A Justificationist View of Disagreement's Epistemic Significance
Abstract and Keywords
This chapter developes a justificationist account of the significance of disagreement between epistemic peers. Whereas current views maintain that disagreement, by itself, either simply does or does not possess epistemic power, this chapter's account holds that its epistemic power, or lack thereof, is explainable in terms of the degree of justified confidence with which the belief in question is held. In this sense, the chapter rejects nonconformism—the absence of doxastic revision in the face of peer disagreement is never justified merely by virtue of the fact that the beliefs in question are either mine or are the product of correct reasoning—and conformism—substantial doxastic revision in the face of peer disagreement is never justified merely by virtue of equal weight being given to my own beliefs and to those held by my epistemic peers. Despite this, however, one advantage of my justificationist account is that it is able to explain why nonconformism provides the intuitively correct result in some cases, while conformism gives the intuitively correct result in other cases. A further advantage is that this chapter's justificationist account is generalizable in a way that neither of these rival views is.
Disagreement with others is a familiar part of our lives. Most of us find ourselves faced with friends who have radically opposing views on the war in Iraq, relatives who explicitly reject our beliefs about religion, and colleagues who dispute our conception of the nature of free will. These are just a few of the more common topics of disagreement, but there are countless others, ranging from the mundane—‘That is a cat by the tree, not a small dog!’—to the extraordinary—‘That I conquered my cancer was not purely a result of my chemotherapy, but was also God answering my prayers.’
In some instances, disagreement can be explained by one party to the dispute being privy to more evidence than the other. You and I may have opposing beliefs about whether Jones committed the murder because only you saw the relevant DNA results linking him to the crime. Other cases of disagreement can be accounted for in terms of various kinds of cognitive asymmetries. You and I may disagree about whether the bird in the backyard tree is a starling because only I am using eyeglasses with an out‐of‐date prescription. Still other instances of disagreement can be explained by one member of the dispute having reason to believe that the other is epistemically inferior in some respect. You and I may have differing attitudes about whether my daughter is the best ballet dancer in her class because you believe that I am biased where my daughter's talents are concerned.1 But adjusting our doxastic states in all of these sorts of cases does not reveal anything significant about disagreement itself, since the fact that you and I disagree drops out of the explanation of this adjustment; it is, for instance, the difference in our familiarity with the relevant evidence or the asymmetry in our cognitive capacities that does the explanatory work. In order to truly assess the significance of disagreement itself, there cannot be any relevant epistemic asymmetries between the parties to the dispute to shoulder the explanatory (p.299) burden;2 in other words, such parties should be epistemic peers.3 The question at issue, then, is this: what is the significance of disagreement between those who are epistemic peers? In particular, what is the rational response to disagreement in situations where there are no relevant epistemic asymmetries between the members involved in the dispute?
There are two answers to this question found in the recent literature. On the one hand, there are those who hold that disagreement itself can be wholly without epistemic significance; thus, one can continue to rationally believe that p despite the fact that one's epistemic peer explicitly believes that not‐p, even when one does not have a reason independent of the disagreement in question to prefer one's own belief. I shall call those who hold this view nonconformists. According to nonconformists, there can be reasonable disagreement among epistemic peers. So, for instance, Gideon Rosen writes:
Similarly, Thomas Kelly claims that:
It should be obvious that reasonable people can disagree, even when confronted with a single body of evidence. When a jury or a court is divided in a difficult case, the mere fact of disagreement does not mean that someone is being unreasonable. (Rosen 2001: 71)
According to nonconformists, then, the mere fact that there is disagreement with one's epistemic peer does not mandate any sort of doxastic revision from either party to the dispute.
The mere fact that others whom I acknowledge to be my equal with respect to intelligence, thoughtfulness, and acquaintance with the relevant data disagree with me about some issue does not undermine the rationality of my maintaining my own view.
