Evidence does not equal knowledge

Evidence does not equal knowledge Philosophical Studies An International Journal for Philosophy in the Analytic Tradition ISSN 0031-8116 Volume 153 Number 2 Philos Stud (2009) 153:235-242 DOI 10.1007/ s11098-009-9488-1 1 23 Your article is protected by copyright and all rights are held exclusively by Springer Science+Business Media B.V.. This e-offprint is for personal use only and shall not be selfarchived in electronic repositories. If you wish to self-archive your work, please use the accepted author’s version for posting to your own website or your institution’s repository. You may further deposit the accepted author’s version on a funder’s repository at a funder’s request, provided it is not made publicly available until 12 months after publication. 1 23 Author's personal copy Philos Stud (2011) 153:235–242 DOI 10.1007/s11098-009-9488-1 Evidence does not equal knowledge Aaron Rizzieri Published online: 5 December 2009 Ó Springer Science+Business Media B.V. 2009 Abstract Timothy Williamson has argued that a person S’s total evidence is constituted solely by propositions that S knows. This theory of evidence entails that a false belief can not be a part of S’s evidence base for a conclusion. I argue by counterexample that this thesis (E = K for now) forces an implausible separation between what it means for a belief to be justified and rational from one’s perspective and what it means to base one’s beliefs on the evidence. Furthermore, I argue that E = K entails the implausible result that there are cases in which a well-evidenced belief necessarily can not serve as evidence for a further proposition. Keywords Epistemology
Evidence
Justification
Knowledge 1 Introduction Timothy Williamson has argued that one proposition e can serve as evidence for a hypothesis h for a subject s under two conditions. The first condition is that the proposition e must be known, and the second condition is that the probability of h on e must be greater than the probability of h alone.1 It is my purpose to challenge this first constraint on evidence (E = K for now). In Sect. 1, I argue by counterexample that (E = K) forces an implausible separation between what it means for a belief to be rational, justified, and properly guided in its formation on the one hand, and evidentially supported on the other. These 1 For Williamson’s views on evidence see his (2000, Chs. 9 and 10). This essay will concern chapter 9 exclusively. A. Rizzieri (&) Humanities Department, LaGuardia Community College, 31-10 Thompson Avenue, Long Island City, NY 11101, USA e-mail: aaron.rizzieri@gmail.com 123 Author's personal copy 236 A. Rizzieri arguments are similar in spirit to arguments that evidentialists and other epistemic internalists have used against reliabilist views of justification.2 In Sect. 2, I suggest a possible response to these arguments that is in accord with how Williamson handles cases of perceptual illusion. I then argue that this response leads to the implausible conclusion that a false belief can have the property of being evidentially supported and yet necessarily cannot itself serve as evidence for further beliefs. Furthermore, I argue that there are cases involving ‘‘forgotten evidence’’ (Goldman 1999) in which it becomes especially awkward to insist that the false belief that is serving as an immediate ground of an inferred belief is not part of the evidence base for that belief. 2 An argument by counterexample Let us consider a situation in which it can be argued that a person’s evidence base contains a false belief. In cases of inferred beliefs, it is especially plausible to hold that a belief that p is justified if and only if one properly bases this belief on evidence that one possesses in favor of that p. A reliabilist, for example, might deny that having justification for our basic beliefs entails having evidence for these beliefs. It should be noted that Williamson holds that having a justified belief that p entails having evidence that p.3 The examples that will occupy our attention involve inferred beliefs that are justified because there is evidence for them.4 Consider the following case. I believe that nobody can enter my office (O for now) because I believe that I have just locked the door (LD for now). Let us stipulate that I have inferred (O) from (LD). I pushed the lock in and gave it a quick twist to the left, which usually does the trick; however, my lock is damaged and does not work. Hence, (LD) is false. If Williamson’s proposal that (E = K) is correct then (LD) can not serve as an evidential ground for (O). This generates problems for (E = K). The first difficulty is that it is very plausible that (LD) does partially constitute my evidence for (O). After all, I am justified in believing (LD), (LD) supports (O), and an explicit inference from (LD) is my most immediate basis or ground for (O). I have other indirect grounds for (O) as well. For example, whatever my evidence for (LD) is (see below), is also evidence for (O). Why should my indirect grounds for (O), count as evidence for (O) when (LD) does not? Obviously, Williamson’s answer is going 2 I have in mind Laurence Bonjour’s clairvoyance argument and Stewart Cohen’s ‘‘new evil demon’’ argument. See Bonjour (1980) and Cohen (1984). 