westview

Press.

Ginet, C. (1990). On Action. Cambridge: Cambridge University Press.

Goetz, S. (1988). ‘A Noncausal Theory of Agency’. Philosophy and Phenomenological

Research, 49: 303–16.

Haji, I. (2004). ‘Active Control, Agent-Causation, and Free Action’. Philosophical

Explorations, 7(2): 131–48.

Heil, J. (2003). From an Ontological Point of View. Oxford: Oxford University Press.

Hitchcock, C. (2001). ‘The Intransitivity of Causation Revealed in Equations and

Graphs’. Journal of Philosophy, 98: 273–99.

Jacobs, J. D. (2011). ‘PowerfulQualities, Not Pure Powers’.TheMonist, 94(1),81–102.

Kane, R. (1996). The Significance of Free Will. New York: Oxford University Press.

Lewis, D. (1986). ‘Causation’. In Philosophical Papers: Volume II. Oxford: Oxford

University Press: 159–213.

Lipton, P. 1990. ‘Contrastive Explanation’. In D. Knowles (ed.), Explanation and its

Limits. Cambridge: Cambridge University Press: 247–66.

Lowe, E. J. (2008). Personal Agency: The Metaphysics of Mind and Action. Oxford:

Oxford University Press.

McCann, H. (1998). The Works of Agency. Ithaca: Cornell University Press.

Markosian, N. (1999). ‘A Compatibilist Version of the Theory of Agent Causation’.

Pacific Philosophical Quarterly, 80(3): 257–77.

Martin, C. B. (2008). The Mind in Nature. Oxford: Oxford University Press.

AGENT CAUSATION IN A NEO-ARISTOTELIAN METAPHYSICS 191

Menzies, P. (2008). ‘Counterfactual Theories of Causation’. The Stanford Encyclo-

pedia of Philosophy. Available at http://plato.stanford.edu/entries/causation-

counterfactual/.

Mumford, S. (2004). Laws in Nature. London: Routledge.

—— (2009). ‘Passing Powers Around’. The Monist, 92(1): 94–111.

O’Connor, T. (2000). Persons and Causes: The Metaphysics of Free Will. New York:

Oxford University Press.

—— (2008). ‘Agent-Causal Power’. In T. Handfield (ed.), Dispositions and Causes.

Oxford: Clarendon Press: 189–214.

—— (2009). ‘The Significance of Persons’. In J. Aguilar and A. A. Buckareff (eds.),

Philosophy of Action: 5 Questions. Birker�d, Denmark: Automatic Press/VIP.

O’Connor, T. and J. R. Churchill (2006). ‘Reasons Explanation and Agent

Control: In Search of an Integrated Account’. Philosophical Topics, 32: 241–54.

O’Connor, T. and J. D. Jacobs (2003). ‘Emergent Individuals’. Philosophical Quar-

terly, 53: 540–55.

Pearl, J. (2000). Causality. Cambridge: Cambridge University Press.

Pink, T. (2004). Free Will: A Very Short Introduction. Oxford: Oxford University

Press.

Shoemaker, S. (1980). ‘Causality and Properties’. In P. van Inwagen (ed.), Time and

Cause: Essays Presented to Richard Taylor. Dordrecht: D. Reidel: 109–36.

—— (1998). ‘Causal and Metaphysical Necessity’. Pacific Philosophical Quarterly,

79(1): 59–77.

Sider, T. (2011). Writing the Book of the World. Oxford: Clarendon Press.

van Inwagen, P. (2000). ‘Free Will Remains a Mystery’. Philosophical Perspectives,

14: 1–20.

Woodward, J. (2003). Making Things Happen: A Theory of Causal Explanation.

Oxford: Oxford University Press.

192 JONATHAN D. JACOBS AND TIMOTHY O’CONNOR

8

Mental Causation and DoublePrevention

S. C. GIBB

Each of the following four claims seems individually plausible and yet they

appear to form an inconsistent set:

(1) Relevance: Mental events are causally relevant in the physical domain.

(2) Closure: Every physical event contains only other physical events in its

transitive causal closure.

(3) Exclusion: As a general rule, events are not causally overdetermined.

(4) Distinctness: Mental events are not physical events.

The apparent inconsistency of these claims gives rise to the problem of

mental causation. Responses to this problem typically provide reasons to

reject one of the claims. Indeed, the various positions in the mental caus-

ation debate can, to a large extent, be distinguished by the claim that they

reject. Hence, eliminativism and epiphenomenalism both reject Relevance.

Interactive substance dualism and anti-physicalist forms of property dualism

typically deny Closure. Most forms of non-reductive physicalism deny (or

disambiguate) Exclusion. Psychophysical reductionism rejects Distinctness.

An alternative way of responding to the problem of mental causation is to

argue that, contrary to appearances, the four claims are in fact consistent.

Indeed, given that the rejection of any one of the claims encounters serious

difficulties, if there were a way of reconciling them, this would seem to be

the most desirable option. E. J. Lowe has presented one way of doing this

which rests upon the idea that the causal role of mental events in the physical

domain is that of making the fact that a causal tree of neural events converge

upon a particular bodily movement non-coincidental (Lowe 1993, 2000,

2008). In this paper I want to sketch an alternative way of reconciling the

claims. The account that I offer is, in many respects, sympathetic to Lowe’s.

Like Lowe’s account, it resolves the problem of mental causation by

providing mental events with a specific causal role in the physical domain.

Moreover, like Lowe, I deny that this causal role is that of initiating any

single physical event or set of physical events in the chains of neurophysio-

logical causation that terminate in bodily movement. Finally, like Lowe,

I consider the resulting account to be one that is anti-physicalist in nature—

according to it, the mental must be something over and above the physical.1

Despite these similarities, there are two crucial differences in our

accounts. First, the causal role that I shall suggest that mental events play

in the physical domain is very different from the one that Lowe proposes.

According to the account that I shall propose, the causal role of mental

events in the physical domain is to serve as ‘double preventers’. Secondly,

unlike Lowe’s account, which is to a large extent neutral between various

accounts of the causal relation, the account of psychophysical causation that

I wish to advance is not. It is an account of psychophysical causation which

emerges from the acceptance of a powers theory of causation—a theory of

causation that is gaining increasing popularity in recent metaphysics.2

}1 outlines the problem of mental causation. }2 provides an account of thepowers theory of causation and examines how it deals with cases of double

prevention. Assuming this theory of causation, }3 argues that an understand-ing of mental events as double preventers in the physical domain provides

a possible way of reconciling Relevance, Closure, Exclusion, and Distinctness.

Finally, }4 considers why the resulting account of mental events as double

preventers is also attractive from a phenomenological point of view.

1. The Problem of Mental Causation

When we consider our relation to the physical world, little seems more

obvious than the claim that mental events are causally relevant in the

1 For the purpose of this paper, one can take a neutral stance between the various forms of anti-

physicalism. That is, one need not choose between substance dualism and full-blooded property dualism.

2 For recent formulations of the powers theory of causation, see Heil (2003), Martin (2008), Mumford

(2009), and Bird (2010). The idea that powers can provide the basis for a theory of causation is also foundin Harre and Madden (1975), Cartwright (1989), Ellis (2001), and Molnar (2003).

194 S . C. GIBB

physical domain. (Note, mental events are here to be taken as things such as

beliefs and desires, that is, as things that have intentional content.) For

example, my desire to catch someone’s attention and my belief that I can

do this by raising my arm seem to be causally responsible for my arm’s raising.

It is precisely because I had this desire and this belief that my arm raised.

Moreover, in normal circumstances, had I not wanted to raise my arm,

my arm would not have raised. The acceptance of Relevance is, however,

problematic if one also accepts Closure, Exclusion, and Distinctness—the

combination of Closure and Exclusion appears to rule out the causal relevance

of mental events in the physical domain, unless, contrary to Distinctness,

mental events are physical events. Before explaining how I think that the

four claims can be reconciled, these additional premises call for a few explana-

tory comments.

I take events to be the causal relata, where an event is the instantiation of a

property by a substance at a time. (For the purpose of this paper, I am not

assuming any particular metaphysical account of substance.) A mental event

is the instantiation of a mental property by a substance at a time. A physical

event is the instantiation of a physical property by a substance at a time.

The identity of events requires the identity of their properties (as well as their

substances and times). Hence, for mental events to be identical with physical

events, mental properties must be identical with physical properties. As

the identity of events requires the identity of their properties, Distinctness—

the claim that mental events are not identical with physical events—

straightforwardly follows from the argument from multiple realizability.

According to this well-known argument, mental properties are multiply

realized by and, hence, cannot be identical with, physical properties.

While my formulation of Exclusion is quite standard, the formulation of

Closure requires some explanation. The formulation of Closure is Lowe’s

(Lowe 2000, 581). By the ‘transitive causal closure’ of event P, Lowe means

the set of events ‘which includes every event which stands in the ancestral of

the “immediate cause” relation to P’. That is, the set of events which

includes ‘the immediate causes of P, the immediate causes of those causes,

the immediate causes of those causes . . . and so on’. GivenClosure, where P is

a physical event, every event in this set must be physical. Now this is a very

strong version of the causal closure principle—indeed, given its strength,

one may wonder whether the mere combination of Closure with Relevance

and Distinctness is inconsistent. However, I have chosen this formulation of

MENTAL CAUSATION AND DOUBLE PREVENTION 195

Closure precisely because of its strength. In what follows, I shall demonstrate

that, even if one allows the physicalist a causal closure principle as strong as

this, Closure, Exclusion, Relevance, and Distinctness can in fact be reconciled.

As noted earlier, my solution to the problem of mental causation has

emerged from my acceptance of a powers theory of causation. It is therefore

this theory of causation to which I now turn.

2. The Powers Theory of Causationand Double Prevention

I shall not attempt to defend the powers theory of causation here for this

would take me too far from the central aims of this paper. Instead, I shall

simply provide one way of outlining this account of causation and consider

its analysis of cases of double prevention.

According to the powers theory of causation, an account of causation can

be provided in terms of powers or dispositions. (I use these terms inter-

changeably.) The powers theory of causation therefore requires a realist

stance towards dispositions. Given this realist stance, all intrinsic properties

are dispositional, where a property is dispositional if, solely in virtue of

instantiating it, a substance possesses a certain power. Because of its fragility

a porcelain vase is disposed to break when dropped on a hard surface.

Because of its sphericity a ball is disposed to roll when placed on a slope.

The realist stance takes this talk ontologically seriously. The power to break

is built in to some property (or set of properties) of the vase, and it is in

virtue of instantiating this property that the vase is disposed to break when

dropped. Similarly, the power to roll is built into some property of the ball,

and it is because the ball instantiates this property that the ball is disposed to

role when placed on a slope.

The realist stance is in opposition to Lewis’ conditional account of

dispositions which claims that the ascription of fragility to the vase is wholly

analysable in terms of one or more statements about the vase’s behaviour in

a set of counterfactual circumstances (Lewis 1997). It is also in opposition to

those that maintain that dispositional properties reduce to the qualitative

properties of their bearers, and thus, for example, those that consider the

vase to be fragile because of its microstructural properties where these

196 S . C. GIBB

are wholly non-dispositional in nature (Armstrong 1997, ch. 5). However,

to maintain a realist stance towards dispositions is not necessarily to maintain

that properties are exhausted by their dispositionality—properties need not

be clusters of powers. It is also consistent with those accounts which claim

that every property is both dispositional and qualitative.3 Here I shall main-

tain a neutral stance between these different forms of realism.

If one is to take dispositions ontologically seriously then it is essential that

one distinguishes between a disposition and its manifestations. (Although

note that the manifestation of a disposition is itself dispositional in nature—

no intrinsic property, according to this approach, is in pure act.) A specific

disposition is either actual or it is not. To be actual a disposition need not be

manifesting any manifestation. Indeed, it need never manifest any manifest-

ation. Thus a porcelain vase that is never dropped is still fragile despite the

fact that it never manifests this fragility, and so is a porcelain vase perman-

ently encased in bubble wrap. Unmanifesting dispositions are not, therefore,

unactualized possibilia—a description which, as Martin has commented, is

more fitting of unmanifested manifestations (Martin 2008, 12).

A particular manifestation of a disposition nearly always depends on the

presence of other dispositions. (This is precisely one of the reasons why a

disposition might never manifest a particular manifestation, as the other

dispositions upon which this manifestation depends might be permanently

absent.) If a vase is fragile, then the breaking of the vase when it is dropped

on a surface depends not only on the fragility of the vase but also the

hardness of the surface. It follows that the breaking of the vase when it is

dropped on a hard surface is a manifestation not only of the vase’s fragility

but also the surface’s hardness. The vase’s fragility and the surface’s hardness

are, in Martin’s words, ‘reciprocal disposition partners’ and the breaking of

the vase is their mutual manifestation (Martin 2008; see also Heil 2003).

I shall assume, along with Martin and others, that a disposition might—and,

indeed, often will—manifest itself differently with different disposition

partners. Thus, for example, while the surface’s hardness and the vase’s

fragility are reciprocal disposition partners for the mutual manifestation

which is the vase breaking, the surface’s hardness and a rubber ball’s

3 For the first approach towards powers, see Shoemaker (1980). For the second, see Martin (2008),Heil (1998) and Heil (2003).

MENTAL CAUSATION AND DOUBLE PREVENTION 197

bounciness are instead reciprocal disposition partners for the mutual mani-

festation which is the ball bouncing.

These considerations about dispositions and their manifestations lend

themselves to an account of causation in terms of the mutual manifestation

of reciprocal disposition partners.4 The porcelain vase being dropped on a

hard surface causes it to break. In such a case, the vase’s breaking is the

mutual manifestation of the vase’s fragility and the surface’s hardness. My

standing on the broken pieces of the vase causes my foot to bleed.

According to the powers theory of causation, my foot’s bleeding is the

mutual manifestation of the porcelain’s sharpness and my foot’s softness.

The rubber ball being dropped on a hard surface causes it to bounce.

Here, the ball’s bouncing is the mutual manifestation of the ball’s bounci-

ness and the surface’s hardness.

Observe that, given this account, the questionable distinction between

‘the cause’ and its ‘background conditions’ should be dispensed with. Given

the distinction between causes and background conditions, the porcelain

vase’s fragility is the cause of the vase’s breaking when it hits the hard

surface, while the surface’s hardness is a mere background condition that

is necessary for this causal relation to take place. The powers theory of

causation rejects such talk. The breaking of the vase is a mutualmanifestation

of the vase’s fragility and the surface’s hardness. Both play an equal role in

contributing to the effect. Neither should be relegated to the background.

This is merely a sketch of one way of formulating the powers theory of

causation—a sketch which obviously needs to be further developed for it to

provide a satisfactory account of causation. However, it is sufficient for the

purpose of this paper, as its primary interest is with the powers theory’s

analysis of double preventers.

Let us therefore turn to the topic of double prevention. Double preven-

tion occurs when an event that would prevent another event from having a

certain effect is itself prevented from doing so. To give an example, let us say

that one of the attractions at a fair is a simple game in which to win the prize

the player must break a glass bottle by hitting it with a ball. But the game is

rigged. Directly in front of the bottle, and blocking the bottle from any ball

that is thrown at it, is a small barrier that is invisible to the player. The barrier

prevents a ball from ever breaking the bottle. Fred is about to take aim at the

4 See Heil (2003) and Martin (2008).

198 S . C. GIBB

bottle and Sally who is running the fairground attraction suddenly takes pity

on him. She presses a button and this destroys the barrier at the moment that

he releases the ball. (For argument’s sake, let’s say that the destruction of the

barrier is also invisible to the player.) Fred’s ball smashes the bottle and he

wins the prize. Sally’s pressing of the button is a double preventer. It

prevents the barrier from preventing the ball breaking the bottle. This can

be represented diagrammatically as follows:5

In Figure 8.1, a is Fred’s throwing of the ball, b is the breaking of the

bottle, y is the barrier being in front of the bottle and x is Sally’s pressing of

the button. An arrow with a solid line depicts a causal relation. A solid line

ending in a dot depicts an inhibitory connection. A broken line ending in a

dot depicts an inhibitory connection that failed to occur. (And, for later

diagrams, a broken line ending in an arrow shall be used to depict a causal

connection that failed to occur.) A circle around a letter signifies the non-

existence of the relevant event.

How should cases of double prevention be analysed given the powers

theory of causation? Well, to answer this, let us start with the question of

how cases of prevention should be analysed given this theory. Just as a

particular manifestation of a disposition typically depends on the presence of

certain reciprocal disposition partners, it also typically depends on the

absence of others. This is because one disposition may be disposed to

prevent the manifestation of another. It may do so in one of two ways.

A disposition may prevent the manifestation of another disposition because

the manifestation of the first disposition leads to the actual loss of the second

disposition. Alternatively, the disposition might be retained but its mani-

festation blocked by the manifestation of the other disposition. The above

example happens to involve both types of prevention. (Although this is

yx

a b

Fig. 8.1

5 To represent this as simply as possible, I use the kind of neuron diagrams that Lewis (1986) does.

MENTAL CAUSATION AND DOUBLE PREVENTION 199

certainly not true for all cases of double prevention.) The glass bottle is

disposed to break if a ball is thrown at it. However, if a barrier is placed in

front of the bottle, this disposition is not manifested. The solidity of the

barrier prevents the mutual manifestation that is the bottle’s fragileness and

the ball’s momentum and hardness. In this particular case the disposition is

retained—the bottle is still fragile—but as a result of another disposition the

bottle is prevented from manifesting its fragility. The other preventer—the

button being pressed—prevents the barrier’s solidity from standing in a

reciprocal disposition partnership with the ball’s hardness by destroying

the barrier and hence its powers.

In cases of double prevention a disposition that is disposed to prevent the

manifestation of another disposition, is itself prevented from doing so by the

presence of a third disposition. Hence, in the above example, the solidity of

the barrier is disposed to prevent the breaking of the bottle, but is itself

prevented from doing so by the pressing of the button. The consequent

breaking of the bottle is a mutual manifestation of a complex set of dispos-

ition partners which include the ball’s momentum, the bottle’s fragility, and

so on.

What is particularly noteworthy about the powers theory of causation’s

analysis of double prevention is that, according to it, double preventer

events are not causes of the event that they have prevented from being

prevented.6 Hence, taking the above example, according to the powers

theory, the button being pressed is not a cause of the breaking of the bottle.

Why is this?

According to the powers theory, the pressing of the button obviously

cannot be an immediate cause of the breaking of the bottle. If causation is the

manifestation of powers, then a cause and its immediate effect must be

spatially and temporally simultaneous or overlapping.7 And, yet, the pressing

of the button and the breaking of the bottle might be both spatially and

temporally remote from one another. Nor, according to the powers theory,

could the pressing of the button be an indirect cause of the breaking of the

bottle. That is, it cannot be the case that the pressing of the button causes the

breaking of the bottle, by causing the destruction of the barrier. To allow

6 For further defence of this claim, see Mumford and Anjum (2009). Mumford and Anjum do not

formulate the powers theory of causation in exactly the way that I do. However, the central points that

they make regarding the powers theory and double prevention still stand.

7 See, for example, Heil (1998), 187, Martin (2008), 46, and Mumford and Anjum (2009), 287.

200 S . C. GIBB

this causal chain of events, one must accept that the destruction of the

barrier is a cause of the bottle’s breaking—more specifically, one must

accept that the absence of the barrier is a cause of the bottle’s breaking.

Absences cannot be causes according to the powers theory—an absence

cannot bear powers and hence cannot be disposed to act in any way.8

Similar considerations apply in the case where disposition D prevents dis-

position D1 from preventing the manifestation of disposition D2 by merely

blocking D1’s manifestation (as opposed to bringing about D1’s non-exist-

ence). Given the power’s theory, D1’s not manifesting itself—an absence of

a manifestation—cannot be a cause of the manifestation of D2. Hence, there

can be no chain of continuous causation from the event involving D to the

event involving D2.

The powers theory of causation’s conclusion that double preventers are

not causes of the event that they prevent from being prevented differs from

the one drawn by many of the other theories of causation that are dominant

in the philosophy of causation. Hence, to give but one example, according

to the counterfactual theory of causation, Sally’s pressing of the button is a

cause of the breaking of the bottle because there is a chain of counterfactual

dependence linking the two events. If Sally hadn’t pressed the button, then

the barrier wouldn’t have been destroyed and if the barrier hadn’t been

destroyed then the bottle wouldn’t have broken.9

The fact that the powers theory of causation does not count double

prevention as causation is arguably one of its advantages. If double preven-

tion is causation then this has various worrying consequences. Mumford

and Anjum observe that it requires one to accept, not only that there is wide

scale macro-causation at a distance and that absences are causes, but also that

causation is not an intrinsic matter (Mumford and Anjum 2009, 280). That

said, to leave double preventers out of the causal story would be to miss

something crucial out. Isn’t it precisely because Sally pressed the button that

Fred’s ball broke the bottle? An explanation of why Fred’s ball broke the

8 See, for example, Martin (1996), 64.9 My suggestion is not that the powers theory of causation is the only theory of causation that

concludes that double preventers are not causes of the event that they prevent from being prevented.