Now, there are two central explanations of the nonconformist response to peer disagreement.5 First, there is what we may call the egocentric view. On this view, I am justified in giving my belief extra weight6 in the face of peer disagreement because the belief in question is mine. So, for instance, Ralph Wedgwood claims:
(p.300) According to the egocentric version of nonconformism, then, the epistemic symmetry involved in peer disagreement can be broken by virtue of the extra weight afforded to my belief in virtue of the belief's being mine.8
Perhaps, quite generally, it is rational for one to place greater trust in one's own intuitions, simply because these intuitions are one's own, than in the intuitions of other people. In other words, perhaps it is rational for each of us to have an egocentric epistemic bias in favour of our own intuitions. (Wedgwood 2007: 261)7
The second explanation of the nonconformist response to peer disagreement is what we may call the correct reasoning view. On this view, I am justified in giving my belief extra weight in the face of peer disagreement because the belief in question is in fact the product of correct reasoning. Thus, Thomas Kelly writes:
So, according to the correct reasoning version of nonconformism, peer disagreement's epistemic symmetry can be broken by virtue of the extra weight afforded to my belief in virtue of its being in fact best supported by the evidence.9
The rationality of the parties engaged in [peer disagreement] will typically depend on who has in fact correctly evaluated the available evidence and who has not. If you and I have access to the same body of evidence but draw different conclusions, which one of us is being more reasonable (if either) will typically depend on which of the different conclusions (if either) is in fact better supported by that body of evidence.
(Kelly 2005: 180)
In contrast to nonconformists, there are those who hold that disagreement itself possesses enormous epistemic significance; thus, unless one has a reason that is independent of the disagreement itself to prefer one's own belief, one cannot continue to rationally believe that p when one is faced with an epistemic peer who explicitly believes that not‐p. The requirement that the reason for preferring one's own belief be ‘independent of the disagreement’ at issue is meant to include both independence from the belief in question and from the reasoning grounding this belief. Without this requirement, begging the question would presumably be an appropriate way to respond to a case of peer disagreement—for example, one could settle a dispute over the question whether p by appealing to p itself. Moreover, since any instance of disagreement will provide one disputant with a reason to believe that the other has failed at some point in her reasoning, parties to a dispute would never qualify as epistemic peers if appealing to the reasoning grounding the belief in question were permissible. I shall call those who hold this view conformists.10 According to conformists, there cannot be reasonable disagreement among epistemic peers. Thus, Richard Feldman claims that:
In a similar spirit, David Christensen says that:
in situations of full disclosure, where there are not evident asymmetries, the parties to the disagreement would be reasonable in suspending judgement about the matter at hand. (p.301) There are, in other words, no reasonable disagreements after full disclosure, and thus no mutually recognized reasonable disagreements. (Feldman 2006: 235)11
And Adam Elga writes:
in [cases of disagreement with epistemic peers], I should change my degree of confidence significantly toward that of my friend (and, similarly, she should change hers toward mine). (Christensen 2007: 189)12
Conformists, then, argue that equal weight should be given to one's own beliefs and to those held by one's epistemic peers, and thus significant doxastic revision is required in the face of peer disagreement. What kind of doxastic revision is necessary? Answers to this question vary. Feldman, for instance, casts the debate in terms of an all‐or‐nothing model of belief, and so he argues that disagreement with an epistemic peer regarding the question whether p requires that both parties to the dispute withhold belief relative to p. Christensen and Elga instead frame the issues in terms of degree of belief, and so they argue that disagreement with an epistemic peer regarding the question whether p requires splitting the difference in the degrees of their respective beliefs. Thus, where 1 represents maximal confidence that p is true and 0 represents maximal confidence that p is false, if I give credence 1 to the proposition that Smith committed the murder and you give credence 0 to the proposition that Smith committed the murder, our attitude towards p should converge in the middle—we should give credence .5 to this proposition and become perfect agnostics with respect to Smith's guilt. But regardless of the details, conformists all agree that when epistemic peers disagree, substantial adjustment is required in their respective beliefs. (p.302)
When you count an advisor as an epistemic peer, you should give her conclusion the same weight as your own . . . call it the ‘equal weight view’. (Elga 2007: 478)
Suppose that before evaluating a claim, you think that you and your friend are equally likely to evaluate it correctly. When you find out that your friend disagrees with your verdict, how likely should you think it that you are correct? The equal weight view says: 50%. (Elga 2007: 488)13
Despite the differences between nonconformists and conformists, they appear to share a commitment to a thesis that I shall call Uniformity,14 which can be characterized as follows:
According to this thesis, it doesn't matter whether one's beliefs conflict with an epistemic peer's over a confidently held perceptual experience or a dubious political conclusion, a necessary mathematical proof or a supernatural religious doctrine, simple directions to the store or a complicated philosophical view—disagreement with epistemic peers either always does or does not require doxastic adjustment.
