knowing how to believe with justification

21
Knowing How to Believe with Justification Author(s): Steven L. Reynolds Source: Philosophical Studies: An International Journal for Philosophy in the Analytic Tradition, Vol. 64, No. 3 (Dec., 1991), pp. 273-292 Published by: Springer Stable URL: http://www.jstor.org/stable/4320262 . Accessed: 25/06/2014 06:50 Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at . http://www.jstor.org/page/info/about/policies/terms.jsp . JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of content in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new forms of scholarship. For more information about JSTOR, please contact [email protected]. . Springer is collaborating with JSTOR to digitize, preserve and extend access to Philosophical Studies: An International Journal for Philosophy in the Analytic Tradition. http://www.jstor.org This content downloaded from 185.2.32.60 on Wed, 25 Jun 2014 06:50:50 AM All use subject to JSTOR Terms and Conditions

Upload: steven-l-reynolds

Post on 01-Feb-2017

214 views

Category:

Documents


2 download

TRANSCRIPT

Page 1: Knowing How to Believe with Justification

Knowing How to Believe with JustificationAuthor(s): Steven L. ReynoldsSource: Philosophical Studies: An International Journal for Philosophy in the AnalyticTradition, Vol. 64, No. 3 (Dec., 1991), pp. 273-292Published by: SpringerStable URL: http://www.jstor.org/stable/4320262 .

Accessed: 25/06/2014 06:50

Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at .http://www.jstor.org/page/info/about/policies/terms.jsp

.JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range ofcontent in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new formsof scholarship. For more information about JSTOR, please contact [email protected].

.

Springer is collaborating with JSTOR to digitize, preserve and extend access to Philosophical Studies: AnInternational Journal for Philosophy in the Analytic Tradition.

http://www.jstor.org

This content downloaded from 185.2.32.60 on Wed, 25 Jun 2014 06:50:50 AMAll use subject to JSTOR Terms and Conditions

Page 2: Knowing How to Believe with Justification

STEVEN L. REYNOLDS

KNOWING HOW TO BELIEVE WITH JUSTIFICATION

(Received 19 February, 1991)

I will defend the unfashionable view that justified perceptual beliefs are justified in part by a relation to the appropriate sorts of perceptual experiences, where experiences are conceived as non-propositional and indeed non-intentional in character. Experiences are like itches and tickles, and unlike beliefs, in not "saying" anything or even being about anything. They are complex ordered masses of sensations produced in perception. There are reasons to doubt that we have such experiences, but I will not address them here. Instead I will show that, if we have such non-intentional experiences, they can help justify our perceptual beliefs.

It has been recently maintained (by, for example, Donald Davidson and Laurence Bonjour) that no relation of a belief to a non-propo- sitional experience could be epistemically justifying.' Those who hold this view usually concede to the experience a role in producing the belief (if they allow the existence of experiences at all), but they deny that a mental state or process which consists of sensations, and which is, consequently, not about any of the things the belief is about, could play any role in justifying that belief. In this paper I will argue against this claim, by describing a relation of belief to experience which plau- sibly could be justifying.

The only recent attempt to respond directly to the Davidson- Bonjour objection presents the transition from experience to belief as a kind of argument, in which the experience plays something like the role of a premise. I will argue that this response won't work. I argue instead for a view which can be expressed in the following thesis:

A perceptual belief is justified if and only if there are no undermining beliefs, and it was arrived at in response to an experience through an adequate exercise of properly learned recognitional skills.

Philosophical Studies 64: 273-292, 1991. ? 1991 KluwerAcademic Publishers. Printed in the Netherlands.

This content downloaded from 185.2.32.60 on Wed, 25 Jun 2014 06:50:50 AMAll use subject to JSTOR Terms and Conditions

Page 3: Knowing How to Believe with Justification

274 STEVEN L. REYNOLDS

I shall leave the notion of undermining beliefs vague, though I recognize that a full defense of the thesis would require clarification of it. I will attempt to give non-circular explications of the normative terms 'adequate exercise' and 'properly learned', though we will not be prepared for that until toward the end of the paper. In addition to citing evidence for the thesis of a more or less ordinary language sort, I will also try to answer several natural objections to regarding ordinary perceptual processes as exercises of skills.

Although the view I develop in defending this thesis has points of contact with recent work by John Pollock, it is, I believe, a substantially new account of perceptual justification.

I

First some remarks about epistemic justification: Justification evidently has some close relation to knowledge. Thus many recent theories make justified true belief necessary for knowledge. It is also apparently uncontroversial that some beliefs are justified, at least in part, by inferential relations to other beliefs.

Two further assumptions about justification help generate the problems about the experiential justification of beliefs. These assump- tions are somewhat more controversial, but they are plausible and widely accepted, and I won't challenge them. I will try to show that they are compatible with experiential justification.

The first is that justification is a normative notion. Justified beliefs are epistemically acceptable, and may thus be retained, while unjustified beliefs should be changed. Or at least the existence of an unjustified belief indicates some sort of epistemic fault on the part of the believer - at some point he should have done, or believed, otherwise than he did.

The second is that justification has a pronounced "internalist" character. One must be able to tell whether one's beliefs are justified, or at least be able (in some sense) to recognize, and so to attempt to avoid, the things that lead to having or to retaining unjustified beliefs. That justification has this internalist character may be argued for from a deontological conception of epistemic acceptability, on the ground that the right sort of ignorance obviates obligation. Or it might be stipulated,

This content downloaded from 185.2.32.60 on Wed, 25 Jun 2014 06:50:50 AMAll use subject to JSTOR Terms and Conditions

Page 4: Knowing How to Believe with Justification

KNOWING HOW TO BELIEVE WITH JUSTIFICATION 275

more or less, by picking out justified belief as the subjectively accessible aspect of knowledge (understanding that a mental state may wear the aspect of knowledge without actually being knowledge). In any case, whether or not one's belief is justified is commonly thought to depend only on subjectively accessible matters, such as one's own beliefs and experiences.

