Internalism in the Epistemology of Testimony Redux

Internalism in the Epistemology of Testimony Redux B.J.C. Madison (Forthcoming in Erkenntnis) Draft Version – Do Not Cite Without Approval Abstract: In general, epistemic internalists hold that an individual’s justification for a belief is exhausted by her reflectively accessible reasons for thinking that the contents of her beliefs are true. Applying this to the epistemology of testimony, a hearer’s justification for beliefs acquired through testimony is exhausted by her reflectively accessible reasons to think that the contents of the speaker’s testimony is true. A consequence of internalism is that subjects that are alike with respect to their reflectively accessible reasons are alike with respect to what they have justification to believe. Testimony should be thought no different: hearers that are alike with respect to reflectively accessible reasons to think that a speaker’s testimony is true are alike with respect to their justification for beliefs based upon that testimony. But it has been recently argued that this view faces powerful counterexamples (Wright, 2015). So the central question is this: assuming that a hearer can acquire justification to believe a proposition through the testimony of a speaker, can epistemic internalism provide the resources to explain how such justification is possible? My aim in this paper is to address these counterexamples, and in so doing, defend epistemic internalist accounts of testimony. 1) Introduction: Epistemic Internalists, despite their differences, stress the epistemological importance of the first-person perspective. In its more traditional forms, internalism holds that the only factors that can contribute to a belief’s justification are those that a subject has conscious access to, and that factors external to such awareness can play no justificatory role1. Arguments for such positions vary, but variations of two widely discussed thought experiments are taken to motivate this kind of view: cases of 1 For a recent survey of ways of formulating the epistemic internalism / externalism distinction, see Madison 2010. 1 clairvoyance are taken to support an awareness condition on justification, and the New Evil Demon thought experiments are offered in support of the idea that factors external to the subject’s awareness, such as the reliability of the mechanism that give rise to the belief, are not necessary for the belief to be justified2. If internalists about a particular epistemic status hold that conscious awareness of one’s grounds are necessary for that status obtaining, and that factors external to such awareness are not relevant to determining that epistemic status, traditional externalism can be taken to be the denial of these two claims. A simple process reliabilism, for example, is a paradigm of such an externalist approach. According to this kind of view, a belief is justified iff it is formed via a reliable belief forming method. What reliablists take to determine if a belief is justified, for example, is that the belief results from a process which is actually reliable, not that the subject is aware that the belief is reliably produced (e.g. Goldman 1979)3. Complicating this traditional internalism / externalism dichotomy, however, is a fairly recent family of pluralist approaches in epistemology. According to some authors, such as Tyler Burge, Peter Graham, and Mikkel Gerken, what they called epistemic warrant is a genus with two species: an epistemically internalist one, which they call justification, and an epistemically externalist one, which they call entitlement4. It is stipulated that justification is an internalist notion that requires 2 For the clairvoyance cases, see Bonjour 1985. For the classic presentation of the New Evil Demon argument, see Cohen 1984. 3 The most extreme forms of externalism might hold that what a subject may or may not be aware of is justificationally immaterial. More modest forms of externalism can hold that while there is no positive requirement to be aware of any considerations that favor their belief, they can still hold a “no-defeat” clause. Such a clause means that if a subject is, or perhaps ought to be, aware of considerations that count against their belief, then this can undermine or rebut the justification her belief might have otherwise enjoyed. Thanks to a referee for Erkenntnis for impressing upon me the importance of qualifying the view in this way. 4 The use of this terminology begins with Burge 1993; 2003, and continues, among other places, with Graham 2011, and specifically in the epistemology of testimony with Gerken 2013. 2 cognitive access to reasons or grounds for one’s belief, whereas entitlement attaches to a belief in virtue of the belief being reliably formed in the subject’s normal environment. There are plenty of in-house disputes among pluralists, of course, including how the notion of an epistemically normal environment is to be understood. One might endorse epistemic internalism about different positive epistemic statuses (e.g. knowledge, Plantinga-warrant5, Foley-rationality6, entitlement – which is stipulated as being externalist, etc.), but my focus here shall be specifically on epistemic justification. Unlike Burge, Graham, and Gerken, I do not stipulate that justification is correctly accounted for in internalist terms. Rather than taking the issue to be settled by definitional fiat, I take it to be an open question what the nature of epistemic justification consists in. This is because I do not intend to use the notion of justification as a technical term of art. Instead, by “justification” I intend to refer to epistemic reasonableness, which I take to be more or less interchangeable with epistemic rationality. I take it that whether epistemic reasonableness or rationality is internalist or externalist is a substantive philosophical question. I intend to remain neutral on the question of pluralism about other epistemic statuses, a kind of pluralism understood as the view that there are a variety of positive epistemic statuses, and that some of them might be internalist, while others might be externalist. To make my commitments clear, if there is only one property of epistemic rationality, I contend that it is internalist, which makes me an internalist monist. If pluralism about epistemic rationality is true, however, I similarly contend that the justification 5 Alvin Plantinga stipulates that warrant is that which converts true belief into knowledge. See Plantinga 1993. 