This article was downloaded by: [Indiana Universities]On: 08 September 2012, At: 01:11Publisher: RoutledgeInforma Ltd Registered in England and Wales Registered Number: 1072954Registered office: Mortimer House, 37-41 Mortimer Street, London W1T 3JH,UK

Australasian Journal ofPhilosophyPublication details, including instructions for authorsand subscription information:http://www.tandfonline.com/loi/rajp20

Remembering without knowingSven Bernecker aa University of California, Irvine

Version of record first published: 27 Mar 2007

To cite this article: Sven Bernecker (2007): Remembering without knowing, AustralasianJournal of Philosophy, 85:1, 137-156

To link to this article: http://dx.doi.org/10.1080/00048400601176460

PLEASE SCROLL DOWN FOR ARTICLE

Full terms and conditions of use: http://www.tandfonline.com/page/terms-and-conditions

This article may be used for research, teaching, and private study purposes.Any substantial or systematic reproduction, redistribution, reselling, loan, sub-licensing, systematic supply, or distribution in any form to anyone is expresslyforbidden.

The publisher does not give any warranty express or implied or make anyrepresentation that the contents will be complete or accurate or up todate. The accuracy of any instructions, formulae, and drug doses should beindependently verified with primary sources. The publisher shall not be liablefor any loss, actions, claims, proceedings, demand, or costs or damageswhatsoever or howsoever caused arising directly or indirectly in connectionwith or arising out of the use of this material.

REMEMBERING WITHOUT KNOWING

Sven Bernecker

This paper challenges the standard conception of memory as a form ofknowledge. Unlike knowledge, memory implies neither belief nor justification.

In the last six decades, most philosophers have come to advocate a view ofpropositional memory according to which remembering that p entailsknowing that p. The so-called epistemic theory of memory holds that toremember something is to know it, where this knowledge was previouslyacquired and preserved. Memory is long-standing or continuing knowledge.Despite its widespread acceptance, I believe that the epistemic theory isfundamentally mistaken. What passes into memory may be merely arepresentation or thought, not knowledge. Memory works in a myriad ofways, sometimes neither beginning with nor ending with knowledge.Knowledge supervenes on some but not all cases of remembering, forsometimes memory, though hitting the mark of truth, succeeds in anepistemically defective way.

Section I identifies the kind of memory the epistemic theory is concernedwith, namely propositional memory. Section II explains the epistemicanalysis of memory. Section III argues that it is not necessary forpropositional memory that the proposition in question be justifiablybelieved when it was originally acquired or that it be justifiably believedwhen it is recalled. Sections IV and V challenge the epistemic theory ofmemory by arguing that one can remember something without believing it.Section VI deals with the analysis of impure memory and Section VII offerssome concluding remarks.

I. Propositional Memory

Proponents of the epistemic theory of memory differentiate betweenpropositional, personal (or experiential), and practical memory. Practicalmemory is remembering how to do something, where this refers to apreviously acquired and retained skill. Since the epistemic theory does notconcern practical memory we may prescind from this kind of memory.

Instances of propositional memory are substituents of the schema ‘Sremembers that p’, where ‘S’ stands for a person and ‘p’ stands for a trueproposition. We can remember propositions about the past (e.g., thatBrutus stabbed Caesar), the present (e.g., that my partner is now shopping),the future (e.g., that I have a dentist’s appointment tomorrow morning) as

Australasian Journal of Philosophy

Vol. 85, No. 1, pp. 137 – 156; March 2007

Australasian Journal of Philosophy

ISSN 0004-8402 print/ISSN 1471-6828 online � 2007 Australasian Association of Philosophy

http://www.tandf.co.uk/journals DOI: 10.1080/00048400601176460

Dow

nloa

ded

by [

Indi

ana

Uni

vers

ities

] at

01:

11 0

8 Se

ptem

ber

2012

well as timeless truths (e.g., E¼mc2). Propositional memory is not limited tothings with which one has had direct or personal acquaintance. One neednot have witnessed the event to remember that Brutus stabbed Caesar.Consequently propositional memory does not require qualitative experi-ences and imagery. Though the objects of propositional memory need notdeal with the past, one’s learning of what one propositionally remembersmust precede the remembering. You cannot remember that p if you haveonly just learned that p. Moreover, to remember a proposition you must notonly have thought about it before but your present thought must beconnected, in the right sort of way, to your past thought. The connectionbetween a past thought and its subsequent recall is usually taken to be acausal process involving intermediary memory traces. It should beemphasized that it is propositional memory which the epistemic theory ofmemory focuses on.

Personal memory is said to have two characteristics. When an agentpersonally remembers something, his memory of it is caused by theexperience he had of that object. The first characteristic, then, is that, in thissense, one can personally remember only what one has personallyexperienced. That is, one’s memory of something is a personal memoryonly if one has experienced that thing. The second characteristic is thatpersonal memory represents the remembered content from the first-personperspective (from ‘within’) and thus personal memory involves qualitativeexperiences and imagery. As Don Locke [1971: 76] explains, ‘personalmemory consists in bringing some previously experienced thing to mind,thinking about it again, and going over what it was like’.1

As it stands, the distinction between propositional memory and personalmemory is not sharp. Consider, for example, my remembering that I hadeggs for breakfast this morning. Is it an instance of personal memory ordoes it belong to the class of propositional memory? To answer thisquestion, proponents of the standard tri-partite classification scheme willpresumably inquire whether the content of the memory consists merely of aproposition or whether it also includes images and qualia. But the problemwith this strategy is that the frequency of memory images varies greatly fromone person to another [Nigro and Neisser 1983]. Some people report thattheir mental lives are replete with imagery as vivid and detailed as the actualscenes they recall. For others, imagery is uniformly vague, dim, and fleeting.And, for a final group, there seems to be no imagery at all.

Propositional and personal memory are governed by different criteria.The criterion for identifying propositional memories is primarily agrammatical one: the memory content must have the form of a that-clause.Yet the criterion for identifying personal memories is, in the first instance, apsychological one: they present themselves to the mind as images andqualitative experiences. Since the criterion for identifying propositional

1Martin and Deutscher [1966:162 – 3] explain: ‘If someone is asked whether he remembers what he did lastFriday at lunchtime, he may be able to say that he went down the street. Yet he may feel scarcely in a positionto say that he remembers actually going down the street. What he needs in order to be able to say that he does[personally] remember going down the street is at least more detailed remembering that . . . certain thingshappened when he went down the street . . . [T]his addition of detail must be due to the original perception’.

138 Sven Bernecker

Dow

nloa

ded

by [

Indi

ana

Uni

vers

ities

] at

01:

11 0

8 Se

ptem

ber

2012

memory is of a different kind than that for identifying personal memory it isnot surprising that the boundary between these kinds of memory is anythingbut sharp.2

In light of this difficulty some employ Bertrand Russell’s distinctionbetween knowledge by acquaintance and knowledge by description todifferentiate between personal and propositional memory [Locke 1971: 70;Malcolm 1963: 207; Pollock 1974: 184]. Something is known byacquaintance when there is direct experience of it; it is known by descriptionif it can only be described as a thing with such-and-such properties. Personalmemory is taken to be memory by acquaintance and its objects are notpropositions but people, places, things, events, and situations. Propositionalmemory, on the other hand, is supposed to be analogous to knowledge bydescription.