So much for the assumptions about justification. Now for a few taxonomic stipulations. Coherentists hold that all justified beliefs are in effect justified by inferential relations to other beliefs. (This requires a broad conception of inferential support to be plausible.) Founda- tionalists hold on the contrary that some beliefs would be justified even if they had no such inferential supporting connections to other beliefs. The view that I will defend is foundationalist in this sense. It holds that some beliefs are or could be justified, not by the existence of any supporting beliefs, but by a certain sort of relation to non-propositional experiences, together with the absence of undermining beliefs. Founda- tionalism proper also requires that all inferential support ultimately depends upon such basic beliefs, but I will not be discussing this further thesis. So much for taxonomy.

A foundationalist view doesn't have to say that beliefs are justified by relations to experiences. There are views that are foundational in the sense just described but which do not assign any justifying role to experiences. One such view is that all beliefs that have a certain kind of content are justified (e.g., those that are about one's own subjective perceptual states). Another is that all beliefs for which one has no countervailing reasons are justified. Still another view holds that there is something about the process of acquiring the belief apart from its relationship to experience that confers justification on it. One popular version of this last view says, roughly, that a belief is justified if and only if it was acquired by a process that reliably produces true beliefs.

But it is natural to think that a relationship to perceptual experience justifies at least some of our basic beliefs. On any plausible founda- tionalist view, many of the basic beliefs will be perceptual beliefs. That is, they will be beliefs caused by ordinary perceptual processes, such as seeing, hearing, smelling, tasting, and touching. In those processes, complex arrays of sensations are produced (we are assuming). It seems that these complex arrays of sensations, which I am calling "experi-

This content downloaded from 185.2.32.60 on Wed, 25 Jun 2014 06:50:50 AMAll use subject to JSTOR Terms and Conditions

Page 5: Knowing How to Believe with Justification

276 STEVEN L. REYNOLDS

ences", should have some role to play in justifying the resulting beliefs. If similar processes produced the same beliefs, but without producing any visual, auditory, etc., sensations, as perhaps happens in the phe- nomenon of "blind-sight", we would not want to regard the resulting beliefs as justified. (And indeed patients who exhibit blind-sight abilities apparently do typically regard themselves as "just guessing".2)

Some argue on the contrary that the sensational component of perception is just window dressing, and that the (reliable) causation of appropriate sorts of beliefs is the only epistemically relevant aspect of the perceptual process.3 But there is a strong temptation to think of the sensational aspect of perception as epistemically relevant. Before we arm ourselves with philosophical arguments to resist this temptation, perhaps we should consider whether we can respectably embrace it.

What sort of justifying relation could a non-doxastic experience have to a belief? A first step toward answering this question would be to claim that justification requires an appropriate causal connection between the experience and the belief. For example, if, while at a party, I look absent-mindedly across the room, and just guess that Sam is present, not noticing that he is in fact in view, then my belief is not justified. It doesn't have the appropriate causal connection to my visual experience.

But the requirement of an appropriate causal connection is not specific enough to soothe doubts about the possibility of experiences justifying beliefs. Why should we think that that a causal connection can be justifying? Causal connections are naturalistically described, and justification is normative. We need to describe the experience-belief relation in enough detail to see how it captures the normative aspect of justification.

II

Alan Millar has attempted to show how non-propositional experiences can help justify beliefs, by comparing the transition from experience to belief to the transition from beliefs to belief in inference. (John Pollock had earlier sketched a similar view of the relation of experience to justified belief, but not while trying to answer (at least explicitly) the worries that concern us.)4 On Millar's view, an experience of a certain type, and a belief that there are no undermining beliefs, play the role of

This content downloaded from 185.2.32.60 on Wed, 25 Jun 2014 06:50:50 AMAll use subject to JSTOR Terms and Conditions

Page 6: Knowing How to Believe with Justification

KNOWING HOW TO BELIEVE WITH JUSTIFICATION 277

premises in an argument-like structure whose conclusion is the justified belief. Millar classifies experiences according to the situations and objects that in fact typically produce them. An experience that could have been produced by a normal person's looking at a square yellow piece of paper in appropriate conditions is a square-yellow-piece-of- paper type experience. (This description of types of experience should not mislead us into thinking that, for Millar, the experience is about square yellow pieces of paper. The description just picks out a pheno- menal type by its normal causal antecedents.) According to Millar, people who have such experiences, and have mastered the appropriate concepts, will tend to arrive at beliefs about square yellow pieces of paper in response to them. If such a person also believes that there are no facts that should undermine his belief, and this belief about the absence of undermining facts appropriately affects the transition from the experience to the perceptual belief, then the resulting perceptual belief is justified. The transition from experience and belief to belief is called a "quasi-inference". We learn to make appropriate quasi- inferences from experiences to beliefs, Millar says, as a necessary part of mastering the concepts.