3 For Williamson’s discussion of the relationship between justification and evidence see his (2000, pp. 207–208). This might be a little bit surprising because Williamson’s account of evidence is externalist in nature in both the conceptual and epistemic senses of the term ‘external.’ 4 It is common for epistemologists to distinguish between propositional justification and doxastic justification. Imagine that I see a chair in front of me, but I fail to form a belief that a chair is present on the basis of that evidence. Instead, I base this belief on the superstition that every time I am in this room on a Thursday a chair will be in the room. The proposition ‘‘a chair is present’’ is justified for me in the sense that I have ready access (the chair is in my visual field) to evidence for this belief. My belief that a chair is present remains unjustified. It is doxastic, as opposed to propositional justification, that is the sense of justification that is a candidate for being a necessary condition for knowledge. 123 Author's personal copy Evidence does not equal knowledge 237 to be, ‘‘because (LD) is false but the beliefs that provide evidence for (LD) and ultimately (O) as well are true.’’5 However, there are features of the evidential support relation and the details of the case at hand that make it difficult to deny that (LD) is part of my evidence base for (O). Consider the following features of the evidential support relation and the present case: (F1) Proposition e is evidence for S’s belief that p only if that p is based on e. (see Footnote 4) (F2) Inferring one belief from another is a paradigmatic way of basing one belief on another. (F3) (O) was inferred from (LD). (F4) (LD) is itself justified and reasonable for me to accept. (F5) (LD) renders (O) more probable than (O) would be otherwise; this is Williamson’s probability constraint on evidence. When all of these features are laid out it is easy to understand the intuitive pull behind including (LD) in my evidence base for (O). Some of these features highlight the significance of the basing relation in regards to the evidential support relation and the accrual of epistemic justification (F1–F3). Others make explicit the intimate connections that hold between the family of terms justified, probable, reasonable, and evidenced (F4–F5). 3 A possible rejoinder What could someone that accepts (E = K) say in response to this argument by counterexample? The (E = K) theorist can deny that (LD) is a part of my evidence base for (O) and yet not surrender the intuition that (O) has evidence in its favor which on Williamson’s own view is necessary for one’s having a justified belief that (O). Perhaps my evidence for (O) is constituted by these propositions: (1) (2) I appear to have locked my door. In the past when I have appeared to have locked my door, I have actually locked my door.6 Williamson adopts a similar strategy in regards to cases that involve perceptual illusions: If perceptual evidence in the case of illusions consists of true propositions, what are they? The obvious answer is: the proposition that things appear to be that way.7 5 I go into more detail later concerning what constitutes this indirect, or perhaps implicit, evidence. Basically, I have in mind my grounds for affirming (LD) itself. This would be my belief that the door appears to be locked and my belief that doors that appear to be locked are likely to be locked. 6 (2) might not be necessary. This depends on whether or not (1) justifies the conclusion that the door is locked all by itself, without one having to make any additional justified assumptions. The ‘‘problem of easy knowledge’’ is relevant here. See White (2006) and Cohen (2002). 7 Williamson, ibid. p. 198. 123 Author's personal copy 238 A. Rizzieri In regards to (1) and (2), it is plausible that these two propositions are the grounds of my false belief (LD) that I inferred (O) from. Furthermore, if I retained (1) and (2) but eliminated (LD) from my noetic structure, I would still be in an equally strong position to evaluate the truth of (O). (1) and (2), unlike (LD), have the property of being known. The conjunction of (1) and (2) also meet Williamson’s probability constraint on evidence in relation to (O). Hence, thus far we have not unearthed anything approaching an absurdity for the (E = K) theorist since the (E = K) theorist can still give a full defense of (O) on her theory of evidence. 