6 According to Richard Foley, very roughly, a belief is epistemically rational just in case a subject would hold the belief upon deep reflection. An epistemically rational belief is immune to self-criticism in the sense that it meets the subject’s most deeply held epistemic standards. 3 component is internalist, in agreement with authors like Burge, Graham, and Gerken. As will become clear, however, both options are compatible with my defense of internalism – indeed, internalism is easier to defend for the pluralist. So for dialectical purposes, my discussion shall presuppose internalist monism. I shall show that even this strong form of epistemic internalism remains unscathed by the externalist arguments I consider below. In sum, I shall use the term justification throughout to refer to epistemic rationality (noting that authors like Burge, Graham, and Gerken use the term ‘warrant’ to denote this property). Besides the question of whether pluralism or monism is true of different epistemic statuses, one might endorse internalism with regard to certain classes of beliefs, but not others. For example, one might think that the correct epistemology of perceptual belief is internalist, but accept a form of externalism in one’s account of self-knowledge, or in one’s moral epistemology. My concern here is the epistemology of testimony. With this background in place, we can now turn to the central question of this paper: assuming that a hearer can acquire justification to believe a proposition through the testimony of a speaker, can epistemic internalism provide the resources to explain how such justification is possible? In a recent paper, Stephen Wright argues against the following internalist thesis: (TI) A listener’s justification for her belief that P is a matter of her reflectively accessible reasons for thinking that the speaker’s testimony that P is true. (Wright, 2015, p. 2) 4 This thesis is worthy of careful consideration, since it is general enough to capture many different forms of internalism, of both monist and pluralist varieties, and different approaches to the epistemology of testimony, but specific enough to capture the essence of traditional epistemic internalism, which turns on conscious awareness and reflective accessibility. Wright attacks (TI) by arguing that one of its consequences is false. That important internalist consequence is thus: (TI*) Listeners that are alike with respect to reflectively accessible reasons for thinking that a speaker’s testimony is true are alike with respect to justification for their beliefs based on testimony. (Wright, 2015, p. 3) Wright argues that internalist theories of justification from testimony are incomplete on the grounds that, at least sometimes facts about the epistemic competence of the speaker make a difference to the epistemic justification of the hearer – even if these facts are not ones that the hearer has reflective access to. If this is true, it will create serious difficulties for internalist theories of the justification of beliefs acquired through testimony, and as such, Wright’s argument ought to be taken very seriously indeed. In this paper I shall argue in defense of epistemic internalist accounts of justification of testimony based belief, and specifically, in defense of the two related internalist claims introduced above, TI and TI*. My aim is to neutralize Wright’s intuitive case against internalist theories of justification from testimony, and to explain the intuitions that seem to count against internalism. 5 My plan is as follows: in section 2 I shall present Wright’s argument against internalist accounts of justification through testimony. With the argument in place, in section 3a I shall then cast doubt on the reasons given to think that the epistemic difference between the pairs of cases given is a difference in justification. But I shall concede that Wright has correctly identified important differences between the subjects in the cases he presents. In section 3b, therefore, I shall explain what this epistemic difference is, if not a difference of justification. As I shall show, this difference, while significant, is consistent both with epistemic internalism in general, and with the internalist approach to testimony in particular. Epistemic internalism, therefore, remains very much a live option in the epistemology of testimony7. 2) The Argument: The argument for the conclusion that epistemic internalism cannot give a complete account of the epistemology of testimony begins from the following pair of cases8: DOCTOR: One day Eric discovers a rash on his arm and goes to visit his doctor Ernie. Eric has various good reasons for thinking Ernie a competent doctor and a sincere testifier. These reasons consist in both inductive evidence of Ernie previously having correctly diagnosed Eric’s dermatological conditions in the past as well as evidence of the general accuracy of what doctors say. Ernie correctly tells Eric that the rash on his arm means he has condition A. Ernie correctly tells Eric 7 The locus classicus of a kind of epistemic internalism in the epistemology of testimony is of course Fricker 1994. 