Uniformity: Disagreement with epistemic peers functions the same epistemically in all circumstances.
But what, one might ask, is involved in being an epistemic peer with someone? Answers to this question typically involve requiring the satisfaction of at least the following two conditions:
(p.303) In addition to evidential and cognitive equality, Richard Feldman adds what he calls full disclosure to the conditions relevant to the disagreement at issue. More precisely:
Evidential equality:15 A and B are evidential equals relative to the question whether p when A and B are equally familiar with the evidence and arguments that bear on the question whether p.
Cognitive equality:16 A and B are cognitive equals relative to the question whether p when A and B are equally competent, intelligent, and fair‐minded in their assessment of the evidence and arguments that bear on the question whether p.17
Let us say that when there is both evidential and cognitive equality between A and B in situations of full disclosure with respect to the question whether p, they are epistemic peers.
Full disclosure: A and B are in a situation of full disclosure relative to the question whether p when A and B have knowingly shared with one another all of their relevant evidence and arguments that bear on the question whether p.18
I shall call disagreement involving epistemic peers in this sense idealized, which can be understood as follows:
The ‘aware that’ clause of condition (1) is included to rule out the relevance of the following sort of case: I believe that p and some person in China, who happens to be my epistemic peer relative to this question, believes that not‐p, but we are entirely unaware both of each other and of our disagreement. Surely, it is not even clear that such a case properly involves a disagreement, let alone one that should be at the centre of this discussion. Similarly, condition (2) is included to preclude the relevance of the following sort of case: I believe the rather complicated mathematical conclusion that p and a three‐year‐old, whom I (p.304) have just met, believes that not‐p. While I am aware of our disagreement, I have absolutely no idea that this three‐year‐old is a math whiz for her age and thus my epistemic peer regarding this question. Once again, such a case clearly does not represent the sort of disagreement at issue in this debate.
Idealized disagreement: A and B disagree in an idealized sense if and only if, relative to the question whether p, (1) A and B are aware that they hold differing doxastic attitudes, (2) prior to recognizing that this is so, A and B take themselves to be epistemic peers with respect to this question,19 and (3) A and B are epistemic peers.
Now, idealized disagreement is to be distinguished from what I shall call ordinary disagreement. Ordinary disagreement does not require the parties to the dispute to actually satisfy the conditions of evidential equality, cognitive equality, and full disclosure. In particular:
So, whereas idealized disagreement occurs when the parties to the dispute are, as a matter of fact, epistemic peers, ordinary disagreement takes place when such parties at least take themselves to be roughly such peers.21 There are, then, two different questions that may be at issue: first, can one continue to rationally hold a belief in the face of idealized disagreement and, second, can one continue to rationally hold a belief in the face of ordinary disagreement? Which of these is the focus of the debate between nonconformists and conformists?