Hence, for example, the energy transference theory of causation also shares this conclusion because there

is no transfer of energy between the relevant events. (Elsewhere I argue that the energy transference

theory of causation should not provide the basis for a theory of mental causation, as it provides an

account of causation that is physically biased. See Gibb (2010).)

MENTAL CAUSATION AND DOUBLE PREVENTION 201

bottle which focused only on the ball, the bottle, and the throwing of the

ball would be incomplete. Why, according to the powers theory of caus-

ation, is this? For this theory of causation to be plausible, a positive account

of the role of double preventers needs to be provided. Here, I shall offer my

account which hinges upon the distinction between causing an event and

permitting an event to be caused.

Although, according to the powers theory of causation, a double pre-

venter event does not cause the event that it prevents from being prevented,

it permits (or, in other words, allows) the event to be caused. This role is

an objective one, not a merely explanatory one—the fact that a further

event is required to permit the relevant causal relation to take place is quite

independent of our attitudes and interests. Hence, for example, although

Sally’s pressing of the button is not a cause of the breaking of the

bottle, Sally’s pressing of the button permits the breaking of the bottle. It

permits the breaking of the bottle by permitting a causal relation involving

Fred’s ball and the bottle to take place. It permits the causal relation to take

place by causing something else—the destruction of the barrier.

This grounds the counterfactual dependence relation between Sally’s

pressing of the button and the breaking of the bottle—if Sally had not

pressed the button, and hence not permitted Fred’s ball to hit the bottle,

then the bottle would not have broken.10

It also explains why Sally’s pressing of the button should appear in an

explanation of why Fred’s ball broke the bottle. Sally’s pressing of the

button plays an essential role in its occurrence. Indeed the role of an

event that permits a cause to bring about an effect is no less important

than the role of the cause.11

10 Note, therefore, that given my account, causes are not always sufficient for their effects. Take all of

the events that are, in this particular instance, causes of the breaking of the bottle. These are not

collectively sufficient for the breaking of the bottle. In other words, given that all of these causes exist,

it does not follow that the breaking of the bottle occurs. The double preventer event—Sally’s pressing of

the button—must also exist for the breaking of the bottle to occur. But the double preventer event is not

a cause of the effect. As a consequence, given this account, the formulation of the causal closure principle

according to which every physical effect has a sufficient physical cause is misconceived because of its

underlying assumption that every physical effect has a sufficient cause. Note that the acceptance of

Closure does not involve any such assumption.

11 I should emphasize that I am not suggesting that this gives rise to two concepts of causation, which

is a conclusion that Hall (2004) comes to in his discussion of double prevention, distinguishing between

causation as production and causation as counterfactual dependence. According to my account there is

only one concept of causation and it is that described by the powers theory of causation. To permit an

event to be caused is not to cause it.

202 S . C. GIBB

Other theories of causation, such as the counterfactual theory of caus-

ation, do not allow one to recognize the crucial distinction between an

event that causes another event and an event that permits an event to cause

another event—the distinction between causing and permitting a causing.

Both count as causation under the counterfactual theory of causation

because both involve counterfactual dependence. The fact that the counter-

factual theory does not recognize this distinction is arguably an error on its

part, for the events appear to play very different roles.

Assuming a powers theory of causation, let us now return to the mental

causation debate.

3. The Double Prevention Solution to the Problemof Mental Causation

Let us refer to the event that is the firing of neuron N1 as ‘n1’, the event that

is the firing of neuron N2 as ‘n2’ and the event that is Fred’s hand’s moving

as ‘b1’. Consider a possible world in which N1’s firing is disposed to make

N2 fire and this, in turn, is disposed to make certain muscles in Fred’s body

contract and hence make his hand move. For the sake of simplicity let us

assume that no other dispositions are required for these manifestations.

Hence, n1 causes n2 and n2 causes b1. Now let us add that n2’s causing

b1 is prevented by the presence of mental event m2, where m2 is Fred’s

desire to keep his body still. More specifically, that neuron N2 retains its

disposition to make Fred’s hand move, but its manifestation is blocked by

the presence of Fred’s desire. Let us say that Fred has this desire and his hand

is therefore still. (Call this last physical event b2.) In such a world, Closure is

clearly false. This can be represented diagrammatically as follows:

Now consider a second possible world. It is identical with the first in

many respects. Neuron N1 fires in the brain of Fred’s counterpart. n1 causes

b2

n1

m2

n2 b1

Fig. 8.2

MENTAL CAUSATION AND DOUBLE PREVENTION 203

n2. n2 causes b1 but its doing so would be prevented by the presence of m2.

In this possible world Fred has m2, but he also has the conflicting, stronger

desire to move his hand (m1)—let us say due to a bad case of pins and

needles. Fred therefore moves his hand. It might be the case that Fred retains

the desire to keep his body still, but its manifestation is blocked by the

presence of Fred’s overriding desire to move his hand to get rid of the pins

and needles. Alternatively, it might be the case that gaining the desire to move

his hand causes Fred to lose the desire to keep his body still. Either way, the

presence of Fred’s desire to move his hand prevents the manifestation of his

desire to keep his body still. m1 is therefore a double preventer. Fred’s desire

to keep his body still (m2) is disposed to prevent his hand moving (b1), but is

itself prevented from doing so by his desire to move his hand (m1). Taking

the second case, it can be represented diagrammatically as follows:

In this second possible world there is no violation of Closure. By prevent-

ing m2 from preventing n2 from causing b1, m1 is not thereby a cause of

b1—according to the powers theory of causation, double preventers are not

causes of the event that they prevent from being prevented. The cause of b1

is n2 whose cause is in turn n1. Both b1 and n2 therefore contain only other

physical events in their transitive causal closure, as Closure demands.

There is, however, an obvious problem with the proposal as it stands.

Although in the second possible world there is, as a matter of fact, no

violation of Closure, there seems to be great potential for such a violation.

What if n1 causes n2, but m1 isn’t there to prevent m2 from preventing n2

from causing b1 and, hence, we have a similar situation to the one described

in the first possible world? For Closure to be true, in any case where there is

n2 and m2, m1 must be there to prevent m2 from preventing n2 from

causing b1.

This would be the case if the existence of some event in the chain of

neurological events leading to n2 entailed the existence of m1. Hence, for

m1 m2 b2

b1n1 n2

Fig. 8.3

204 S . C. GIBB

example, if the existence of n1 entailed the existence of m1. But what would

explain the entailment relation between n1 and m1? The existence of n1

would obviously entail the existence of m1 if n1 was identical with m1. But

this would be to reject Distinctness which is one of the claims upon which

this discussion is premised. Alternatively, the existence of n1 would entail

the existence of m1 if m1 was distinct from but ‘realized’ by n1. Again,

I think that this suggestion must be rejected. The notion of realization is a

notoriously elusive one, but common to most accounts of realization is the

idea that every causal power of the realized event is also a causal power of

the realizer event (but not vice versa).12 However, this is not the case with

m1 and n1. Given the proposal under consideration, m1 has a causal power

that n1 does not have, namely, the power to affect m2. m1 cannot,

therefore, be realized by n1.

Of course, if these were the only two ways of explaining the entailment

relation between n1 and m1, then one might be forced to accept that any

causal power that m1 has must in fact also be a causal power that n1 has. But

fortunately one is not forced to accept either of these ways of explaining the

proposed entailment relation. A third proposal, which is entirely consistent

with both the distinctness of m1 and n1 and their causal independence, is

that the existence of n1 entails the existence of m1 because whatever causes

n1 also causes m1. This proposal is represented by the following diagram,

where n0 is some further neurological event.

Let me make a number of clarificatory points in defence of this sugges-

tion. First, although whatever event causes n1 must also cause m1, it is not

the case that whatever event causes m1must also cause n1. (Quite clearly for

Closure to be true, it is not necessary that m1’s existence entails n1’s.)

m1 m2 b2

b1n1 n2

n0

Fig. 8.4

12 An important exception to this is Peter Menzies’ account of realization (Menzies 2003). For thepurpose of this paper, I do not have the space to explore this interesting account.

MENTAL CAUSATION AND DOUBLE PREVENTION 205

Second, one might object that even though m2 does not actually have

any physical effect, the claim that m2 would prevent b1 and that it would

bring about b2 if it weren’t for the presence of m1 is still in conflict with the

causal closure of the physical domain. More specifically, there must be

irreducibly psychophysical laws relating m2 and b1 and relating m2 and

b2, even though these laws are never implemented because of the presence

of m1. The claim that there are irreducibly psychophysical laws violates the

causal closure principle, for according to it all of the basic laws must be

purely physical ones.13

To reply, the claim that there are no irreducibly psychophysical laws is

certainly not a consequence of the causal closure principle as it has here been

formulated. Closure states that every physical event contains only other

physical events in its transitive causal closure. This is quite consistent with

the existence of the kind of irreducibly psychophysical laws that I would

have to allow. To rule out such laws, an even stronger closure principle

would be required. However, I fail to see what plausible support could be

mustered—either metaphysical or empirical—for a closure principle of the

required strength. Given the powers theory of causation, laws are nothing

more than generalized claims about causal relations and causal relations are

to be accounted for in terms of powers. Therefore to rule out the kind of

irreducible psychophysical law that I would have to allow, one must

basically provide a reason to rule out the claim that a mental entity could

be disposed to bring about or to prevent a physical entity. But I fail to see

what plausible reason one could give. It is certainly not a consequence of the

powers theory of causation—the theory of causation upon which this

discussion is premised. Nor is it suggested by any fact of physics. Yes, if a

mental event actually did, for example, prevent a physical event, one would

have good grounds for saying that this was, for example, a violation of the

laws of conservation of energy and momentum. But the point is that the

mental never actually does prevent the physical even though it is disposed to

do so. Hence, the conservation laws are never actually violated.

Thirdly, in the causal system that has been presented, one may be uneasy

about the fact that it is not naturally possible for Fred’s desire to keep still to

manifest all that it is disposed for—that is, it is not naturally possible for it to

prevent n2 from causing b1 and for it to cause b2.

13 I’m grateful to E. J. Lowe for raising this possible objection.

206 S . C. GIBB

In response, first note that the claim is not that it is naturally impossible

for Fred’s desire to keep still to ever be satisfied. Rather, the suggestion is

that it is naturally impossible for Fred’s desire to keep still to be satisfied if

neuron n0 fires and hence if Fred has the stronger and conflicting desire to

move his hand. Prior to having the desire to move his hand, Fred’s desire to

keep still might, for example, itself have been playing the role of double

preventer in which case Fred’s desire to keep still would have resulted in

him keeping still.

The point still stands, however, that it is naturally impossible for Fred’s

desire to manifest all that it is disposed for. It is naturally impossible, for

example, for Fred’s desire to keep still to prevent n2 from causing Fred’s

hand’s moving. However, this is arguably not a worry if one takes dispos-

itions seriously, as indeed one must if one is to accept the powers theory of

causation. Dispositions are not made real by their manifestations—they exist

regardless of whether they are manifesting any manifestation. Whether a

disposition can manifest a manifestation depends on both the presence and

absence of other dispositions. Nature rules out certain combinations of

dispositions and allows others. As a consequence, it permits certain mani-

festations of certain dispositions but prevents others. Martin comments that:

‘Salt in a world lacking H2Owould have many of its readinesses unfulfilled’

(Martin 2008, 6). If we accept Martin’s point then, just as in a world lacking

H2O, salt could never manifest all that it is disposed for, in a world in which

n1 is the only cause of n2 and n1 is always accompanied by m1, m2 could

never manifest all that it is disposed for.14

In the proposed system, Closure is therefore true. Indeed it is precisely

because of a mental event, namely m1, that it is true. However, Distinctness,

Exclusion, and Relevance are also true. The proposed system is obviously

compatible with Distinctness—m1 and m2 are not physical events. It is also

clearly compatible with Exclusion—neither m1 nor m2 cause any physical

event and hence do not threaten to causally overdetermine any physical

14 Indeed, Lowe has pointed out to me that the laws of physics provide a good reason to think that

there are unmanifestable dispositions. Quark confinement is the hypothesis that quarks cannot be

isolated singularly. Quarks only exist in combination with other quarks—a total of three in the case of

baryons and two in the case of mesons. Up quarks have a positive charge of two-thirds. Down quarks

have a negative charge of one-third. But, given quark confinement, only charges that are multiples of

these unit charges can be manifested in the form of attraction or repulsion as dictated by Coulomb’s law.

I’m very grateful to Lowe for raising this point.

MENTAL CAUSATION AND DOUBLE PREVENTION 207

event. Finally, regarding Relevance, Fred’s desire to move his hand is no less

important in an account of why Fred’s hand moved, than Sally’s pressing of

the button is in an account of why Fred’s ball broke the bottle. Just as Sally’s

pressing of the button permits the breaking of the bottle, Fred’s desire to

move his hand permits his hand to move. In more detail, m1 (Fred’s desire

to move his hand) permits b1 (the moving of his hand) by permitting the

causal relation between n2 and b1 to take place. It permits the causal relation

between n2 and b1 by preventing m2 from preventing the causal relation

between n2 and b1. If it had not done this, m2 would have prevented n2

from causing b1 and b1 would not have occurred. Fred’s desire therefore

plays an essential role in the occurrence of the hand moving. This role is just

as important as that played by n2. Obviously, it also follows from this

account that it is true that if Fred had not had the desire to move his hand

then his hand would not have moved. That is, b1 counterfactually depends

on m1. This is because if m1 had not occurred, and hence not prevented m2

from preventing n2 causing b1, then b1 would not have occurred.

Given this account, an explanation of Fred’s hand’s moving which simply

focused on the chain of neurological events leading to the hand’s moving

would be incomplete. It is true that one would have specified the complete

set of causes of the hand’s moving. But something crucial would have been

left out of one’s account, because it is in virtue of Fred’s desire to move his

hand that the causal relation between n2 and b1was allowed to take place in

the first place. In this respect, my account of psychophysical causation shares

a very important feature with that of Lowe’s. As Lowe explains, the role that

he gives mental events in the physical domain—that of making the fact that

a causal tree of neural events converge upon a particular bodily movement

non-coincidental—is completely invisible to the scientist who studies only

the physical events and their causal relationships (Lowe 2008, ch. 3). This is

also true of my account of psychophysical causation. An empirical investi-

gation of the chain of neurological events that gives rise to b1will not reveal

any role for non-physical events in the explanation of why b1 occurred.

This is because there are no gaps in this chain of neurological events—every

event in it has an immediate physical cause. Because of this it would be quite

reasonable for the neuroscientist to conclude that a complete account of

why b1 occurred can be given in terms of these neurological events. But, if

my account is correct, this conclusion would be mistaken for it is a non-

physical event (m1) that permits the causal chain of neurological events to

208 S . C. GIBB

give rise to b1 in the first place. This role will be invisible to the neuroscien-

tist. To give an analogy, going back to the example of Sally and Fred, a

passerby focusing solely on the movement of the ball from Fred’s hand to

the bottle will be completely unaware of Sally’s role in the breaking of

bottle, no matter how carefully he looks. Similarly, focusing on the chain of

neurological events leading to b1 will not allow one to detect m1’s role in

bringing about b1, no matter how carefully one studies this chain.

This is not to suggest that my proposal has no empirical consequences. If

empirical evidence established that our decision to perform a bodily move-

ment coincided with (or, indeed, came before) the initiation of a causal

chain of neurological events leading to the bodily movement, this would

prove my account false. In the model that I have presented, neurological

event n0 occurs before Fred has the desire to move his hand and it is

presumably n0 (or, perhaps, some prior neurological event that causes n0)

that initiates the chain of physical events that leads to his hand’s moving. But

empirical evidence establishes no such thing. Indeed, according to Libet’s

experimental findings, it seems to establish quite the opposite (Libet 1985).

Empirical evidence suggests that a ‘readiness potential’ begins in the brain

up to a second or more before a voluntary bodily movement. This raises the

question of whether the conscious experience of deciding to perform the

bodily movement also appears so far in advance. Libet’s famous neurological

experiments seem to demonstrate that it does not. In these experiments

subjects were asked to spontaneously flick their wrist whenever they wanted

to and were asked to note the position of a moving dot when they first

became aware of their decision to move their wrist. Libet found that the

onset of the readiness potential commonly began several hundred millisec-

onds before the subject’s conscious experience of deciding to move. If these

findings are correct then, as Libet observes, the role of mental events cannot

be to initiate chains of causation that give rise to bodily movement. If this

were their role, then one would expect the conscious experience of decid-

ing to move one’s wrist to occur before the onset of the readiness potential

that mediates the bodily movement. Libet’s experiments suggest that it is

instead an unconscious cerebral process that initiates the bodily movement.

In cases of voluntary bodily movement, rather than initiating the chain of

physical events that gives rise to the bodily movement, Libet considers that

the role of a mental event must be to somehow ‘select and control it, either

by permitting or triggering the final motor outcome of the unconsciously

MENTAL CAUSATION AND DOUBLE PREVENTION 209

initiated process or by vetoing the progression to actual motor activation’

(Libet 1985, 529).

Both Libet’s experimental results and the conclusions that he draws from

them clearly fit with the model of psychophysical causation that I have

presented. For the purpose of argument, let us say that n0 is the neurological

event that marks the onset of the readiness potential that initiates the chain of

physical events that give rise to Fred’s hand’s moving. As is consistent with

Libet’s experimental results, Fred’s decision to move his hand occurs after n0,

and hence after the onset of the readiness potential. Moreover, my account

provides a way of explaining how mental events permit the final motor

outcome. A mental event permits the motor outcome by permitting a

neurological event to cause the motor outcome. It permits this causal relation

by preventing a further mental event from preventing this causal relation.

4. Phenomenology and the DoublePrevention Account

I have argued that, within the framework of a powers theory of causation,

understanding mental events as double preventers provides a possible way

of making the combination of Closure, Relevance, Exclusion, and Distinctness

consistent. It is, however, one thing to present a solution to the problem

of mental causation that is plausible from a metaphysical point of view,

and quite another to show that it provides the correct description of

the actual world. And, indeed, from a phenomenological point of view,

one may at first baulk at the idea of mental events playing the role of

double preventers. But in fact the more that one considers it, the more

attractive it becomes. I shall finish this discussion by briefly considering

some of the reasons why.

In a famous passage from The Principles of Psychology, William James

describes the experience of getting out of bed on a cold morning:

We know what it is to get out of bed on a freezing morning in a room without a

fire, and how the very vital principle within us protests against the ordeal [ . . . ] We

think how late we shall be, how the duties of the day will suffer; we say, ‘I must get

up, this is ignominious’, etc.; but still the warm couch feels too delicious [ . . . ]

Now how do we ever get up under such circumstances? If I may generalize from my

210 S . C. GIBB

own experience, we more often than not get up without any struggle or decision at

all. We suddenly find that we have got up [ . . . ] It was our acute consciousness of

both the warmth and the cold during the period of struggle, which paralyzed our

activity then and kept our idea of rising in the condition of wish and not of will. The

moment these inhibitory ideas ceased, the original idea exerted its effects. ( James

1981, 1133)

In James’ example, the desire to get up and start the day and the desire to

stay in the warm bed are in conflict. It is presumably, according to James,

the overriding desire to get up which causes the temporary suspension of

our desire to stay in the warm bed. When we cease to have the desire to stay

in bed, we ‘suddenly find that we have got up’. Here James is not suggesting

that the bodily behaviour is automatic—we get up because of our desire to

get up. But nor is it the case that the desire to get up has re-exerted itself all

over again. Rather, according to James, we find ourselves acting on our

original desire to get out of bed when the conflicting desire to stay in bed

momentarily ceases. As James goes on to comment; ‘the immediate point of

application of the volitional effort lies exclusively within the mental world.

The whole drama is a mental drama’ (James 1981, 1168).

This fits in nicely with an account of mental events as double preventers.

My desire to stay in the warm bed would have prevented me from getting

up. However, my desire to get up causes me to temporarily cease to have

this desire. My desire to get up therefore does not exert itself on some

neurological event but rather on some antagonistic mental event—my

desire to stay in the warm bed. The ‘drama’ is, therefore, as James puts it,

a wholly mental one. Once my desire to stay in the warm bed is suspended

as a result of my overriding desire to get up, this permits the relevant causal

chain of physical events to proceed and I find myself getting up.

Now as James makes clear, not all examples of voluntary action are like

this. According to him, decisions with effort merge into those without it.

These in turn merge into ideo-motor actions—that is those movements that

follow unhesitatingly and immediately from our notion of them in the

mind—which in turn merge into involuntary reflex acts (James 1981,

1178). In the case of ideo-motor acts, James explains that when we perform

the act we are ‘aware of nothing between the conception and the execution.

All sorts of neuro-muscular processes come between, of course, but we

know absolutely nothing of them. We think the act, and it is done [ . . . ]’

MENTAL CAUSATION AND DOUBLE PREVENTION 211

( James 1981, 1131). Thus, for example, to give one of James’ examples,

whilst talking I become aware of some dust on my sleeve, and without

interrupting the conversation, brush the dust from my sleeve (James 1981,

1131). As James explains ‘[i]n all this the determining condition of the

unhesitating and resistless sequence of the act seems to be the absence of any

conflicting notion in the mind. Either there is nothing else at all in the mind, or

what is there does not conflict’ (James 1981, 1132).