8 The pair of cases Wright introduces is very similar to a pair of cases introduced in Schmitt 1999 p.370. Interestingly, Schmitt thinks these examples make a case against epistemic externalism, not internalism, as Wright presents it. Schmitt’s verdicts about the cases align with my own, namely that there is a sameness of justification between the pairs of subjects. While Schmitt originally presents the cases, he says very little there in defense of his verdicts about them. My contribution in this paper, therefore, can in part be seen as an extended defense of Schmitt’s judgments about these cases, which in turn contributes to a defense of internalism in the epistemology of testimony, against Wright’s recent attack. 6 that he has condition A because, whilst the rash on Eric’s arm looks similar to a rash associated with condition B, Ernie can tell that the rash on Eric’s arm is the one associated with A rather than the one associated with condition B. (Wright, 2015, p. 4) DOCTOR*: One day Eric* discovers a rash on his arm and goes to visit his doctor Ernie*. Eric* has just as many and varied good reasons for thinking that Ernie* is a competent doctor and a sincere testifier as Eric (in DOCTOR) has for thinking Ernie is a competent doctor and a sincere testifier. Ernie* tells Eric* that the rash on his arm means that he has condition A. Like Ernie’s diagnosis of Eric, Ernie*’s diagnosis of Eric* is correct. Unlike Ernie, however, Ernie* is unable to distinguish between the rash associated with condition A and the rash associated with condition B. This is because Ernie* has a more limited experience and range of expertise than Ernie and as a result is unaware of condition B. (Ibid.) With these cases in place, Wright presents the following argument: 1) Eric and Eric* are not alike with respect to justification for their beliefs that they have condition A. 2) Eric and Eric* are alike with respect to reflectively accessible reasons for thinking that they have condition A. Of course, since (I*) [“Individuals that are alike with respect to reflectively accessible reasons are alike with respect to justification.”] holds that this isn’t possible, it follow from (1) and (2) that: 3) (TI*) is false. And, as observed above, since (Ti*) is a consequence of (TI), it follows from (3) that: 4) (TI) is false. (Wright, 2015, p. 4) 7 My aim is to defend epistemic internalist accounts of testimony, and specifically, (TI): a listener’s justification for her belief that P is a matter of her reflectively accessible reasons for thinking that speaker’s testimony that P is true. To do so, I shall defend (TI*): Listeners that are alike with respect to reflectively accessible reasons for thinking that a speaker’s testimony is true are alike with respect to justification for their beliefs based on testimony. But given Wright’s valid argument that aims to show that (TI*) is false, however, I shall be questioning the argument’s soundness by resisting its first premise. That is, in defense of the internalist position, I shall first argue that we have been given no good reason to hold that Eric and Eric* are not alike with respect to justification for their beliefs that they have condition A. I intend to show that Wright’s claim that they are not alike with respect to justification for their belief is unmotivated. I shall then go on to offer a diagnosis of what the epistemic difference is in these cases, if not a difference in justification as Wright maintains. 3) A Reply: 3a) Why Think that the Epistemic Difference Between Eric and Eric* is a Difference in Justification? Besides appeal to intuition, what is Wright’s main argument that there is a difference of justification between Eric and Eric*? The main answer comes from subjects looking backwards, so to speak, and asking: if each were made aware of the full facts of their situation, when reflecting at a latter time on their earlier beliefs, does such reflection change the epistemic status of their belief, or does it merely reveal what epistemic status the belief had all along? Wright argues that consideration of 8 cases reveals that such awareness merely reveals justificatory status, it does not change it. My proposal is that the argument that there is a difference in justification between Eric and Eric* relies on a highly unreliable method for determining whether justification is present, and to what degree. Therefore, we have been given no good reason to think that Eric and Eric* differ in justification. We will also see that we have good reason to think that they share sameness of justification. Wright asks us to suppose that the full facts of the case are made available to both Eric and Eric*: both learn that Ernie can distinguish between conditions A and B, but Ernie* cannot. At this point, all should agree that when such facts are revealed, Eric’s belief that he has condition A is more justified than Eric*’s corresponding belief. (Wright, 2015, p.6) But as Wright notes, this is no problem for internalism, since at this later time, there are now reflectively accessible differences between Eric and Eric* (Ibid.); in particular, both are now aware of reasons to think that Ernie* is unreliable in his ability to correctly diagnose condition A. But this should not be surprising since it is clear that at the moment of revelation, Eric* acquires a defeater for his belief that he formerly lacked. According to Wright, what is meant to be a problem for internalism, however, is this: “What internalists aren’t able to explain, however, is the idea that, when the facts of the case are made apparent to Eric and Eric*, they realize that they never were alike with respect to justification in the first place.” (Ibid.) Wright takes this to be established by Alvin Goldman’s fake barn case (Goldman 1992). This now-familiar case involves a subject, Henry, who is driving through the countryside and looks into a field. In the field there appears to be many barns. Unbeknownst to Henry, however, every barn in the field is a subjectively indistinguishable fake, except for one. Luckily for Henry, however, he looks at that very barn, and believes truly, that it 9 is a barn. Here the subject’s belief is true, and it is justified. Despite this, most judge that this is insufficient for