Ordinary disagreement: A and B disagree in an ordinary sense if and only if, relative to the question whether p, (1) A and B are aware that they hold differing doxastic attitudes, and (2) prior to recognizing that this is so, A and B take themselves to be roughly epistemic peers with respect to this question.20
Given that the distinction between idealized and ordinary disagreement is not drawn in the literature, combined with the fact that some theorists seem to emphasize the former while others rely on the latter, it is not entirely clear that there is a single answer to this question. For instance, Peter van Inwagen's emphasis on evaluating disagreement in terms of ‘all objective and external criteria’ (van Inwagen 1996: 275) and Elga's requiring that peers ‘have the same evidence’ (Elga 2010: ms p. 2) suggest idealized disagreement, while Feldman's talk of there being no ‘evident asymmetries' (Feldman 2006: 235) between the parties to the debate indicates that he is concerned with ordinary disagreement. And some theorists appear to focus on both kinds of disagreement, but at different points in their discussion. For instance, Kelly motivates the problem of epistemic disagreement with the question, ‘Can one rationally hold a belief while knowing that that belief is not shared (and indeed, is explicitly rejected) by individuals over whom one possesses no discernible epistemic advantage’ (Kelly 2005: 168), which sounds very similar to what I am calling ordinary disagreement. But when he explicitly argues against conformists, Kelly requires (p.305) evidential and cognitive equality between the parties to the debate, which parallels what I am calling idealized disagreement.22 So, while both kinds of disagreement figure in the debate, they do not always do so explicitly or consistently. In what follows, I shall, when relevant, be clear that there is the distinction between idealized and ordinary disagreement and I shall specify which kind is at issue when necessary.
In this chapter, I shall argue that neither nonconformism nor conformism provides a plausible account of the epistemic significance of peer disagreement. I shall first show that in some cases, nonconformism provides the intuitively correct result, and in other cases, conformism does. This leads to the rejection of Uniformity: disagreement with epistemic peers does not function epistemically the same in all circumstances. I shall then develop my justificationist account of peer disagreement's epistemic significance. Whereas current views maintain that disagreement, by itself, either simply does or does not possess epistemic power, my account holds that its epistemic power, or lack thereof, is explainable in terms of the degree of justified confidence with which the belief in question is held. I shall then show that my justificationist account has two central advantages: first, it is able to provide a principled explanation for why nonconformism provides the intuitively correct result in some cases, while conformism gives the intuitively correct result in other cases and, second, it is generalizable in a way that neither of these rival views is.
There are two questions that are at the centre of the debate between nonconformists and conformists: (1) does disagreement with an epistemic peer require substantial doxastic revision, and (2) can there be reasonable disagreement between epistemic peers? As we saw above, an answer to the former is taken to dictate an answer to the latter:23 nonconformists respond negatively to (1) and thus affirmatively to (2),24 while conformists answer affirmatively to (1) and thus negatively to (2).25 In this section, I shall begin with (1), focusing primarily on the response nonconformists have given to this question. (p.306) After doing so, I shall return to a more general consideration of both of these questions.
To my mind, the most promising line of defence for the nonconformist begins with the following type of case:
Now, PERCEPTION can be read as involving either idealized disagreement or ordinary disagreement. Let us evaluate these in turn.
PERCEPTION: Estelle, Edwin, and I, who have been room‐mates for the past eight years, were eating lunch together at the dining room table in our apartment. When I asked Edwin to pass the wine to Estelle, he replied, ‘Estelle isn't here today’. Prior to this disagreement, neither Edwin nor I had any reason to think that the other is evidentially or cognitively deficient in any way, and we both sincerely avowed our respective conflicting beliefs.