Four comments. First, many autobiographical data are remembered bydescription. Second, our memory of propositions is frequently due to ushaving been acquainted with the things they are about. Third, construingthe personal/propositional distinction along the lines of the acquaintance/description distinction yields the counterintuitive consequence that, strictlyspeaking, I cannot personally remember that p. Norman Malcolm [1963:216], a proponent of the tri-partite typology of kinds of memory, bites thebullet. He claims that one cannot personally remember that the city hallburned down ‘since it would be senseless to speak of having beencontemporaneous with the fact that the city hall burned down’. LikewiseJohn Pollock [1974: 184] holds that personal memory always has the form ‘Iremember its being the case that p’ rather than ‘I remember that p’. Thisposition, I reckon, is too implausible to be acceptable. Fourth, in onerespect, propositional memory is analogous to knowledge by acquaintance.Russell himself thought that in entertaining a proposition, one is acquaintedwith that proposition. So appealing to Russell’s notion of acquaintance doesnot have the consequence which the philosophers in question want.

Granted that the schema ‘S remembers that p’ can express bothpropositional and personal memory, what, if anything, distinguishes thesekinds of memory? In the end, the only real difference between personal andpropositional memory is that the former, but not the latter, is limited toitems with which one has had direct acquaintance. In what follows I willtake ‘propositional memory’ (or ‘memory’ for short) to refer to substituentsof the schema ‘S remembers that p’, irrespective of whether ‘p’ refers tosomething one has personally experienced.

II. The Epistemic Analysis of Memory

The epistemic theory has come to be the standard contemporary account ofmemory, drawing support from a wide variety of philosophers and only

2Locke [1971: 48], a principal in devising the tri-partite classification scheme, concedes this point when hewrites: ‘although the three forms are each associated with a particular grammatical construction—remembering that such-and-such, remembering how to do such-and-such, remembering such-and-such itself—grammar provides only a rough guide to which form of memory is involved’.

Remembering without Knowing 139

Dow

nloa

ded

by [

Indi

ana

Uni

vers

ities

] at

01:

11 0

8 Se

ptem

ber

2012

infrequent criticism. In his classic monograph on memory Don Locke [1971:39] declares:

Memory consists . . . in immediate knowledge of the past; memory is to be

thought of . . . as a form of knowledge. Certainly, this has come to be thestandard contemporary account of the nature of memory; to remembersomething is to know it, where this knowledge has been acquired in the past.

For example, Robert Audi [1998: 67] says that ‘if you remember thatwe met, you know that we did. Similarly, if you remember me, youknow me’. Following this view, Michael Dummett [1993: 420 – 1] writes:‘Memory is not a source, still less a ground, of knowledge: it is themaintenance of knowledge formerly acquired by whatever means’. AndMalcolm [1963: 223; 1977: 102 – 8] defines propositional memory as follows:‘A person B remembers that p if and only if B knows that p because he knewthat p’.3

By and large, advocates of the epistemic theory of memory assumethe following analysis of propositional memory: at t2 S remembers thatp only if

(1) S knows at t2 that p.

(2) S knew at t1 that p.

(3) S’s knowing at t2 that p is suitably connected to S’s knowing at t1 that p.4

Condition (1) may be labelled the present knowledge condition, condition (2)the past knowledge condition, and (3) the connection condition. The presentknowledge condition requires that to remember something is to occupy astate of knowing. The past knowledge condition ensures that one can onlyremember what one previously knew. Since knowledge implies truth, (1) and(2) guarantee that one can only remember what is the case. Finally, thepurpose of the connection condition is to exclude re-learning from the ranksof remembering and to guarantee that the knowledge had in remembering isretained knowledge. Given (3) it is possible for S to know at t2 that p, and tohave known it previously at t1, and still fail to remember at t2 that p. S maysimply have learned that p at t1, forgotten it entirely in the interval betweent1 and t2 and then re-learned it afresh at t2.

Traditionally, epistemologists have held that the concept of propositionalknowledge has three individually necessary and jointly sufficient conditions:

3Other proponents of the epistemic theory of memory are Annis [1980: 324], Anscombe [1981: 127], Ayer[1956: 138, 147 – 8], Dancy [1985: 187, 195], Dretske and Yourgrau [1983: 361], Evans [1982: 235], Grice[1941: 344], Holland [1974: 359], Huemer [1999: 346], Landesman [1961: 59, 61], Margolis [1977: 188], Moore[1959: 214], Munsat [1967: 15 – 17], Owens [2000: 156], Pappas [1980: 129], Pollock [1974: 196; 1986: 55], Ryle[1949: 272 – 9], Shoemaker [1970: 282], Squires [1969: 185], Unger [1972: 304], Williams [1973: 142],Williamson [2000: 37 – 8] and Zemach [1968: 529].4Here and throughout paper the value of the index in the subscript to ’t’ determines whether the point of timeis in the past or the present: the relatively biggest number indicates the present. So here ‘t2’ is the present and‘t1’ is the past. When there is more than one past time involved, ‘t1’ indicates the distant past, ‘t2’ the closepast, and ‘t3’ the present.

140 Sven Bernecker

Dow

nloa

ded

by [

Indi

ana

Uni

vers

ities

] at

01:

11 0

8 Se

ptem

ber

2012

justification, truth, and belief. Knowledge is said to be justified true belief.The truth condition states that if you know that p, then p is true. The beliefcondition claims that knowing that p implies believing that p. A person neednot be absolutely certain that something is true in order to know that it is.The belief condition only requires some kind of acceptance in the interest ofobtaining truth. (Whether knowledge implies belief is an issue I will comeback to in Section V.) Finally, the justification condition requires thata known proposition be evidentially supported. The justification conditionis there to prevent lucky guesses from counting as knowledge when theguesser is sufficiently confident to believe his own guess. Furthermore, thejustification condition is supposed to rule out Gettier cases, i.e., cases wherethe truth of p and the justification of p are not only independent (as it alwaysis the case with fallibilist knowledge) but where they are not suitably related:what accounts for the justification of the belief is not connected, in the rightkind of way, to what accounts for its truth.5 Yet in the present context thereis no need to discuss the Gettier problem.

Given that knowledge implies truth and justification (however construed)and given the transitivity of implication, the present knowledge condition(1) entails three conditions:

(1.i) p is true at t2.

(1.ii) S believes at t2 that p.

(1.iii) S is justified at t2 in believing that p.

Likewise the past knowledge condition (2) implies three conditions:

(2.i) p was true at t1.

(2.ii) S believed at t1 that p.

(2.iii) S was justified at t1 in believing that p.

Conditions (1.i) and (2.i) are truth conditions, (1.ii) and (2.ii) are beliefconditions, and (1.iii) and (2.iii) are justification conditions. If any one ofthese conditions is false or untenable, the epistemic theorist of memory isforced to conclude that S does not remember that p. The task of evaluatingthe epistemic account of memory is a matter of determining the tenability ofthe belief conditions and the justification conditions.

Section III challenges the justification conditions by arguing that not onlyis it possible to remember something one did not justifiably believe in thepast but also one might acquire between t1 and t2 some misleading but

5This interpretation of the Gettier problem differs slightly from the standard reading. On my reading, theGettier problem is about the interrelation between the justification-maker of Bp and the truth-maker of p.Yet the Gettier problem is commonly taken to be about the covariation between the justification of Bp andthe truth of p. In other words, on the standard reading, the reasons a gettierized belief doesn’t qualify asknowledge is because there is at least one close possible world which is indistinguishable vis-a-vis the subject’sevidential situation but where the belief is false.

Remembering without Knowing 141

Dow

nloa

ded

by [

Indi

ana

Uni

vers

ities

] at

01:

11 0

8 Se

ptem

ber

2012

reasonable evidence that destroys the status as justified belief of the once-genuine justified belief which one still remembers. Sections IV and Vchallenge the present belief condition by arguing that S can remember at t2that p without believing at t2 that p.