knowledge. Reflecting on this case, Wright says, We take it that Henry doesn’t know that he’s looking at a barn. In the same way that we don’t want to say that Henry’s status with respect to knowledge changes when the full facts of the situation come into view, I don’t think that we should say that Eric and Eric*’s statuses with respect to justification change when the facts of their situation come into view. Justification is that which puts someone in a position to know things. [Emphasis added] And the idea is that, upon finding out the facts of his situation, Henry should think that he was never in a position to know that there was a barn in the first place, not that he used to be in a position to know but now isn’t. (Wright, 2015, p. 7) All should agree that Henry lacks knowledge, and this fact would be revealed to him if he learned the full facts of the situation. But there is no reason to suppose that the reason he lacked knowledge is that something was wrong with his justification. Wright asserts that justification is that which puts one in a position to know things. This is ambiguous, and on disambiguation it is false, and on the other it might be true, but it creates no problem for internalism. One way of understanding the idea that justification is that which puts one in a position to know is that justification, together with truth and belief, is sufficient for knowledge. But all should reject this understanding in light of the Gettier problem, which forcefully shows the inadequacy of the tripartite analysis of knowledge. No one should think that justification is what puts someone in a position to know, in this sense. Another way of understanding the idea that justification is that which puts one in a position to know is that justification, together with truth and belief, is necessary for knowledge. Two options present themselves to internalists here: they can accept the 10 necessity of justification for knowledge, or they can deny it. If the internalist accepts that justification is merely necessary for knowledge, they can allow Henry lacks knowledge for some other reason (for reasons I shall explain below in section 3b), but that his justification remains unaffected. If, on the other hand, internalists deny that justification is necessary for knowledge, then they can also allow that Henry’s justification remains unaffected in fake barn country. Wright seems to acknowledge this option, though claims that it “amounts to a serious concession from the internalist, however, since it devalues justification, as internalists conceive of it, detaching it from knowledge.” (Wright, 2015, p. 5, footnote 10) This is no concession for internalism, properly understood. The charge that this amounts to a serious concession relies on two unargued for assumptions which are in need of defense: the first is that the value of justification is exhausted by its instrumental value, and that it is instrumental in gaining knowledge. The internalist, however, can hold that justification is valuable for its own sake9. The second assumption behind this charge is that this approach “detaches” justification from knowledge. This does not follow, however. Consistent with justification being 9 See Madison 2015 for an argument that the value of justification is not exhausted by its instrumental value, and that there is more of epistemic value than true belief or knowledge. The central argument in that paper comes from reflecting on a version of the New Evil Demon problem. By reflecting on a value specific version of New Evil Demon cases, I argue that we can find justified false beliefs that are still of epistemic worth. I try to show that justified false belief, which is no means to truth, can nevertheless still be of epistemic value. This in turn makes a good prima facie case that justification is valuable for its own sake. If this is right, we have also found reason to think that truth value monism is false: assuming that true belief does have value, there is more of final epistemic value than mere true belief. But as I hope to have made clear in the body of the text, the cogency of this particular line of argument is not essential to my overall response to Wright. It is only one move out of at least three possible responses the internalist can make at this juncture. As I have noted, Wright argues from the premises that justification is that which puts someone in a position to know things, as well as the fact that Henry lacks knowledge in fake barn country, to the conclusion that the facts of Henry’s justification were revealed to him, not changed. But this simply does not follow unless justification, together with truth and belief, are sufficient for knowledge – but this is a claim that no one should accept in the post-Gettier era. As noted, there is also the other interpretation of Wright’s claim, namely that justification is that which puts someone in a position to know, taken to mean that justification is merely necessary for knowledge. In that case, it is open to internalists to hold that knowledge is lacking for some other reason in the example of Henry, other than because of defective justification (a point I go on to develop at length below in section 3b). 