If Edwin and I are in an idealized disagreement over the presence of Estelle, then we must be epistemic peers with respect to this question, which requires evidential and cognitive equality in a situation of full disclosure. But it is unclear how to make sense of disagreement occurring in PERCEPTION under these conditions. For recall that two people are evidential equals relative to a question when they are equally familiar with the evidence and arguments that bear on this question. However, if I have a phenomenologically vivid experience of it seeming to me that Estelle is at the dining room table, and Edwin does not, then how could we be equally familiar with the evidence that bears on whether Estelle is present? Perhaps evidential equality could be glossed as follows: two people are evidential equals when they have an equal amount of evidence supporting their given beliefs. So, Edwin and I need not be equally familiar with the same relevant evidence; we just need to possess equal amounts of evidence for our respective, conflicting beliefs. If this weaker notion of evidential equality is granted, however, then pressure begins to build against the plausibility of the cognitive equality condition obtaining. For at least one of us must be hallucinating or experiencing some sort of cognitive malfunction in order to plausibly explain how one of us claims to see Estelle while the other does not when we are all presumably inches from each other. It is, therefore, quite difficult to even grasp how there could be idealized disagreement between Edwin and me in this situation. For these reasons, PERCEPTION seems best understood as a case of ordinary disagreement.
Let us, then, regard the disagreement between Edwin and me as ordinary in nature. Now, consider the situation in question from my perspective: it clearly seems to me that Estelle is sitting at the dining room table with me—indeed, suppose that minutes earlier we were engrossed in conversation while eating our pasta. Moreover, I have never in my life hallucinated an object, I have not been drinking or taking any drugs, I have my contact lenses in, I have ample evidence of my eyesight functioning reliably when my nearsightedness is corrected, and I know all of this to be true of myself. How, then, should I rationally respond to (p.307) Edwin's claim that Estelle is not dining with us? Despite the fact that, up to now, I have had good reason to regard Edwin as an epistemic peer, it seems clearly rational for me to continue to believe just as strongly that Estelle is present at the table. Indeed, even after full disclosure—where Edwin explains that he does not seem to see anything in the chair that Estelle purportedly occupies—I still seem rational in being fully convinced that she is dining with us at the table. For given the extraordinarily high degree of justified confidence with which I hold my belief about Estelle's presence, Edwin's disagreement seems best taken as evidence that something has gone awry with him, either evidentially or cognitively. In other words, I seem justified in concluding that Edwin is no longer an epistemic peer, even if he was prior to the disagreement in question. This conclusion is further evidenced by considering whether it would be appropriate for my belief to continue to guide my responses and actions in ways paradigmatic of confident belief. Imagine, for instance, that immediately after Edwin's full disclosure of his reasons for disagreeing with me, the doorbell rings and I open the door to find Estelle's mother asking if she is at home. Surely, it would be rational for me to respond affirmatively to this question. Indeed, Estelle's mother would rightly be utterly perplexed if I were to say, ‘Although I can apparently see her, I really do not know if she is here since I am withholding belief in light of my disagreement with Edwin on the topic’. Thus, when the disagreement in question is ordinary, nonconformism seems to deliver the appropriate intuitive response in PERCEPTION.26 (p.308)
Similar considerations apply in the following case:
As was the case in PERCEPTION, there seem to be two general points that can be made about the situation described in ELEMENTARY MATH. First, when this case is said to involve idealized disagreement, it becomes rather inexplicable. How could two adults—both of whom are functioning normally—who possess equal evidence relevant to the question at hand, disagree on whether 2 + 2 = 4? Surely, at least one of us is either confused or cognitively deficient in some way. Second, when the disagreement is ordinary, it intuitively seems quite rational for me to retain my belief even in the face of disagreement with Harry, whom I have had very good reason to believe is an epistemic peer. For given my extraordinarily high degree of justified confidence in my belief that 2 + 2 = 4, Harry's disagreement seems rightly regarded by me as evidence that he is not well, either evidentially or cognitively. In other words, as was the case in PERCEPTION, Harry's disagreement with me over the truth of 2 + 2 equalling 4 seems appropriately taken by me as evidence that we are no longer epistemic peers, and thus nonconformism again provides the intuitively plausible response.