III. Memory without Justification

Given that memory implies knowledge and that knowledge impliesjustification, I cannot remember that p unless I am justified in believingthat p and I have been justified in believing that p in the past. Audi [1998:37], for example, claims that ‘if I remember that I met you, I am justified inbelieving I met you’. The task of this section is to show that one mightremember that p without being justified in believing that p.

Counterexamples to the justification condition of memory can be dividedalong three lines. First, counterexamples may challenge the presentjustification condition (1.iii) or the past justification condition (2.iii).Second, a rememberer may either not be aware of his being justified inbelieving p (absent-justification cases) or he may possess positive counter-evidence to p and thus be aware of his not being justified in believing p(defeated-justification cases). Third, counterexamples may presume aninternalist or an externalist account of epistemic justification. When theepistemic theory of memory is coupled with internalism about justification,S remembers at t2 that p only if he has cognitive access at t1 and t2 to thefactors that justify his belief that p.6 Externalism about justification is simplythe denial of internalism. So when the epistemic theory of memory iscombined with externalism, S remembers at t2 that p only if his belief that phas at t1 and t2 the property of being truth-effective; whether or not thesubject takes his belief to be truth-effective doesn’t add anything to thebelief’s epistemic status.

The cognitive capacities required by internalism about justification aretoo complex to be plausibly ascribed to higher animals, small children, andunsophisticated adults. Thus if the epistemic theory of memory is combinedwith internalism, we cannot speak of such subjects as remembering things;an absurd consequence.7 That might explain why the majority ofproponents of the epistemic theory of memory are externalists—at least

6Internalists explicate this cognitive access in terms of the route by which one has access: whether one isjustified in believing p is wholly determined by factors which one is in a position to know by reflection alone.By ‘reflection’ internalists mean a priori reasoning, introspective awareness of one’s own mental states, andone’s memory of knowledge acquired in those ways.7Malcolm [1963: 239n], however, declares: ‘I do not believe there is any sense in which a dog or infant can besaid to know that it has some sensation. I accept the consequence that a dog cannot be said to remember thathe had a painful ear, and also the more interesting consequence that a human being cannot be said toremember that he had one, if he had it at a time before he knew enough language to be able to tell anyonethat he had it’. Malcolm’s reason for drawing this conclusion are independent from epistemic internalism.Instead Malcolm claims—following Wittgenstein—that it is senseless to speak of someone knowing hissensations. Granted this and granted that memory is a form of knowledge, one cannot speak of someoneremembering his sensations. Malcolm realizes that this conclusion is too counterintuitive to be acceptable.He therefore concedes ‘that a sense can be given to saying that a person knows that he has a sensation at thetime he has it. He knows it in the sense that he can tell you that he has it’. The reason animals and childrencannot be said to know or remember their sensations is due to their lack of a language in which to describethe sensations.

142 Sven Bernecker

Dow

nloa

ded

by [

Indi

ana

Uni

vers

ities

] at

01:

11 0

8 Se

ptem

ber

2012

about the present justification condition (1.iii). They hold that one’s initialjustification for a belief continues, so long as one merely continues to holdthis belief—regardless of whether one is aware of one’s initial grounds. Thedoctrine according to which a belief may inherit its justificatory status iscommonly called the principle of ongoing justification: at t2, S’s memorybelief that p from t1 is continuously justified if and only if S continues tobelieve that p at t2—even if S lost his original knowledge-producing evidenceand has acquired no fresh evidence in the meantime.8 Since advocates of theepistemic theory of memory tend to be epistemic externalists and sinceabsent-justification cases only work against an internalist construal of thepresent justification condition, there is no point in discussing these caseshere. Instead I will focus on defeated-justification cases.

Defeated-justification cases are concerned with subjects who do rememberthat p but who have reasons to believe that their belief in p is not justified.Whether such cases pose a threat to an externalist construal of the epistemictheory of memory depends on the assumed version of externalism. Given anextreme form of epistemic externalism, the question of whether a belief isjustified (truth-effective) is quite distinct from the question of whether thesubject takes his belief to be justified (truth-effective). A belief is or isn’tjustified regardless of what the subject thinks about the justificatory statusof the belief. If one relies on what is, in point of fact, a conclusive reason forp, one is justified in believing that p, despite being convinced that p is false.Fred Dretske [2000: 595] calls such an austere form of epistemic externalismmad-dog reliabilism.

Mad-dog reliabilism has been challenged by Laurence BonJour’s [1985:chap. 3] clairvoyance case. Imagine someone who possesses a sixth sense ofclairvoyance. Further suppose that this clairvoyant sense is remarkablyreliable. Given mad-dog reliabilism, S might suddenly know that p, eventhough he has no evidence either way on whether p is true, on whether hepossesses the power of clairvoyance, and on whether such a power is evenpossible. Internalists, like BonJour, claim that the belief in question fails tobe justified, despite the claim of reliabilism to the contrary. The belieffails to be justified because, from S’s perspective, it is an accident that it istrue. The upshot is supposed to be that epistemic justification requiresthat the acceptance of the belief in question be rational, which in turnrequires that the believer be aware of a reason for thinking that the beliefis true.9

In response to the clairvoyance cases, many externalists hold that eventhough a subject need not be aware of the factors that justify his belief, hemust not be aware of evidence that undermines his belief. The negativecoherence condition ensures that for a belief to become knowledge it must

8Among the proponents of the principle of ongoing justification are Dretske [1982], Dretske and Yourgrau[1983], Owens [2000: 153], Pappas [1980; 1983], Pollock [1974: chap. 7; 1986: chap. 2], Senor [1993], andShoemaker [1967: 271 – 2].9Some externalists (myself included) deny that the clairvoyance case poses a threat to mad-dog reliabilism.For the intuitive plausibility of the clairvoyance case hinges on the presumption that clairvoyance is notreliable. But if a clairvoyant faculty actually existed, then either it would prove itself reliable or not. If itproved itself reliable, then intuitively there would be no reason to deny clairvoyants justification andknowledge. The internalist interpretation of the thought experiment presupposes a bias against clairvoyance.Yet this is not the place to defend mad-dog reliabilism.

Remembering without Knowing 143

Dow

nloa

ded

by [

Indi

ana

Uni

vers

ities

] at

01:

11 0

8 Se

ptem

ber

2012

not be incoherent with the background information the subject possesses.Given this mitigated version of epistemic externalism (or ‘mitigatedexternalism’ for short), defeated-justification cases do pose a threat to theepistemic theory of memory. Now let’s consider two defeated-justificationcases.

First, consider a counterexample to the past justification condition(2.iii)10: sometime in the past, at t1, S believed that something was occurringin front of him (e.g., the Loch Ness monster poking its head out of thewater), something which he was, in fact, witnessing, but when he hadplausible but misleading reasons to suppose that he had just been given astrong hallucinogenic drug. Given mitigated externalism about justification(or internalism), S wasn’t justified at t1 in believing that such a thing wastaking place in front of him and thus he didn’t know that it was. At t2, Slearns that, in spite of his past evidence, he had not actually been given ahallucinogenic drug but only a placebo. He had really seen what he hadirrationally believed to be seeing while falsely thinking to be hallucinating.At t3 S is able to give an accurate account of what he witnessed at t1.

The question is whether S may be said to remember at t3 that the eventtook place or whether he cannot be remembering what he witnessedbecause, at the time, he wasn’t justified in believing his eyes. I embrace thefirst option. To motivate this view it is important to be clear about whatprecisely S is supposed to remember at t3 and not to have known at t1.