Millar's claim that the recognitional abilities involved are required for the mastery of the concepts seems dubious, for, as everyone now says, it seems that I might have beliefs about, say, elms, without being able to recognize elms. Perhaps however Millar intends to claim a connection between recognition and concepts, not for elms and other natural kinds, but only for concepts such as those of colors. But then what is the justificational status of recognizing elms? Does the lack of a recognition-concept connection imply that the quasi-inferences one makes in recognizing elms won't justify beliefs about elms? Or is the recognition-concept connection not really necessary for experiential justification? (Although he doesn't say so, it may be that he wants that connection because concepts are naturally connected to truth, and he wants to connect justification and truth.)

John L. Pollock also claims that the epistemic norms relevant to the justification of a belief are determined by the concepts involved (where the norms could be expressed as rules for arriving at beliefs, apparently including perceptual recognitional beliefs). He motivates this claim in part by arguing that it rules out a kind of epistemic relativism that he dislikes.S According to this sort of epistemic relativism, it is possible for

This content downloaded from 185.2.32.60 on Wed, 25 Jun 2014 06:50:50 AMAll use subject to JSTOR Terms and Conditions

Page 7: Knowing How to Believe with Justification

278 STEVEN L. REYNOLDS

there to be two people who hold the same belief, on the same grounds (such as being in the same non-doxastic perceptual state), but one is justified and the other is not. Their justification in holding the belief differs because they differ in the epistemic norms to which they happen to be committed. This sort of relativism is ruled out by Pollock's thesis (and presumably also by Millar's), for if the relevant epistemic norms for a belief are determined by the concepts involved in that belief, everyone who holds a particular belief is thereby committed to the same epistemic norms for it. So this sort of relativism is false.

Pollock handles the difficulty about elms by holding that two people may both have the same belief about elms, in one sense of "same belief", without having the same concept of elms.6 There are different ways of thinking of the attribute of being an elm, that is to say, different concepts of being an elm, and people who have different concepts in this sense may nevertheless have the same belief. The relativism he rules out then applies only to beliefs individuated very narrowly - it will still be possible for two people to have exactly similar experiences, and the same belief in response (in the coarser, but much more usual sense of "same belief"), and yet one is justified and the other is not. So there may be ways to successfully finesse the apparent counterexamples to theses connecting recognitional abilities and concept possession.

The other problem with Millar's (and Pollock's) account seems to run much deeper. Millar wants us to think of the transition from experience to belief as somehow like an inference. He calls it a "quasi- inference", and he even presents it in an argument-like format. It is fairly clear why he does. He wants us to accept the transition from an experience to a belief as justifying that belief. Appropriate inferential transitions are uncontroversially justifying, so, naturally, he wants to present the experience-belief transition as analogous, in its normative aspects, to such inferences. But how is the experience-belief transition like an inference?

Inferences are typically transitions from old beliefs to new. (The exceptions are cases where an already existing belief acquires new support, but they don't pose any special problems here.) One starts off, for example, with the beliefs that business executives tend to be impatient, and that Sam is a business executive, and one moves from these pre-existing beliefs to the new belief that Sam is likely to be

This content downloaded from 185.2.32.60 on Wed, 25 Jun 2014 06:50:50 AMAll use subject to JSTOR Terms and Conditions

Page 8: Knowing How to Believe with Justification

KNOWING HOW TO BELIEVE WITH JUSTIFICATION 279

impatient. But not just any transition from old beliefs to new beliefs counts as an inference. The old beliefs must be regarded (in some sense) by the believer as having a relevant logical or evidential relation to the new belief.7

The norms for inferential transitions among beliefs are apparently analogous to the norms by which we evaluate arguments (although there are important differences between the norms of argument and the norms for inferential transitions among beliefs8). Thus in order to apply the norms of inference to determine the quality of a particular infer- ence, one needs to know (at least) what relevant logical or evidential relations there are between the old and new beliefs, just as one needs to know the logical or evidential relations among premises and conclusion in order to determine the quality of an argument.

But the analogy to arguments, which seems fairly strong in the case of genuine inferences, completely breaks down for Millar's quasi- inferences from experiences and beliefs to beliefs. For experiences just aren't like the premises of arguments. They don't, and can't, have evidential or logical relations to the resulting beliefs, because they can't be true. So the truth-preserving norms that we apply to arguments evidently cannot be applied, even with modifications, to the experience- belief transition. But if those kinds of norms can't be applied, then it is not clear yet whether any other sort can be. Thus Millar's attempt to present the experience-belief transition as normative, by analogizing it to the undoubtedly normative inferential transition, fails.

III

But Millar's general strategy for answering doubts about whether experiences can justify beliefs seems to me to be very promising. The idea is to find a process which everyone accepts, or can be brought to accept, as normative in the right way, and then show it to be relevantly analogous to the transition from experience to belief. The comparison with inference doesn't work. But perhaps we can find a more appro- priate comparison.

We might compare epistemic evaluation to the moral evaluation of actions. It has been held, for example, that one has epistemic duties to

This content downloaded from 185.2.32.60 on Wed, 25 Jun 2014 06:50:50 AMAll use subject to JSTOR Terms and Conditions

Page 9: Knowing How to Believe with Justification

280 STEVEN L. REYNOLDS

believe, or to refrain from believing, certain things in certain circum- stances.9 This suggests that the processes that lead to unjustified belief can be understood as analogous to actions in violation of moral duties.

But it is difficult to see how to develop this analogy in order to make it plausible that the transition from experience to belief has a similar normative aspect, perhaps because we are not very clear about the details of the normative structure of our moral evaluations. Are moral norms expressible in rules? If so, what kinds of rules? What are the relevant underlying properties of actions in terms of which such rules should be framed? We don't have uncontroversial answers to these questions, and the attempted answers we do have (utilitarianism, rights theories, virtue theories etc.) don't initially seem very promising as analogies to perceptual justification.