ELEMENTARY MATH: Harry and I, who have been colleagues for the past six years, were drinking coffee at Starbucks and trying to determine how many people from our department will be attending the upcoming APA. I, reasoning aloud, say, ‘Well, Mark and Mary are going on Wednesday, and Sam and Stacey are going on Thursday, and, since 2 + 2 = 4, there will be four other members of our department at that conference’. In response, Harry asserts, ‘But 2 + 2 does not equal 4’. Prior to this disagreement, neither Harry nor I had any reason to think that the other is evidentially or cognitively deficient in any way, and we both sincerely avowed our respective conflicting beliefs.
Of course, even when the disagreement in PERCEPTION and ELEMENTARY MATH is ordinary, rather than idealized, it may be argued that the kind of disputes in these cases is so peculiar that it is unclear whether any general conclusions about nonconformism follow from them. For instance, how often does it happen that people disagree about whether their friend is sitting in the chair next to them, or about whether 2 + 2 = 4? So, let us consider a case that is similarly mundane, but where the ordinary disagreement in question is more likely to obtain:
What response should I rationally have to Jack's ordinary disagreement with me about My Thai's location? In particular, must I withhold, or at least significantly reduce my confidence in, my belief because a neighbour whom I believe to be an epistemic peer claims that the restaurant is on State Street? To my mind, nonconformism once again seems to give the correct intuitive result. For if I have lived in Chicago for fifteen years, know the city extremely well, frequently eat at My Thai, have not been drinking or taking any drugs, have substantial evidence that my memory is functioning reliably, and know all of this to be true of myself, then I seem perfectly justified in my confidence about My Thai's location, even after Jack fully discloses his vivid memory of the restaurant being on State Street. Indeed, given the substantial amount of credence and epistemic support enjoyed by my belief, Jack's disagreement seems appropriately regarded as evidence that something is not right with him. I may, for instance, suspect that he has been drinking, is delusional, or is suffering from some kind of memory loss; in any case, I seem rational in concluding that Jack is no longer an epistemic peer regarding the location of My Thai. Thus, as was the case in PERCEPTION and ELEMENTARY MATH when the disagreement in question is ordinary, nonconformism seems to deliver the intuitively correct result in DIRECTIONS.27
DIRECTIONS: I have lived in Chicago for the past fifteen years and during this time I have become quite familiar with the downtown area. Of the many restaurants that I enjoy frequently dining at, My Thai on Michigan Avenue is among my favourites. Jack, my neighbour, moved into the same apartment building the very weekend that I did fifteen years ago and he, too, has become quite competent in his acquaintance with the city. Indeed, it is not uncommon for us to bump into each other at various places, My Thai being one of them. Today, when I saw Jack coming out of his apartment, I told (p.309) him that I was on my way to My Thai on Michigan Avenue, after which he responded, ‘My Thai is not on Michigan Avenue—it is on State Street’. Prior to this disagreement, neither Jack nor I had any reason to suspect that the other's memory is deficient in any way, and we both rightly regarded one another as peers as far as knowledge of Chicago is concerned.
But, the conformist may object, in order for you to rationally retain your fully confident belief in the face of disagreement with an epistemic peer, there needs to be what we may call a ‘symmetry breaker’.28 A symmetry breaker is something that indicates that the epistemic position of one of the parties to the disagreement in question is superior to the other's. For instance, in DIRECTIONS, I have had good reason for believing that Jack is my epistemic peer relative to My Thai's location, and thus I have had good reason for believing that our epistemic situations are symmetrical regarding this topic. In order for me to fully retain my confidence in my belief in the face of disagreement with Jack, then, there needs to be a symmetry breaker that justifies my nonconformity. Otherwise, my resistance to doxastic revision is little more than dogmatic egoism.