11

Presumably, at t1 S had both justified belief and knowledge of how the LochNess monster looked to him. What he did not justifiably believe is that therewas a monster in front of him. Once S learns that he has only been given aplacebo, it is reasonable to suppose that he remembers (and maybe evenknows) not only how the monster looked to him but also that there was amonster.12

If the suggestion is that S remembers at t3, but did not justifiably believeat t1, that there was a monster, the thought experiment is open to theobjection that this is a case of impure rather than genuine memory. In such acase, there is a mixture of remembering and inference, which results in a useof ‘remembering-that’ which is only elliptical for a different concept.According to the objection at hand, when S claims at t3 ‘I remember thatthere was a monster’ this involves a use of ‘remember’ that is elliptical fortwo separate claims: (i) I remember that it looked to me as if there was amonster and (ii) I am now justified in believing (and know) that there was amonster. (Section VI will propose a more sophisticated analysis of impurememory.) Thus S does not, strictly speaking, remember at t3 that there was amonster; or so a critic could argue.

This objection fails. The example at hand is not an instance of impurememory. In the case of impure memory the content of the state of recallexceeds the content of the original representation. The paradigm example

10The example is adapted from Shope [1973: 308 – 9]. Essentially the same example can be found in Martinand Deutscher [1966: 191 – 2].11I am grateful to an anonymous referee for raising this issue and urging me to make myself clearer.12Even if S never learns that he has only been given a placebo, and we are inclined to deny him rememberingthat there was such a scene, we would still grant him remembering how the monster looked.

144 Sven Bernecker

Dow

nloa

ded

by [

Indi

ana

Uni

vers

ities

] at

01:

11 0

8 Se

ptem

ber

2012

of impure memory is due to Malcolm [1963: 223]: when a person originallysees a bird and only later learns that it was a bluejay, he may claim‘I remember that I saw a bluejay’ but this is elliptical for two claims:(i) I remember that I saw a bird of this kind and (ii) I now know that itwas a bluejay. What the person claims to remember is more than what heoriginally knew. But now compare this example to the one about Nessy.The propositional content of S’s memory claim—that there was amonster—is the very same as the content of his original belief. By beinginformed that he has only been given a placebo S is given no newinformation about what he witnessed. The issue of whether S remembershaving distrusted his eyes is quite distinct from the question whether heremembers seeing what he did see. Thus the Nessy-example cannot beconsidered a case of impure memory.

Next, consider a counterexample to the present justification condition(1.iii)13: at t1 S met his sister and noticed that she was wearing a blueblouse. Because S then knew this fact about his sister, he holds at t2 theveridical memory that his sister wore a blue blouse. At t3 S is told by allthe people who were present at t1 that his sister wore a green blouse.Moreover, he is presented with plausible but misleading evidence to theeffect that his sister wore a green blouse. Given mitigated externalismabout justification (or internalism), S doesn’t know at t3 that his sisterwore a blue blouse, for he is unable to rule out the relevant alternativethat she wore a green blouse. He fails to know that his sister wore a blueblouse, despite the fact that he still remembers it. Thus, S knows at t1 thatp, remembers at t3 everything he knew at t1, and yet fails to know at t3that p—even though he continues to believe that p and p continues to betrue—for the reason that he isn’t justified in believing that p. To thinkotherwise is to confuse conditions for remembering with conditions forsaying that one remembers.

What makes a mental state a state of remembering is the fact thatcertain conditions are met; whether or not the subject takes himself tomeet these conditions is of no importance. Remembering is, in this respect,like seeing. Just as I can see something without being aware that I see it,I can remember something even if I think I am suffering a hallucination.Thus the fact that S isn’t justified at t3 in believing that p and thusdoesn’t know that p shouldn’t prevent us from attributing to him thememory that p.

It should be noted that just as Nessy-example is not a case of impurememory nor is the blouse-example. The propositional content of S’smemory claim—that his sister wore a blue blouse—does not exceed thecontent of his original knowledge. The two diachronic content-tokens aretype-identical. Thus there is no admixture of inference or present realizationinvolved in the process of remembering.

In sum, memory, unlike knowledge, does not imply justification. Not onlyis it possible to remember something at t2 that one didn’t justifiably believe

13The example is adapted from Saunders [1965a: 282 – 3]. Similar examples can be found in Dretske andYourgrau [1983], Ginet [1988: 160], and Naylor [1986: 298].

Remembering without Knowing 145

Dow

nloa

ded

by [

Indi

ana

Uni

vers

ities

] at

01:

11 0

8 Se

ptem

ber

2012

at t1 but also one may acquire between t1 and t2 some plausible butmisleading evidence that destroys the status as justified belief of the once-genuine justified belief that one still remembers. What passes into memorymay be merely a representation or thought, not knowledge.

IV. Memory without Belief

Granted that knowing implies believing, the epistemic theory has it thatremembering too implies believing. Malcolm [1963: 224], a leadingproponent of the epistemic theory puts the point as follows: ‘Being unsurewhether p is true counts both against knowing that p and againstremembering that p’. In this section I will argue that S can remember thatp without believing that p.

Prima facie, a proponent of the epistemic theory of memory may dismissthe possibility of memory without belief on the grounds that ‘I rememberthat p; but I don’t believe that p’ is equally incoherent as G. E. Moore’sfamous paradoxical statement ‘It is raining; but I don’t believe that it israining’. The idea is that the incoherence of ‘I remember that p; but I don’tbelieve that p’ cannot be explained unless one assumes that rememberingimplies believing. This objection doesn’t work. It is possible to explain thepragmatic incoherence of the statement ‘I remember that p; but I don’tbelieve that p’ while maintaining that memory does not imply belief: When Iclaim to remember that p, I am convinced that p is the case and hence believethat p. This is what the first part of the statement expresses. Yet the secondpart of the statement denies that I believe that p. Thus ‘I remember that p;but I don’t believe that p’ is incoherent not because one cannot rememberthat p without believing that p but because one cannot claim to rememberthat p while claiming to not believe that p. And since the conditions forclaiming to remember are distinct from the conditions for remembering, itdoes not follow that remembering that p implies believing that p becauseclaiming to remember that p implies claiming to believe that p.

Before I can attempt to argue that memory does not imply belief I need tosay a few words about what it takes to believe something and whatdistinguishes believing something from merely thinking about it. Beliefs arethe attitudes we have, roughly, when we take something to be the case,regard it as true, or accept it.14 The state of believing that p can manifestitself in different forms. One may be convinced by the evidence of its beingtrue that p, one may be surprised to learn of an event that is evidence againstits being true that p, one may feel pleased at its being true that p, and so on.Furthermore, an individual who believes that p, who has no reason to

14Here I employ the common-sense notion to accept. The verb ‘to accept’ has been given various technicalmeanings by various authors. For example, Cohen [1992: 4] declares: ‘[B]elief that p is a disposition . . . nor-mally to feel it true that p and false that not-p . . . But . . . to accept that p is to have or adopt a policy ofdeeming, positing, or postulating that p . . . whether or not one feels it to be true that p’. Cohen’s notion ofacceptance can be characterized by means of the following theses: (i) acceptance is voluntary or intentional,unlike belief; (ii) belief aims at truth, but one can accept things one believes to be false; (iii) beliefs come indegree while acceptance is an all or nothing affair. My intuitive notion of acceptance does not satisfy any ofthe conditions that govern Cohen’s technical notion. Lehrer [1990: chap. 2] argues that knowledge requiresacceptance (in the technical sense of the term) rather than belief.