So I don't think that either of these paradigms, the logical evaluation of arguments, or the moral evaluation of actions, is likely to help us understand, by analogy, the normative aspect of perceptual believing.

But there is another normative paradigm worth considering seriously for this sort of project. It is our evaluations of the correctness of particular exercises of learned skills.10

Think of a student playing a piece at the piano from sheet music. We can talk about whether he is playing the piece correctly - whether he gets all of the indicated notes, in the right order, with the appropriate rhythm, dynamics, accents and phrasing. This sort of evaluation does not involve the difficulties and vaguenesses of true aesthetic evaluation - we're not asking whether his performance counts as music, but just whether it's correctly played, according to the generally accepted standards of proper piano technique.

Or think of someone speaking a natural language such as English. We can evaluate the sentences he produces for phonetic and gram- matical correctness. There is clearly a normative dimension to the exercises of these skills. The performances are reliably evaluated as acceptable or unacceptable according to publicly known standards.

The standards for correctness in the exercises of these skills are taught to new performers and critics primarily by bringing to their attention examples of good performance and secondarily by expressing disapproval of performances that don't meet the standards. Rules for correct performance are rarely stated. But if we were to try to articulate

This content downloaded from 185.2.32.60 on Wed, 25 Jun 2014 06:50:50 AMAll use subject to JSTOR Terms and Conditions

Page 10: Knowing How to Believe with Justification

KNOWING HOW TO BELIEVE WITH JUSTIFICATION 281

in words the standards involved, we would presumably have to express them in rules. Thus books on piano technique usually include lots of advice phrased as rules, and attempted grammars for natural languages take the form of systems of rules for forming acceptable sentences. Of course it is understood that the performers usually cannot state such rules in anything like sufficient detail, and they certainly don't explicitly consider such rules in the course of exercising their skills. Nevertheless, if we had a complete system of rules, one that really articulated all of the standards of correct performance at the piano, say, we could characterize any particular mistake (that a fully competent performer or critic would recognize as such) as a violation of the rules. And we could see the performer's attempts to meet the standards of acceptable performance as a matter of trying to follow the rules.

There is a bit of metaphor involved in calling this "rule following", but only a bit. If we are going to talk about skills in any detail, it seems that characterizing them in terms of rules is unavoidable. So the relation to these rules of the person who intentionally performs correctly may as well be called "following" them, provided we keep in mind that he doesn't (and often can't) state or consult them.

Another very important point for our purposes is that the exercise of these skills does not itself require one to have beliefs, either about the rules, or about the circumstances of the antecedent clauses of those rules. "Whenever you are in such and such circumstances, do so and so' says the rule. But to "follow" it, in the sense we have in mind, a performer need not believe that he is in such and such circumstances, at least not in any ordinary sense of "believe", such as concerns us while we are doing epistemology." His ability to perform correctly is a matter of knowing how, which is not to be reduced to any kind of knowing, or even believing, that. The possession of a skill for speaking a language or performing music is not plausibly regarded as a matter of possessing great heaps of wonderfully detailed and instantly accessible propo- sitional knowledge. (This point is obviously important for an account of perceptual justification that makes it analogous to correctness in exercising a skill, because denying that such exercises involve beliefs will avert otherwise threatening regresses of justification.)

Although he doesn't have beliefs that would enable him to usefully consult a rule book of the sort we have been imagining, the performer

This content downloaded from 185.2.32.60 on Wed, 25 Jun 2014 06:50:50 AMAll use subject to JSTOR Terms and Conditions

Page 11: Knowing How to Believe with Justification

282 STEVEN L. REYNOLDS

is nevertheless constantly aware of whether he is performing correctly. He monitors the process of performing, in some sort of non-doxastic way, avoiding or correcting mistakes. So correctness in exercising a skill seems to be "internal" in the way that we expect justification to be.

I claim that the normative aspect of perceptual justification is best understood by analogy with the correctness of such exercises of skills. We have learned how to respond to particular sorts of experiences by acquiring the appropriate sorts of beliefs. To put it more in the language of the thesis stated above: We have recognitional skills, which we exercise in arriving at our perceptual beliefs. If the skill has been properly learned, and the particular exercise of it is up to certain public standards, and the believer does not have other undermining beliefs, then the resulting perceptual belief counts as justified.

IV

I doubt that anyone is going to object to calling our perceptual re- cognitional abilities "skills". It is, after all, only ordinary language, or something pretty close. Ordinary language admits the existence of skilled judges of music or horses or wines or paintings. Part of what makes them skilled judges is that they are more capable of arriving at justified perceptual beliefs about their subject matter than are others who lack their talent and training. They are more sensitive to the relevant perceptual differences than are people who haven't had similar instruction and practice. But the differences to which they are sensitive are relevantly similar to the features that we become sensitive to in learning to recognize people and objects. So no one could reasonably refuse to call our recognitional abilities "skills", though some may want to claim that there are important differences between these recog- nitional skills and such skills as skill in playing the piano.