11 valuable for its own sake, and being present and unaffected in the fake barn case, is that justification is connected with knowledge, by being necessary for it. All that is required for this key point is the non-controversial assumption that justification is not sufficient for knowledge, something that all parties to the dispute should now accept in the post-Gettier era. So we have been given no reason to suppose that justificatory status is merely revealed in the fake barn case, and it does not follow from the fact that the subject lacks knowledge in that case. This too is consistent with the fact that this lack of knowledge would be revealed if all the facts of the situation were made known. That is, sometimes epistemic statuses are genuinely revealed, but sometimes they are changed when all the facts are revealed. For example, consider a case of putative inductive knowledge: Suppose Bill has incredibly good reasons for thinking that there will be a sand storm tomorrow. He has observed all relevant weather forecasts, consulted historical records about the weather at this time of year, and so on. In virtue of this, Bill takes himself to know that there will be a sandstorm tomorrow. But now suppose that there is no storm. When Bill finds out that he is mistaken, there are two things that we could think. The first is that he used to know, but now he does not. The second is that he never knew in the first place, he just thought that he did10. We should agree that the above case is one where, when all the facts are made known, the subject comes to realize that they never had knowledge, but only thought that they did. This is a case of an epistemic status being revealed, not changed, upon becoming aware of all the facts. But this is a very different case from the ones at issue that involve testimonial justification. The differences between knowledge and justification impact upon how to understand these cases. 10 This example is based on one that Wright gave in an early draft of Wright 2015. 12 One relevant difference is that knowledge entails truth, whereas justification does not11. So if it false that there will be a sandstorm tomorrow, then no subject, not even an omniscient god, can know that there will be (though of course subjects can be justified in believing it, or justified in believing that they know it, etc.). Both internalists and externalists can agree that in this case, the status of knowledge was revealed, it did not change from a state of knowledge to a state of ignorance. But what relevance does this have specifically for justification? If a subject has incredibly strong reasons to think that there will be a sand storm tomorrow, then in virtue of this, their belief that there will be a storm is justified. Once they find out that they were mistaken, however, they are no longer justified in believing that there will be a storm. But it does not follow from this that, before acquiring the relevant defeater, they lacked justification for believing as they did. This seems straightforwardly like a case where a subject has strong (inductive) justification, acquires defeating evidence, and then loses the justification that he previously had. Wright’s proposed test boils down to this: “[…] if Eric and Eric* find out about the difference in competence between Ernie and Ernie*, they’ll think that they aren’t alike with respect to justification. And because they are just finding out facts about the epistemic situation, they won’t think that their beliefs have changed with respect to justification. Rather, they will take themselves to be finding out that they were different all along.” (Ibid. 2015, pp. 9-10) While sometimes epistemic statuses are revealed, sometimes they are changed, and this test is an unreliable method of distinguishing between them. If this test were 11 Of course, some deny the undeniable. For some examples of those who hold that justification is factive, see Sutton 2005 and 2007, Littlejohn 2012. 13 consistently applied, it would predict it to be impossible for there to be justification present in two key kinds of cases, cases where justification is indeed possible, namely: Gettier cases and cases of justified false belief. First, consider again Gettier cases. Compare Roderick who seems to see a sheep in field, his perception is veridical, and on that basis justifiably believes that there is a sheep in the field. Compare that with Roderick* who justifiably believes that there is a sheep in the field because he seems to see one, though what he actually sees is a cleverly disguised dog – his belief that there is a sheep in the field is true, of course, since there is one sheepishly hiding in the back of the field out of sight. Here we ought to say that Roderick has knowledge, but Roderick* does not. But now what would happen if Roderick and Roderick* find out the differences in the situations, including the facts that Roderick is enjoying a veridical perception, whereas Roderick* had been looking at a cleverly disguised dog, although there had also been a sheep in the field? The two will now think that they are not alike with respect to justification. And while they are just finding out the facts of the epistemic situation, they ought to think that in one sense the epistemic status of their beliefs have changed, whereas in another sense they have not: while the two find out that they differed in knowledge all along, they shared sameness of justification right up until they found out about the differences in their environments12. If we are required to say that Roderick* is now finding out that he lacked justification before his discovery that he had been looking at a dog disguised as a sheep, then we are in effect saying that Gettier cases are impossible: a consequence of Wright’s account is that it is impossible to have justified true belief that does not 12 Mikkel Gerken makes a similar point against Sandy Goldberg in his review of Goldberg 2010. See especially sections 3 and 4 of Gerken 2012. 