146 Sven Bernecker

Dow

nloa

ded

by [

Indi

ana

Uni

vers

ities

] at

01:

11 0

8 Se

ptem

ber

2012

deceive, and who is asked, in a language he understands, whether p is thecase, will generally say that it is. Believing that p involves holding p true, butit does not involve actively reflecting on p or an especially high degree ofconfidence with respect to p.15

Acceptance is a central component not only of occurrent belief but also ofdispositional belief. Occurrent beliefs are in the forefront of one’s mind,while dispositional beliefs lie dormant in one’s ‘belief box’ or memory,waiting to be turned into occurrent beliefs under certain circumstances. ForS to dispositionally believe that p is for S to possess the behaviouraldisposition to act as though p is the case. For example, S will assent toutterances of p in the right sorts of circumstances (if he understands thelanguage, wishes to reveal his true opinion, is not physically incapacitated,etc.). Thus to dispositionally believe a proposition implies that one isdisposed to hold it true.

Both occurrent and dispositional beliefs are representations that areactually tokened in one’s brain (in the right sort of way). In addition toexplicit beliefs some philosophers postulate implicit (or tacit) beliefs.Implicit beliefs are not explicitly represented and don’t inhabit one’s ‘beliefbox’. According to the widespread formation-dispositional account, toimplicitly believe that p is to be disposed to have an explicit belief that p, insuch-and-such circumstances.16 Since, on this account, implicit belief isexplained by way of dispositional explicit belief and since explicit beliefimplies acceptance it follows that implicit belief also presupposesacceptance. Thus all forms of believing that p require that one takes p tobe the case or would take p to be the case in the right sorts of circumstances.Given these preliminary remarks on the nature of belief, we can address thequestion whether one can remember something without believing it.

Prima facie, there are four (putative) reasons why S may remember at t2that p without believing at t2 that p and/or without having believed at t1 thatp. One such possible reason is that at t1 S didn’t yet possess the conceptsnecessary to believe that p. Such cases are commonly called impure memory(or elliptical memory). Another reason why S may remember at t2 that pwithout having believed at t1 that p is that he was too busy doing somethingelse and was not paying attention to the fact that p. In lack of a better term Iwill refer to this kind of remembering as inattentive remembering. A thirdreason for remembering without past believing is that p states something onedidn’t do or hasn’t happened (e.g., one didn’t lock the door). This kind ofmemory is usually called negative memory. Finally, a fourth reason why Smay remember at t2 that p without believing at t2 that p is that heerroneously takes p to be a figment of his imagination. I will refer to thiskind of remembering as ignorant remembering. While cases of impure,

15A subject’s degree of confidence in p is usually taken to be the maximum amount he would be willing towager on a bet that pays nothing if p is false and one unit if p is true.16[De Sousa 1971; Lycan 1988: chap. 3; Sellars 1969]. As it stands, this characterization of the formation-dispositional account of implicit belief is insufficient. Counterexamples show that to implicitly believe that prequires something more than merely being disposed to explicitly believe that p, and less than representing pinternally. Two strategies have been considered for supplementing the initial account. The first strategyrequires the disposition to have an explicit belief to be grounded in an extrapolator-deducer mechanism (i.e.,a device which operates on occurrent beliefs and generates obvious consequences when the occasion arises),and the second strategy requires the disposition be based on reasons (good or bad).

Remembering without Knowing 147

Dow

nloa

ded

by [

Indi

ana

Uni

vers

ities

] at

01:

11 0

8 Se

ptem

ber

2012

inattentive, and negative memory challenge the past belief condition (2.ii),cases of ignorant remembering are directed against the present beliefcondition (1.ii).

The discussion of impure memory is deferred for Section VI. Here Iconcentrate on the other three kinds of memory without belief. The goal isto critically examine examples for each of the three cases of memory withoutbelief. The conclusion is going to be that instances of inattentive andnegative memory fail to refute the past belief condition (2.ii), but that casesof ignorant remembering are successful in casting the present beliefcondition (1.ii) into doubt. Obviously, cases of memory without belief donot count against the epistemic theory of memory, unless knowledge impliesbelief. In Section V I will defend the idea that belief is a necessarycomponent of knowledge.

Keith Lehrer and Joseph Richard [1975: 122] construct the followingexample of inattentive remembering: suppose that while S is avidlylecturing, a bell rings indicating the end of the lecture. S doesn’t noticethat the bell is ringing and continues his lecture. After some time, a studentasks S whether he has forgotten the time. At that moment S remembers thatthe dismissal bell had rung, even though he didn’t previously believe that ithad rung. The auditory experience had failed to break into S’s consciousnessat the time it occurred, but passed into memory nonetheless. Lehrer andRichard take this example to show that there are instances whereremembering something means believing (and knowing) it for the first time.Granted that such instances are possible, the epistemic theory of memory ismistaken in identifying memory with retained knowledge.17

The problem with this alleged counterexample to the epistemic theory ofmemory is the implausible presumption that to believe (and know) that pone must be conscious that p. According to Lehrer and Richard, the person,not having been conscious of the dismissal bell, did not believe (or know)that the bell rang. But on a dispositional reading of ‘to believe’, S did believethat the bell rang since this belief was manifested in subsequent action, thatis, the remembering. Moreover, if at the time the bell rang someone hadasked S whether he thinks that the bell is ringing he would, no doubt, haveanswered in the affirmative. Since it is implausible to restrict knowledge toconscious and occurrent belief, cases of inattentive remembering fail toundermine the epistemic analysis of memory.

William James [1890: I, 649] develops the following case of negativememory: after having left his house and walked down the street Sremembers that he didn’t lock the door. He returns, verifies that the dooris unlocked and locks it. This is taken by some to be an example ofremembering without believing (and knowing) because if, at the time Sinitially left the house, he had believed that the door is unlocked he wouldhave locked it right away. While James’s example concerns a negativespecific fact, Carl Ginet [1975: 149] gives an example of remembering a

17Similar thought experiments can be found in Locke [1971: 54], Pollock [1974: 196 – 7], and Traiger [1978:111]. Locke’s example involves a professor not noticing a certain student in a lecture hall, Pollock’s exampleconcerns a driver in a dangerous situation not noticing a billboard beside the road, and Traiger’s example isabout a skier during a slalom ski race not noticing a patch of blue snow.

148 Sven Bernecker

Dow

nloa

ded

by [

Indi

ana

Uni

vers

ities

] at

01:

11 0

8 Se

ptem

ber

2012

negative general fact: S can recall that during the last twenty years he hasnever been to a dog show without once during this period having believedthat he has not been to a dog show. Thus, S is able to remember aproposition that he has never before believed or known.

Cases of negative memory fail to undermine the epistemic theory ofmemory for the same reason as cases of inattentive remembering fail toundermine the epistemic theory. It is natural to suppose that at t1 S had adispositional belief to the effect that the door is unlocked or that he hasn’tbeen to a dog show, respectively. And since knowledge allows fordispositional belief, a proponent of the epistemic theory could argue thatthe reason S remembers at t2 that the door is unlocked or that he hasn’t beento a dog show, respectively, is because S dispositionally knew this at t1. Socases of inattentive and negative remembering do not conflict with the beliefcondition of the epistemic analysis of memory for the simple reason that tobelieve and know that p one need not be conscious that p. To avoidmisunderstanding, the past belief condition of the epistemic analysis ofpropositional remembering

(2.ii) S believed at t1 that p.

should be changed to

(2.ii*) S (occurrently or dispositionally) believed at t1 that p.

This small correction is sufficient to render the epistemic theory compatiblewith cases of inattentive and negative remembering understood as counter-examples to the past belief condition.18 (In [2001: 156 – 7] I have mistakenlyclaimed that the epistemic theory of memory has no way of accounting forcases of inattentive and negative remembering.)