As in the cases of playing piano or speaking a natural language, there are public standards for correctly arriving at one's perceptual beliefs. Think of a defense attorney cross-examining a witness to a crime. She may ask questions designed to show that he lacks the relevant recog- nitional skills (the ability to recognize a particular person, or a certain kind of gun, for example), or that circumstances were such that he could not, or at least did not, effectively exercise those skills (distrac-

This content downloaded from 185.2.32.60 on Wed, 25 Jun 2014 06:50:50 AMAll use subject to JSTOR Terms and Conditions

Page 12: Knowing How to Believe with Justification

KNOWING HOW TO BELIEVE WITH JUSTIFICATION 283

tions, obstructions of his view, etc.). If the questions succeed, they will show that the witness's beliefs were not justified, for they will show that, in the circumstances, he should not have acquired the perceptual beliefs he has expressed. Very likely he will then feel some embarrassment about having held those beliefs. The jury can tell from the appropriate answers to the questions whether or not the standards for the acqui- sition of perceptual beliefs are likely to have been violated. So there must be public standards, public not in the sense that others can follow the rules by responding to the same experiences the performer does, but in the sense that others have ways of telling whether the recogni- tional standards the witness has have been properly learned and adequately complied with.

The witness's capacity to respond to the questions in a revealing way also seems to indicate that these standards are internally applicable, at least at the time of acquiring the beliefs. ("The shape of his chin did seem a little odd at the time, I must admit.") Memory may fail to reveal the incorrectness later, but the believer was able to tell, at the time, whether he was arriving at his perceptual beliefs correctly.

Some people may still doubt however whether these so-called perceptual skills are relevantly similar to the skills of piano playing and speaking a natural language.

There are three apparent disanalogies: 1) The normative aspect of exercises of skills such as playing the piano or speaking English can be seen as a matter of "following rules". But how can we ever hope to see perceptual judgments as a matter of following rules? How could one hope to write rules for perceptual judgment? 2) Arriving at perceptual beliefs on having experiences doesn't seem to be something that one does, in the way that one speaks or plays the piano. It seems to be something that just happens, like digestion. But if so, then, again, how could it be a matter of "following" prescriptive rules? 3) Correctness in exercises of the skills of piano playing and speaking English seems to be largely a matter of arbitrary cultural history. But epistemic justi- fication is not a matter of arbitrary cultural history; cultural variation in standards of justification seems to be constrained by a close relation- ship between justification and truth. But how could there be such a relationship on the skills account?

This content downloaded from 185.2.32.60 on Wed, 25 Jun 2014 06:50:50 AMAll use subject to JSTOR Terms and Conditions

Page 13: Knowing How to Believe with Justification

284 STEVEN L. REYNOLDS

V

Let us take these apparent disanalogies in order. For the first question, as to whether any system of rules could describe our transitions from the patterns of experience to beliefs, we have a model at hand in the attempts to write visual recognition programs for computer driven robots. Presumably those programs are roughly akin to the sets of rules that would (if we could only write them) capture the recognitional skills of a human being. The rules for the experience-belief transition would collectively prescribe beliefs of certain types as responses to appro- priate patterns in sensory input, or in experience, where the patterns might be described, perhaps, in terms of sensory qualities occupying regions of a sensory field.

We can now try to explain the term 'adequate exercise'. An adequate exercise of a recognitional skill conforms to the rules that would capture that skill.

Most of the complications found in describing the patterns of experience to which one responds in recognition presumably have no particular philosophical interest, and are best left to experimentally oriented cognitive scientists. One of the advantages of giving up the claim that having recognitional abilities is somehow constitutive of having the concepts, or of understanding the meanings of the corre- sponding words, is that we aren't tempted to suppose, as perhaps some of the sense-data theorists did, that describing our recognitional abilities is equivalent to providing definitions, or elucidating ideas, and should therefore be capable of being done a priori.

I don't think recognitional skills can be described a priori. But I do have some speculations to offer about the structure of our recognitional abilities.

Experience is amorphous stuff, not easy to divide up into separate experiences. But suppose we agree to think of a person's total visual sensations over a period of a few seconds, as organized, as a single visual experience. We will distinguish types of such experiences by phe- nomenal quality - say, if it's phenomenally indistinguishable in (accurate) memory, then it's the same type of experience. Note that Millar's classification of experiences is much coarser than this - on my classification scheme square-yellow-piece-of-paper Millar-type experi-

This content downloaded from 185.2.32.60 on Wed, 25 Jun 2014 06:50:50 AMAll use subject to JSTOR Terms and Conditions

Page 14: Knowing How to Believe with Justification

KNOWING HOW TO BELIEVE WITH JUSTIFICATION 285

ences are broken down into a vast number of phenomenally distin- guishable experiences that might be caused, in part, by visual contact with a square yellow piece of paper. The finer individuation is desirable for talking about leaming recognitional skills.

It seems that I am constantly having experiences of phenomenal types I have not had before, in response to which I arrive at beliefs of types that I have not had before. And it seems that the number of types of experiences that I can have, and the number of resulting types of beliefs, are practically unlimited. But how can any finite and learnable skill have this sort of protean creativity?

This question echoes the familiar question about how we can understand an indefinitely large number of sentences that we haven't heard before. The natural answer to it is similar - we must understand our skills for arriving at perceptually justified beliefs as being com- pounded of lesser skills which can be combined and re-combined in a practically unlimited number of ways. My ability to be perceptually justified in believing that Sam is standing in the doorway will somehow be composed of my several abilities to recognize Sam, the posture of standing, the doorway, and instances of the relation of being in. Each of these abilities can also be used in responding to many other experiences of types that I have never had before - such as when I come to believe that Irving, whom I have previously only seen in a seated posture, is standing.