14 amount to knowledge, since on this account, justification is impossible in such cases. This is because if the subject were to find out all the relevant facts about the situation, including, therefore, that he is in an abnormal Gettier situation, he would not deem himself to be justified in such cases. But justification, by definition, is present in Gettier cases: a Gettier case just is an instance of justified true belief that does not amount to knowledge13. The general point that Wright’s method of determining whether or not justification had been present (and to what degree) before the facts of the epistemic situation were revealed to pairs of subjects is unreliable can be seen by considering not only the possibility and nature of Gettier cases, but also the fact that subject’s can be justified in believing falsehoods. Suppose that at time t1 Al has lots of evidence for the proposition that P. Despite this overwhelming good evidence, however, P is false. Now consider time t2: Al is made aware of all of the relevant facts, including therefore, that P is false. Does that entail that his belief that P was never justified? On pain of saying that false propositions can never be justifiably believed, we ought to hold Al had believed with justification, but now that he has acquired defeating counterevidence, he no longer has justification to believe that P. To be sure, upon 13 A referee for Erkenntnis raised the following question: is the situation being revealed in a Gettier case analogous to the case that Wright offers? After all, in Wright’s case, the individuals discover the relevant facts about their informants—that one is more competent than the other. In the cases I present, the individuals discover facts about their sources plus the fact that their beliefs are true. Is this not a significant epistemic difference? While this is a difference, it is an innocuous one. Wright’s proposed test amounts to seeing if subjects would revise judgments about the previous presence or absence of epistemic justification upon learning all the facts of the epistemic situation. Presumably, it is only the revelation of relevant epistemic facts that should make a difference to their judgments. But the truth or falsehood of a token belief is not itself relevant to whether or not it is justified. Unlike knowledge, which entails truth, justification does not. This is why becoming aware that a belief is false can reveal that it was not known, but becoming aware of its truth-value does not itself reveal whether it was justifiably held. If it did, then justified false beliefs would be impossible, an unwelcome consequence of a strict application of Wright’s test -- a result I discuss below. 15 acquiring new information, the epistemic status of his belief changed, it was not merely revealed. If a subject were to find out that they were Gettiered, or that what they believe is false, they should always revise in the following sense: they ought to now judge that holding the relevant belief would now be unjustified. But they can consistently hold that, until that point, their beliefs had been justified. To hold that making subjects aware of all the epistemic facts merely reveals what justificatory status a belief has, instead of changing it, as Wright does, is in effect to deny that subjects can be justified in believing falsehoods, as well as to deny that justification can be present in Gettier cases – two very unwelcome consequences indeed. So while the backwards looking test might fare better in establishing whether knowledge was present, it is an unreliable way of trying to determine whether a subject had believed with justification. 3b) Explaining the Difference Between Eric and Eric*: an Epistemic Difference, But Not A Difference in Justification: So then what are we to make of Wright’s central cases of Eric and Eric*? He is right that the cases clearly do show some kind of epistemic difference between the subjects. But if the epistemic difference is not a difference in justification, as internalists shall maintain, then how is that difference best explained? Reflecting on the cases of DOCTOR and DOCTOR*, Wright notes that “I suggest that we should think that a listener confronted by testimony from a more competent speaker is better off in terms of justification in virtue of the speaker’s 16 enhanced competence than a listener confronted by a less competent speaker[…]” (emphasis added) (Wright, 2015, p. 3) It does seem correct that in Wright’s cases one subject is epistemically better off than the other. Internalists can happily agree with this, however, since there are many ways that one subject can be epistemically better off than another. In the cases in question, Eric is indeed epistemically much better off than Eric*: as presented, it seems that Eric knows that he has condition A, whereas Eric* does not. While they differ in knowledge, however, there is no reason to suppose that they differ in justification – which is consistent with the internalist position14. The reason that Eric has knowledge, whereas Eric* does not, is explained by the fact that the testifying doctors differ in knowledge, too: while Ernie knows that Eric has condition A, Ernie* does not know that Eric* has condition A. While both diagnoses are correct, and both patients believe truly on the basis of their doctor’s testimony, Ernie*, and therefore Eric*, arrived at a true belief in a way that is too lucky to count as knowledge. 14 There are other ways of explaining the epistemic difference between Eric and Eric* consistent both with internalism and with the explanation I give here in terms of a difference in knowledge. For example, pluralists such as Burge, Graham, and Gerken can allow a sameness of justification, but might allow a difference in entitlement. Depending on how the details of the case are spelled out, pluralists can allow either a sameness of entitlement, or a difference. The question of sameness of externalist entitlement in Wright’s cases will depend on the relative frequency of the occurrences of rashes of type A and B, and so whether the doctors are locally or globally reliable / unreliable in making the correct diagnosis. For more on sameness or difference in entitlement in twin-cases, see Gerken 2012 and 2013, especially section 4. See also Malmgren’s 2011 discussion of Goldberg 2010. A referee for Erkenntnis rightly pointed out that even externalist monists of certain sorts can reply to Wright’s cases by holding that there is a sameness of justification, while allowing a difference in knowledge. In considering a similar case and responding on behalf of Alvin Goldman, Mikkel Gerken writes, “[…] the Orthodox Reliabilist will claim that the processes involved in accepting a doctor’s testimony about a medical matter are globally reliable. The fact that the particular doctor in the not-so-good case [one where a doctor merely appears to be performing a test, but is irresponsibly guessing and saying that the result is certain] is unreliable does not change the global reliability of the relevant process.” (Gerken, 2012, p. 87) An orthodox reliabilist like Alvin Goldman can hold that a belief is justified if it is the product of a belief-forming process that is globally reliable, but that in addition, knowledge requires that the process is also locally reliable. See Goldman 1986 for a sophisticated development of a version of orthodox process reliabilism. 