Here is where the ordinary nature of the disagreement is relevant. For notice: even when you and I have had excellent reasons for believing that we are epistemic peers, I will, in ordinary situations, often have access to information (p.310) about myself that I lack with respect to you—let us call this personal information. Personal information is information that one has about the normal functioning of one's own cognitive faculties. I may, for instance, know about myself that I am not currently suffering from depression, or not experiencing side effects from prescribed medication, or not exhausted, whereas I may not know that all of this is true of you. Of course, given that I have excellent reason for believing that you and I are epistemic peers, I cannot, apart from the disagreement itself, have evidence for believing that you are currently suffering from depression, or that you are experiencing side effects from prescribed medication, and so on. But this is certainly compatible with my not knowing that these things are not true of you. Consider: two oncologists, both of whom are widely regarded as the best in the field, can surely appropriately regard one another as epistemic peers relative to cancer diagnoses, even if they do not at any given time know that the other is not depressed, not overtired, not suffering from a personal crisis, and so on. Now, applying these considerations to DIRECTIONS, note that the very disagreement at issue indicates that at least one of us—that is, either Jack or myself—is seriously cognitively malfunctioning. However, I know about myself that I have not been drinking, have not suffered from any recent delusions, and do not have any evidence for questioning the reliability of my memory. And while I do not have prior reason to question Jack's capacities, the fact of the disagreement itself, in conjunction with the personal information that I possess about myself, now gives me reason to think that there is a serious problem with his cognitive faculties. In particular, my personal information, when it combines with the already extraordinarily high degree of justified confidence that I have in my belief in DIRECTIONS, is able to serve as the requisite symmetry breaker here. Where Jack had previously seemed to be an epistemic peer of mine, I no longer regard him as one—his epistemic standing has collapsed. This permits me, then, to retain my belief about the restaurant's location with the same high degree of justified confidence I had before the disagreement occurred.
This same point could be cast in the language of defeaters. An instance of ordinary disagreement regarding the question whether p provides me with a defeater for my belief that p. When I am very highly justified in holding this belief, the personal information that I possess about myself and lack about my interlocutor can provide me with a defeater‐defeater for this belief. And, so long as I do not then acquire a defeater‐defeater‐defeater, I am thereby permitted to rationally retain my belief that p with the same degree of credence.
Perhaps the conformist will here object that the above considerations reveal precisely why it should be idealized disagreement that is at issue. For stipulating that the parties to the disagreement are, in fact, epistemic peers who have fully disclosed their equal evidence effectively rules out the sort of symmetry breaker provided by personal information in cases where there is a high degree of justified confidence, thereby enabling us to focus entirely on the epistemic significance of the disagreement itself. In particular, it can be built directly into evidential and (p.311) cognitive equality that both parties to the debate know of each other that there are no relevant asymmetries in their respective epistemic situations.
Notice, first, however, that when the case is idealized to this extent, it becomes quite difficult to make sense out of how disagreement is even possible. For if everything even remotely relevant to the topic at hand must be equal, then epistemic peers begin to sound much more like epistemic clones. It then becomes perplexing how epistemic clones relative to a question can even be engaged in a disagreement. Second, even if we were able to render coherent idealized disagreement between epistemic peers in DIRECTIONS, this concept has virtually no connection to the very disagreements that breathe life into this debate. For it is very common for philosophers writing on this topic to motivate interest in it by citing debates in history, philosophy, politics, religion, and other areas where disagreement is widespread and impassioned. But these debates bear very little resemblance to the hyper‐idealized scenarios under consideration here. It would rarely, if ever, happen that two people continue to disagree with one another about, say, gay marriage where there is evidential and cognitive equality with known epistemic symmetry in situations of full disclosure.29 Typically, there are all sorts of asymmetries at work, such as in the background assumptions that are being tacitly relied upon—for example, only one party to the dispute may regard what the Bible says as relevant to social institutions like marriage—or in different character traits—for example, only one party may be risk‐averse—or in different values—for example, only one party may value tradition more than equality, and so on. Moreover, it is even rarer for the two parties to the debate to know of one another that there are no epistemic asymmetries, particularly when personal information is at issue. How often does it happen, for instance, that I know that my colleague, with whom I disagree about the Iraq war, is not depressed, exhausted, distracted, and so on, on any given day? Thus, conclusions drawn from hyper‐idealized situations involving disagreement ultimately tend to have very little connection to the disagreements we face every day.30 For these reasons, I think that, at least for the most part, ordinary disagreement ought to be the focus of this debate.31