So it seems that our perceptual skills for appropriately arriving at beliefs must be composed of re-combinable sub-skills, which, ap- parently, roughly correspond to some of the nouns and predicates of our language. Arriving at the justified perceptual belief that Sam is standing requires recognizing the referent of 'Sam', by, for example, a pattern of visual qualities produced by light reflected from his face; it requires recognizing an instantiation of the predicate 'is standing', by another pattern of visual qualities; and (to justify the combination) it requires recognizing the appropriate arrangement in the experience of the (facial) pattern for 'Sam' and the (bodily) pattern for 'is standing'. Since the ability to recognize a pattern is presumably also the ability to tell when it is absent, justifiedly coming to believe that Sam is not standing also requires an exercise of the same abilities. One will be responding to the facial pattern for 'Sam', and the absence, in any

This content downloaded from 185.2.32.60 on Wed, 25 Jun 2014 06:50:50 AMAll use subject to JSTOR Terms and Conditions

Page 15: Knowing How to Believe with Justification

286 STEVEN L. REYNOLDS

appropriate relation to the pattern for 'Sam', of the pattern for 'is standing'. Perhaps for the perceptual belief that someone is standing one would only need to exercise the ability to recognize the (humani- form) visual pattern for 'is standing'.

These analogies between the structure of our recognitional abilities and the linguistic structures in which we express the resulting beliefs may help explain why we are tempted to say that we see what makes a certain sentence true, or that we perceive the fact that corresponds to the sentence. It also suggests some tenuous connection between truth and justification by recognitional skills, although I won't be relying on this sort of connection in reply to that objection.

This account may appear to be at odds with Quine's epistemology. He connects observation sentences, not terms, with experiences.12 But the provision for undermining perceptual justification retains the most considerable part of Quine's holism; on this view empirical justification is not independent of systematic considerations. And Quine himself has taken a small step in the direction of connecting terms with sensory stimulations in his suggestion that some predications, such as 'This pebble is blue', are compoundings of observation sentences.13

VI

The second objection holds that, unlike playing the piano or speaking a language, perceptually coming to believe is not something that one does."4 It just happens, like digestion, so it doesn't make sense to call it a skill in the full sense of the word 'skill'. Of course we can evaluate it as working well or ill, as we could evaluate the workings of someone's digestive system, but we cannot evaluate it as a performance, and so regard it as correct or incorrect. On this view the rules that might be stated by cognitive scientists studying our recognitional processes should not be regarded as rules in the sense of prescriptions; instead they are generalizations about the process. They are more like scientific laws for normal biological systems than they are like the laws of a nation. We follow such rules only in the sense in which we might be said to follow rules that describe the process of (normal) digestion. But then the normative aspect of the transition from experience to belief disappears, since the prescriptive rules in which we had hoped to

This content downloaded from 185.2.32.60 on Wed, 25 Jun 2014 06:50:50 AMAll use subject to JSTOR Terms and Conditions

Page 16: Knowing How to Believe with Justification

KNOWING HOW TO BELIEVE WITH JUSTIFICATION 287

represent it now appear to be instead only idealized factual generaliz- ations.

Presumably the reason for holding that perceptual processes just happen is that they seem to be quite automatic, and not under our conscious control. But a process can be quite automatic, and even out of one's conscious control, and yet still be an instance of the exercise of a skill in the full, normatively governed sense. Indeed, the transition from deliberate, controlled, "doing" of constituent actions to their "automatic" performance is typically required for mastery of a skill. Compare a beginning piano player's laborious and deliberate search for the correct keys with the expert player's automatic reaching for them, or the novice fencer's slow, deliberate actions with the expert's "re- flexive" responses. Or compare a chess novice's deliberate assessment of the material advantage by tabulating the pieces on each side with the expert's instantaneous and practically involuntary assessment.

The process of acquiring new perceptual skills evinces a similar distinction between novice and expert. A novice first identifies elms slowly and deliberately, looking for particular distinguishing features. With practice, this process becomes quicker, until at last she can't help but recognize an elm at a glance.

The objection thus seems to be claiming, absurdly, that we should stop regarding recognition as normatively governed just at the point where full proficiency is reached.

But it may be replied that control is necessary for normative evaluation to be appropriate. If the unjustified believer lacks control, he could not be at fault. In answer we may say that sufficient control is exercised in acquiring, or failing to acquire, appropriate instruction and practice, and in controlling the circumstances - e.g. making sure one gets a good look.

So I don't think it is plausible to hold that perceptual processes fail to be performances in any way that threatens our normative charac- terization of them as exercises of skills.

VII

The final objection holds that the skills account cannot or will not include a close relation of justification to truth, and that it consequently

This content downloaded from 185.2.32.60 on Wed, 25 Jun 2014 06:50:50 AMAll use subject to JSTOR Terms and Conditions

Page 17: Knowing How to Believe with Justification

288 STEVEN L. REYNOLDS

threatens to leave justification too dependent on the vagaries of cultural and individual development of epistemic norms.

A more or less standard account of the relation of truth and justi- fication sees truth as a goal, and justification as an evaluation relative to that goal. One is justified if and only if one believes as one ought. This 'ought' is understood on the model of the 'ought' that occurs in 'If you want go to the market, you ought to turn right at the second stop sign'. It indicates advice about how to achieve a goal. Roughly speaking, the epistemic goal is to acquire true beliefs and avoid false beliefs. This goal is not often explicitly stated, since it is assumed that everyone has it, or, perhaps, more cautiously, that striving to realize this goal is a defining characteristic of the epistemic enterprise, so that advice in epistemic contexts always presupposes it.15

One might try to depict exercises of recognitional skills as always being focused, somehow, on a goal of truth, as one might try to depict exercises of piano playing skills as always aiming at a goal of beauty. But that seems most implausible. I doubt that most perceptual believers have a goal of truth, in any straightforward sense of 'have a goal', and I know for a fact that not all piano students have beauty as their goal. Often they're only seeking to avoid embarrassing mistakes. They just want to play it correctly, as they were taught.