If James’ account is correct, then the existence of ideo-motor actions

might be thought to present a problem for an account of mental causation in

terms of double prevention—after all, such cases, according to James,

involve the absence of any conflicting notion in the mind. However,

these kinds of cases do in fact fit with the account that I have offered—

the difference being that in such cases the relevant mental event has no

mental event to prevent to permit the causal relation between the relevant

physical events to take place. If I desire to remove some dust from my sleeve

and I have no conflicting desires, then there is nothing that my desire must

do to permit the causal relation between the relevant physical events to take

place. It is precisely for this reason that the sequence of acts is, as James puts

it, ‘resistless’.

For reasons that I have only had the space to touch upon here, from a

phenomenological point of view, understanding mental events as double

preventers provides the basis for a rich and potentially diverse account of our

actions.

To conclude, I have argued that the powers theory of causation reveals

an important distinction between the role of causing an event and the role

of permitting an event to be caused. Given the powers theory, double

preventers are not causes of the event that they prevent from being

prevented. Rather, they permit the event to be caused, and they do so

by preventing the event that would have prevented it. This role is a

crucial one and allows us to see why double preventers should play a

central role in our explanation of the occurrence of certain events and also

in our counterfactual judgements regarding them. I went on to suggest

that if mental events are double preventers then this provides a way of

reconciling Relevance, Closure, Exclusion, and Distinctness. According to this

claim, a mental event permits a certain bodily movement to take place by

permitting a neurological event to cause the bodily movement. It permits

212 S . C. GIBB

this causal relation by preventing a mental event that would have pre-

vented the causal relation between the neurological event and the bodily

movement. The resulting account is not only consistent with the empir-

ical evidence, but also coincides with some of James’ crucial insights into

the phenomenology of will.15

References

Armstrong, D. M. (1997). A World of States of Affairs. Cambridge: Cambridge

University Press.

Bird, A. (2010). ‘Causation and the Manifestation of Powers’. In A. Marmodoro

(ed.), Powers: Their Grounding and their Manifestations. Abingdon: Routledge:

160–8.

Cartwright, N. (1989). Nature’s Capacities and their Measurement. Oxford: Oxford

University Press.

Ellis, B. (2001). Scientific Essentialism. Cambridge: Cambridge University Press.

Gibb, S. C. (2010). ‘Closure Principles and the Laws of Conservation of Energy and

Momentum’. Dialectica, 64: 363–84.

Hall, N. (2004). ‘Two Concepts of Causation’. In J. Collins, N. Hall, and L. Paul

(eds,), Causation and Counterfactuals. Cambridge, MA: MIT Press: 225–76.

Harre, R. and E. H. Madden (1975). Causal Powers: A Theory of Natural Necessity.

Oxford: Blackwell.

Heil, J. (1998). Philosophy of Mind: A Contemporary Introduction. New York:

Routledge.

——(2003). From an Ontological Point of View. Oxford: Clarendon Press.

James, W. (1981). The Principles of Psychology, ed. F. H. Burkhardt. Cambridge, MA:

Harvard University Press.

Lewis, D. (1986). ‘Postscript C to Causation’. In Philosophical Papers, Vol. 2.

Oxford: Oxford University Press: 184–8.

——(1997). ‘Finkish Dispositions’. Philosophical Quarterly, 47: 143–58.

Libet, B. (1985). ‘Unconscious Cerebral Initiative and the Role of Conscious Will

in Voluntary Action’. Behavioural and Brain Sciences, 8: 529–66.

Lowe, E. J. (1993). ‘The Causal Autonomy of the Mental’. Mind, 102: 629–44.

15 I’m very grateful to Jonathan Lowe for helpful comments on earlier drafts of this paper. I would also

like to thank James Clarke and audience members at the Durham Philosophy Department’s Research

Seminar. This paper was completed with support from the Arts and Humanities Research Council’s

Research Grant AH/F009615/1 ‘The New Ontology of the Mental Causation Debate’.

MENTAL CAUSATION AND DOUBLE PREVENTION 213

——(2000). ‘Causal Closure Principles and Emergentism’. Philosophy, 75: 571–86.

——(2008). Personal Agency: The Metaphysics of Mind and Action. Oxford: Oxford

University Press.

Martin, C. B. (1996). ‘How It Is: Entities, Absences and Voids’. Australasian Journal

of Philosophy, 74: 57–65.

——(2008). The Mind in Nature. Oxford: Oxford University Press.

Menzies, P. (2003). ‘The Causal Efficacy of Mental States’. In S. Walter and

H. D. Heckmann (eds.), Physicalism and Mental Causation. Exeter: Imprint

Academic: 195–224.

Molnar, G. (2003). Powers: A Study in Metaphysics, ed. S. Mumford. Oxford:

Oxford University Press.

Mumford, S. (2009). ‘Passing Powers Around’. The Monist, 92: 94–111.

Mumford, S. and R. L. Anjum (2009). ‘Double Prevention and Powers’. Journal of

Critical Realism, 8: 277–93.

Shoemaker, S. (1980). ‘Causality and Properties’. In P. van Inwagen (ed.), Time and

Cause. Dordrecht: D. Reidel: 109–35.

214 S . C. GIBB

9

The Identity Theory as a Solutionto the Exclusion Problem

DAVID ROBB

This chapter is about a proposed solution to the exclusion problem, one I’ve

defended elsewhere (Robb 1997, 2001; Heil and Robb 2003). Details aside,

it’s just the identity theory: mental properties face no threat of exclusion

from, or preemption by, physical properties, because every mental property

is a physical property. Here I elaborate on this solution and defend it from

some objections. One of my goals is to place it in the context of a more

general ontology of properties, in particular, a trope ontology.

The exclusion problem takes several forms. The version I confront here is

generated by three principles:

Efficacy: Mental properties can produce physical effects.

Closure: Only physical properties can produce physical effects.

Dualism: Mental properties are not physical.

The problem is that each seems true, but the triad is apparently inconsistent.

These principles depart in some ways from Robb (1997). For example,

I here explicitly frame Efficacy (what I used to call Relevance) and Closure in

terms of causal production (compare Kim 2007). Properties here are powers

to produce characteristic physical manifestations in the appropriate circum-

stances (Molnar 2003; Heil 2003, ch. 8). It seems to me that it’s here the

exclusion problem is most clearly a problem of metaphysics, and even, as

I hope to show, ontology. There are, granted, less metaphysically loaded

versions of the problem, along with corresponding solutions. These appeal

not to powers or production, but to, say, explanation (Burge 1993), coun-

terfactual dependence (Yablo 1992; Loewer 2007), or causal intervention

(Woodward 2008). I hope that much of what I have to say below sheds light

on these alternative conceptions of the problem, but I will not explore these

connections here.

I rejectDualism, at least on the most straightforward reading on which the

principle is incompatible with the other two (more on this below). Every

mental property is physical. In particular, every mental property is the

physical property that would otherwise threaten to preempt it with respect

to physical (and especially behavioral) effects. More simply, every mental

property is what some would call its physical “base” or “realizer.” So this is

an identity solution to the problem. This proposal, however, needs consid-

erable refinement and defense, my aims in the rest of this chapter. I’ll

proceed by responding to a number of objections.

O1: Psychophysical property identity was undermined decades ago by the multiple

realizability argument (Fodor 1974; Boyd 1980; Putnam 1980). Indeed, the contem-

porary debate is really just about how to save mental causation in the face of the non-

reductive physicalism established by this argument. Proposing an identity theory at

this point ignores these results and disengages from the contemporary debate.

R1: According to the multiple realizability argument, mental properties

are not physical because they are multiply realizable in the physical. Put

another way, when we ascribe mental properties, we abstract away from

details of physical implementation. On the functionalist version of this

argument, to instantiate a given mental property is to be in some state or

other that plays the defining causal role of that mental property. Since this

realizer or role-filler state can be any of a variety of physical properties, the

mental property cannot be identified with any one of them. Mental prop-

erties are thus second-order properties, at best realized in, but distinct from,

any physical property.

The objector is right that this argument is almost universally endorsed

among those contributing to the mental causation literature. This may be

what leads Yablo (1997, 255), for example, to say that accepting Dualism is

included in “the price of admission” to the mental causation debate. But

while I grant that the argument is sound, its relevance to the exclusion

problem is not so clear.

Start with a few roles properties are thought to play (compare Campbell

1990, 29; Oliver 1996). For example, properties are features; they are, that is,

the truthmakers for (some) predications. The truthmaker for “This apple is

216 DAVID ROBB

round,” for example, is a property, roundness. Properties also sometimes

appear as types. Here properties play the role of a one over many, something

shared by objects of the same type or kind: various round things all have the

same property, roundness. And properties also appear as powers, where this

role can be indicated by the qua locution: an apple causes a certain kind of

impression in soft clay qua round thing—in virtue of its roundness—but not

qua red thing, in virtue of its color.

Now it’s at least not obvious that the same sort of entity answers to

“property” or “roundness” in each case. The three roles—feature, type,

power—may not be filled by the same thing. An ontology of properties

would work this out. But for now, it’s enough to point out that the multiple

realizability argument is aimed (and, I grant, succeeds) at distinguishing

mental and physical types. What the argument shows is that there is no

one-to-one match between mental and physical types. So the psychophys-

ical “property dualism” warranted by this argument is type dualism. It

remains open whether there is any such dualism of features or powers.

But this question is central to the exclusion problem, for it’s clearly proper-

ties as powers that appear in Closure and Efficacy. For all that’s been said so

far, type dualism is compatible with these principles.

Now it could turn out that features, types, and powers coincide, that the

same sort entity fills all three roles. But this is a heavyweight ontological

claim, one that goes far beyond O1’s seemingly innocent appeal to the

multiple realizability argument. And this is where a trope ontology becomes

directly relevant, for according to this ontology, tropes are features and

powers, but something else is a type. Typical among trope theorists is to say

a type is a resemblance class of tropes (Williams 1966; Campbell 1990; Bacon

1995) or, what may amount to the same thing, a collection or plurality of

resembling tropes. The mental type elation, for example, will be a class

of tropes, all of which resemble one another closely enough to count as

tropes of that mental type. But it’s compatible with this that (1) all of the

tropes making up elation are physical, (2) not all of these tropes resemble

one another closely enough to count as a physical type, and (3) some of

them do resemble one another closely enough to count as a physical type:

each cluster of closely resembling tropes will make up a physical type that

“realizes” elation. (For more on this, see R6 below and Robb 1997.) What

results is type dualism but trope monism: a psychophysical identity theory of

features and powers.

THE IDENTITY THEORY AS A SOLUTION 217

Return then to the exclusion problem and the principles that drive it. Let

the “properties” in these principles appear consistently as powers, so that the

principles are incompatible. The proposed solution is to reject Dualism in

favor of an identity theory: mental powers are physical. If O1 protests that

the multiple realizability argument undermines this identity theory, my

reply is that O1 conflates properties as types with properties as powers.

(Or if the objector insists on stipulating that the properties appearing in

Dualism are types, then my reply is that the three principles are consistent

after all, as the other two explicitly concern powers: the principles no longer

generate a problem.) These remarks are not intended to establish a trope

ontology, but just to show that O1’s appeal to multiple realizability is not

ontologically innocent: the nature of properties must be confronted directly

if one is to claim, as so many do, that the multiple realizability argument

blocks an identity solution to the exclusion problem.

O2: Even if the classic multiple realizability argument concerns types, a similar argument

can be deployed concerning tropes, and thus powers. A mental trope is more compos-

itionally plastic than any physical trope. Consider an elation trope E and the complex

physical trope P with which E is allegedly identical. E could survive the change of a

single neuron or particle, while P could not. The tropes are thus distinct, resulting again

in Dualism, but this time it’s explicitly a dualism of tropes, and so powers (compare

Boyd 1980; Pereboom 2002).

R2: Unlike the original multiple realizability argument, this newer ver-

sion threatens an identity solution to the exclusion problem, for it claims to

directly establish a dualism of powers. But the intuitions driving this newer

argument are far more controversial than those behind the original version.

While it’s no doubt true that an elation trope could exist in the absence of P,

by what right does O2 claim that E itself, that very trope, could exist

without P? If this is supposed to be just a brute intuition, it is one that

I do not share.

Still, perhaps I can advance a bit beyond a clash of basic intuitions (though

it’s admittedly hard to avoid question-begging on this front). The judg-

ments of type identity and distinctness in the classical multiple realizability

argument are fundamentally driven by judgments of similarity, and espe-

cially imperfect similarity: physically diverse creatures, while not exactly

resembling, are similar enough to fall under the same mental type. We can

rely here on the fact that higher-level types by their nature abstract away

218 DAVID ROBB

from micro-differences, permitting less than perfect resemblance. But O2

cannot help itself to an analogous claim about tropes, for it’s not nearly as

clear that what’s at the ontological ground level will, like types, permit

higher, more abstract layers. Being ground-level and maximally determinate

would seem to go hand in hand. Conflicting intuitions aside, then, there

seem to be general reasons to be suspicious of the claimed plasticity of

mental tropes.

O3:Multiple realizability is not the only reason to be a property dualist. There are several

other arguments against the identity theory. For example: Mental properties are

irreducibly subjective or private, while no physical property is (compare Jackson 1982).

When ascribing mental properties, we are subject to normative or holistic constraints

that do not bind us when we ascribe physical properties (Malcolm 1970; Davidson

1980). A being—my ‘zombie twin’—could duplicate all of my physical properties yet

lack my mental properties (Chalmers 1996). I might have existed disembodied, with

all of my mental properties but with no physical properties (Yablo 1990).

A psychophysical property identity would have to be necessary, but it appears

contingent, and there is no plausible way to explain away this appearance (Kripke

1980). And there are others. These are directly aimed at distinguishing mental

and physical features, which, at least for the trope theorist, are powers. One

cannot, then, shrug off such arguments as irrelevant to Dualism and the

exclusion problem.

R3: These do indeed threaten an identity theory of mental and physical

features, and thus of powers.1 If any one of these arguments is sound, then it

looks as if Dualism is unavoidable, so that either Efficacy or Closure must be

abandoned. I have no Master Reply to these arguments: they must be

confronted individually, something I won’t attempt here. (I’ve taken on

the zombie in Robb 2008.) However, I will make this more limited reply:

much of the mental causation literature is conducted in the context of non-

reductive physicalism, the pairing of Dualism with the thesis that the mental

is always realized in the physical. The driving argument for Dualism in this

context is the multiple realizability argument, which I hope to have shown is

not nearly as favorable to a dualism of powers as some imagine. But

the arguments in O3 are importantly different, for if they are sound, non-

reductive physicalism is false, and a more robust form of property dualism—

1 I’ll assume in what follows that features and powers coincide: the same sort of “property” (namely, a

trope) fills both rolls.

THE IDENTITY THEORY AS A SOLUTION 219

feature and power dualism—takes its place. In that case the dialectic changes

considerably, and the aim of an identity theorist such as myself is not to

accommodate the dualist argument—as I try to accommodate (classical)

multiple realizability—but to confront it directly.

O4: The identity solution merely relocates the exclusion problem. While psychophysical

trope identity may rescue the causal efficacy of mental powers (tropes), it still leaves

open whether they are causally efficacious qua mental. Put another way: if mental

powers are both mental and physical, why not think they are causally efficacious only

in virtue of being physical? Closure would motivate this problem, and we’re back

with, if not the same exclusion problem, at least very a similar one (Noordhof 1998;

Shoemaker 2003, 434; Macdonald and Macdonald 2006, 552–3).

R4: The general thought behind this objection seems to be that we will

have solved the exclusion problem only if we finally arrive at something—

of whatever ontological category—that’s only mental, that is, mental but

not physical. Anything that’s both mental and physical, the thought goes,

invites exclusion worries all over again. So, for example, Davidson’s (1980,

1993) desired stopping point is at mental events, which, he says, are physical.

But since mental events are both mental and physical, exclusion worries

arise, so that we must show that mental events are causes in virtue of their

mental features (powers). Similarly, the line goes, if our desired stopping

place is with these mental powers, it had better turn out that they’re not also

physical, for then we get the same problem all over again, so that we must

show that mental powers are causally efficacious in virtue of their mental

features (higher-order powers). And so on.

But there’s something suspect in this general line of objection. If psycho-

physical identity at the desired stopping place continues to invite exclusion

worries, why shouldn’t other psychophysical relations at the desired stop-

ping place invite similar worries? Suppose, for example, that mental powers

are in fact only mental—that is, mental but not physical—yet are immanent

in (Yablo 1997, 275) or nothing over and above (Wilson 2005) the physical,

where this may be spelled out in terms of, say, realization (Boyd 1980), the

determinable–determinate relation (Yablo 1992), constitution (Pereboom

2002), metaphysical necessitation (Bennett 2008), de re, a priori determin-

ation ( Jackson 2006), or something else. And suppose that immanence, in

whatever form it takes, does in fact secure the efficacy of mental powers

with respect to physical effects. Is there still a lingering worry that these

220 DAVID ROBB

powers are efficacious, not in virtue of being mental, but merely in virtue of

being immanent in the physical (Lowe 1993, 632–3)? Maybe mental powers

are efficacious, but only because they piggyback on their physical base

powers. This appears to be, if not the original exclusion problem, one

that’s very similar to it.

Now the immanence theorists cited above may suspect that these worries

are somehow ill-conceived. But then why should it be that when we move

to the most intimate form of immanence—namely, identity—these worries

are legitimate? It seems to me that whatever one’s view of immanence, the

way to stop these recurring qua questions is not to ban immanence in the

physical at the desired stopping place. It’s to show that qua questions—and

the exclusion worries that threaten—at that place are somehow illegitimate.

And this is where I think an ontology of properties will again be relevant.

In the work cited earlier, I’ve argued that at the level of powers, qua

questions are illegitimate because powers do not themselves have higher-

order features or powers. More simply: there are no tropes of tropes. Such

higher-order tropes, I’ve argued, are explanatorily idle, threaten to start a

vicious regress, and are just plain odd. Here I’ll make just the oddity point.

Start again with how the qua questions arise for an event identity theorist

such as Davidson: even if mental events cause behavior, they should do so in

virtue of their mental features. For example, the feel of a pain or the content of

a belief must be efficacious. For this purpose, mental features (powers) are

recognized. But now suppose these mental powers are physical, and one

wants to raise qua questions again. Is the motive still that we want the mental

features of a mental power to do some causal work? But this is what strikes

me as odd. A phenomenal trope, for example, doesn’t have a qualitative

feel—at least not in the sense of having a qualitative feature—it is a

qualitative feel. A mental power to cause a bit of behavior doesn’t have a

power to cause such behavior: it is such a power. To raise qua questions at

this level, invoking features of features (powers of powers), looks strange,

and appears to commit a category mistake.

A worry closely related to O4 is that while psychophysical property

identity may secure the causal efficacy of mental powers, it doesn’t secure

‘distinctively’ mental causation, or causal efficacy for mental powers ‘in their

own right’ (e.g., Lowe 1993, 632; Wilson 2009, 150). But again, this seems

to assume that mental powers on the identity theory have a dual nature, so

that (in accordance with some suitably modified version of Closure) only the

THE IDENTITY THEORY AS A SOLUTION 221

physical nature is engaged when mental properties are causally efficacious.

But on the trope ontology, there is no such division within a trope: a mental

power, its mental nature, and its physical nature are all one and the same. In

this sense, a mental trope is both fully mental and fully physical.

O5: The identity solution advanced here combines psychophysical trope identity with type

dualism: one and the same trope can be a trope of two types. But this is impossible,

since tropes are individuated by their constituent types. (For more detailed discussion

of this argument, see Ehring 1996; Whittle 2007.)

R5: The objector here is apparently thinking of tropes on analogy with

Kimian events (Kim 1993), which are individuated, in part, by their con-

stituent types. On such an analogy, if mental and physical types are distinct

(and if a trope can have only one constituent type), no mental trope can be a

physical trope.

But tropes here are not complex entities with types as constituents.

Distinguish two conceptions of a trope (Daly 1997): as a complex entity (a

substance’s instantiating a universal), or as a fundamental entity. On the

latter conception, which I endorse, tropes are of types, but they don’t have

types as constituents. Such a view is required by a trope ontology, on which

tropes are the basic building blocks, the “alphabet of being.” I take it no

building block can have, as a constituent, the derived entities it grounds.

And if tropes don’t have types as constituents, there’s no general barrier to a

trope’s being of more than one type.2

However, while there’s nothing in general to prevent a trope’s being of

distinct types, there would be if the types in question were incompatible.

For example, no trope can be a red trope and a green trope. Returning to

the issue at hand, O5 might insist that mental and physical types are

incompatible, again resulting in psychophysical trope dualism. The dualist

arguments from O3, if sound, would deliver this result: if to be mental is,

say, to be irreducibly subjective, and to be physical is to be irreducibly

objective, then these types would exclude one another. But what about the

anti-reductionist argument in play here, the multiple realizability argument?

This, I claim, shows only that mental types are not physical; it doesn’t show

that they are anti-physical, that is, that they exclude the physical.

2 An analogy to Davidsonian events is suggestive, though it’s just an analogy: I would not want to say

that tropes are events, Davidsonian or otherwise.

