Evidence and causation in biology and economics

Michael Joffe  Imperial College London

Canterbury, September 2012

Structure of the presentation • a concept of causation – epidemiology – physiology

• systems in biology – physiological systems – population biology – evolutionary biology

• systems in economic theory • conclusion

Structure of the presentation • a concept of causation – epidemiology – physiology

• systems in biology – physiological systems – population biology – evolutionary biology

• systems in economic theory • conclusion

A concept of causation I

• a causal relationship is one that has a mechanism that by its operation makes a difference 

• difference-making: a change in the probability and/or timing of an event, or in its magnitude or severity – includes partial/multiple, stochastic and deterministic

causation, plus counterfactual and manipulationist accounts

– but excludes chains, webs, cycles – single link only • this is compatible with classic accounts of

causation in epidemiology (Bradford Hill, Rose)

A concept of causation II • not all differences correspond directly to a causal

relationship – something has to make a difference, over time – no direct mechanism responsible for the sex difference

in breast cancer incidence – it is due to metabolic difference between the sexes – these do play a causal role over time, i.e. upstream causes

• mechanism and difference-making are properties of the causal relation, and seen as complementary – discovery of either can come first – a totally convincing explanation includes both

A concept of causation III • mechanism: powers/capacities [Cartwright; MDC]• combining mechanism and difference-making is

similar to the Russo/Williamson epistemic theory but takes an ontic view: asking what is the source of evidence in the real world – science can uncover evidence of the structure (“what”)

and mode of operation (“how”) of a mechanism – and of the difference it makes (“that”) – quantitative

or qualitative

• asymmetry requires an ontic perspective: beliefs do not alter reality, but reality can alter beliefs

“Physiology” • takes a primarily mechanistic view – how the

body works – but typically starts from an observed difference (often qualitative): – dietary protein is broken down – how? → pepsin, its

structure, etc – not necessarily complicated – a nerve impulse crosses a synapse – how?

• the mechanistic evidence is then juxtaposed with more difference-making evidence, e.g. many-one synapses and their quantitative characteristics

• physiological systems are evolved => regularity; the difference they make controls their evolution

Epidemiology • a difference-making approach: demonstrating

that the rate of a disease differs in groups defined by their exposure

• causal inference: the demonstrated difference is not due to e.g. chance, confounding or selection – mechanisms are often sought e.g. biomarkers

• evidence of mechanism is complementary: how the damage occurs

• any postulated cause must be plausible (yellow fingers); in due course the complementary biochemical pathways need to be elucidated

Structure of the presentation • a concept of causation – epidemiology – physiology

• systems in biology – physiological systems – population biology – evolutionary biology

• systems in economic theory • conclusion

Physiological systems I • until now, we have focused on a single link • organisms are composed of chains, webs and

cycles of causal links – each being a mechanism that makes a difference

• the dominant feature of physiological systems is homeostasis: a ± constant internal environment – core body temperature – numerous chemical concentrations

• the system property is constancy – a difference-making property

Physiological systems II • how does the causal concept of mechanism and

difference making fit with a system of this kind? • body temperature maintenance (homeothermy) – shivering is muscle activity (mechanism) that raises

temperature (the difference) – sweating is fluid secretion (mechanism) that lowers

the temperature (the difference)

• so: each link has mechanism + difference-making • the system only has its own difference-making

properties: relative constancy due to balancing or compensating (negative) feedback

Balancing fb response to external “shock”

Physiological systems III • in general, a system of this kind has numerous

links, each a mechanism that makes a difference, as well as a characteristic mode of operation – the difference-making of the system as a whole

• this can be regarded as system or endogenous causation – the system is relatively insensitive to initial conditions [Forrester 1970; Lane 2007]

• a focus on individual links is “reductionist”, in contrast to system “emergent” properties

• evidence is obtainable for all three categories

Other biological systems I • in population biology/ecology, population growth

is exponential (reinforcing, or positive, feedback) • carrying capacity: logistic growth (–ve fb too) • Lotka-Volterra classic predator-prey model: – predation and reproduction are its component links,

each with mechanistic and difference-making aspects – the system property is that the population sizes

fluctuate systematically – a property of observed ecosystems and of the systems model

– this pattern results from balancing feedback with delay, a classic pattern in system dynamics

Some biological systems

Other biological systems II • in evolutionary biology, a different pattern is

frequently observed: – the growth propensity of each tree threatens to

obscure the sunlight of other trees, leading them to compete over evolutionary time – they all grow tall