It seems more plausible to regard 'epistemically justified', not as an evaluation relative to a goal of achieving true and avoiding false beliefs, but instead as an evaluation indicating an acceptable degree of con- formity to epistemic norms. In the case of perceptual beliefs, it indicates an acceptable degree of conformity to the rules that would describe the appropriate recognitional skill.

What relation does justification then have to truth? I think that a tendency to produce true beliefs causally explains why we are com- mitted to certain epistemic norms. It is true, on my view, that certain practices of belief acquisition are justifying because they tend to produce true beliefs. But the 'because' indicates, not a goal, but the salient aspect of a causal explanation of our adoption of the epistemic norms that permit such belief acquisition.

First, notice that recognitional skills, if they are to be justifying, have to be properly learned. For example, ordinarily I would learn to recognize a particular person, Sam Smith, when he is pointed out to me,

This content downloaded from 185.2.32.60 on Wed, 25 Jun 2014 06:50:50 AMAll use subject to JSTOR Terms and Conditions

Page 18: Knowing How to Believe with Justification

KNOWING HOW TO BELIEVE WITH JUSTIFICATION 289

or introduced to me. Let us suppose instead that I know him by reputa- tion, but not by sight. I acquire a recognitional skill associated with his name in the following way. At a large meeting where I expect him to be present I notice, among many people whom I do not know, someone whom I think may be Sam Smith. I look at this person several times, fixing his appearance in my memory.

Later, not recollecting these events, I see this same man coming out of Mae's restaurant, and, in response to the visual experience thus produced, I come to believe that Sam Smith is coming out of Mae's. I have a recognitional skill which I use on this occasion, yet my belief that Sam Smith is coming out of Mae's is not justified. The reason seems to be that I haven't properly learned how to recognize Sam Smith (even if this man is Sam Smith).

Similar stories could be told about my abilities to recognize instan- tiations of various predicates. It thus seems that recognitional skills have to be properly learned to be capable of helping justify the per- ceptual beliefs they give rise to.

Now think of our earliest acquisitions of recognitional skills, when, as children, we learned to distinguish and name such things as colors, shapes, common objects, and a few people. It is most implausible that a small child has any goal of truth in learning these skills. He has various other goals: finding Mama, taking food, seeing a certain color, avoiding falling, chewing on something soft but resistant, gaining parental approval.

He responds to a particular sort of experience, when he is hungry, by reaching and grabbing. If the experience is of a certain kind, by so doing he successfully satisfies his hunger. As this is repeated he learns to believe that there is food in front of him in response to this sort of experience. If such beliefs are likely to be true when he is having that sort of experience, the actions guided by the beliefs thus arrived at will frequently be successful. The successes encourage similar beliefs in response to similar experiences in the future. The behavior guided by these beliefs also brings approval or correction from others. (I mention the role of other people because I think there is something to the Kripke-Wittgenstein claim that prescriptive rules are possible only in a community.16) He thus gradually adopts as a perceptual norm this pattern of believing in response to that sort of experience.

This content downloaded from 185.2.32.60 on Wed, 25 Jun 2014 06:50:50 AMAll use subject to JSTOR Terms and Conditions

Page 19: Knowing How to Believe with Justification

290 STEVEN L. REYNOLDS

The likelihood of developing true beliefs when responding to experiences as a norm requires is thus typically a part of the causal explanation of someone's adopting that perceptual norm.

Being likely to produce true beliefs is not essential to a properly learned recognitional skill however. If someone fraudulently introduces himself to me as Sam Smith, then I will acquire a recognitional skill that will typically lead me to have false beliefs about Sam Smith. But so long as I exercise this skill properly, and do not acquire any new reason to doubt it, the resulting beliefs will be justified. A properly learned recognitional skill is just a skill confirmed by successful actions and (a certain kind of) approval from other members of the community.

Should this soothe fears that the skills approach is prone to a bad sort of cultural or individual relativism? If my conjectures are roughly correct (and no doubt they need refinement), then, presumably, accepted epistemic practices which seem to us not to justify the resulting beliefs, fail to do so, in our opinion, because we think that some other pressures guided their adoption. They were adopted for some reason other than success in the resulting actions and the appro- priate approval from the community. The dubious epistemic practices of primitive societies are seldom purely recognitional in character. But perhaps the following case sufficiently illustrates this point. A man has a powerful desire to find rare minerals easily. After finishing a good mineral identification course involving frequent correction by experts, this man, although he has normal visual acuity and color vision, con- fidently identifies various worthless stones as rare minerals. Evidently he has failed to benefit from correction while he was learning. The norms he has in fact adopted were shaped not by success in his actions or by the approval or disapproval of his instructors, but by his desire to see himself as easily finding rare minerals. So his recognitional skills have not been properly learned, and, consequently, exercises of them are not justifying. Other societies' adoption of epistemic norms which seem to us not to be justifying will also, on examination, be found to have been wrongly influenced, perhaps by religion or politics, and so to have been improperly learned. The skills approach will thus rule that these practices are not epistemically justifying.

Now we are in a position to answer a final question about the relation between experience and truth. We assumed that an experience

This content downloaded from 185.2.32.60 on Wed, 25 Jun 2014 06:50:50 AMAll use subject to JSTOR Terms and Conditions

Page 20: Knowing How to Believe with Justification

KNOWING HOW TO BELIEVE WITH JUSTIFICATION 291

isn't propositional and isn't about anything. But by way of it, it seems, we can learn the truth. How is that possible?