222 DAVID ROBB

As a way of fleshing out this point, consider a version of O5 from

Shoemaker (2003, 434, note omitted):

If we think of the instantiation of a property as the conferring on something of

the conditional powers associated with that property, then when properties

confer different sets of conditional powers, the instantiation of one of them is

not identical with the instantiation of the other.

If “property” here means type and “instantiation” means trope (power) this

looks like an argument that mental types are anti-physical in the relevant

sense: no trope can be a trope of both a mental and physical type. In reply,

I grant that no mental type has the same set of causal powers associated with

it as any physical type. (Indeed, one might take this to be precisely the lesson

of multiple realizability.) But this does not show that mental types are anti-

physical. Suppose the causal powers of one type may be, as Shoemaker

himself believes, a proper subset of the causal powers of another (see also

Wilson 1999; Whittle 2007). Types standing in this intimate relation won’t

be incompatible, and this in fact entails that their corresponding tropes will

be identical (assuming the background of a trope ontology). Returning to

the case at hand, if the causal powers associated with mental type M are a

subset of the causal powers associated with one of its physical realizers P,

then trope identity follows: every P-trope will be an M-trope. And if M has

only physical realizers, every M-trope will be some physical trope or other.

O6: Trope identity and type dualism are nevertheless incompatible. We can show this by

assuming trope identity and deriving type identity. Take types to be resemblance

classes of tropes, and assume trope identity, so that a given mental type, say elation, is

a class of physical tropes. Now if this is to be a genuine type, its tropes must exactly

resemble, since (Gibb 2004, 471):

It is only those classes of tropes with the greatest possible degree of unity, that is,

sets of exactly resembling tropes, which can be substituted for universals. This

can be seen by the formal properties of the relation of resemblance. Whilst all

resemblance relations are reflexive and symmetrical, it is only in the case of exact

resemblance that the relation of resemblance is transitive.

From here, the route to type identity is quick. The physical tropes making up elation

must exactly resemble. But then elation is itself a physical type, for it’s no doubt

true—perhaps analytic—that exactly resembling physical tropes belong to the same

physical type. So if elation collects all and only those tropes belonging to a given

physical type, elation is that physical type. Moreover, if all of the tropes making up

THE IDENTITY THEORY AS A SOLUTION 223

elation exactly resemble, we’ve lost any sense in which this mental type is multiply

realizable in physically diverse creatures. The upshot is that the trope identity solution

is forced into type identity: elation, and every mental type, is a physical type.

R6: It seems to me that this objection goes wrong at the beginning:

physical tropes making up mental types needn’t exactly resemble. In fact, for

the trope theorist, this is the lesson of multiple realizability. Color provides a

standard analogy: while any determinate shade of red (shade type, that is)

consists of exactly resembling tropes of that shade, red itself, the determin-

able, is a class of tropes with more relaxed resemblance standards: some red

tropes exactly resemble one another—these exact-resemblance classes form

the maximally determinate shades that realize red—but exact resemblance

isn’t required. For example, the type red contains, say, the scarlet trope of

my blanket and the crimson trope of my chair, and these two inexactly

resemble. Mental types work in much the same way: some of the tropes in

elation exactly resemble. These classes of exactly resembling physical tropes

form all of the physical types that realize elation. But exact resemblance isn’t

required for membership in elation. For example, a trope of human elation

won’t exactly resemble a trope of Martian or dolphin elation.

By appealing to inexact resemblance, I run up against the passage from

Gibb quoted earlier. Following Armstrong (1989, 122–3), Gibb says that it is

“only those classes of tropes with the greatest possible degree of unity, that

is, sets of exactly resembling tropes, which can be substituted for universals.”

But granting this, I reply that mental types—that is, classes of mental

tropes—don’t substitute for universals (compare Whittle 2007, 71). Let

them instead substitute for what Armstrong calls second-class properties,

properties (types) that, while not universals, need to be recognized in any

ontology as part of the manifest image. (Armstrong sometimes suggests

colors as an example.) The idea is not foreign to trope theory. For example,

it seems to have been Williams’ (1966, 81) point when he says that classes of

inexactly resembling tropes “provide a less definite universal.” And Camp-

bell (1981, 484) notes that “The closeness of resemblance between the tropes

in a set can vary. These variations correspond to the different degrees to

which different properties [types] are specific.”3 A second-class property, or

a less definite universal, or a less specific property, is still a natural—not

3 Both this and the Williams passage are quoted in Bacon (1995, 17), though Bacon himself insists on

exact resemblance, as does Macdonald (1998, 334).

224 DAVID ROBB

conventional, not gerrymandered—class of tropes, though it may be, to

borrow a term from Lewis (1983), less than perfectly natural. It’s just that it’s

a class of tropes whose requirements for membership are more relaxed than

those more determinate classes that, I grant, are uniquely qualified to

substitute for universals.

This may be the best place to address the following question: If all of the

tropes in elation (or in any alleged mental type) are physical, what makes this

a mental type? Put another way, if all of the tropes in this class are physical,

what makes them mental as well, and in particular, elation tropes? Here I’m

neutral and say: deploy your favored theory of mentality. Functionalists, for

example, will say that what makes all of these elation tropes is that they all

have the causal profile definitive of elation. Those inclined toward qualia

can say that what makes them elation tropes is their qualitative feel. Those

who think intentionality is the mark of the mental can say that what makes

them elation tropes is their representational content. I take no stand here

on which is these is correct, only that each is compatible with inexact

resemblance between tropes of the same mental type.

O7: But why tropes? If the exclusion problem calls for an identity solution, type identity is

available, for there are versions of the type identity theory, such as Kim’s (1998, 2005),

that accommodate multiple realizability. Since an ontology of properties already requires

types to fill the role of “one over many,” types might as well be pressed into service as

powers as well.

R7: There are some close similarities between the trope identity solution

here and the type identity solution favored by Kim,4 which itself has some

affinities with type identity theories from Lewis (1994) and Armstrong

(1968). A full comparison between the views would take its own paper,

but here I sketch what I take to be the main advantages of a trope identity

solution over a Kim-style type identity solution.

First (Robb and Heil 2008, }6.5), the trope identity solution does not

require, as Kim’s type identity does, that mental types fragment into many

structure-restricted types.On the trope identity solution, there is a single type

elation, and many tropes of that type. What unifies them into a single mental

type is their (inexact) resemblance. I consider this a slim advantage at best,

4 Kim presents this brand of type identity as the best way to save mental causation, but he stops short

of endorsing it: see Kim (2005, 161).

THE IDENTITY THEORY AS A SOLUTION 225

however, for it seems that the plurality of structure-restricted “elations”

recognized by Kim might also be united into a single type by inexact

similarity, even if Kim himself (1998, 111) is not inclined to recognize such

unity.

Second, Kim concedes that his own reductivist account of mental caus-

ation is second-best:

The best, or the most satisfying, outcome would have been a vindication of

mental causation along the lines of nonreductive physicalism; that would have

allowed us to retain mentality as something that is causally efficacious and yet

autonomous vis-a-vis the physical domain.

But the best outcome, as we saw, is not to be had. The next best outcome, in

fact our only hope at this point if mental causation is to be saved, is physical

reductionism. Physical reduction would save causal efficacy for mentality, at the

cost of its autonomy. Reductionism allows only one domain, the physical

domain, but the mental may find a home in that domain. (Kim 2005, 159)

The suggestion here is that something valuable is lost in the move to a

reductionist (or identity) solution to the exclusion problem. Kim says that

what’s lost is autonomy from the physical, but if autonomy here is just taken

to be Dualism, then it’s not clear why losing that should be mourned.

Perhaps Kim has in mind here the distinctively mental contribution of mental

powers, the efficacy of the mental as such mentioned above in R4. But if

that’s what is lost in Kim’s account, then this is a reason to favor the trope

identity solution, for there is no such loss on a trope ontology: for reasons

given in R4, nothing distinctively mental is missing in either mental tropes

or their causal efficacy.

A third reason to favor trope identity solution over its type identity

sibling is a familiar ontological worry: it’s not clear that types are the right

sorts of the things to be powers, as they would have to be if a Kim-style

solution is to be tenable. Types play the “one over many” role for properties,

and so whatever exactly types are, they seem to be “spread out” over their

various instances. (This is the case, I take it, even for Kim’s structure-

restricted types.) Powers, however, are local, in re, here-and-now. The

point is clearest if types are taken to be resemblance classes, for a class, in

addition to being abstract, is not local in a way a power must be (for a bit

more on this, see Heil and Robb 2003, 175–6). Types may be useful in

explanation, but this epistemic role should not be confused with the meta-

physical role of powers, a role for which types seem to be ill-suited.

226 DAVID ROBB

O8: Still, why tropes? What’s essential to this identity solution is that mental powers are

local, in re, and above all, physical. But for all that tells us, mental powers are not

tropes but Armstrong-style universals (Armstrong 1989, 1997). Like tropes, Arm-

strong-style universals are both features and powers. The main difference is that

Armstrong-style universals are, well, universals: they are wholly present in multiple

instances, and moreover (in some cases) are types. But this difference doesn’t seem to

make a difference as far as an identity solution to the exclusion problem goes. Tropes

or Armstrong-style universals would do just as well (Whittle 2007, 70; Heil 2008;

Maurin 2008).

R8: There are indeed multiple and substantive similarities between a

trope ontology and Armstrong’s ontology of universals. For example, in

their respective ontologies, both tropes and Armstrong-style universals are:

(1) features; (2) powers;5 (3) spatiotemporal, at least when we restrict our

attention to physical properties; (4) instantiated in re; (5) dependent beings,

depending, in particular, on the objects that instantiate them; (6) maximally

determinate: neither ontology permits the sort of “layering” one finds among

types; (7) responsible for sameness of type; that is, they are the grounds or

truthmakers when two objects are of the same type or kind.

The similarities between the two ontologies are so striking that it’s

tempting to deny there is a deep difference here. Indeed, at one point

Armstrong (1989, 139) wonders, following a suggestion from H. H. Price,

whether the difference between these apparent rivals is merely one of

alternative languages: what the trope theorist describes as exactly resembling

features, the Armstrong describes as one and the same. Both describe, in

their own ways, the same underlying facts. There are a number of way this

suggestion could play out. Let S1 and S2 be two exactly resembling features.

The suggestion could be the relatively innocuous claim that S1 and S2,

while strictly distinct, are nevertheless identical in a looser sense. This seems

to be the lesson of Williams’ (1986) later view on universals, what Campbell

(1990) calls ‘painless realism’. Alternatively, one might appeal to the doc-

trine of relative identity (Geach 1980), so that S1 and S2 are distinct tropes,

but the same universal. Or the suggestion could be a conventionalism of

sorts, so that there’s no objective fact of the matter of whether S1 is S2: the

5 There is a difference here: tropes in themselves bestow causal powers, while Armstrong-style

universals do so only given the laws of nature. But both are powers in the sense that they are those

entities that are causally efficacious, whether or not such efficacy depends on laws of nature.

THE IDENTITY THEORY AS A SOLUTION 227

apparently distinct ontologies result from the neutral facts plus either of two

conceptual or linguistic overlays chosen for, say, pragmatic reasons (Carnap

1956; Sidelle 2002). I can’t follow through on these lines here, or maybe

anywhere. But supposing it turns out that one of them is correct, what

becomes of O8? If the worry is that Armstrong’s ontology would do just as

well as the trope ontology in response to the exclusion problem, my reply

would be that it would do just as well, but only because it’s the same

ontology, differently described.6

This is as far as I want to pursue this option. Suppose there is a genuine

ontological difference, as there appears to be, between Armstrong-uni-

versalist and trope-theoretic versions of the identity solution. Is there any

reason to favor one over the other? I doubt an advantage will be found

within the confines of the exclusion problem. The ontologies appear

isomorphic in any respect relevant to this problem. If either has advantage,

it would have to come from more general considerations.

One minor edge the trope ontology has is that it can give a uniform

account of types. As before, call the ground-level “properties,” whether

tropes or universals, features. For the universalist, all types are resemblance

classes of features except for maximally determinate types. Every maximally

determinate type is itself a feature, a single universal. But for the trope theorist,

there’s no such abrupt change at the most basic level: all types, even the most

determinate, are classes of resembling features. What’s distinctive of the basic

level of types is that the features of those types exactly resemble. But there’s no

categorial difference at this level for the trope theorist. Here ontological

continuity strikes me as a virtue, though this is admittedly a small advantage.7

Another potential advantage for the trope ontology may be found in the

metaphysics of causality. It’s typical for a trope theorist to insist that tropes

are better suited than universals—even Armstrong-style universals—to play

6 I take causal efficacy to be, like causation itself, extensional: a power to produce a certain effect is so

no matter how described. Matters would be different if, say, explanation rather than causal efficacy were

in play. Explanation is intensional, so that even if the two ontologies are equivalent, there may be

explanatory reasons to favor the conceptual apparatus of one over that of the other.

7 The advantage for the trope theorist would be greater if the universalist were forced to say that the

higher-level types are not genuine (because not universals). In that case, the trope theorist but not the

universalist could admit elation, e.g., as a legitimate type: compare Ehring’s (2003, 384) response to O8.But it seems to me that a universalist can be a realist about higher-level types (whether or not Armstrong

himself is). They just will not count as genuine universals, but something else, such as the ‘second-class

properties’ mentioned in R6.

228 DAVID ROBB

the role of powers: Campbell (1981, 480), for example, makes the point, as

does Honderich (1992, 246–7). Representative is this passage from Camp-

bell (1990, 23):

It is not the stove, the whole stove, that burns you; not even the whole stove

here now. For its solidity, iron structure, enamel surface and smoothness have

nothing to do with it. It is the temperature that does the damage. Moreover, it is

not any temperature, or temperature in general, but this particular case of

temperature, among the myriads in the world, and even among the many the

stove has during its life. Yesterday’s stove temperature is quite innocuous. It is

today’s that burnt you.

Again, accommodation of the ontology of causes into the trope scheme is so

smooth because what is required is an element that combines particularity with a

very restricted qualitative character, since causes are always features (almost

always a small selection from the host of features present) and every particular

cause is a particular feature or constellation of features.

Campbell’s point here is complicated by his taking tropes to be the causal

relata, while I’m neutral on this matter. But the argument applies equally as

well if a causally efficacious property of a cause—a power—must, like causes

themselves, combine elements of particularity and qualitativity.8 Moving

beyond Campbell’s argument, one might also motivate the trope ontology

in the context of a more general metaphysics of causation. Here I have in

mind, for example, Ehring’s (1997) recent defense of the transference

theory, one which makes essential use of tropes. In any case, however one

evaluates Ehring’s trope-based metaphysics of causation, its relevance to O8

helps to confirm a primary theme of this chapter: the ontology of properties

is not optional when confronting the exclusion problem.9

References

Armstrong, D. M. (1968). A Materialist Theory of the Mind. London: Routledge &

Kegan Paul.

—— (1989). Universals: An Opinionated Introduction. Boulder, CO: Westview Press.

8 Combination, however, is only a metaphor here: for reasons given earlier, a trope is its particularity

and qualitativity.

9 A version of this chapter was read in September 2009 at Davidson College and at “The New

Ontology of the Mental Causation Debate,” a conference held at Durham University. I thank the

audiences on those occasions for helpful comments, questions, and objections.

THE IDENTITY THEORY AS A SOLUTION 229

Armstrong, D. M. (1997). A World of States of Affairs. Cambridge: Cambridge

University Press.

Bacon, J. (1995). Universals and Property Instances: The Alphabet of Being. Oxford:

Blackwell.

Bennett, K. (2008). ‘Exclusion Again’. In J. Hohwy and J. Kallestrup (eds.), Being

Reduced: New Essays on Reduction, Explanation, and Causation. Oxford: Oxford

University Press: 280–306.

Boyd, R. (1980). ‘Materialism with Reductionism: What Physicalism Does Not

Entail’. In N. Block (ed.),Readings in Philosophy of Psychology, Vol. 1. Cambridge,

MA: Harvard University Press: 67–106.

Burge, T. (1993). ‘Mind–Body Causation and Explanatory Practice’. In J. Heil and

A. Mele (eds.), Mental Causation. Oxford: Clarendon Press: 97–120.

Campbell, K. (1981). ‘The Metaphysic of Abstract Particulars’. Midwest Studies in

Philosophy, 6: 477–88.

—— (1990). Abstract Particulars. Oxford: Basil Blackwell.

Carnap, R. (1956). ‘Empiricism, Semantics, and Ontology’. In Meaning and Neces-

sity, 2nd edn. Chicago: University of Chicago Press: 205–21.

Chalmers, D. A. (1996). The Conscious Mind: In Search of a Fundamental Theory.

Oxford: Oxford University Press.

Daly, C. (1997). ‘Tropes’. In D. H. Mellor and A. Oliver (eds.), Properties. Oxford:

Oxford University Press: 140–59.

Davidson, D. (1980). ‘Mental Events’. In Essays on Actions and Events. Oxford:

Clarendon Press: 207–28.

—— (1993). ‘Thinking Causes’. In J. Heil and A. Mele (eds.), Mental Causation.

Oxford: Clarendon Press: 3–18.

Ehring, D. (1996). ‘Mental Causation, Determinables and Property Instances’.

Nous, 30: 461–80.

—— (1997). Causation and Persistence: A Theory of Causation. Oxford: Oxford

University Press.

—— (2003). ‘Part–Whole Physicalism andMental Causation’. Synthese, 136: 359–88.

Fodor, J. A. (1974). ‘Special Sciences (Or: the Disunity of Science as a Working

Hypothesis)’. Synthese, 28: 97–115.

Geach, P. T. (1980). Reference and Generality, 3rd edn. Ithaca: Cornell University

Press.

Gibb, S. C. (2004). ‘The Problem of Mental Causation and the Nature of Proper-

ties’. Australasian Journal of Philosophy, 82: 464–76.

Heil, J. (2003). From an Ontological Point of View. Oxford: Clarendon Press.

—— (2008). ‘Modes and Mind’. In S. Gozzano and F. Orilia (eds.), Tropes,

Universals and the Philosophy of Mind. Frankfurt: Ontos Verlag: 13–30.

230 DAVID ROBB

Heil, J. and D. Robb (2003). ‘Mental Properties’. American Philosophical Quarterly,

40: 175–96.

Honderich, T. (1992). ‘Causation: One Thing Just Happens After Another’. In

L. E. Hahn (ed.), The Philosophy of A. J. Ayer. La Salle: Open Court: 243–70.

Jackson, F. (1982). ‘Epiphenomenal Qualia’. Philosophical Quarterly, 32: 127–36.

—— (2006). ‘On Ensuring that Physicalism is Not a Dual Attribute Theory in

Sheep’s Clothing’. Philosophical Studies, 131: 227–49.

Kim, J. (1993). ‘Events as Property Exemplifications’. In Supervenience and Mind.

Cambridge: Cambridge University Press: 33–52.

—— (1998). Mind in a Physical World. Cambridge, MA: MIT Press.

—— (2005). Physicalism, or Something Near Enough. Princeton: Princeton Univer-

sity Press.

—— (2007). ‘Causation and Mental Causation’. In B. P. McLaughlin and J. Cohen

(eds.), Contemporary Debates in Philosophy of Mind. Oxford: Blackwell: 227–42.

Kripke, S. A. (1980). Naming and Necessity. Cambridge, MA: Harvard University

Press.

Lewis, D. (1983). ‘New Work for a Theory of Universals’. Australasian Journal of

Philosophy, 61: 343–77.

—— (1994). ‘Lewis, David: Reduction of Mind’. In S. Guttenplan (ed.),

A Companion to the Philosophy of Mind. Oxford: Blackwell: 412–30.

Loewer, B. (2007). ‘Mental Causation, or Something Near Enough’. In

B. P. McLaughlin and J. Cohen (eds.), Contemporary Debates in Philosophy of

Mind. Oxford: Blackwell: 243–64.

Lowe, E. J. (1993). ‘The Causal Autonomy of the Mental’. Mind, 102: 629–44.

Macdonald, C. (1998). ‘Tropes and Other Things’. In S. Laurence and

C. Macdonald (eds.), Contemporary Readings in the Foundations of Metaphysics.

Oxford: Blackwell: 329–50.

—— and G. Macdonald (2006). ‘The Metaphysics of Mental Causation’. Journal of

Philosophy, 103: 539–76.

Malcolm, N. (1970). ‘Scientific Materialism and the Identity Theory’. In

C. V. Borst (ed.), The Mind/Brain Identity Theory. London: Macmillan: 171–80.

Maurin, A. (2008). ‘Does Ontology Matter?’ In S. Gozzano and F. Orilia (eds.),

Tropes, Universals and the Philosophy of Mind. Frankfurt: Ontos Verlag: 31–56.

Molnar, G. (2003). Powers: A Study in Metaphysics. Oxford: Oxford University

Press.

Noordhof, P. (1998). ‘Do Tropes Resolve the Problem of Mental Causation?’

Philosophical Quarterly, 48: 221–6.

Oliver, A. (1996). ‘The Metaphysics of Properties’. Mind, 417: 1–80.

THE IDENTITY THEORY AS A SOLUTION 231

Pereboom, D. (2002). ‘Robust Nonreductive Materialism’. Journal of Philosophy,

99: 499–531.