17 In the wake of the Gettier problem, it is widely acknowledged that knowledge is incompatible with too much of certain kinds of luck. It has been proposed that the presence of so-called veritic luck is sufficient to prevent an otherwise justified true belief from amounting to knowledge (e.g. Pritchard 2005)15. While specifying the exact anti-luck condition on knowledge has proven difficult, two prominent accounts bring out the central idea that knowledge must exhibit a modally robust connection to the truth.16 The Sensitivity Principle holds that, roughly, if a believed true proposition had been false, the subject would not have believed it17. A sensitive belief “tracks the truth”, as it were (e.g. Nozick 1981). The Safety Principle, by contrast, maintains the following as necessary for knowledge: not only does the subject hold a true belief, but in most (if not all) of the nearby worlds where the subject forms her belief in the same way as in the actual world, the belief remains true18. We can see that Eric*’s belief that he has condition A is neither Sensitive nor Safe. First, his belief is insensitive: if it were false that he had condition A, he would still believe that he did, on the basis of Ernie*’s testimony. Second, his belief is also unsafe: while Eric*’s belief is actually true, in nearby worlds where he believes that he has condition A on the basis of Ernie*’s testimony, those beliefs are false. This lack of sensitivity and safety is passed on to Eric* from Ernie* whose belief that Eric* has condition A is likewise neither sensitive nor safe. Given that Ernie* is unable to 15 In fact, that knowledge is incompatible with (veritic) luck is a foundational principle of the anti-luck epistemology movement. The movement, championed by Duncan Pritchard, among others, aims to articulate an epistemology which rules out too much of certain kinds of luck, thus explaining the possibility and nature of knowledge as non-lucky true belief (which may well meet further conditions). 16 Endorsing merely necessary conditions on knowledge is consistent with the Knowledge-First program’s contention that knowledge cannot be reductively analyzed in terms of conditions that are individually necessary and jointly sufficient. The locus classicus for the Knowledge-First program is of course Williamson 2000. 17 For classic expressions of sensitivity-based epistemologies, see Dretske 1971 and Nozick 1981. For a recent collection of papers discussing the merits of Sensitivity, see Becker and Black 2012. 18 See Sosa 1999; Pritchard 2005; Pritchard 2007 for Safety-based accounts. 18 distinguish condition A from condition B, given that they present with very similar rashes, and given that Ernie* is entirely unaware of condition B, Ernie*s correct diagnosis is quite lucky indeed. Had Ernie* been presented with the rash symptomatic of condition B, he would have falsely believed that the patient was suffering from condition A. So while in the case presented Ernie* has a true belief, that true belief is too lucky to count as knowledge. Ernie, by contrast, knows what condition Eric suffers from, given his ability to discriminate condition A from condition B on the basis of their very similar symptoms. Given the above discussion, we can see that there is a perfectly good sense in which Eric is epistemically better off than Eric*: he knows what condition he has on the basis of his doctor’s testimony, whereas Eric* does not. Despite these differences in knowledge, however, we should feel no pressure to say that the subjects differ in what they have justification to believe. In general, modal differences, such as differences in safety and sensitivity, are taken to affect knowledge, but not justification. We can be reminded of this crucial point by reflecting on some classic Gettier cases, cases whose mere existence depends on this fact. Take Chisholm’s classic case of the sheep in the field, alluded to above (Chisholm, 1977). In that case we are asked to suppose the following: a subject looks into a field and sees a cleverly disguised dog, which appears to be a sheep. On the basis of what he sees, he forms the belief that there is a sheep in the field. This justified belief would have been false, but for the following twist: there is a sheep in the field, hidden behind a tree. So the subject holds a justified belief, which is also true, but intuitively it does not amount to knowledge. While explanations of why knowledge is lacking vary, one popular diagnosis is that, given how the belief was formed, it is just too lucky that it is true (for instance, the belief is neither safe nor 19 sensitive). Here is the important point for our purposes, however: while the oddness of the environment and the presence of too much luck robs the subject of knowledge, his justification remains intact. In fact, for Gettier cases to work, we have to assume that the Gettiered subject is as justified as the paradigm knower. Otherwise, one could respond to Gettier cases by claiming that the subject violates the justification condition19. If the orthodox verdict is to be maintained that this is a Gettier case, it is required that it is a case of justified true belief, without being a case of knowledge. Modal factors, and factors outside the agent’s ken, can be epistemically dangerous: but their