I think that the experience conveys information of the truth in question, in somewhat the way that a dinosaur footprint in rock conveys information about the dinosaur. The rock isn't about the dinosaur, but by looking at it we can learn about the dinosaur.

Experiences differ from fossils in that experiences are typically not objects of thought for the person who obtains beliefs in response to them. He doesn't have beliefs about the experience as the paleontologist has beliefs about the fossil. So perhaps in some ways my experience is typically more like the electronic signal in a telephone line - although it brings me information, I don't think about that signal when receiving the telephone messages. Likewise, although the experience brings me information, I don't usually think about the experience itself. I just process the information from the experience into a belief that really is about the objects the information comes from.

This distinction between merely carrying information and being genuinely about something is a slippery one, but I think that the examples strongly suggest that it is genuine. The visual experience I have carries the information that Bernie is standing in the door, but I don't have any mental state that is about Bernie just in virtue of having that experience. Even if I do acquire a belief about Bernie in response to it, it will still be true that that very same complex mass of sensations was carrying lots of other information about things that I probably didn't acquire any beliefs, or other propositional mental states, about lighting patterns, shapes etc. There is a mass of detailed information in that array of sensations, only some of which was processed into pro- positional mental states.17

The information contained in the experience has to be processed to result in a perceptual belief. I have argued that this processing is normatively governed in such a way that it is appropriate to think of it, and so of the experience, as capable of helping to justify the resulting belief."8

NOTES

1 Donald Davidson, 'A Coherence Theory of Truth and Knowledge' in Kant Oder Hegel: Stuttgart: Klett-Cotta 1983, p. 428, reprinted in E. Le Pore (ed.) Truth and

This content downloaded from 185.2.32.60 on Wed, 25 Jun 2014 06:50:50 AMAll use subject to JSTOR Terms and Conditions

Page 21: Knowing How to Believe with Justification

292 STEVEN L. REYNOLDS

Interpretation (Oxford, 1986) p. 311. Laurence Bonjour, The Structure of Empirical Knowledge, Cambridge, Mass: Harvard University Press, 1985, pp. 69, 75. 2 L. Weiskrantz, 'Varieties of Residual Experience' Quarterly Journal of Experimental Psychology, XXXII, 3 (1980): 365-386. See especially 371, 374, 378. 3Donald Davidson, 'A Coherence Theory' p. 428, Laurence Bonjour, The Structure of Empirical Knowledge, 111-1 38. 4 Alan Millar, 'Experience and the Justification of Belief', Ratio (New Series) II 2 December 1989, 138-152. John L. Pollock, Contemporary Theories of Knowledge, Savage, Maryland: Rowman & Littlefield, 1986, 175-7. 5John L. Pollock, Contemporary Theories of Knowledge, 148, 175-7. It is also supposed to be useful in replying to skepticism. John L. Pollock, Knowledge and Justification, (Princeton: Princeton University Press, 1974) 20-22. 6 John L. Pollock, The Foundations of Philosophical Semantics, Princeton: Princeton University Press, 1984, Chapter 2. 7Millar, 'Experience', p. 140. 8 Gilbert Harman, Change in View: Principles of Reasoning, Cambridge, Mass.: MIT Press 1986, pp. 1-20. 9 Laurence Bonjour, The Structure of Empirical Knowledge, chapter 1, John L. Pollock, Contemporary Theories of Knowledge, pp. 7-8, Roderick M. Chisholm, Theory of Knowledge, Englewood Cliffs, New Jersey: Prentice Hall, 1989, pp. 58-60. For argument against, see William P. Alston 'The Deontological Conception of Epistemic Justification' in his Epistemic Justification: Essays in the Theory of Knowl- edge. Ithaca: Cornell University Press, 1989, pp. 1 15-152. 10 The analogy has been used for other purposes. John L. Pollock, Knowledge and Justification 15, and Contemporary Theories of Knowledge, 126-132. 11 For senses that are not ordinary, see Daniel C. Dennett, The International Stance, Cambridge Mass.: MIT Press 1987). 12 Willard Van Orman Quine, Word and Object Cambridge Mass.: The MIT Press 1960. 13 W. V. Quine, Pursuit of Truth, (Cambridge, Mass.: Harvard University Press, 1990) 4. 14 For discussion of the related issue of doxastic voluntarism, see William P. Alston, 'The Deontological Conception of Epistemic Justification' 119-133, Richard Feldman 'Epistemic Obligation' in James E. Tomberlin ed. Philosophical Perspectives, 2, Atascadero: Ridgeview, 1988, p. 239, Alvin I. Goldman, Epistemology and Cognition, Cambridge, Mass.: Harvard, 1986, pp. 26, 384 n. 6. 15 One statement of this standard account is in Bonjour, The Structure of Empirical Knowledge pp. 7-8. 16 Saul A. Kripke, Wittgenstein on Rules and Private Language, Cambridge, Mass.: Harvard University Press, 1982, 88-93. 17 Fred Dretske, Knowledge and the Flow of Information, Cambridge Mass.: MIT Press, 1981, chapter 6. 18 Thanks to Arizona State University College of Liberal Arts and Sciences for a summer grant supporting my research. Rogers Albritton warned me against a number of errors in my early thinking about this approach to epistemic justification. Stewart M. Cohen, Gregory W. Fitch, Bernard W. Kobes, and Michael J. White provided helpful comments on earlier drafts.

Department of Philosophy Arizona State University Tempe, AZ 85287-2004 USA

This content downloaded from 185.2.32.60 on Wed, 25 Jun 2014 06:50:50 AMAll use subject to JSTOR Terms and Conditions