4 Cases that involve a false essential premise Let us introduce a more sophisticated case. Consider a subject s that has a set of beliefs that constitutes her evidence base e. Suppose that e supports a theory t1 much better than a rival theory t2. There is a proposition, that p, that is entailed (or at least well-supported) by t1, but not by t2. Hence, it would not be reasonable for s to believe that p without also believing t1 and understanding that this theory supports that p. In such a situation t1 is an essential premise for s’s belief that p. By ‘‘essential premise’’ I mean that it would not be reasonable for s to accept that p without also accepting t1. This makes it especially difficult to deny that t1 is a part of s’s evidence base even if t1 is false. Whether or not a premise is essential for a person will depend on what other beliefs they hold and what inferences they are capable of making. For example, s might believe a set of propositions that happen to support a further proposition (that s also believes) and not be able to make the relevant inference. In such a case s could not use that set of beliefs to accrue doxastic justification for her further belief. Even if s had the cognitive skill to pull off such an inference, if s had not in fact made the inference from the set in question to her further belief, then the members of that set are not what is furnishing s with doxastic justification for the latter belief. Assuming (E = K), if t1 is known then t1 must be included in s’s evidence base for that p. However, if t1 is not known (because t1 is false for example) then t1 is not evidence for that p even though s must believe that t1 on the basis of her more foundational beliefs and understand that t1 supports that p in order for those more foundational beliefs to provide s with evidence for that p. Hence, on (E = K) it is possible for an essential premise to fail to be a part of one’s evidence for that p. An (E = K) theorist is forced to hold that a well-evidenced false belief can be necessary for an agent to have a reasonable belief that p on a particular occasion even though that false belief is not itself evidence for that p. Again, we have a result that is counterintuitive. Rules of inference such as modus ponens have similar properties. When we use the modus ponens argument form we do not consider this inference pattern itself as part of the evidence for our conclusion; yet, unless we are capable of using inference rules such as modus ponens we will not be able to have well-evidenced inferred beliefs. On (E = K) well-evidenced false beliefs that serve as essential premises 123 Author's personal copy Evidence does not equal knowledge 239 would function a lot like our general inference rules in that they would serve to transmit, but would fail to constitute, evidence for beliefs that are inferred from them. That a specific contingent belief or theory could function in this manner is another odd result. 5 The asymmetry argument This brings us to my next (and related) concern that I will call ‘‘the asymmetry argument’’. Williamson agrees that a set of known propositions can lend inductive support to a further proposition, even if that proposition is false and therefore unknown. Williamson is stuck with the conclusion that well-evidenced false beliefs necessarily have a strange asymmetry in regards to their evidential warrant. They can be well-evidenced but necessarily cannot serve as evidence for a further belief that p. How bad is this result? Must every belief that has evidential warrant be capable of serving as evidence for a further belief? Here I think the answer is no. Williamson provides an example (2000, Ch. 8). Suppose that I draw balls n (assume that n is a relatively large number) times from a bag (with replacement) that contains a small number of balls and all the drawn balls are red. I have a well-evidenced belief that the next ball n ? 1 that I draw will be red also, but this well-evidenced belief can not be used as evidence for the conclusion that ball n ? 