danger consists of defeating knowledge, not justification. To reinforce the point that modal differences, such as differences in safety and sensitivity, are taken to affect knowledge, but not justification, let us briefly return to one final classic Gettier case: the case of Goldman’s fake barn that Wright relies on in defending his intuition that there is a difference in justification between Eric and Eric*. Recall that despite Henry’s belief being true and justified, this is insufficient for knowledge. Again, while explanations of this vary, one common diagnosis is that, given how the belief was formed, it is just too lucky that it is true (here again, the belief is neither safe nor sensitive). The important point for our purposes is thus: while the oddness of the environment and the presence of too much luck robs the subject of knowledge, his justification remains in tact. Indeed, if the orthodox verdict is to be maintained that the fake barn case is indeed a Gettier case, it is required that it is a case of justified true belief, without being a case of knowledge. Reflecting on Gettier cases, we can see that in general modal differences and other factors outside the agent’s conscious awareness affect the presence of knowledge, not justification. Given this general truth, there is no specific reason to 19 Thanks to Mikkel Gerken for impressing this point upon me. 20 suppose anything different here in the case of testimonial justification. While there are principled reasons to judge that Eric and Eric* are not epistemically identical, Eric and Eric* are alike with respect to justification for their beliefs that they have condition A. 4) Conclusion: In general, epistemic internalists about the epistemology of testimony are committed to the following: hearers that are alike with respect to reflectively accessible reasons to think that a speaker’s testimony is true are alike with respect to their justification for beliefs based upon that testimony. But as we have seen, this view is thought to face important counterexamples. I have addressed the counterexamples to this view. But I also concede that the examples show something: we should accept the externalist’s intuition that Eric and Eric*, and pairs of cases like them, are not epistemically identical. Plausibly, the difference between them is a difference in knowledge. But there is no pressure to suppose that epistemic difference between them is a difference in justification. This difference, while significant, is consistent both with epistemic internalism in general, and with the internalist approach to testimony in particular.20 20 Thanks to an audience at The Petroleum Institute, Abu Dhabi. Thanks also to anonymous referees for Erkenntnis, Gwen Bradford, Mikkel Gerken, Dick Foley, and especially Rhiannon James for helpful discussion and written comments on earlier drafts of this paper. 21 Sources Cited: Becker, Kelly and Black, Tim. Ed. (2012) The Sensitivity Principle in Epistemology. Cambridge: Cambridge University Press. Bonjour, Laurence. (1985) The Structure of Empirical Knowledge. Cambridge, Mass.: Harvard University Press. Burge, Tyler. (1993) “Content Preservation”. Philosophical Review 103, pp. 457488. Burge, Tyler. (2003) “Perceptual Entitlement”. Philosophy and Phenomenological Research 67, pp. 503-548. Chisholm, Roderick. 1977. Theory of Knowledge, 2nd edn. Englewood Cliffs, NJ: Prentice-Hall. Cohen, Stewart. (1984) “Justification and Truth” in Philosophical Studies 46, pp. 279-295. Dretske, Fred. (1971) “Conclusive Reasons” in Australasian Journal of Philosophy 49/1: 1-22. Fricker, Elizabeth. (1994) “Against Gullibility” in A. Chakrabarti & B. K. Matilal (eds.) Knowing from Words. Kluwer. Foley, Richard. (1987) The Theory of Epistemic Rationality. Cambridge, Mass.: Harvard University Press. Gerken, Mikkel. (2012) “Critical Study of Goldberg’s Relying on Others”. Episteme 9, pp. 81-88. Gerken, Mikkel. (2013) “Internalism and Externalism in the Epistemology of Testimony”. Philosophy and Phenomenological Research LXXXVII, pp. 532-557. Goldberg, Sanford. (2010) Relying on Others. Oxford: Oxford University Press. Goldman, Alvin. (1992) “Discrimination and Perceptual Knowledge” reprinted in Liaisons. Cambridge, Mass: The MIT Press. Goldman, Alvin. (1979) “What is Justified Belief?”. Justification and Knowledge. Ed. George S. Pappas. Dordrecht: D. Reidel. Goldman, Alvin. (1986) Epistemology and Cognition. Cambridge, MA: Harvard University Press. Graham, Peter. (2012) “Epistemic Entitlement”. Nous 46:3, pp. 449-482. Littlejohn, Clayton. (2012) Justification and the Truth-Connection. (Cambridge: Cambridge University Press). Madison, B.J.C. (2010) “Epistemic Internalism” in Philosophy Compass 5/10, 840853. 22 Madison, B.J.C. (2015) “Epistemic Value and the New Evil Demon” in Pacific Philosophical Quarterly. doi: 10.1111/papq.12054 Malmgren, Anna-Sara. (2011) “Review of Relying on Others by Sanford Goldberg”. MIND 120, pp. 1251-1258. Nozick, Robert. (1981) Philosophical Explanations. Oxford: Clarendon Press. Plantinga, Alvin. (1993) Warrant and Proper Function. Oxford: Oxford University Press. Pritchard, Duncan. (2005) Epistemic Luck. Oxford: Clarendon Press. Pritchard, Duncan 2007. “Anti-Luck Epistemology” in Synthese 158/3: 277-297. Pryor, James. (2001) “Highlights of Recent Epistemology” in British Journal for the Philosophy of Science 52, pp. 95-124. Schmitt, Frederick. (1999) “Social Epistemology” in J. Greco and E. Sosa (eds.), The Blackwell Guide to Epistemology, pp. 354-382. Oxford: Blackwell. Sosa, Ernest. (1999) “How to Defeat Opposition to Moore” in Philosophical Perspectives 13: 141-154. Sutton, Jonathan. (2005) “Stick To What You Know” in Nous 39, pp. 359-396. Sutton, Jonathan. (2007) Without Justification. Cambridge, Mass.: The MIT Press. Williamson, Timothy. (2000) Knowledge and Its Limits. Oxford: Oxford University Press. Wright, Stephen. (2015) “Internalism in the Epistemology of Testimony”. Erkenntnis. DOI: 10.1007/s10670-015-9729-y 23