Putnam, H. (1980). ‘The Nature of Mental States’. In N. Block (ed.), Readings in

Philosophy of Psychology, Vol. 1. Cambridge, MA: Harvard University Press:

223–31.

Robb, D. (1997). ‘The Properties of Mental Causation’. Philosophical Quarterly, 47:

178–94.

—— (2001). ‘Reply to Noordhof on Mental Causation’. Philosophical Quarterly, 51:

90–4.

—— (2008). ‘Zombies from Below’. In S. Gozzano and F. Orilia (eds.), Tropes,

Universals and the Philosophy of Mind. Frankfurt: Ontos Verlag: 155–74.

—— and J. Heil (2008). ‘Mental Causation’. In E. N. Zalta (ed.), The Stanford

Encyclopedia of Philosophy, Fall 2008 edition, available at http://plato.stanford.

edu/entries/mental–causation/.

Shoemaker, S. (2003). ‘Realization and Mental Causation’. In Identity, Cause, and

Mind, Expanded Edition. Oxford: Clarendon Press: 427–51.

Sidelle, A. (2002). ‘Is There a True Metaphysics of Material Objects?’ Philosophical

Issues, 12: 118–45.

Whittle, A. (2007). ‘The Co-instantiation Thesis’. Australasian Journal of Philosophy,

85: 61–79.

Williams, D. C. (1966). ‘The Elements of Being’. In Principles of Empirical Realism.

Springfield, IL: Charles Thomas: 74–109.

—— (1986). ‘Universals and Existents’. Australasian Journal of Philosophy, 64: 1–14.

Wilson, J. (1999). ‘How Superduper Does a Physicalist Supervenience Need to

Be?’ Philosophical Quarterly, 49: 33–52.

—— (2005). ‘Supervenience–based Formulations of Physicalism’.Nous, 39: 426–59.

—— (2009). ‘Determination, Realization and Mental Causation’. Philosophical

Studies, 145: 149–69.

Woodward, J. (2008). ‘Mental Causation and Neural Mechanisms’. In J. Hohwy

and J. Kallestrup (eds.), Being Reduced: New Essays on Reduction, Explanation, and

Causation. Oxford: Oxford University Press: 218–62.

Yablo, S. (1990). ‘The Real Distinction Between Mind and Body’. Canadian

Journal of Philosophy, Supplement 16: 149–201.

—— (1992). ‘Mental Causation’. Philosophical Review, 101: 245–80.

—— (1997). ‘Wide Causation’. Philosophical Perspectives, 11: 251–81.

232 DAVID ROBB

10

Continuant Causation,Fundamentality, and Freedom

PETER SIMONS

the main element in the notion of a thing or continuant is the permanency of

functioning that can be discerned in a series of characterised manifestations,

presented in the course of time, as they may be observed in a temporally

continuous, or discrete, series of acts. Thus the notion of a continuant is

constructed in terms of temporal connection and causal determination

W. E. Johnson1

Continuants and Occurrents

Objects in time divide into continuants and occurrents.2 Occurrents are

events, processes, and states. They have temporal as well as spatial extension.

The temporal extension allows us to speak of their temporal parts, that some

temporal parts are earlier or later than others, that longer temporal parts

include shorter, and so on. Continuants by contrast have no temporal

extension, only spatial. A continuant exists at different times in its own

right, and not because it has a temporal part that exists at that time. So when

a continuant changes it has first one property and then another property

incompatible with the first. By contrast when one temporal part of an

occurrent is one way and a later temporal part is another, the property in

question inheres in different items, the temporal parts in question. For

1 Johnson (1924), 98. 2 Johnson (1924), 78–101.

example, the noise of a car engine gets louder and higher in pitch when the

accelerator is depressed. So while occurrents may vary over time in how

they are, because their different temporal parts vary, only continuants

genuinely change. The modes of existing in time of continuants and

occurrents are called ‘enduring’ and ‘perduring’ respectively by David

Lewis.3 Though I shall treat continuants and occurrents as equally accept-

able denizens of the world it will be important to my account of continuant

causation that occurrents have metaphysical priority over continuants.

I shall say that a continuant participates in an event or other occurrent

when this event is part of the life of the continuant in question.

Further Assumptions

Additional assumptions I shall be making but not arguing for in this paper are

the following. Firstly, that determinism is false, so that at any given time

certain aspects of the future with respect to that time have not at that time

been fixed or determined. I call this the open future assumption. Of course the

future is not totally open: some things are bound to happen, at least some

things of a certain kind. For example at some time in the future, we know not

when, current estimates say in about 5 billion years, the sun will become a red

giant and burn up the earth. The precise details of that process have yet to be

determined, but some process of that sort appears inevitable. The second

assumption is that we have some freedom of action: we are able ourselves to

determine in some respects how things turn out. We are familiar with this

phenomenologically from everyday experience, and while some philoso-

phers have denied or doubted it, I do not and I shall not here be arguing for

freedom, which is another enterprise. Rather the task is to see how freedom

may be accommodated in the account of causation. Much of the point of

stressing causation by continuants among the adherents of continuant caus-

ation, in particular agent causation, lies in its role in securing the acceptance of

a coherent account of the phenomenon of freedom. So a denier of freedom is

not the addressee of this paper, although the account of causation we offer is

compatible with the denial of freedom: I am simply trying to establish that it

does not entail lack of freedom. Finally, and more speculatively, I shall pursue

3 Lewis (1986), 202. The terminology is due to Mark Johnston.

234 PETER SIMONS

mydiscussionwithin the framework of amonistic naturalism, the view that all

entities in the universe are within spatiotemporal nature. This rules out

abstract entities, an eternal deity, and any other entity, such as an immortal

soul, which is not by its nature spatiotemporal. One may call this position

‘physicalism’ provided one does not mean by this that everything about the

world can be expressed in the impoverished idiom of physics. Since I regard

all entities in nature to be generated from an ontological basis that includes

fundamental physical kinds and a suite of basic ontological factors4 I prefer the

term ‘generative naturalism’. Finally, and this is consonant with generative

naturalism, that mental states and events are physical: mind/body monism.

Occurrent Priority and Continuant Participation

One could treat participation of a continuant in an occurrent as a meta-

physically basic and unanalysable relationship. This would be the natural

position to adopt if continuants and occurrents are equally primitive and

mutually irreducible categories of entity. The problem is that participation

looks too intimate to be anything like an external relation such as being next

to, but nor is it so intimate as to be identity or part–whole. While not

denying that continuants exist, I claim they are metaphysically posterior to

occurrents, and their nature as persistent enduring individuals has a straight-

forward explanation.5

A continuant like a ball, a rifle, a spacecraft, a human being, exists for a

period of time, changing in various ways throughout its life. For it to

continue as one and the same individual across time certain things about it

have to remain in existence. A ball cannot be cut up or flattened, a rifle

cannot be sawn into pieces, a spacecraft cannot break apart, and a human

being cannot long stop respiring and having electrical activity in the brain.

When we look more closely at what keeps the continuant in existence,

however, we come across processes, more saliently in the case of organisms,

but present also in the more passive cases. The continuous exchange of

photons among the atoms of a body hold it together against forces which

might tend to pull it apart. The constituent atoms are held together by the

exchange of gluons mediating the strong nuclear force. And so on. At the

4 Simons (2009). 5 Simons (2000).

CONTINUANT CAUSATION, FUNDAMENTALITY, AND FREEDOM 235

most physically basic level of apparently partless objects like quarks and

leptons, it might be thought that these simply passively endure, rather than

persist via a constituting process. But quantum theory tells us that at the

most basic level there are no particles, only field values at spatiotemporal

locations. Whether the ultimate ontological assay of this situation will

replace the continuant/occurrent duality with something more basic is

not yet clear, so the claim of occurrent priority partout is presently a

speculative conjecture, and is here acknowledged as such.

The importance of processes to the existence and persistence of continu-

ants also comes out via a different route. When we say truly that a certain

object exists, we may raise the question as to what makes this statement true.

A truth-maker is an entity which makes a statement true because it exists.

Therefore a singular existential statement such as ‘Socrates exists’ would

appear to be made true by the entity its subject term designates, in this

case, Socrates. But there is a snag. The truth-maker for a statement is

supposed to necessitate its truth: that is to say, necessarily, the statement is

true if the truth-maker exists. And indeed Socrates quite happily fulfils this

role with respect to his own existential statement. But qua continuant,

Socrates exists if and only if he exists at some time. So consider any statement

of the form ‘Socrates exists at t’, where t is an arbitrary time. At those times at

which Socrates does not exist, e.g., 600 bce or 600 ce, the statement in

question is false. It is only true for those times at which he exists. So take one,

e.g., 400 bce, and take one particular point in that year. Call this time ‘T’.

What makes it true that Socrates exists at T (or, taking account of the fact that

it is in our past, that Socrates existed at T)? It cannot be Socrates, since it is not

essential to Socrates that he existed at T. He might have died earlier. So

Socrates himself does not necessitate that he exists at T. What then might do

that job? It must be something that, unlike Socrates, has to exist at T if it exists

at all, and has to be intimately related to Socrates. This can only be something

that it has its locations in time essentially, and somehow by existing makes

Socrates exist. This is a process vital to Socrates: its occurring at T is sufficient

for Socrates to exist at T. In fact it is something like a temporal section of

those processes which are necessary and sufficient for Socrates to be alive at

T.Whatever such a process is, it is by its occurring that Socrates exists at T, so

it is a truth-maker for that temporally relativized existence statement.

Continuants exist because they exist at some time. They exist at some

time by virtue of something whose existence at that time is necessary to it

236 PETER SIMONS

and which hand on the sustaining role from one time to succeeding times.

These it seems must be vital processes, whatever these are in detail. The

continuant exists from one time to another by virtue of the continuation of

processes vital to its existence. The relation between phases of such pro-

cesses may be called genidentity. It is in general an equivalence relation, and

applying abstraction to it we arrive at the continuant itself as the invariant

which is identical among the genidentical processes. Although there is more

to genidentity than causal continuity, causal continuity from one phase of a

process to later phases is an important aspect of genidentity. Other things

going on within and to a continuant that are not parts of its vital processes are

nevertheless parts of the total process of what is going on with the continuant

at the time. These include in the case of a person their thoughts and actions,

and within the total process there are smaller strands with causal connections

going back and forth, including the strands leading from thoughts and voli-

tions to actions. These being all processes or other occurrents and their parts,

this is all O-causation. That is what participation consists in: processes that are

part of the total process for the continuant are alongside and rendered possible

by the vital processes in virtue of which the continuant exists.

Again it is an open question whether this analysis applies at the basic

physical level of fields and field values, but for ourselves and the things that

concern us with regard to our lives and actions, the analysis appears tenable.

Continuant and Occurrent Causation

Arguably the most important phenomenon of the natural world is causation,

which I take to be the determining of what happens by something else in

time. This only makes real sense under the open future assumption. The

assumption of naturalism rules out the possibility of any supernatural inter-

vention or determination. Within naturalism there are two potential ways

in which this determination might happen: determination by continuants

and determination by occurrents. It is the thesis of this paper that determin-

ation by occurrents, what I call occurrent causation, is the metaphysically prior

form, and that continuant causation is a derivative and secondary form

of causation, which cannot operate except via occurrent causation. The

ultimate reason for this is the ontological priority of occurrents over con-

tinuants, to which we come now before returning to causation. To have

CONTINUANT CAUSATION, FUNDAMENTALITY, AND FREEDOM 237

short expressions available I shall say continuant causation is C-causation and

occurrent causation is O-causation.

Let’s look at some common or garden examples of occurrent and con-

tinuant causation. First, a bat hitting a ball and the ball’s momentum

changing as a result. This can happen whether the ball was previously

moving or not, and it can happen in a dramatic way, the ball drastically

changing its direction and velocity of motion, or in a mere glancing

deflection. The point is that the event of the bat and the ball colliding

causes the ball’s velocity to change. The collision is an event, brief but not

instantaneous, since the change in momentum is caused by internal distor-

tion and restitution of the ball, which takes a short interval. The subsequent

flight of the ball usually lasts much longer than the collision. This is a typical

case of occurrent-to-occurrent causation. There are of course participant

continuants, namely the bat and the ball, but the changes to them are events

or occurrents.

A more complex case concerns a gas explosion. We imagine a gas pipe or

joint begins to leak gas into the surroundings. This may go on for hours.

The concentration of gas in the surrounding air builds up. Then for some

reason there is an electric spark, and this ignites the gas–air mixture, which

combusts and expands rapidly: an explosion. The explosion, a short-lived

event but with clear phases, is caused by the spark, a short-lived event, but

the spark is only a triggering cause because of the leak. So both the long-

lived leaking and the brief spark are parts of the total cause. Both are

occurrents, and so is the resulting explosion. This is typical for occurrent

causation: the contributory factors are several and it is their conjunction or

coming together which determines that the effect occurs.

We now consider continuant causation. Imagine a snooker player who is

aiming to pot the black. He lines up his cue with the cue ball and the black,

hits the cue ball towards the black and as a result the black ball rolls across the

surface of the table and into the pocket. We say ‘He potted the black’, and

‘pot’ is here an agentive achievement verb. Of course what he did in the first

instance was to move the cue in a certain way so that it hit the cue ball at a

certain angle and in a certain position and with a certain velocity, and the

rest was beyond his control, but it was his skill in hitting the cue ball which

ensured the black was sunk, since the ensuing process of rolling, colliding,

and rolling took place without the player’s further interference, and barring

an earthquake or other unexpected disruptive intervention, the outcome

238 PETER SIMONS

was determined from the moment the cue hit the cue ball, indeed a little

before then. We can quite congruously and correctly say the snooker player

caused the black to go into the pocket.

Take another example with a longer timeline. We imagine a river flowing

swiftly downhill. In the middle of the river is a rock which protrudes from the

water surface and disturbs the water flow. Behind and near the rock, the flow

is turbulent, whereas upstream of the rock and well to the sides and down-

stream, the flow is laminar. We quite happily and correctly say the rock causes

the turbulence. Were it not to be there, and nothing else in its place, the river

would flow smoothly over that place. The way in which the obstacle causes

turbulent flow, while familiar, is, like all turbulence, poorly understood. The

rock and the turbulence may coexist happily for a long period.

Priority of Occurrent Causation

If the fundamental metaphysical entities in space and time are occurrents, then

the fundamental form of causation must be O-causation. C-causation exists

only because continuants do, continuants exist only becauseO-causation does,

so O-causation is prior to C-causation. The solution to the problem of

freedom therefore cannot rest with C-causation, since that is a half-way

house to what is metaphysically fundamental. The problem of freedom is for

that reason not as easily resolved as the adherents of continuant causation

would like. That does not mean it is insoluble, but it does help to explain why

the question of determinismof events in and around the person is so difficult to

reconcile with the idea of freedom, since we cannot simply hive off the free

cause to something outside the usual (O-) causal order.

The derivation from C-causation from O-causation explains also some-

thing that the proponents of agent causation cannot. When an agent acts

freely, that action takes place at a certain time. The time at which the agent

causes the action is therefore either at or just before the action. What makes

it true that the agent causes the action at this time? Since it is contingent to

or inessential of the agent that he or she does so, at any time, let alone this

particular time, then the agent cannot be the truth-maker, for the existence

of the agent is not sufficient for the agent to cause this action at this time, the

agent being a continuant, and the action inessential to the agent. If there is a

truth-maker, it would have to be something whose existence necessitated

CONTINUANT CAUSATION, FUNDAMENTALITY, AND FREEDOM 239

that the action occurs. That points back again to occurrents, but unlike in

the more speculative theory that the continuant is itself constituted by

occurrents, here we have unproblematic occurrents to call upon, namely

those immediately preceding and O-causing the action. The causation can

be timed to the moment the action becomes physically unavoidable, and

that will be because the last O-cause comes about, and its time of occurring

is essential to it.

Of course the agent causation theorist may wish to insist that the state-

ment that the free action occurred at such and such a time has no truth-

maker. But if we can offer a theory in which it does, and the truth-maker is

unproblematic, then that is another reason to accept O-causation as more

basic than C-causation.

But frequently when we use the verb ‘cause’ we are talking about

C-causation, a relation between a continuant and some occurrent it causes.

How is this talk to be squared with the metaphysical prority of O-causation?

Agentive and Causative Verbs

Languages contain many verbs whose meanings include the notion of

causation. Usually this comes from the idea of the agent, typically but not

invariably the subject of the sentence, doing something, describable either

by the verb itself or by the verb in conjunction with noun objects and

complements. We give illustrative examples from English but other lan-

guages provide similar examples:

Sean cleaned the car

Sometimes the subject is not a continuant such as a human being but an

event or other occurrent:

The avalanche destroyed the houses

Some languages have causative cases or special causative endings which

allow us to say that an agent caused (usually in an unspecified way) some-

thing to happen or be the case, specified by a clausal complement.6 In

English this is effected by certain verbs such as ‘make’, ‘have’, ‘get’, and

‘cause’, which are accordingly called causative verbs:

6 For a taxonomy of causatives see Dixon (2000).

240 PETER SIMONS

Jane had the car serviced

Sam made Mary get up

Henry got Sam to leave the room

Alfonse caused the table to tip over

As can be guessed from these samples, agentive and causative verbs more

often take continuant subjects than occurrent subjects. But this linguistic

fact cuts no deep metaphysical ice. It is important at the descriptive,

everyday level but only a datum at the level of fundamental metaphysics.

It does not prove that continuant causation is more basic than occurrent,

since it is almost certainly due at least in good part to the fact that we

ourselves are continuants and have a wide range of verbs to talk about

ourselves and our actions. So now we can formulate our

Main Thesis

Continuant Causation as a Way of Explaining

Freedom

Since agent causation is invoked in good part in order to give a metaphysically

defensible account of freedom, let’s take an example of a free action, one well

known because of its historical importance: the assassination of President John

F. Kennedy by Lee Harvey Oswald. The crucial event in the sequence

bringing about the President’s death is the third and fatal head shot fired by

Oswald. The relation betweenOswald and his act is one of agent-causation: it

is a form, arguably the most important form, of continuant causation.

The argument for freedom invoking this analysis is that since Oswald’s

act is not caused by any prior states or events or processes, it is immune to

criticisms to the effect that these and not Oswald are causally responsible for

Kennedy’s death, or that Oswald was merely the passive vessel or conduit

for an ultimately fatal sequence of events. This argument for agent causation

Continuant causation consists in a continuant’s participation in occurrents which are

occurrent causes of further occurrents, which are said by ‘cause’ and by agentive and

other causative verbs to be caused by the continuant in question.

CONTINUANT CAUSATION, FUNDAMENTALITY, AND FREEDOM 241

has suasive power because it offers a way to escape the supposed incompati-

bility of determinism and freedom.

The picture of freedom it portrays is this: the agent, by bearing or being

host to some (typically) mental event C-causes some (typically) bodily event

which has O-effects which (partly) determine the future: i.e., determine

that some things happen that might not have happened and others do not

happen that might have happened. Note that this analysis involves both

agent-causation and mental causation, both C- and O-causation.

This account of free action raises a budget of metaphysical problems,

which must be reasonably resolved before we can be confident we have a

decent account of freedom (more still may be required). They are:

The old problem of free will and determination

The problem of the relation between the mental and the physical

The problem of the relation between continuants and occurrents

The problem of the relation between C-causation and O-causation.

Our position rejects C-causation as primitive and attempts to explicate it via

O-causation. So our take on these problems is: compatibilism with indeter-

minism with regard to freewill, monism with regard to the mental and the

physical, occurrent priority and an abstractionist account of continuants,

and on the main topic of this paper, O-causation priority. While nothing in

this account is original, it is offered as a coherent and naturalistically

acceptable package.

Three Examples

To illustrate our main thesis we consider three examples in greater detail: a

simple Humean billiard ball example, the Challenger accident of 1986, and

again Oswald’s assassination of Kennedy in 1963.

Example 1: Ball hits ball

A white ball, having been hit by a player’s cue, rolls across the baize towards

a stationary black ball. It hits this, causing the black ball to move away.

There are three phases of this scenario. In the first phase, the black ball is

stationary and the white ball is approaching it. The second phase is the

collision, during which the two balls are in contact. We often treat collisions

242 PETER SIMONS

as momentary or instantaneous events, but they cannot be, because it is only

by changes within the two colliding balls that momentum is transferred

from one to the other. When the balls first come into contact, the inertia of

both as the electromagnetic forces holding each together come into oppos-

ition cause both balls to distort internally, becoming flattened along the

plane of their contact. This sets up internal stresses within the balls which act

to restore the balls to their undisturbed spherical shape. These stresses which

operate while the two balls remain in contact slow the incident white ball

down and speed the target black ball up, in a direction and with a speed

determined by various factors. The third phase consists in the black ball

moving away and the white ball either moving differently than before the

collision or now remaining stationary. So when we say:

The white ball propels the black ball away

this is to be analysed metaphysically according to our Main Thesis as:

for some event E, the white ball is the active participant in E and E O-causes the

resulting motion of the black ball

The white ball is the active participant and the black ball the passive

participant in this particular exchange since the white ball is initially moving

but the black ball is not. If both balls were initially moving both would be

active as well as passive. The event E is in our example the collision of the

two balls, but this information, while a reasonable inference, is not analyt-

ically contained in the sentence (e.g. the balls might be both magnetized and

the change caused by magnetic repulsion). We may suppose neither ball is a

more important or major participant than the other.