Finally let’s turn to ignorant remembering and the challenge it poses tothe present belief condition (1.ii). The following example is adapted fromMalcolm [1963: 213 – 14] who developed it for a different goal altogether19:

18Pollock [1974: 197] defines memory in terms of implicit rather than explicit knowledge: ‘S occurrentlyremembers-that-p if and only if (i) S believes-that-p on the basis of recalling-that-p, (ii) S implicitly knew thatp, and (iii) S’s recalling-that-p is caused by his having implicitly known that p.’ And implicit knowledge isdefined as follows: ‘S implicitly knows-that-p if and only if the epistemologically relevant circumstances aresuch that he could justifiably believe-that-p, and that belief would be an instance of knowledge’. Thus, whatPollock calls ‘implicit knowledge’ I label ‘dispositional knowledge’.19According to Malcolm’s version of the example, S has memory images of having been kidnapped ratherthan propositional memories to the effect that he was kidnapped. These memory images are classified byMalcolm as instances of ‘perceptual memory’, i.e., a non-propositional version of personal memory.According to Malcolm [1963: 215 – 19], S perceptually remembers x if and only if (i) S previously perceived orexperienced x, (ii) S’s memory of x is based wholly or partly on his previous perception or experience of x,and (iii) S can form a mental image of x. The point of the kidnapping-example, according to Malcolm, is toshow that perceptual memory implies propositional memory. His argument to this effect can be phrased infive steps [cf. 1963: 212 – 14]: (i) S personally remembers x only if S can form an image of x. (ii) If S has animage of x, then S can correctly describe that image of x. (iii) If S can correctly describe the image of x, then xcan correctly describe facts learned in the past. (iv) If S can correctly describe facts learned in the past, then Sremembers that p, where ‘p’ is a true sentence which describes these facts. (v) If S personally remembers x,then S remembers that p. Saunders [1965b: 111] agrees with Malcolm that S perceptually remembers thekidnapping but denies that S propositionally remembers that he was kidnapped, on the grounds that S doesnot believe that he was kidnapped.Other examples of ignorant remembering are due to Martin and Deutscher [1966: 166 – 70] and Lehrer and

Richard [1975: 121]. Martin and Deutscher’s example involves someone who paints what he takes to be an

Remembering without Knowing 149

Dow

nloa

ded

by [

Indi

ana

Uni

vers

ities

] at

01:

11 0

8 Se

ptem

ber

2012

at t2 S suddenly finds himself with the thought that he has been kidnappedwhen he was a small boy (at t1). The idea that he has been kidnapped justpops into his head; it seems to come ‘out of the blue’. S can’t make sense ofthis idea and takes it to be merely imaginary. After all the likelihood ofbeing kidnapped is rather low. What is more, the idea in question isinferentially isolated from the large body of inferentially integrated beliefs towhich S has access. Nothing of what S knows or believes about his pastconnects with the idea that he has been kidnapped. But now suppose that,unbeknownst to S, it is in fact the case that he has been kidnapped. Theflashbulb thought is an instance of propositional memory. Perhaps becauseof the terror of the experience S can’t allow himself to even consider thepossibility that he had been the victim of kidnapping but instead takeshimself to be making it up.20

What distinguishes this case from the blouse-counterexample to thepresent justification condition (1.iii) discussed in Section III is that S notonly lacks justification for what he remembers, but also lacks belief in whathe remembers. It would be wrong to say that S believes at t2 that he waskidnapped at t1. For only when, at t3, he is presented with the police recordand newspaper clippings about his kidnapping, does he reluctantly acceptthat he had been kidnapped when a small boy. It takes considerableconvincing until S consents to the thesis according to which the thought inquestion springs from his memory rather than his imagination. At t2 S notonly remembers that p without believing that he remembers that p, but heremembers that p without believing that which he remembers, namely p.Thus the present belief condition (1.ii) of propositional memory isdisproven.

At first sight it might seem reasonable to suppose that instances ofignorant remembering can be dealt with along the same lines as instances ofinattentive and negative memory. But is it really the case that ignorantlyremembering that p presupposes dispositionally believing that p? Wouldreplacing the present belief condition

(1.ii) S believes at t2 that p.

by

(1.ii*) S (occurrently or dispositionally) believes at t2 that p.

do the trick of rendering cases of ignorant remembering compatible with theepistemic theory? I don’t think so. At t2, that is, before S has been convinced

imaginary scene of a farmhouse. The person’s parents see the finished painting and recognize it as an accuraterepresentation of an actual farm, often visited by the painter as a boy. Lehrer and Richard’s example is aboutsomeone who has a vivid image occur to him of an elderly woman standing by a stone well next to a red barn.The thought occurs to him that the woman is his grandmother. The rememberer doesn’t considerinformation to support this thought, and has no feeling of conviction that the thought is true. But since thethought is actually true and caused in the right kind of way it qualifies as memory.20A terminological note: Most propositional attitude concepts insinuate that the agent accepts p or at leastthat he takes p to be more likely than non-p. Cases in point are to suppose, to consider, to judge, to hold, toinfer, and to assume. The reason I speak of S ‘thinking that p’ and of him ‘having the idea that p’ is becausethese formulations do not presuppose that S accepts p.

150 Sven Bernecker

Dow

nloa

ded

by [

Indi

ana

Uni

vers

ities

] at

01:

11 0

8 Se

ptem

ber

2012

that his flashbulb thought to the effect that he has been kidnapped springsfrom memory rather than imagination, there is no good reason forattributing to him a dispositional belief to the effect that he was kidnapped.For example, if S was asked at t2 whether he thinks it more likely than notthat he has been a victim of kidnapping he would, no doubt, dismiss thesuggestion that he has been kidnapped. And at t3, when he reluctantlyaccepts the thesis according to which the flashbulb thought is due to himremembering having been kidnapped, rather than reviving a dormant belief,he acquires a novel (dispositional) belief.

Granted that S does not dispositionally believe at t2 that he waskidnapped, maybe he believes it implicitly. To implicitly believe that p is tobe disposed to explicitly believe that p while not internally representing p.Now I don’t see that the introduction of implicit belief helps to render theepistemic theory of memory compatible with ignorant-rememberingcounterexamples to the present belief condition. The reason is that, aswas explained above, implicit belief also presupposes acceptance (in the non-technical sense of the term). But clearly S does not accept at t2 that he waskidnapped. And when S does finally accept that he was kidnapped, at t3,rather than saying that he had implicitly believed it all along it is much morenatural to say that he acquires a new occurrent belief.

V. Knowledge and Belief

So far we have assumed that knowledge implies belief. Yet someepistemologists hold that you can know something without believing it. Ifit were possible to know that p without believing that p, then cases ofmemory without belief would not count against the thesis that to rememberthat p is to know that p.

In the early 1970s, epistemologists avidly debated the question of whetherknowing implies believing. Colin Radford [1966] and Zeno Vendler [1972:89 – 120] answered in the negative while David Armstrong [1970] and Lehrer[1970] answered in the affirmative. Radford presents the following exampleof knowledge without belief: S is forced to take part in a quiz on key dates ofEnglish history. He protests quite sincerely that he doesn’t know anyhistorical dates, but when quizzed is able to answer most questions correctly.Radford contends that S knows some history, for his answers are not just alucky guess. The fact that he answers most of the questions correctlyindicates that he has actually learned, and never forgotten, the basic facts ofEnglish history. But at the same time Radford maintains that S does notbelieve, say, that Queen Elizabeth I died in 1603. The reason is that S thinkshe doesn’t know the answer to the questions. He doesn’t trust his answersbecause he takes them to be mere guesses. Radford [1966: 4] writes:

Although in this situation S knows that p, he is not certain, or sure, orconfident that p. Indeed he is fairly certain that his answer to the question iswrong, i.e., that not-p, since he believes it to be a pure guess in a situation

where only one of many such guesses could be correct.