– a plant evolves the capacity to poison animals that eat it → some of the animals develop the ability to deal with the toxin, likely → further response by the plant

• this is an arms race, a form of reinforcing (positive) feedback, tending to produce exponential growth – although this can be limited by carrying capacity

Summary of biological systems • like all causal systems, they are composed of links,

each being a mechanism that makes a difference • to be called a system (in this sense), they have to

possess an additional “emergent” property of system or endogenous causation – a difference-making characteristic that results from the way that the links combine

• this system property is just as real in its effects as its component links – even in systems that are not organised (outside physiology) – not “just a model”

• systems can also be subject to exogenous causes

Structure of the presentation • a concept of causation – epidemiology – physiology

• systems in biology – physiological systems – population biology – evolutionary biology

• systems in economic theory • conclusion

Economic systems I • the fundamental topic of study in economics is

“the” market • system property of convergence towards a stable

equilibrium has been recognised since Adam Smith

The standard market equilibrium model

price

quantity

P1

Q1

D

S

Economic systems I • the fundamental topic of study in economics is

“the” market • system property of convergence towards a stable

equilibrium has been recognised since Adam Smith

Economic systems I • the fundamental topic of study in economics is

“the” market • system property of convergence towards a stable

equilibrium has been recognised since Adam Smith • this is a typical simple balancing feedback system,

like those of homeostasis – but in a system that is not evolved or deliberately organised – hence it is an idealisation or abstraction from real economic life – it could still capture the underlying essence though – does it? for all types of market?

price A quantity

demanded A-

intensity of competition

quantity supplied A

cost A

--

//profit/incentive

price A quantity

demanded A-

intensity of competition

quantity supplied A

cost A

--

//profit/incentive

A SYSTEM WITH COMPENSATING (NEGATIVE) FEEDBACK – IT TENDS TO MOVE TOWARDS STABLE EQUILIBRIUM

price A quantity

demanded A-

intensity of competition

quantity supplied A

cost A

--

//profit/incentive

a “supply shock” – e.g. a better

harvest than usual

price A quantity

demanded A-

intensity of competition

quantity supplied A

cost A

--

//profit/incentive

a “demand shock” – e.g. a successful

promotion campaign

price A quantity

demanded A-

intensity of competition

quantity supplied A

cost A

--

//profit/incentive

THESE PROCESSES OCCUR OVER TIME, BUT ARE STATIC IN THE SENSE THAT THE ONLY TIME-

DEPENDENT ENDOGENOUS PROCESS IS TOWARDS A STABLE EQUILIBRIUM

a “supply shock” – e.g. a better

harvest than usual

a “demand shock” – e.g. a successful

promotion campaign

Some economic series

Economic systems II • balancing feedback with delay: e.g. construction

cycles in property markets; business cycles? • bubbles: trend extrapolation that causes a self-

fulfilling prophecy – reinforcing feedback • financial markets: bubbles-prone, highly volatile,

and with some endogenous chaotic properties • capitalist growth: firms’ control over the means

of production gives them flexibility over costs and the size of the market they can supply – an arms race – reinforcing feedback

Economic systems III • balancing feedback is a feature of economic

systems, but is frequently joined by other types of feedback, giving the system different properties

• neither the conventional view, that markets are always self-correcting, nor the critical view, that they do not have this property at all, is correct

• this is not the same issue as “market failure” • fluctuations may occur, due to – ve fb with delay• to understand bubbles, or capitalist growth, it is

necessary to understand reinforcing feedback

Economic systems IV • the properties of the system primarily depend on

its feedback structure • rationality vs. realistic behaviour is a secondary

issue – all that is needed to make the system work is some degree of regularity, especially in response to incentives – situational rationality

• rationality is useful for mathematical modelling• many economists see theory/models not based

on optimisation/strict rationality as “ad hoc” – this is a basic category error

Structure of the presentation • a concept of causation – epidemiology – physiology

• systems in biology – physiological systems – population biology – evolutionary biology

• systems in economic theory • conclusion

Conclusion • there are cross-cutting methods of analysis that

are applicable across many disciplines • system dynamics – cyclical combinations of

causes containing feedback loops – is one; another is complex systems (complexity, chaos)

• systems with feedback have characteristic modes of behaviour: endogenous causation, a difference-making property; their constituent links have mechanism + difference-making

• different market types have radically different properties, resulting from feedback structure

THANK YOU!