Here it seems to me we have a perfectly acceptable O-causation account

of the interaction. The balls are participants in events and processes which

stand in O-causal relations.

Example 2: The Challenger Accident

This event was minutely investigated after it occurred. On 28 January 1986 a

space shuttle was launched fromKennedy Space Center at Cape Canaveral on

mission STS-51-L.About73 seconds into the flight a rapid burning of escaping

fuel occurred, causing the shuttle, comprising two solid rocket boosters

(SRBs), an external fuel tank and the space orbiter Challenger to break up.

All seven astronauts on board were killed. Subsequent disaster investigations

by a committee of inquiry set up by President RonaldReagan established that:

CONTINUANT CAUSATION, FUNDAMENTALITY, AND FREEDOM 243

A combustion gas leak through the right Solid Rocket Motor aft field joint

initiated at or shortly after ignition eventually weakened and/or penetrated the

External Tank initiating vehicle structural breakup and loss of the Space Shuttle

Challenger during STS Mission 51-L.7

Investigation of the debris and footage of the launch established the causes

beyond doubt. They were:

Conditions

Low ambient launch temperature of 2ºC (–2ºC at the crucial place)

Low resiliency of SRBO-ring seals at low temperatures (factor 5 compared to 24ºC)

Out-of-round sections due to prior use (wider gap between tang and clevis at the

seal joint).

These standing conditions, coupled with the build-up of forces and tempera-

tures occurring during the launch led via a sequence of events, detailed in the

investigation, to the explosion and disintegration of the vehicle. The narrative

of the Report, whichmakes sobering reading, is simply a string of events, one

after the other. Continuants are active or passive participants in the events.

They figure chiefly as bearers of causal powers (roundness, resiliency, com-

bustibility, exerting pressure, being under pressure . . . ). The report is under-

standably coy about mentioning the crew, whose deaths constituted the

tragedy of the disaster.

Once again, the causal story is one of event after event, with the nature of

the causation made manifest by the natures of the successive events. Most

disaster reports by their nature have this form.

Example 3: The Kennedy Assassination

We now come to a case of human action, so one closer to the issues of

freedom and continuant causation. It is one we mentioned before. Again

this is an event which has been anatomized minutely many times since it

occurred on 22 November 1963. The Warren Commission Report con-

cluded that the causal story is like this:

Conditions

Oswald hates Kennedy

Oswald is a trained marksman

7 Rogers et al. (1986).

244 PETER SIMONS

Oswald owns a rifle

Oswald works in Dallas at the Texas Book Depository.

Events and Processes

Oswald learns of Kennedy’s visit to Dallas

Decides to (forms intent to) attempt to kill Kennedy

Travels to Irving Texas to get rifle

Returns with rifle

Smuggles rifle into ‘Sniper’s Nest’ in Texas Book Depository next to motor-

cade route

Presidential car drives through Dealey Plaza

On seeing Kennedy, Oswald fires twice: one hit (back of neck), one miss

Reloads breech by operating rifle bolt

Aims again

Squeezes trigger

Rifle fires

Bullet traverses intervening space

Bullet hits Kennedy in the head

Impact causes massive brain trauma, as a result of which

Kennedy dies.

The crucial event in this sequence is clearly Oswald’s firing the third shot.

Assuming that this was a free action on his behalf, we may ask what caused

it. Proponents of the agent causation theory claim it wasOswaldwho caused

it, he did that (maybe an act of volition, maybe a basic voluntary movement)

which caused the gun to fire and thereby caused Kennedy to die. Oppon-

ents of C-causation as ontologically basic deny this and claim that the

cause(s) of the crucial event lie in events, processes and states in and around

Oswald leading up to the firing of the fatal shot. That we say Oswald caused

Kennedy to die is because there are events in which Oswald is the active

participant which caused Kennedy to die, as per our main thesis.

There being these causes does not take away fromOswald’s responsibility

for his action, since the occurrent causes do not pre-empt Oswald from

acting: rather they are part of what his so acting consists in. His action is

free because he is in control: that means he consciously initiates, monitors,

and steers the action. We know from our own case what this is like so there

is no point in belabouring it. If someone claims not to knowwhat it is like to

act freely they are either confused, mistaken, or lying. Since Oswald’s act of

pulling the trigger is causally determined by prior events and conditions, and

CONTINUANT CAUSATION, FUNDAMENTALITY, AND FREEDOM 245

yet is free, I am subscribing here to a form of compatibilism. Since the world

is in general indeterministic there is no need to worry about global deter-

minism, but I agree with those determinists and compatibilists who consider

that if there were sufficiently influential random factors at work in Oswald’s

action then he would not be in control, so for these local purposes we may

consider the act as determined, unproblematically O-caused, with the

crucial O-causing events being within Oswald.

The analysis of Oswald’s action makes him the agent as well as the host of

the crucial initiating events. I claim that there is no competition between

Oswald and the causing events for responsibility, since the term ‘responsi-

bility’ is being used in two different ways when we say the causing events

and Oswald were responsible for the action.

The Package

So to account for freedom within our assumptions, we consider that a

person P freely does an action A when:

P considers whether to do A or not,8 P decides to (forms an intent to) do A,

which either directly or in conjunction with later events O-causes A, where

P does not act under external duress or internal processes over which they

exercise no control.

We have not explicated what is involved in detail when P consciously

monitors and has control of A and some of the processes leading up to it.

Nor will we: the empirical detail is not for philosophy, though the philo-

sophical minimal requirements are that P be aware of the intention to do

something of the type of A and that P at some stage or other shortly before

doing A have been capable of not doing A. Freedom of action is an outcome

of considerable complexity in mentally endowed continuants, not a meta-

physical primitive. This view is consonant with monistic naturalism, and it

accepts continuant causation, but treats it as ontologically secondary to

occurrent causation. It is further open to confirmation or refutation in the

way any empirical account should be: by advances in science, in this case

8 Typically: in routine cases or cases where the action is part of an intended train, individual deliber-

ation about A may be lacking, but it is part of an intended sequence.

246 PETER SIMONS

brain science. It is also a hostage to various possible metaphysical (mis)

fortunes: determinism, dualism, idealism, panpsychism, and primitive

agent causation. I think it is a better framework for explaining freedom

than these other theories, but whether it or something recognizably similar

is right, time and research alone will tell.

References

Dixon, R. M. W. (2000). ‘A Typology of Causatives: Form, Syntax and Meaning’.

In R. M. W. Dixon and A. Y. Aikhenvald (eds.), Changing Valency: Case Studies

in Transitivity. New York: Cambridge University Press: 30–83.

Johnson, W. E. (1924). Logic, Vol. 3. Cambridge: Cambridge University Press.

Lewis, D. K. (1986). On the Plurality of Worlds. Oxford: Blackwell.

Rogers, W. P. et al. (1986). Report of the Presidential Commission on the Space Shuttle

Challenger Accident. Available at http://history.nasa.gov/rogersrep/genindex.htm

Simons, P. M. (2000). ‘Continuants and Occurrents’. Proceedings of the Aristotelian

Society, Supp. Vol. 74: 78–101.

—— (2009). ‘OnticGeneration:Getting Everything from theBasics’. InA.Hieke and

H. Leitgeb (eds.), Reduction—Abstraction—Analysis. Frankfurt/M: Ontos: 137–52.

CONTINUANT CAUSATION, FUNDAMENTALITY, AND FREEDOM 247

11

There is No Exclusion Problem

STEINVOR THOLL ARNADOTTIR

AND TIM CRANE

Introduction

Many philosophers want to say both that everything is determined by the

physical and subject to physical laws and principles, and that certain mental

entities cannot be identified with any physical entities. The problem of

mental causation is to make these two assumptions compatible with the

causal efficacy of the mental. The concern is that this physicalist picture of

the world leaves no space for the causal efficacy of anything non-physical.

The physical, as it is sometimes said, excludes anything non-physical from

doing causal work.

The general shape of the problem is not new. Leibniz famously argued

that Descartes’s conception of the relationship between mind and body had

no place for mental causation. On Descartes’s view, according to Leibniz,

the mind can only affect the body by changing the ‘direction of motion’ of

the body’s ‘animal spirits’. Descartes had held that in this way the total

‘quantity of motion’ was conserved in psychophysical interaction. But

Leibniz claimed that what should be conserved in these interactions is not

quantity of motion but (as we would now put it) quantity of momentum

(mass times velocity). So the mind cannot alter the direction of motion of the

animal spirits without altering the quantity of momentum in the physical

world. The physical law that Leibniz took himself to have discovered

excludes the mental from making a causal difference.1

1 For Leibniz’s views, see Leibniz (1695, 1696). For contemporary discussion, including of the

question of whether Leibniz had correctly interpreted Descartes, see Garber (1983) and Woolhouse

(1985).

Leibniz’s objection to Descartes was based on his view of the nature

of the physical world. Contemporary philosophers also see the problem of

mental causation as arising from assumptions about the physical world (see

Papineau 1990). Partly because of the need to accommodate mental caus-

ation given these assumptions, many philosophers have sought to find a

more intimate connection between the mental and the physical, holding

that mental entities are determined by or constituted by physical entities. But

the problem of mental causation has not gone away. The dominant worry

about mental causation in the last few decades is that the physicalist prin-

ciples to which most contemporary philosophers subscribe still leave no

space for mental causation.

First, it seems that anyone who holds that everything is determined by the

physical will have to say that the mental has physical effects if it is to have any

effects at all. Whatever else the determination thesis might involve, it

involves at least the thesis of global supervenience; the thesis that any

minimal physical duplicate of the actual world is a duplicate in every

respect—a duplicate ‘simpliciter’ ( Jackson 1998; Lewis 1986, 1994). In

other words: fix all the physical facts, and you fix all the facts there are.

But if no changes are possible without the appropriate changes in the

physical, then it is very reasonable to think that the mental must bring

about physical changes if it is to bring about any changes at all.2

Second, there is another widely accepted principle known as the causal

closure of the physical or completeness of physics, which claims that any

physical effect has sufficient physical causes.3 What this means is that

physical entities alone fix the occurrence of any physical effect (either

deterministically, or by fixing the chance of the effect). As we might put

it, once all the physical entities are in place, you don’t need to add anything

in order to get the effect to occur. ‘[P]hysics is causally and explanatorily self-

sufficient: there is no need to go outside the physical domain to find a cause,

or a causal explanation, of a physical event’ (Kim 2005, 16). But if that’s true,

according to this line of thought, then there simply is no room for their

2 This doesn’t strictly follow. All that follows is that in order for the mental to bring about any changes

in the world somethingmust cause changes in the physical. But we will ignore this here. We find it highly

plausible, in any case, that if your desire is to cause you to walk to the fridge it must do so by affecting

some physical change in your body.

3 One might reasonably question the truth of the closure principle (see, e.g., Sturgeon 1998;Cartwright 1999) but we will grant it for the purposes of this discussion.

THERE IS NO EXCLUSION PROBLEM 249

having any mental causes as well. Not, at least, if mental causes are distinct

from physical causes.

The exclusion problem, then, results from the alleged fact that the three

following theses cannot be jointly held:

Mental–Physical Efficacy: There are mental causes of physical effects.

Physical Causal Closure: All physical effects have sufficient physical causes.

Non-Reductivism: Mental causes are distinct from physical causes.

This has been particularly forcefully argued by Jaegwon Kim in a body of

work spanning more than two decades (see, e.g., Kim 1989, 1998, 2005).

And the intended moral is that as nobody ought to reject efficacy or closure,

the incompatibility shows non-reductivism to be untenable. Our aim here

will be to defend non-reductivism against this charge.

Philosophers have responded to the exclusion problem in a number of

different ways, some of which involve significant revisions of ontological

and metaphysical assumptions. Some believe that the problem can only be

solved by returning to reductive physicalism (e.g., Papineau 1990; Kim

1998); others that it should be solved by some kind of dualism (Lowe

2008); or that it needs a different account of the relata of causation (Robb

1997; Ehring 1999; Macdonald and Macdonald 1986; Gibb 2004); or that it

needs a different account of causation (Menzies 2008; Raatikainen 2010).

This is the form the debate has taken in the last decade or two. In this paper,

we shall return to the general form of the mental causation problem and

question one of the assumptions on which it rests.

In our view, we do not need to adopt a new metaphysics of causation, or

of the causal relata, in order to defend non-reductivism against the exclusion

problem. What we need is a proper demonstration of the assumptions

underlying the problem. Moreover, in order to resist the problem in the

way we suggest, we do not need to adopt any specific ontological views

about the relata or metaphysics of causation, or on fundamental ontology in

general. The correct metaphysics of causation is, of course, a huge and

important topic, and a full understanding of mental causation obviously

needs a full understanding of causation. But a response to the exclusion

problem does not need this. To the extent, then, that the exclusion problem

is the mental causation problem, the mental causation problem does not

await an answer to the question of what the correct fundamental ontology

should be.

250 STEINVOR THOLL ARNADOTTIR AND TIM CRANE

1. Whose Problem is It?

Standardly, the exclusion problem is thought to be a problem for all and

only non-reductive physicalists. But the problem arises neither for all nor for

only non-reductive physicalists. In its current form, the exclusion problem

originates from a misguided criticism of Davidson’s anomalous monism in

the 1980s. Davidson’s critics complained that although his theory made

room for mental causes by identifying them with physical causes, it did not

give an adequate causal role to the mental ‘as mental’ (see, e.g., Stoutland

1980; Honderich 1982; Johnston 1985). In essence, the complaint was that

on Davidson’s account all the work is done by the physical properties of

events and this leaves no room for a causal contribution by mental proper-

ties.4 This criticism of Davidson is based on a misunderstanding of his theory

of events and causation, as a number of writers pointed out (e.g. Smith 1984;

Crane 1992; McLaughlin 1993). For Davidson himself denies that properties

play any role in causation. Causation, he holds, relates events as such, and

does not hold ‘in virtue of ’ the properties of those events, or in virtue of

how they are described.We are here using the term ‘non-reductivism’ to be

the thesis that mental causes cannot be identified with physical causes. But

sometimes the term is used more narrowly, for the thesis that mental

properties cannot be identified with physical properties. And if the term is

used in this latter sense, then the exclusion problem does not arise for all

non-reductivists. For Davidson is a non-reductivist in this sense and does

not face the problem.5

The exclusion problem, as it is standardly put forward, is a problem for

those who deny the identification of mental causes with physical causes, while

accepting supervenience and closure. This includes, but is not restricted to,

non-reductive physicalists. For non-reductivists may accept supervenience

4 More specifically, this is because causation implies laws, but mental events can be seen as instances of

laws only under their physical descriptions. But Davidson (1970) had argued that all mental events are

physical events, based on a particular theory of causation—that causation is a relation between events and

that events can be seen to instantiate laws only under some descriptions.

5 It should be noted, perhaps, that although the exclusion problem misses its mark as a criticism of

Davidson, it might nonetheless be thought to be a serious worry for anyone who holds that the (primary)

causal relata are Davidsonian events, while wanting to give a causal role to mental properties. As the

problem is standardly formulated, however, in terms of the sufficiency of one cause and the redundancy

of another, it misses the mark against such views. There may be some alternative formulation of the

problem, appealing to in virtue of locutions, on which a tension can be forced. But we shall not explore

this here. It is our belief that if the standard formulation fails, any such weaker formulation will fail also.

THERE IS NO EXCLUSION PROBLEM 251

and closure without subscribing to physicalism. What makes a non-reductive

view physicalist are two things: its commitment to the core physicalist thesis of

global supervenience, and its commitment to the claim that any fundamental

or ‘brute’ facts must be facts within physics (Horgan 1993). A view that accepts

the former claim but rejects the latter deserves the name ‘emergentist’ rather

than physicalist (see Crane 2001; cf. Kim 2010). Emergentists of this kind take

the mental to be intimately dependent on the physical, but they take this to be

a matter of brute fact, rather than a fact explicable by the facts of physics.

Although emergentists need not accept the closure principle, they may well

do so. And if they do, they too need a response to the exclusion problem.

2. What Exactly is the Problem?

But what exactly is the exclusion problem? Supposedly, the truth of the

closure principle precludes non-reductivists from saying that some physical

effects have mental causes. But why should that be?

The closure principle does not entail that physical effects have no non-

physical causes. For although the physical causes by themselves suffice to fix the

physical effects, this does not rule out that certain physical effects (certain bodily

movements, for instance) might have both physical and non-physical causes.

Indeed, it is perfectly compatible with closure that certain physical effects

(actions, for instance) always have both physical and non-physical causes. So

non-reductivism, it would seem, is not in fact incompatible with efficacy and

closure. To get from the claim that all physical effects have sufficient physical

causes to the conclusion that physical effects have only physical causes, Kim

relies crucially on two further assumptions:

Denial of Overdetermination. Mental causes do not overdetermine their effects.

The Exclusion Principle: ‘No single event can have more than one sufficient cause

occurring at any given time—unless it is a genuine case of overdetermination.’

(Kim 2005, 42)

It is against the backdrop of these assumptions that non-reductivism can be

seen to be incompatible with closure and efficacy. Indeed, if we combine

these assumptions with closure and efficacy, we get a valid argument for the

falsity of non-reductivism:

252 STEINVOR THOLL ARNADOTTIR AND TIM CRANE

1. Mental–Physical Efficacy: There are mental causes of physical effects.

2. Physical Causal Closure: All physical effects have sufficient physical causes.

3. Denial of Overdetermination: Mental causes do not overdetermine their effects.

4. The Exclusion Principle: ‘No single event can have more than one sufficient

cause occurring at any given time—unless it is a genuine case of overdeter-

mination.’ (Kim 2005, 42)

Therefore,

C. Identity of Causes: Mental causes are identical to some physical causes.

The argument so formulated is neutral on what types of entities causes are. If

you think causes are events, then you will take this as an argument for the

identity of mental and physical events. If you think causes and effects are

properties, you will take it as an argument for the identity of mental and

physical properties. Kim’s own view is that instantiations of properties (or

what he calls ‘events’) are causes, and he argues via the identity of such

property instantiations to the identification of mental and physical proper-

ties (see e.g., Kim 1998). We will not assume any particular account of the

causal relata in our discussion. We think the argument fails to establish

the identity of causes, regardless.

A number of philosophers have recently rejected Kim’s exclusion argu-

ment (see especially Bennett 2003; Loewer 2007; Raatikainen 2010) by

questioning some of the assumptions behind it. While these responses have

their merits, they depend on adopting some more or less controversial theses

in the philosophy of language or metaphysics (e.g. Lewis’s semantics

for counterfactuals, causation as counterfactual dependence, or causation as

difference-making). Our approach, by contrast, is to undermine the argument

by making the weakest possible assumptions about causation and ontology. We

will argue that the exclusion principle is contrary to our ordinary judgements

about causation, has no strong independent defence, and ought to be rejected.

3. Overdetermination

Before getting on to our criticism of the exclusion principle, we need to

explain first what is meant by overdetermination and why mental causes and

physical causes do not overdetermine their effects.

THERE IS NO EXCLUSION PROBLEM 253

The cases most naturally thought of as cases of genuine causal overdeter-

mination are cases where there are two or more causes, each of which

would have been sufficient to bring about the effect in the absence of the other.

A classic morbid example is death by two assassins working independently,

where a man is killed by two bullets hitting him at once. Another is where a

window is shattered by two rocks making impact with it at the very same

time. What such cases have in common is that in whatever way the causes

are sufficient, they are sufficient independently of each other. This is shown by

the fact that had either of the causes been deleted, without anything being

added in its place, the other cause would still have caused the effect. What

makes the death of the assassinated man overdetermined, for instance, is that

had one of the assassins not shot him, he would still have been shot dead by

the other assassin. In his recent work, Kim has been explicit that in

formulating the exclusion principle and denying overdetermination he has

in mind overdetermination of this standard type.6

We should agree with Kim that mental and physical causes do not

overdetermine their effects in this way. This is because mental and physical

causes are not independent of one another. This leaves open, however, that

mental and physical causes are nonetheless numerically distinct. For distinct

sufficient causes need not be independent sufficient causes. Indeed, that there

might be distinct but dependently sufficient causes is what we should expect

if physicalism is true; and also if emergentism is true. Physicalists and

emergentists hold that the mental is very intimately dependent on the physical,

but identity is only one way in which this need be so. One popular way of

spelling out this dependence is to say that mental properties are realized by

physical properties, where realization may be understood as follows:

To realize is to ‘make real’ in a sense of ‘makes’ that is constitutive rather than

causal. So a property-realizer of a property is a property whose instantiation

constitutively makes real an instantiation of the realized property. (Shoemaker

2007, 10)

How exactly a theory of realization should be developed is a question that

we need not address here. The important thing in the present context is just

6 ‘The usual notion of overdetermination involves two or more separate and independent causal

chains intersecting at a common effect. Because of Supervenience, however, that is not the kind of

situation we have here. In this sense, this is not a case of genuine causal overdetermination’ (Kim

2005, 48).