Remembering without Knowing 151

Dow

nloa

ded

by [

Indi

ana

Uni

vers

ities

] at

01:

11 0

8 Se

ptem

ber

2012

Radford concludes that knowledge without belief is indeed possible.The basic problem with Radford’s alleged example of knowledge without

belief is that it presupposes, without argument, that you can know that pwithout being sure that p but that you cannot believe that p without beingsure that p. If belief is a necessary condition for knowledge—as I think itis—then the conditions for believing are eo ipso conditions for knowing.Either both knowledge and belief require subjective confidence or neitherdo. Given my commitment to epistemic externalism, I maintain that neitherbelief nor knowledge demand the feeling of being sure. So while I agree withRadford that S may know, say, that Queen Elizabeth I died in 1603, I thinkthat we should also say that S believes that Queen Elizabeth I died in 1603.

In recent years the idea that knowledge does not involve belief has againbecome fashionable. Timothy Williamson [2000: chap. 1] suggests reversingthe order of explanation between knowledge and belief: instead of analysingknowledge in terms of belief, the concept of knowledge should be used toelucidate the concept of belief. The idea that knowledge is a fundamentaland irreducible mental state is summed up in the slogan ‘knowledge first’.

I am suspicious of the idea that knowledge is an irreducible mental state.It is strange to maintain that knowing is a purely mental state, given thatknowing entails believing truly, which isn’t a purely mental state (at least notif the truth of the proposition is independent of the believer). Williamson[2000: 27] uses an example from geometry to argue that there is nothingstructurally incoherent in the idea that the non-mental state of believingtruly is ‘sandwiched between’ the mental state of believing and the mentalstate of knowing. The non-geometrical property of being a triangle whosesides are indiscriminable in length to the naked human eye is also‘sandwiched between’ the geometrical property of being an equilateraltriangle and the geometrical property of being a triangle. FollowingJonathan Lowe [2002: 484 – 5], I doubt that this analogy is pertinent.Knowing essentially involves the truth of what is known, which is somethingthat is non-mental. Being an equilateral triangle, however, does notessentially involve having sides which are indiscriminable in length to thenaked human eye. The reason for the latter entailment is the purely logicalfact that if lines are of the same length, then they cannot be distinguished inrespect of their length—not by the human eye and not by any other means.

VI. The Analysis of Impure Memory

As was explained at the beginning of Section IV, a putative reason why anagent may remember p without having believed p in the past is because helacked the conceptual means necessary to believe p. To illustrate this kind ofcase reconsider the bluejay-example due to Malcolm that was first discussedin Section III: at t1 S sees a bird without knowing what sort of bird it is, andthen later, at t2, discovers from a book that such birds are bluejays. At t2 hesays ‘I remember that I saw a bluejay at t1’. But it is not true that S believedat t1 that it was a bluejay. He has only just learned that it was a bluejay hesaw. Does this mean that we have a counterexample to the past belief

152 Sven Bernecker

Dow

nloa

ded

by [

Indi

ana

Uni

vers

ities

] at

01:

11 0

8 Se

ptem

ber

2012

condition (2.ii) of the epistemic analysis of remembering? Malcolm answersin the negative. He suggests that when S asserts ‘I remember that I saw abluejay at t1’ he is using this expression in an elliptical sense. Malcolm [1963:223] writes:

I believe that the man in our example would agree to substitute for his sentence‘I remember that I saw a [bluejay]’ the conjunctive sentence ‘I remember that

I saw this bird (or: a bird of this kind) and now I know it was a [bluejay]’. Thesentence he originally uttered was an ellipsis, in the grammarian’s sense,the meaning of which is given by the conjunctive sentence. In this conjunction

the first conjunct expresses factual memory, the second conjunct expresses thenew information.

Since the epistemic theory of memory is intended to capture only purememory, that is, memory with no admixture of inference or presentrealization, the bluejay-example does not count against it.

I agree with Malcolm that cases of impure memory do not pose a threat tothe epistemic analysis of memory. For this reason I have been at pains inSection III to show that neither the Nessy-example nor the blouse-exampleare instances of impure memory. Yet I disagree with Malcolm’s analysis ofimpure memory. On Malcolm’s reading,

(4) S remembers at t2 that he saw a bluejay at t1.

amounts to saying

(5) S remembers at t2 that he saw a bird of this kind at t1 & S comes to knowat t2 that it was a bluejay.

The problem with this analysis of impure memory is that it glosses over thepossibility where S knows that he saw an F, knows that Fs are G, and yetfails to know that he saw a G because he doesn’t draw the proper inference.But if S doesn’t know that he saw a G, it follows that he doesn’t rememberimpurely that he saw a G. Thus to remember that he saw a bluejay S notonly needs to have learned what bluejays are and what they look like, but healso needs to apply this piece of knowledge to the bird he saw at t1. (Thelinguistic expression of S’s applying the newly acquired knowledge to theprevious experience is that the indefinite pronoun ‘it’ in (5) refers back to ‘abird of this kind’.) If, for some reason, S failed to put his knowledge aboutbluejays together with the bird he saw at t1, then it would be wrong to saythat he remembers having seen a bluejay.

Against the background of the ‘application problem’, Eric Stiffler [1980]suggests analysing sentence (4) as an existential generalization:

(6) (9x) (x is a bird of this kind & S remembers at t2 that he saw x at t1 & Scomes to know at t2 that x was a bluejay).

Unlike statement (5), (6) preserves the proper application link between laterknowledge and earlier memory. According to Stiffler, rather than reducing

Remembering without Knowing 153

Dow

nloa

ded

by [

Indi

ana

Uni

vers

ities

] at

01:

11 0

8 Se

ptem

ber

2012

to pure propositional memory plus present knowledge, impure memoryamounts to memory (and knowledge) de re.

Despite the advantages of (6) over (5), Stiffler’s analysis of ellipticalremembering is both too strong and too weak. The analysis is too strongbecause S can remember impurely that he saw a bluejay in the garden eventhough he is not in a position to visualize the particular bird that he saw inthe garden at t1. The analysis is at the same time too weak because the factthat S’s earlier memory and his later knowledge concern the same objectdoes not insure that S ‘puts two and two together’. S can remember withrespect to a particular bird (de re) that he saw it at t1, know now with respectto this bird (de re) that it is a bluejay and yet fail to remember ellipticallythat he saw a bluejay at t1. Thus, analysis (6) doesn’t really solve theapplication problem.

Following Arnold Cusmariu [1980: 307 – 8], I reckon that the applicationproblem can be solved within the confines of Malcolm’s originalconjunctive, de dicto analysis of elliptical remembering. All we need to dois to add a third conjunct to Malcolm’s two-part analysis of statement (4):

(7) (i) S remembers at t2 that he saw a bird of this kind at t1, & (ii) S comesto know at t2 that any bird of this kind is a bluejay, & (iii) S comesto know at t2 that he saw a bluejay inferentially from (i) and (ii).

More generally, where p, q, and r are logically inequivalent propositions, Simpurely remembers that p only if

(8) (i) S remembers at t2 that q, & (ii) S comes to know at t2 that r, & (iii) Scomes to know at t2 that p inferentially from (i) and (ii), & (iv) theconjunction of q and r entails but is not entailed by p.

Yet, as it stands, (8) is too strict, for it assumes that impure memory isnecessarily a form of inferential knowing. Granted that my argumentsagainst the epistemic theory of memory work, (8) must be replaced withsome weaker condition:

(9) (i) S remembers at t2 that q, & (ii) S comes to truly think at t2 that r, &(iii) S comes to truly think at t2 that p inferentially from (i) and (ii),& (iv) the conjunction of q and r entails but is not entailed by p.