2 will be red. Williamson’s diagnosis is that I can not make this move because I do not know that the next ball will be red; I merely have a justified/well-evidenced true belief that it will. I agree that some well-evidenced but unknown beliefs can not serve as evidence for the further beliefs to which they bear a logical/probabilistic relationship. However, I do not think that none of them can. Contra (E = K), sometimes a known belief can not serve as evidence for a further belief even though it renders that belief more probable than it otherwise would be. Let us consider such a case. We can imagine a non-skeptical context in which someone asks what my evidence is that the sun will rise the day after tomorrow out of curiosity, and not out of a motivation to undermine my inductive knowledge that this will happen.8 Even though I know that the sun will rise tomorrow, if I know anything by induction, I can not use this bit of knowledge in response to a challenge to establish that the sun will rise the day after tomorrow. This is because the proposition that the sun will rise tomorrow, even though known, has an epistemic status for me that is similar enough to the status of the belief that the sun will rise the day after tomorrow, that it would border on begging the question to use this former belief in defense of the latter. Anyone who is curious about my evidence that the sun will rise two days from now will also be curious about what my evidence is that the sun will rise tomorrow. Furthermore, I have available to me even more well-credentialed beliefs concerning the sun’s rising in the past that support the truth of this further proposition. It may be conversationally 8 I have tried to set up the example in such a way as to avoid the dispute between contextualists and noncontextualists. 123 Author's personal copy 240 A. Rizzieri inappropriate to refer to lower quality evidence than one can refer to. This argument establishes that at best knowledge that p, combined with Williamson’s probability constraint, is necessary but not sufficient for that p to be a part of one’s evidence base for a particular further conclusion. I also want to show that meeting these two constraints is not necessary for one belief to serve as evidence for another. Let us compare Williamson’s ‘‘ball in urn’’ case with our original locked door case. (LD) was so well justified for me, that if it were true it would have been known. This is the key difference between this case and Williamson’s ball case. Williamson’s case involves a belief that is unknown because it lacked knowledge level evidence/justification; whereas my case involved a situation in which a belief is unknown because it is false. Hence, we can account for Williamson’s case without accepting his diagnosis. My view is that beliefs that are insufficiently justified to count as known can not serve as evidence for further beliefs, but beliefs that are sufficiently justified to count as known but are in fact false can serve as evidence for further beliefs. This diagnosis avoids the asymmetry problem, and accounts for both the ball in urn and locked door cases. 6 The problem of forgotten evidence I want to look at one more issue. That issue is the problem of forgotten evidence. The problem of forgotten evidence was originally forwarded by Alvin Goldman as an argument in favor of reliabilism concerning justification (Goldman 1999). We can imagine a case in which I am in my office all day and I can no longer call to mind what my original grounds were for affirming that I locked the door. I can only become aware of my conviction that the door is locked. For the sake of argument let us say that my original grounds were (1) and (2) above: (1) (2) I appear to have locked my door. In the past when I have appeared to have locked my door, I have actually locked my door. It is implausible that I have zero evidence for (O). This is because (O) is a justified inferred belief. Yet, I have forgotten what my original known grounds for (LD) were, and (LD) itself is false. Does it follow that I now have zero evidence for (O) assuming (E = K)? One possible way around this problem is to hold that (LD) is itself currently grounded by evidence that the agent has, but no longer has access to, and that this evidence also serves to ground (O). The problem with this suggestion is that it is not clear that one can have evidence that they do not have any access to whatsoever. What exactly does it mean to have evidence that one does not have any conscious access to? Perhaps this means that I have a ‘‘memory trace’’ of the original experience of attempting to lock my door that could be stimulated and made conscious under the right conditions. I just do not happen to be in those conditions. Let us stipulate that for whatever reason there is no memory trace of my having attempted to lock the door that is either encoded in my brain states or is a part of my subconscious. On 123 Author's personal copy Evidence does not equal knowledge 241 this assumption, the information that I locked my door earlier is completely lost to me. Under such conditions, it makes more sense to hold that (LD) initially received its positive epistemic status by being inferred from (1) and (2) and that this