254 STEINVOR THOLL ARNADOTTIR AND TIM CRANE

that in claiming that the relationship between mental properties and their

physical realizers is constitutive, the physicalist requires a tighter relationship

than mere correlation (even if this correlation is nomological). The rela-

tionship that holds between mental property instantiations and their phys-

ical realizers is not akin to that which holds between smoke and fire, for

instance. It is more closely analogous to the sort of relationship that holds

between a statue and the lump of clay that makes it up. (That, of course, is a

relationship between particulars; but realization in the relevant sense is

intended to be an analogous relationship between properties.)

This analogy between realization for properties and constitution for

particulars is very helpful in evaluating both the exclusion principle and

the denial of overdetermination. Given this understanding of the relation-

ship between the mental and the physical, mental causes and their physical

realizers could not overdetermine their effects in the ‘independent assassins’

way. For holding everything else fixed, you could not delete one of them

from a given context without thereby deleting the other. If such a relation

were to hold between properties, then this is an excellent reason for denying

that mental causes and their realizers overdetermine their effects.

The important thing to stress here is that even when combined with

mental–physical efficacy and the closure principle, the denial of overdeter-

mination gives us no obvious reason to reject non-reductivism. For an effect

may have distinct causes without being overdetermined, granted that the

causes in question are suitably dependent (see Mellor 1995, 103–4 for a

similar point).

4. The Exclusion Principle

It is the fourth premise of the exclusion argument, the exclusion principle,

that challenges the compatibility of non-identity with premises (1)–(3).

According to this principle ‘no single event can have more than one

sufficient cause occurring at any given time—unless it is a genuine case of

overdetermination’. And if that’s true, then it is clear that non-reductivism

cannot make room for mental causation, given the truth of closure and

the denial of overdetermination. It is on this principle that the case against

non-reductivism rests and the remainder of our discussion will be devoted

to examining it.

THERE IS NO EXCLUSION PROBLEM 255

4.1. Causal Exclusion and Explanatory Exclusion

How did the exclusion principle enter this debate? In his first significant

intervention in this debate, ‘Mechanism, Purpose and Explanatory Exclu-

sion’ (1989), Kim talked in terms of what he called the principle of explana-

tory exclusion: ‘no event can be given more than one complete and

independent explanation’ (1989, 79). But over the years, Kim has changed

from talking about explanatory exclusion to talking about causation. His

causal exclusion principle states that ‘[n]o single event can have more than

one sufficient cause occurring at any given time—unless it is a genuine case

of overdetermination’ (Kim, 2005, 42).

Kim does not say why he has moved from talk of explanation to talk of

causation. But this move is certainly a move in the right direction, since the

problem of mental causation is a problem about causation and not explan-

ation.7 There is not even a prima facie difficulty of mental explanations being

incompatible with, or ‘crowded out by’ physical explanations (see Burge

1993). Any occurrence can be explained in countless ways, and there is no

incompatibility between any physical explanation of an event and a mental

explanation of the same event. So there seems to be little plausibility to the

idea that one explanation ‘excludes’ another.

However, Kim’s ‘explanatory exclusion’ principle does talk about

complete explanations. What is a complete explanation? Suppose it is the

conjunction of all the many different true explanations of the event—if we

can make sense of this idea. Then of course, no event can have more than

one complete explanation in this sense. So it seems that the explanatory

exclusion principle is either obviously false or trivially true. The same is not

true of the causal exclusion principle. Though perhaps it is the connection

with the (possibly illusory) idea of a ‘complete’ explanation of an event that

lies behind Kim’s thought that the exclusion principle is an obviously true

principle that requires no defence.

7 It should be said that how closely related explanatory and causal exclusion are depends very much on

your account of causation and of causal explanation. On Kim’s account the same things can serve as the

relata of causation and causal explanation (i.e., property instantiations or facts) while for others, e.g.,

Davidson, causation and explanation relate very different things. It is only on the latter sort of view that

causal explanatory exclusion is much less plausible than causal exclusion. It is fair to say, however, that if

one wants to press a perfectly general problem of mental causation, one ought to avoid building in

assumptions about the relation between causation and explanation. And on its face the principle of causal

explanatory exclusion seems a lot less plausible than the principle of causal exclusion.

256 STEINVOR THOLL ARNADOTTIR AND TIM CRANE

4.2. Why the Causal Exclusion Principle is Substantive

Kim remarks that the causal exclusion principle is ‘virtually an analytic truth

with not much content’ (2005, 51). But given what he means by ‘genuine

overdetermination’, and given what he has to mean by ‘more than one’, this

claim is clearly mistaken. There is a principle in the vicinity of the exclusion

principle that is a better candidate for being ‘virtually an analytic truth’;

namely, that no single event can have more than one independently sufficient

cause occurring at any given time—unless it is a genuine case of overdeter-

mination. That much just falls out of the notion of overdetermination under

discussion. However, that principle entails no stronger conclusion than that

mental causes are not sufficient independently of physical causes, and as we

have seen that is not a claim that is in tension with non-reductive physical-

ism or emergence.

Kim’s causal exclusion principle is not specified in terms of independently

sufficient causes, but only in terms of distinct sufficient causes. It says that

there cannot be distinct sufficient causes of an event occurring at any given

time except in cases of genuine overdetermination. This claim, we argue, is

false. We have two main objections. First, it seems to us that far from being

an analytic truth, the exclusion principle is not even plausible on its face. It

conflicts with our causal judgements even before any physicalist commit-

ments enter the picture and is subject to a number of counter-examples.

Given this, we ought to demand some very good arguments to persuade us

that the principle is true. Our second objection is that we lack such

arguments.

4.3. On the Implausibility of the Exclusion Principle

Notice, to begin with, that the principle as stated involves the important

qualification ‘at the same time’. Without this qualification it would be

refuted by the simple fact that every effect has many causes, stretching

back across time. Think of a causal ‘chain’ where A causes B and B causes

C. Even without assuming the transitivity of causation, A and B can both be

causes of C. And if all causes are sufficient causes, then an effect can

therefore have many sufficient causes across time: everyone should accept

this. This is presumably why Kim adds the qualification, ‘at the same time’.

But why should it be more plausible with this qualification added? Even

at a time, events have many causes, and not just in cases of genuine

THERE IS NO EXCLUSION PROBLEM 257

overdetermination. J. L. Mackie’s (1965) famous discussion of a short-circuit

causing a fire illustrates this nicely. Putting to one side the distinction

between cause and background conditions, it is natural to say that the

presence of oxygen and the presence of flammable material are causes of

the fire, just as much as the short-circuit is. But these are states of affairs, or

property-instances, that exist at the same time as the fire. So the short-

circuit can have many causes occurring at the same time.

But it may be said (as Mackie himself did) that none of these are sufficient

causes. If ‘sufficient’ means absolutely sufficient on its own, then they are not

sufficient causes. But is there anything which is absolutely sufficient on its

own for the occurrence of an event? Those who believe in sufficiency in

this sense might appeal to the fact that there is an entire state of the universe

before the occurrence of the event which is sufficient for that event’s

occurrence. Now if determinism is true, then there must be such a state.

Whether or not this gives any plausibility to the claim that no event can

have more than one sufficient cause at the same time depends on the

relationship between this idea of the entire state of the universe, and the idea

of something’s being a cause, or a sufficient cause. Of course, there have been

theories which make a close connection between these ideas—Mill’s notion

of the ‘whole cause’ is the most famous—but these are specific accounts of

causation, and are not uncontroversial.

By contrast, the exclusion argument is not supposed to rely on any

particular account of causation. The exclusion principle is intended as a

general principle that one ought to accept whatever one’s account of

causation and the causal relata: so we should expect the principle to accord

with ordinary causal claims. But it does not appear to do so. Here is an

example. We are in general happy to attribute causal powers to ordinary

objects (see Lowe 2008). We say things like ‘the furniture scratched the

floor’, ‘my shoe gave me blisters’, or ‘the hammer made an indentation

in the clay’, for instance. But where we are happy to say that objects

caused things we are often happy to say at the same time that their parts

caused things. Suppose, for instance, that the indentation that my hammer

makes in the soft clay on top of which I place it, is made by the hammer’s

head and not by its shaft. There seems no tension in saying both that the

hammer caused the indentation in the clay and that its head did. But the

hammer and its head are numerically distinct things, so this would violate

the exclusion principle—so long as they are sufficient causes.

258 STEINVOR THOLL ARNADOTTIR AND TIM CRANE

Are they sufficient causes? Not, of course, in the sense of ‘absolutely

sufficient’—the sense in which Mill’s entire state of the universe is suffi-

cient. But they are sufficient in the sense in which any cause we can know

about is sufficient. The causes that we cite as sufficient or necessary are only

sufficient or necessary given other factors, including other causes and maybe

the laws of nature. This is what people mean when they sometimes say that

causes are sufficient ‘in the circumstances’. Our everyday commonsense

way of thinking about causes—part of the data which the metaphysics of

causation is arguably supposed to explain—only requires causes to be suffi-

cient in this sense.

4.4. On the Lack of Argument

Given the above considerations, if we are going to defend the exclusion

principle, we need a strong argument. Our second main claim is that

we lack such an argument. Kim suggests in a number of places that it

would be very odd if actions always had two sufficient causes.8 But why is

it odd? One answer is that it is odd if the causes are not just distinct but also

independent and absolutely sufficient. It would, of course, be an astonishing

coincidence if our actions always had two independent and absolutely sufficient

causes, and this would cry out for some explanation. But the non-reducti-

vist can deny both that the mental and physical case is a case of two

absolutely sufficient causes, and that it is a case of independent causes.

First, non-reductivism can deny that the causes are absolutely sufficient.

Physicalists will want to say, of course, that the physical causes are as close to

absolute sufficiency as any cause gets. For example, they may say that given

determinism and the transitivity of causation, the state of the universe at the

Big Bang is a cause of today’s weather in Iceland. Of course, even this is only

sufficient given the laws of nature, on the usual way of thinking about these

matters. But nonetheless it is as close to absolute sufficiency as we get.

Second, the important point for non-reductivists is that the mental cause

of a physical effect is not a candidate for being sufficient in anything like this

absolute sense, given that it is dependent on its physical basis. Moreover,

this dependency explains why the mental cause is not sufficient in the

sense of being absolutely sufficient. So once we acknowledge this intimate

8 E.g. ‘[i]t is at best extremely odd to think that each and every bit of action we perform [has] two

distinct sufficient causes’ (Kim 1989, 86).

THERE IS NO EXCLUSION PROBLEM 259

relationship between the mental and the physical—assumed both by

physicalists and emergentists—there is nothing strange about our actions

having both mental and physical causes. It certainly wouldn’t be a coinci-

dence nor would it be something for which we lacked an explanation.

So what is the argument for the exclusion principle? Why should we

accept that there cannot be distinct sufficient causes except for in cases of

genuine overdetermination? It seems to us that the above considerations

seriously undermine this contention. First, we seem only too happy to

speak as if there were distinct sufficient causes where overdetermination

clearly does not apply; i.e., where the causes in question are not independ-

ently sufficient. Second, once we acknowledge that the causes in question

are not independent the reservations that one might have about allowing

distinct sufficient causes ought to let up. Curiously, however, it turns out

that Kim thinks it is precisely because the two causes are not independent that

there is a problem in acknowledging both of them:

[O]ur problem is not exactly that of causal overdetermination, although both

have to do with an overabundance of causes. It is important to see that the

problem that we face arises because the two putative causes are not independent

events. The difficulty is exactly that the causal status of the dependent event is

threatened by the event on which it depends. (1998, 53)

But how exactly is the causal status of the dependent event threatened by

the event on which it depends? Kim’s answer seems to be that given that the

causal powers of the dependent event are determined by the event on which

it depends, it couldn’t bring any additional causal powers to the picture, and

this, he thinks, means that it cannot play any causal role. (This is the basis of

what he came to call in his 1998 book the ‘supervenience argument’.) In

Mind in a Physical World, Kim claims ‘there is a real problem, the exclusion

problem, in recognizing second-order properties as causally efficacious in

addition to their realizers’ (1998, 53, our emphasis). He goes on to explain

what the problem is, as follows:

For there is nothing in the instantiation of [the second-order property] F on this

occasion over and above the instantiation of its realizer H. Given this, to think

that this instance of F has causal powers in excess of these of H is tantamount to

belief in magic. (1998, 54–5, our emphasis)

260 STEINVOR THOLL ARNADOTTIR AND TIM CRANE

Again, the idea is that where an effect already has a sufficient cause, any

simultaneous distinct sufficient cause would have to bring some additional

causal powers to the scene. Let us call this the motivating principle.

We find this motivating principle no more plausible than the exclusion

principle. We do not have to admit that instantiations of mental properties

have additional causal powers (on any given occasion) to those bestowed on

them by their realizers. Indeed, it is difficult to see how anybody who believes

in global supervenience could think this possible. For if the thesis of global

supervenience is true, then fixing the physical facts fixes all the facts there are,

including causal facts. If we have to accept that this rules mental properties out

from being causally efficacious, we still need to know why. Just stating that a

thing could not be causally efficacious unless it brought additional causal

powers to those already determined by the physical seems straightforwardly

to beg the question against those who believe both that global supervenience

holds and that there are causally efficacious things that are neither identical nor

separate entities, but rather stand in the relation of constitution or realization

to one another. In other words, the motivating principle begs the question

against the very people against whom the exclusion argument is directed.

The comparison with particular objects is instructive at this point. Many

philosophers hold that statues are constituted by rather than identical to the

lumps of matter that make them up. Statues and lumps, they claim, are distinct

material objects that share all their matter and microphysical properties. If this

is right, then when you place a copper statue in a tub of water with the effect

that the water level rises, you place numerically distinct material objects in the

water: namely, the statue, and the lump of copper that constitutes it. What’s

more, it seems that the statue is sufficient (given the other causes) for raising

the water level and also that the lump is sufficient (given the other causes) for

raising the water level. But the statue and the lump certainly do not overdeter-

mine the raising of the water level. Nor does the statue add anything to water-

raising powers already put in play by the lump. Those who believe that statues

are distinct from lumps should take this as reason to reject Kim’s principles.9

It might be replied, of course, that we should deny that statues are distinct

from lumps: they are identical. But our present point is that we should not

9 The example can also be reconstructed in terms of property instantiations. Given that property

instantiations are individuated by the individuals instantiating the properties, the lump being submerged

in the water and the statue being submerged in the water are distinct property instantiations, each of

which is sufficient for the water level’s rising; and neither of which adds any causal powers to those of the

other one.

THERE IS NO EXCLUSION PROBLEM 261

insist that this is so simply because at any given time the statue has no causal

powers in addition to those of the lump. Because if statues were in fact

constituted by lumps we should expect them to have no causal powers in

addition to those of the lump at any given time. Note that the ‘at any given

time’ clause is important here. There being a statue can be relevant in all

sorts of ways. And on occasions where the lump does not constitute a statue,

coming to do so might well give it new causal powers. But on any given

occasion where the lump is such that it constitutes a statue, the statue brings

no extra causal powers to those already determined by the lump.

Similarly, we should not argue that mental property instantiations must

be identical to, rather than realized by, physical property instantiations on

grounds that mental property instantiations don’t have any causal powers in

excess of their realizers. For if mental property instantiations were in fact

realized by, rather than identical to, physical property instantiations, that’s

exactly what we should expect.

Our conclusion is that we have no good reason to accept the exclusion

principle. It fits badly with ordinary causal judgements, it is unsupported by

argument, and it begs the question against those who believe in consti-

tutively related causes. Since non-reductivists—non-reductive physicalists

and emergentists alike—agree that the causes in question are intimately

related, there is no reason why they should be moved by the appeal to the

exclusion principle.

5. Concluding Remarks

Once we have rejected the exclusion principle, we have disarmed the

exclusion argument. A non-reductive physicalist or an emergentist can

accept premises (1)–(3) of the argument, so long as they accept that mental

and physical causes are intimately dependent. (It is a further question of

whether this dependence can be explained—i.e., whether dependence is a

‘brute fact’ as some emergentists claim.) Intimate dependency is also de-

pendency of causal powers, and this is why non-reductivists can reject

Kim’s motivating principle that non-identical causes would have to bring

additional causal powers to those determined by the causes they depend on.

262 STEINVOR THOLL ARNADOTTIR AND TIM CRANE

These points are illustrated by the example of constitution between distinct

objects.

What is more, we can disarm the exclusion argument without making

very many heavy-duty assumptions about causation. The only substantive

claims we are making about causation is that every effect can have many

causes, both over time and at a time, and that these causes can be sufficient

(in one of a number of senses typically appealed to by theories of causation).

According to physicalists, closure entails that the physical causes are as

absolutely sufficient for their effects as any cause can be. Mental causes

will not be sufficient in this sense: they will only be sufficient given the

other causes and other factors. But contra Kim, this is not something which

undermines the mental cause’s status as a cause. For one thing, most of the

causes we know about are sufficient ‘in the circumstances’ and not abso-

lutely sufficient. But more importantly in this context, the mental cause is

dependent on the physical cause. This is the essence of what it means to be a

non-reductive physicalist, or an emergentist in our sense. So if these forms

of non-reductivism are true, then we should not expect that mental causes

would be absolutely sufficient.

If the closure principle is true, then the physical cause suffices for the

physical effect. But since the mental cause is dependent on the physical

cause, the latter also suffices for the former. Since the mental cause is, by

hypothesis, a cause of the physical effect, this shows that the physical effect

can have more than one cause. The mental cause is sufficient for this effect

too; it’s just that it is sufficient given the physical cause. But this is the sense

in which most causes are sufficient, or at least the sense in which most

theories of causation allow themselves to talk of ‘sufficient causes’.

It would be a mistake to conclude from this that ‘all that matters’ is the

physical cause. For this would assume that the physical cause could be there

without the mental cause. But this contradicts the supposition, common to

both non-reductive views being considered, that the physical cause neces-

sitates the mental cause (see Loewer and LePore 1987; Loewer 2007). So we

should reject the idea that all that matters is the physical cause. Compare our

analogy with the statue again: just because the statue is determined by the

arrangement of the clay, this does not mean that all that matters is that there

is clay there. It matters that the clay gives rise to a statue.

Our response is available to those holding a wide variety of views about

the relata of causation. Causes can be ordinary objects, or substances (cf.

THERE IS NO EXCLUSION PROBLEM 263

Lowe 2008). They can be events as conceived by Davidson (1967) or

property-instantiations as conceived by Kim (1980) or differently conceived

by Macdonald and Macdonald (1986). They can be facta in Mellor’s (1995)

sense. Or they can be tropes, as Ehring (1999) and Robb (1997) think. None

of these views makes any difference to the way the exclusion problem

should be treated. Once the causal exclusion principle is rejected, then it

is clear that non-reductivism is not threatened by the conjunction of (1)–(3).

The correct ontology can be argued about at a later stage.10

References

Bennett, K. (2003). ‘Why the Exclusion Problem Seems Intractable and How, Just

Maybe, to Tract it’. Nous, 37(3): 471–97.

Burge, T. (1993). ‘Mind–Body Causation and Explanatory Practice’. In J. Heil and

A. Mele (eds.), Mental Causation. Oxford: Clarendon Press: 97–120.

Cartwright, N. (1999). The Dappled World: A Study of the Boundaries of Science.

Cambridge: Cambridge University Press.

Crane, T. (1992). ‘Mental Causation and Mental Reality’. Proceedings of the Aristo-

telian Society, 92: 1–18.

——(2001). Elements of Mind. Oxford: Oxford University Press.

Davidson, D. (1967). ‘Causal Relations’. Repr. in (2001), Essays on Actions and

Events, 2nd edn. Oxford: Clarendon Press, 149–62.

——(1970). ‘Mental Events’. Repr. in (2001), Essays on Actions and Events, 2nd edn.

Oxford: Clarendon Press, 207–24.

Ehring, D. (1999). ‘Tropeless in Seattle: The Cure for Insomnia’. Analysis, 59(261):

19–24.

Garber, D. (1983). ‘Mind, Body and the Laws of Nature in Descartes and Leibniz’.

Midwest Studies in Philosophy, 8(1): 105–33.

Gibb, S.C. (2004). ‘The Problem of Mental Causation and the Nature of Proper-

ties’. Australasian Journal of Philosophy, 82(3): 464–76.

Honderich, T. (1982). ‘The Argument for Anomalous Monism’. Analysis, 42:

59–64.

10 We would like to thank: the Leverhulme Trust and the Isaac Newton Trust for their generous

support for Steinvor Arnadottir’s research; two anonymous reviewers for OUP for their helpful

comments; and Jonathan Birch for his insightful remarks when some of this material was presented by

Tim Crane in a seminar in Cambridge.

264 STEINVOR THOLL ARNADOTTIR AND TIM CRANE

Horgan, T. (1993). ‘From Supervenience to Superdupervenience: Meeting the

Demands of the Material World’. Mind, 102: 555–86.

Jackson, F. (1998). From Metaphysics to Ethics. Oxford: Clarendon Press.

Johnston, M. (1985). ‘Why Having a Mind Matters’. In E. LePore and

B. McLaughlin (eds.), Actions and Events: Perspectives on the Philosophy of Donald

Davidson. Oxford: Basil Blackwell: 408–26.