VII. Conclusion

We have seen that it is possible to remember something in the present thatone didn’t justifiably believe in the past. Likewise one may acquire in themeantime some plausible but misleading evidence that destroys the status asjustified belief of the once-genuine justified belief that one still remembers.Moreover, cases of ignorant remembering show that one can remembersomething that one does not believe. In sum then, knowledge supervenes on

154 Sven Bernecker

Dow

nloa

ded

by [

Indi

ana

Uni

vers

ities

] at

01:

11 0

8 Se

ptem

ber

2012

some but not all cases of propositional remembering. Unlike knowledge,memory implies neither belief nor justification.21

University of California, Irvine Received: June 2005Revised: November 2005

References

Annis, David B. 1980. Memory and Justification, Philosophy and Phenomenological Research 40: 324 – 33.Anscombe, Elizabeth 1981. Memory, ‘Experience’ and Causation, in Collected Philosophical Papers 2:

Metaphysics and the Philosophy of Mind, Oxford: Blackwell: 120 – 30.Armstrong, David M. 1970. Does Knowledge Entail Belief?, Proceedings of the Aristotelian Society 70: 21 –

36.Audi, Robert 1998. Epistemology, London: Routledge.Ayer, A. J. 1956. The Problem of Knowledge, Harmondsworth: Penguin.Bernecker, Sven 2001. Impliziert Erinnerung Wissen?, in Erkenntnistheorie, ed. Thomas Grundmann,

Paderborn: Mentis: 145 – 64.BonJour, Laurence 1985. The Structure of Empirical Knowledge, Cambridge MA: Harvard University Press.Cohen, L. Jonathan 1992. An Essay on Belief and Acceptance, Oxford: Clarendon Press.Cusmariu, Arnold 1980. A Definition of Impure Memory, Philosophical Studies 38: 305 – 8.Dancy, Jonathan 1985. An Introduction to Contemporary Epistemology, Oxford: Blackwell.De Sousa, Ronald 1971. How to Give a Piece of Your Mind, Review of Metaphysics 25: 52 – 79.Dretske, Fred 1982. A Cognitive Cul-de-Sac, Mind 91: 109 – 11.Dretske, Fred 2000. Entitlement: Epistemic Rights Without Epistemic Duties, Philosophy and Phenomen-

ological Research 60: 591 – 606.Dretske, Fred and Palle Yourgrau 1983. Lost Knowledge, Journal of Philosophy 80: 356 – 67.Dummett, Michael 1993. Testimony and Memory, in Seas of Language, Oxford: Clarendon Press: 411 – 28.Evans, Gareth 1982. The Varieties of Reference, Oxford: Clarendon Press.Ginet, Carl 1975. Knowledge, Perception, and Memory, Dordrecht: Reidel.Ginet, Carl 1988. Memory Knowledge, in The Handbook of Western Philosophy, ed. George H. R.

Parkinson, New York: Macmillan: 159 – 78.Grice, H. P. 1941. Personal Identity, Mind 50: 330 – 50.Holland, Alan 1974. Retained Knowledge, Mind 83: 355 – 71.Huemer, Michael 1999. The Problem of Memory Knowledge, Pacific Philosophical Quarterly 80: 346 – 57.James, William 1890. The Principles of Psychology, 2 vols., London: Macmillan.Landesman, Charles 1961. Philosophical Problems of Memory, Journal of Philosophy 59: 57 – 65.Lehrer, Keith 1970. Believing that One Knows, Synthese 21: 133 – 40.Lehrer, Keith 1990. Theory of Knowledge, Boulder: Westview Press.Lehrer, Keith and Joseph Richard 1975. Remembering Without Knowing, Grazer Philosophische Studien 1:

121 – 6.Locke, Don 1971. Memory, London: Macmillan.Lowe, E. Jonathan 2002. Review of T. Williamson, Knowledge and Its Limits, International Journal of

Philosophical Studies 10: 483 – 503.Lycan, William G. 1988. Judgement and Justification, Cambridge: Cambridge University Press.Malcolm, Norman 1963. Knowledge and Certainty, Ithaca: Cornell University Press.Malcolm, Norman 1977. Memory and Mind, Ithaca: Cornell University Press.Margolis, Joseph 1977. Remembering, Mind 86: 186 – 205.Martin, C. B. and Max Deutscher 1966. Remembering, Philosophical Review 75: 161 – 96.Moore, G. E. 1959. Four Forms of Scepticism, in Philosophical Papers, London: Muirhead Library: 196 –

226.Munsat, Stanley 1967. The Concept of Memory, New York: Random House.Naylor, Andrew 1986. Remembering Without Knowing—Not Without Justification, Philosophical Studies

49: 295 – 311.Nigro, Georgia and Ulric Neisser 1983. Point of View in Personal Memories, Cognitive Psychology 15: 467 –

82.Owens, David 2000. Reason Without Freedom, London: Routledge.Pappas, George S. 1980. Lost Justification, Midwest Studies in Philosophy 5: 127 – 34.Pappas, George S. 1983. Ongoing Knowledge, Synthese 55: 253 – 67.Pollock, John L. 1974. Knowledge and Justification, Princeton: Princeton University Press.Pollock, John L. 1986. Contemporary Theories of Knowledge, Savage: Rowman and Littlefield.Radford, Colin 1966. Knowledge—by Examples, Analysis 27: 1 – 11.Ryle, Gilbert 1949. The Concept of Mind, London: Hutchinson.Saunders, John T. 1965a. Professor Malcolm’s Definition of ‘Factual Memory’, Theoria 31: 282 – 8.

21I wish to thank Chris Daly and an anonymous referee for helpful comments. Financial support for thisresearch was provided by a Heisenberg-Grant from the Deutsche Forschungsgemeinschaft.

Remembering without Knowing 155

Dow

nloa

ded

by [

Indi

ana

Uni

vers

ities

] at

01:

11 0

8 Se

ptem

ber

2012

Saunders, John T. 1965b. Does All Memory Imply Factual Memory?, Analysis 25: 109 – 15.Sellars, Wilfrid 1969. Language as Thought and as Communication, Philosophy and Phenomenological

Research 29: 506 – 27.Senor, Thomas D. 1993. Internalistic Foundationalism and the Justification of Memory Belief, Synthese 94:

453 – 76.Shoemaker, Sydney 1967. Memory, in The Encyclopedia of Philosophy, ed. Paul Edwards, New York:

Macmillan: V, 265 – 74.Shoemaker, Sydney 1970. Persons and Their Pasts, American Philosophical Quarterly 7: 269 – 85.Shope, Robert 1973. Remembering, Knowledge, and Memory Traces, Philosophy and Phenomenological

Research 33: 303 – 22.Squires, Roger 1969. Memory Unchained, Philosophical Review 78: 178 – 96.Stiffler, Eric 1980. Malcolm on Impure Memory, Philosophical Studies 38: 299 – 304.Traiger, Saul 1978. Some Remarks on Lehrer and Richard’s ‘Remembering Without Knowing’, Grazer

Philosophische Studien 6: 107 – 11.Unger, Peter 1972. Propositional Verbs and Knowledge, Journal of Philosophy 69: 301 – 12.Vendler, Zeno 1972. Res Cogitans, Ithaca: Cornell University Press.Williams, Bernard 1973. Deciding to Believe, in Problems of the Self: Philosophical Papers 1956 – 1972,

London: Cambridge University Press: 136 – 51.Williamson, Timothy 2000. Knowledge and its Limits, Oxford: Oxford University Press.Zemach, Eddy 1968. A Definition of Memory, Mind 77: 526 – 36.

156 Sven Bernecker

Dow

nloa

ded

by [

Indi

ana

Uni

vers

ities

] at

01:

11 0

8 Se

ptem

ber

2012