positive epistemic status is simply preserved by the subjects having a rational degree of confidence in the proposition that matches what the evidence initially supported even though the agent no longer has evidence for (LD). This interpretation is in line with Goldman’s strategy of arguing that in cases of forgotten evidence the agent’s belief is still justified because of the reliability of memory and not because the agent can recall (and therefore ‘‘possess’’ in the epistemically relevant sense) the initial evidence for holding her belief (Goldman 1999). I disagree with Goldman concerning the nature of justification in that I think that the initial accrual of epistemic justification for a belief necessarily involves an epistemically internalist constraint on justification. However, I do think that once justification is accrued it can be preserved by being sustained by reliable belief preserving mechanisms. Mapping out a theory of how justification, reliability, and evidence relate to one another is of course beyond the scope of this article. What is important for our purposes is that one can deny (with Goldman) that being justified that p entails currently possessing evidence in p’s favor. I would argue that this is the best interpretation of the (LD) case. I have no evidence at present for (LD), and yet (LD) retains the positive epistemic status that it received from (1) and (2). This leaves the question on the table, ‘‘is my occurrent belief (O) evidenced by (LD)?’’ Williamson’s view is that justification entails having evidence. Williamson could adopt the view that a belief can be justified if I once had evidence for it and I have maintained a rational degree of confidence in the belief that reflected the original evidence base. This would be a good explanation for why (LD) is justified, but I do not think it helps us explain why (O) is justified assuming (E = K). To return to a well-worn theme, one belief serving as an evidential ground for another is just as much a psychological relation as it is a quasi-logical/evidential relation. Concerning the psychological aspect of the grounding relation, our evidence is what guides the formation of our inferred beliefs in such a manner as to confer positive epistemic status on those beliefs. If (1) and (2) have been truly forgotten then (1) and (2) are not at present guiding the formation of (O); only my commitment to (LD) is. Hence, it would be very odd indeed to insist that (1) and (2) are my evidential grounds for (O) and that (LD) is not. I am only assuming here that if two beliefs have never been held at the same time, then one can not be considered evidence for the other. If it is the case that (O) must have evidence and that neither (1), (2) nor (LD) is evidence for (O), then what evidence could the defender of (E = K) appeal to? Perhaps my current evidence for (O) is the strong sense that I remember having had good grounds for it and an awareness that when I have had this strong sense I have usually been right.9 I do not think that this response solves the problem. For one thing, this response has the problem of the unsophisticated knower. A child might retain a confidence that the door is locked without having either a memory of 9 I thank E. J. Coffman for this suggestion. 123 Author's personal copy 242 A. Rizzieri locking the door or a belief that her own memory is reliable. Indeed a child might not have any beliefs about her memories at all. 7 Conclusion I have urged several arguments against (E = K) that highlight the connections that hold between justification, rationality, and the grounding of one’s beliefs in one’s evidence. I have drawn special attention to the psychological aspect of the evidential basing relation and to the awkwardness of denying that a well-evidenced belief can serve as evidence for a further belief on a broad variety of occasions. What I have not done is canvassed all of the possible responses to the arguments that a defender of (E = K) could appeal to, nor have I addressed the formidable positive case for (E = K) that Williamson provides (Williamson 2000, Ch. 10). Hence, my modest conclusion is that (E = K) faces substantial difficulties. References Bonjour, L. (1980). Externalist Theories of Empirical Knowledge. Midwest Studies in Philosophy, 5, 53– 74. Cohen, S. (1984). Justification and truth. Philosophical Studies, 46, 279–295. Cohen, S. (2002). Basic knowledge and the problem of easy knowledge. Philosophy and Phenomenological Research, 65, 309–328. Goldman, A. (1999). Internalism exposed. The Journal of Philosophy, 96(6), 271–293. White, R. (2006). Problems for dogmatism. Philosophical Studies, 131, 525–557. Williamson, T. (2000). Knowledge and its limits. New York, NY: Oxford University Press. 123