Kim, J. (1980). ‘Events as Property Exemplifications’. In M. Brand and C. Walton

(eds.), Action Theory. Dordrech: Reidel: 159–77.

——(1989). ‘Mechanism, Purpose and Explanatory Exclusion’. Philosophical Per-

spectives, 3: 77–108.

——(1998). Mind in a Physical World. Cambridge, MA: MIT Press.

——(2005). Physicalism, or Something Near Enough. Princeton: Princeton University

Press.

——(2010). ‘Emergence: Core Ideas and Issues’. In Essays in the Metaphysics of

Mind. Cambridge: Cambridge University Press: 66–84.

Leibniz, G. W. (1695). ‘New System of the Nature of Substances’. In

R. S. Woolhouse and R. Franks (trans. and eds.) (1998), G. W. Leibniz: Philo-

sophical Texts. Oxford: Oxford University Press: 143–52.

——(1696). ‘First Explanation of the New System of the Communication between

Substances’ & ‘Third Explanation of the New System of the Communication

between Substances’. In R. S. Woolhouse and R. Franks (trans. and eds.) (1998),

G. W. Leibniz: Philosophical Texts. Oxford: Oxford University Press, 187–93.

Lewis, D. (1986). On the Plurality of Worlds. Oxford: Basil Blackwell.

——(1994). ‘Reduction of Mind’. In S. Guttenplan (ed.), A Companion to Philoso-

phy of Mind. Oxford: Basil Blackwell: 412–31.

Loewer B. (2007). ‘Mental Causation, or Something Near Enough’. In

B. McLaughlin and J. Cohen (eds.), Contemporary Debates in Philosophy of Mind.

Oxford: Basil Blackwell: 243–64.

Loewer, B. and E. LePore (1987). ‘Mind Matters’. Journal of Philosophy, 84: 630–42.

Lowe, E. J. (2008). Personal Agency: The Metaphysics of Mind and Action. Oxford:

Oxford University Press.

Macdonald, C. and G. Macdonald (1986). ‘Mental Causes and Explanation of

Action’. Philosophical Quarterly, 36: 145–58.

Mackie, J. L. (1965). ‘Causes and Conditions’. American Philosophical Quarterly, 2(4):

245–64.

McLaughlin, B. P. (1993). ‘On Davidson’s Response to the Charge of Epipheno-

menalism’. In A. Mele and J. Heil (eds.), Mental Causation. Oxford: Oxford

University Press: 27–40.

Mellor, D. H. (1995). The Facts of Causation. London and New York: Routledge.

THERE IS NO EXCLUSION PROBLEM 265

Menzies, P. (2008). ‘The Exclusion Problem, the Determination Relation and

Contrastive Causation’. In J. Hohwy and J. Kallestrup (eds.), Being Reduced: New

Essays on Reduction, Explanation, and Causation. New York: Oxford University

Press: 196–217.

Papineau, D. (1990). ‘Why Supervenience?’ Analysis, 50(2): 66–71.

Raatikainen, P. (2010). ‘Causation, Exclusion, and the Special Sciences’. Erkennt-

nis, 73: 349–63.

Robb, D. (1997). ‘The Properties of Mental Causation’. Philosophical Quarterly,

47: 178–94.

Shoemaker, S. (2007). Physical Realization. Oxford: Oxford University Press.

Smith, P. (1984). ‘Anomalous Monism and Epiphenomenalism: A Reply to

Honderich’. Analysis, 44(2): 83–6.

Stoutland, F. (1980). ‘Oblique Causation and Reasons for Action’. Synthese, 43:

351–67.

Sturgeon, S. (1998). ‘Physicalism and Overdetermination’. Mind, 107: 411–32.

Woolhouse, R. (1985). ‘Leibniz’s Reaction to Cartesian Interaction’. Proceedings of

the Aristotelian Society, 86: 69–82.

266 STEINVOR THOLL ARNADOTTIR AND TIM CRANE

Index

Bold numbers denote reference to illustrations.

Andersen, R. 62–3, 66Anjum, R. 200 n. 6, 201anomalous monism, see monismArmstrong, D. M. 8, 28, 197, 224–9Arnadottir, S. T. 16autonomyof higher-level properties 61of the mental 60, 88, 226of mental causes 131–5of mental properties 61of the physical 20–2; see also causal

closureof psychology 23of psychological explanation 55

Bacon, J. 224 n. 3Bennett, K. 9 n. 10, 21, 81, 132, 220, 253Bjornsson, G. 176 n. 6, 276 n. 7Block, N. 68Boyd, R. 41, 42, 216, 218, 220Burge, T. 215, 256

Campbell, K. 31 n., 107, 216, 217, 224,227–9

Carnap, R. 227causationagent 9, 162–3, 171, 175, 176, 180–90,

234, 239–47continuant 16, 234–47contrastive 93–4counterfactual 11, 19, 93, 174–5, 201, 203as difference-making 14, 60, 72 n. 5, 75,

77–8, 84–5, 253as transference 11, 12, 201 n. 9, 229event 6, 9, 10, 24 n., 167, 169, 186fact 9–10generalist 11higher level 126–7macroscopic 14, 137, 149 126–51nomological 12; see also genereralist

causation; principle of thenomological character of causality

occurrent 16, 234–47psychophysical 2, 5; see also principle of

causal interactionpowers theory of 15, 194, 196–203, 204,

206, 210, 212pre-emptive 72 n., 93, 95probabilistic 63, 128, 130–4, 159, 178, 188property 92, 111–23singular 6, 11substance 15, 153–71, 183, 185–6, 190supervenient 22and thermodynamics 129–30, 133

causal asymmetry 127–134, 141causal closure 1–5, 12–13, 70, 122, 149–50,

167–9, 193–213, 216–21, 249–53,255, 263

principle of 2, 149, 168causal efficacy 5, 22–3, 46, 50, 64, 71,

134, 148causal exclusionargument from 14, 60, 61–5downward 61, 77–8, 80, 84–5Kim’s argument from 14, 65, 84–5, 88,

90, 253Kim’s principle of 64new principle of 64, 68–9, 71principle of 14, 77–81, 90–1, 193, 252problem of 16, 250revised principle 77–8upward 61, 77–8, 85

causal network 90, 96causal non-overdetermination 2–5, 13

principle of causal non–overdetermination 2

causal overdetermination 2–16, 39, 91,169, 252–5

argument from 2causal power, see powerscausal process 14, 58, 94, 129, 138–42causal profile 36, 38, 39, 41, 43–54, 81–2, 98,

101, 110, 117–20, 146, 225causal reductionism 176–7

causal sufficiency 63, 67, 69, 71–2, 77,82–5, 96

Chalmers, D. 219Clapp, L. 41Clarke, R. 175 n. 3, 180–1colors 30, 39, 224co-manifestation, see mutual manifestationcompatibilism 61, 81–4, 185 n. 19, 242, 246continuant 233–7; see also substancesconstitution 55, 101, 104, 105, 118, 119, 220,

255, 261, 263as co-ordination of the small 121and grounding 121, 123physical 100property 112property instance 112, 119

conventionalism (about exactresemblance) 25

counterfactualdependence 19–21, 94, 115, 174, 175,

201–3, 208, 215, 253circumstances 196co-variation 175, 177

counterfactuals 20, 74, 76–80, 84–5, 94, 116,144, 146, 147, 174, 253

and difference making 85, 94Crane, T. 16, 23, 251, 251

Daly, C. 222Davidson, D. 6–9, 22–9, 32, 188 n. 22,

219–21, 251, 256 n., 264Descartes 18–20, 24 n. 2, 58, 65–9, 248–9determinable-determinate 117, 220, 224disembodied 219double prevention 15, 194, 196–213

Ehring, D. 97, 103, 222, 228, 229, 250, 264Ekstrom, L. W. 179 n. 14emergent dualism 89

epiphenomenal 118–19emergentism 16, 190, 252, 254, 260, 262–3epiphenomenalism

of the mental 21, 37, 60, 62, 65, 111, 149,190, 193

of properties 7–8, 69of types 21, 22

eventsDavidsonian 26, 222 n., 251 n. 5identity of mental and physical 3–7, 22,

24–6, 168, 195, 251 n. 4, 253Kim-style 26, 28, 222, 225–6, 253, 264

mental 2–7, 10, 15, 18–26, 154, 169,194–5, 208–12, 220–1, 251 n. 4

monism 4, 7physical 2–4, 6–10, 12, 18, 21–7, 168–9,

193–5, 203–4, 206–12, 249as property instantiations 4, 8; see also

Kim-style eventsexplanation 51–2, 89, 93, 108, 114, 120, 129,

144, 147–8, 176, 181causal 72, 80, 249complete 256contrastive 187–8and control 175, 186–90and double preventers 212intensional 228 n. 6mental 148physical 139, 141, 144, 256problem of 187psychological 19–20, 23, 55rational 165scientific 31, 187thermodynamic 129, 136

explanatory exclusion 256explanatory power 174

Fodor, J. A. 21, 59, 113, 114,135, 216

Freedommetaphysical 15, 175, 176, 190of action 163, 167, 234, 241–2, 246agent-causal account of 174and moral responsibility 173–4and neo-Humeanism 174–7problem of 239of will 163, 169–171, 242; see also the will,

willingfunctionalism 142–43, 148

Gibb, S. C. 8, 15, 201 n. 9, 223–4, 250Gibbons, J. 21, 94, 105Groff, R. 187 n. 20

Heil, J. 10, 14, 31, 109, 177 n. 8, 190 n. 24,197, 215, 225–7

holistic constraints 219Honderich, T. 7 n. 5, 228, 241

Identityconditions 10, 11, 154–5of causes 253and causal profile 81

268 INDEX

of events 4, 195genidentity 237of the mental and physical 4, 22, 37,

65–6, 215–29, 253–4; see alsopsychophysical identity

and ontological dependence 154primitive 97psychophysical 38, 53, 59–60, 216–17of property instances 4, 92, 96–108,

111, 253relative 227of properties 10, 79, 92, 195, 215–29token 22, 25–7type 22, 25–7, 28, 54, 99, 135, 138, 150,

215–29immanence 220–1incompatibilism 185individuation 31, 93, 102, 104, 153, 156–7,

182, 222, 261 n.intentionalaction 1, 9agents 58, 175, 190behaviour 81, 83cause 185content 165, 195indeterministic causes 185

intentionality 190, 225

Jackson, F. 219, 220, 249Jacobs, J. 15, 177 n. 9, 182 n. 18, 186 n.James, W. 210–13Johnson, W. E. 234

Kane, R. 180 n. 14Kim, J. 6–7, 14, 20–1, 26–8, 38, 46, 58, 60,

64, 69–71, 88–92, 99–100, 127, 149n., 168 n., 190 n. 23, 216, 222, 225–6,249–50, 252–4, 257–64

Kornblith, H. 21Kripke, S. 219

laws 140basic, see fundamental lawscausal 11, 21, 23, 50–1, 113, 170–1and causal profile, see causal roleand causal relevance 113and causal role 115, 117ceteris paribus 20, 113, 114dynamical 126–7, 140exceptionless, see strict lawsfundamental 29, 63, 67, 119, 206

as linguistic items 29macroscopic 141of nature 174, 178, 180–81, 227 n., 259and patterns of variation 113physical, see laws of physicsof physics 55, 59, 63, 67, 89, 206, 248physically reducible 141and powers 206and properties 89psychological 101, 112psychophysical 7, 206as regularities 11, 177strict 7, 29, 113, 177and supervenience 140and thermodynamics 129–30, 141

law-statements 29Leibniz 248–9LePore, E. 21, 132, 135, 263levels 67; see also unilevellersof being 10, 13of description 11fundamental 69hierarchy of 48, 69, 83macro to micro 70ontological 11and supervenience 70

Lewis, D. 36, 74–5, 112, 130–1, 174–5, 196,224–5, 234, 249, 253

libertarianismagent- causal 184–6event-causal 177, 179

Libet, B. 209–10Lipton, P. 187List, C. 61, 78, 80, 93, 95, 115–16, 132,

135, 149Locke, J. 154, 160, 164Loewer, B. 21, 132, 135, 215, 253, 263Lowe, E. J. 2 n. 2, 5 n. 4, 9, 10, 15, 183–4

193–5, 207 n., 208, 220–1, 250,258, 264

Macdonald, C. 21, 27–8, 220, 250, 264Macdonald, G. 21, 27–8, 220, 250, 264McGinn, C. 6Mackie, J. L. 258McLaughlin, B. 22, 42, 251Malcolm, N. 219manifestationsof continuants 233of dispositions, see manifestations of

powers

INDEX 269

manifestations (cont.)mutual 184, 197–8, 200of active powers 159, 161of passive powers, see non-causal powersof powers 164, 199, 200, 201, 203–4,

207, 215of non-causal powers 160, 183of two–way powers 164type 153, 156–61unmanifested 197of volition 160, 163, 183

Markosian, N. 185 n. 19Martin, C. B. 31, n., 118, 177 n., 194 n. 2,

197, 207Maurin, A. 227Mellor, D. H. 23, 255, 264mental

causal relevance of the 71–7, 193dependence on the physical 20distinct from the physical, see autonomyefficacy of the 5, 22, 23, 35, 37, 56,

65–71, 179, 180, 220–1, 226,228 n. 6, 248

privacy of the 219subjectivity of the 8, 15supervenes on the physical 22, 27 n., 60,

65–6, 85, 126, 132Menzies, P. 14, 93, 95, 115–16, 132, 135, 149,

174 n. 2, 205 n., 250metaphysical necessity 89, 111, 118metaphysics: see also ontology

causal powers 178, 182–6, 189–90neo-Aristotelian 15, 174, 177–9, 182, 185,

186, 190neo-Humean 174–7, 178, 179, 190substance causal 190trope 88, 92, 103, 105–9

microphysical realization 40, 43–5, 47, 53,56, 137

Mill, J. S. 258–9Mills, E. 21Mind-body problem 18–20, 32, 59, 246–7Molnar, G. 156 n. 7, 194 n. 2, 216monism 29

anomalous 6–8, 22, 251event 4, 7mind/body 235, 242property 29, 30trope 217of the whole 121

monistic naturalism 235

multiple realizability 4, 9, 36, 40, 53, 59, 60195, 219, 220, 223, 225

argument from 3, 30, 195, 216–18,219, 222

Mumford, S. 177 n., 194, 200 n. 6, 201

naturalism 237generative 235monistic 235, 246

necessitate 99, 101, 120, 178, 236, 239causally 2, 13, 159, 168, 263metaphysically 92, 100, 110, 113, 118–20,

126necessitation-basemetaphysical 89–90, 91–4, 96, 112,

115, 118minimal 111–14, 120

necessity 44, 157, 177of indeterminism for freedom 176metaphysical 89, 111, 118nomological 89

non-causalism 180, 184non-reductive physicalism 5, 10–15, 19–25,

39, 54, 60–1, 67, 88–90, 106, 111,123, 135, 138–42, 150 n., 193,216–19, 257

Noordhof, P. 14, 220nothing over and above 59, 112, 220

objectivechance 166, 175, 178, 187and physical 222preference 28probability 159propensity 181, 187relations 11similarities/differences 25

occurrents 233–47O’Connor, T. 15, 162 n. 11, 181–2, 185Oliver, A. 216ontologicalassay 236basis 108, 235category 6, 8, 10, 20, 220dependence 5, 9difference 228independence 154levels 11; see also levelssystem 6, 8

ontologicallybasic 245

270 INDEX

dependent 154distinct 132emergent 186fundamental 121, 180, 186independent 132, 139, 154innocent 218secondary 246serious 29, 196–7

ontology 1, 8, 13, 224, 253, 264emergent dualism 118of events 27fundamental 250neo-Aristotelian 184New 184–6non-reductivist 20powers 15, 89, 117–19of properties 215, 217, 225, 229trope 88, 215–29and universals 98, 227

Papineau, D. 2 n. 2, 14–15, 128, 249, 250particulars 6, 115, 119, 154, 255; see also

individual substancesas concrete 154, 157, 175 n. 4and constitution 255as basic 154

Paul, L. 132 n., 151 n.Pereboom, D. 21, 81, 132, 218, 220physicalism 18, 30, 49, 52–3, 58–61, 63,

67–9, 85, 88–9, 99, 120, 135–6, 149,167–9, 235, 250, 252, 254

about the mental 59type 35

powers 194, 196–7, 201active/passive 153, 158–9, 160, 161,

162, 180agent-causal 153–4, 160–7, 173–90causal 11, 15, 21, 38, 46–7, 51, 56, 68–71,

81–3, 98–110, 117–20, 122–3, 138–9,153, 157–8, 160, 161–71, 177–8,182–90, 205, 223, 228 n., 244,258–62

of choice and action 182conditional 222dualism of 218–19first-order 156fundamental 186higher-order 156, 220individuated by manifestation type 153intentional 66intrinsic 177mental 218–21, 226

non-causal 153, 157–8, 160physical 31, 169, 220properties as 216–18, 223, 225, 229pure 31quality/power distinction 32rational 154, 165and causal relevance 51spontaneous 15, 153–4, 160–1, 164–5,

169, 183as tokens 155–7as truthmakers 29two-way 15, 154, 164–6as types 155–7, 226as universals 227–8

powers metaphysicsevent-causal 184–6, 189–90substance-causal 184–5, 189–90

Price, H. H. 227principle of the anomalism of the mental 7principle of causal closure 2–3, 5, 12–13,

149, 168–9, 206principle of causal exclusion 14, 256 nprinciple of explanatory exclusion 256; see

also exclusionprinciple of causal interaction 7; see also

causation; psychophysicalprinciple of causal non-

overdetermination 2–3, 5, 13; seealso causal non-overdetermination

principle of the nomological character ofcausality 7

principle of suffient reason 188principle of transmission of causal sufficiency

across realization 69, 83problem of control 189propertieshigher order (or: higher level) 41, 46,

50, 56mental, see qualities of experienceMSE-property 36, 40, 48–54MSE*-property 54–5natural 174, 177, 224as particulars 10phenomenal 13; see also qualities of

experiencephysical 4, 11, 18–20, 22–4, 27, 31, 37–9,

53–5, 59, 63, 65, 68, 89–91, 95,97–101, 105–6, 112–13, 116–23, 126,135, 195, 201–2, 219, 227, 251–4

qualitative 30–2, 118, 197, 221, 229; seealso qualities of experience

second-class 224, 228

INDEX 271

properties (cont.)universals 8–10, 26, 97–8, 103, 105, 107,

223–4, 226–8tropes 6, 9–10, 15, 26 n., 97, 106–9,

217–29, 264higher order 221resemblance classes 106–9, 112,217–18, 223–5, 226, 228

property dualism 4, 7, 65, 193, 217, 219principle of 4–5, 9

property-instantiationsidentity of 4, 253as events, see eventsas facts 256 n.mental 39, 255, 262microphysical 40physical 262as states of affairs 8as states of substances 9

Putnam, H. 3

qualitiesof experience 30–2powerful 30–2primary 30secondary 30

qualia 21, 30–1, 225, see also qualitativeproperties; qualities of experience

realizationphysical 39microphysical 40, 43–5, 47, 53, 56, 137property 39–43, 46, 47, 53, 55–6subset view 41, 56

reciprocal disposition partners 197–200resemblance 107, 223, see also tropes as

resemblance classesexact 106–7, 109, 223nominalism 106–8rough 106–7, 224–5

Robb, D. 9, 15–16, 108, 216–17, 219, 225–6,250, 264

roll-back argument 163, 189

Salmon, W. 72Shoemaker, S. 14, 21, 41–2, 61, 81–3, 85, 98,

101, 132, 177 n. 8, 197 n., 220,222–3, 254

Sidelle, A. 227Simons, P. 16, 325 n.Sosa, E. 22, 130–1

Spinoza 24Stoutland, F. 22, 251Strawson, P. F. 154, 173substance dualism 5, 58, 68–9, 194substances 4, 8and agents 162, 186Aristotelian 8as bare particular 8; see also particularsas causes 183–6, 190individual 6, 9, 153–5, 157–9; see also

particularsindividuated by powers 154kinds of 154ontologically emergent 186ontologically independent 154as property bearer 26–9, 195reducible to bundles of properties 8as state of affairs 9as substratum 8

sufficient causes 90–1, 257–9, 263absolutely 258–9independent 254, 257dependent 16, 254distinct 257–60

Tooley, M. 130–1tropes: see propertiestruth conditions 73–8truthmakers 24, 28–9, 31, 110, 216,

227, 240truthmaking 110–11, 123type diversity 25type-type reduction 136

unilevellers 109–11, 117; see also levelsuniversals 8–10, 26, 97–8, 103, 105, 107, 112,

223–8

Watkins, M. 41Whittle, A. 98, 222–4, 227Williams, D.C. 217, 224, 227the will 153–4, 160–7willing 161–2, 164–7, 183Wilson, J. 118, 220–1, 223Woodward, J. 19, 21, 61, 72, 75, 115, 174 n.

2, 216

Yablo, S. 21, 51, 72, 76, 82–4, 95, 132–3,215–16, 219, 220

zombie 219

272 INDEX