There is another door that is opened for the nonconformist by focusing on ordinary disagreement. Recall that Feldman adds the condition of full disclosure to the kinds of disagreement that are at issue in this debate. But obviously, there are more and less idealized versions of this requirement. At one end of the spectrum, full disclosure requires that epistemic peers knowingly share with one another literally all of the evidence and arguments in their possession that bear in any way on the question under dispute. While no doxastic revision is (p.312) hard to rationally justify under such hyper‐idealized circumstances, it is also difficult to understand how or whether this type of disagreement ever in fact obtains. As Ernest Sosa has emphasized, many of our beliefs are supported by countless pieces of subtle and complex evidence acquired via multiple sources over a number of years.32 When disagreeing over deeply entrenched religious and political beliefs or over memorial beliefs from the distant past, for instance, it is often practically impossible to conjure up all of the evidence and arguments that we have that bear on these topics. Once again, then, idealization worries lead to focusing on ordinary disagreement, which requires only that the parties to the debate take themselves to be roughly epistemic peers on the topic at hand. This, however, opens the door to relevant epistemic asymmetries explaining the disagreements at issue. For if there are subtle and complex pieces of evidence that have not been fully disclosed, then some of these very pieces of evidence may explain why the epistemic peers are engaged in the dispute under consideration. Perhaps underwriting my belief is a phenomenological experience that I cannot adequately convey, or massive amounts of evidence accumulated over many years that I couldn't possibly remember, or data acquired from various sources and contexts that I am unable to articulate. In these types of cases, disclosure would, at best, be only partial. Given that these sorts of epistemic asymmetries may be at work in cases of ordinary disagreement, then, there may be rational reasons for epistemic peers to not engage in doxastic revision.
So, when we focus on ordinary disagreement, it looks as though nonconformists are correct that a negative response to question (1) above is warranted: a subject who disagrees with an epistemic peer is not thereby rationally required to revise her beliefs. Does this thereby necessitate, as nonconformists suggest, an affirmative answer to question (2)—that is, that there can be reasonable disagreement between epistemic peers? Not necessarily.33 For while all three cases above are ones where I am not required to revise my doxastic attitudes in the face of disagreement, they also involve my coming to conclude that something has gone awry—either evidentially or cognitively—with my companion. For instance, upon hearing that Edwin is sincerely denying that Estelle is dining with us in PERCEPTION, I no longer believe that we are evidential and cognitive equals with respect to this question. Thus, the disagreement that Edwin and I have regarding this question is not a reasonable one; that is to say, it is not the case that we are equally reasonable in holding our respective beliefs.
But then is it correct to say that the nonconformist's negative response to question (1) has been defended? For if I no longer regard you as evidentially and cognitively equal in the above cases, then has it been shown that doxastic revision is not rationally required in the face of disagreement between epistemic peers? In (p.313) order to answer this question, recall Elga's characterization of conformism in the following passage:
According to conformism, then, if pre‐disagreement (at t1) A believes that she and B are equally likely to be correct regarding the question whether p, then post‐disagreement (at t2) A should believe the same. In other words, the disagreement itself should not change one's beliefs about the probability that one is right. Yet this is precisely what I, along with the nonconformist, deny in the above cases. For in PERCEPTION, ELEMENTARY MATH, and DIRECTIONS, I believe at t1 that in the case of disagreement with my epistemic peer, the probability that I am right regarding the question whether p is 50% and then at t2, in light of the nature of the disagreement itself and the positive personal information that I possess, I believe that the probability that I am right is dramatically higher.34 Given this, it would be more accurate to frame the original questions with which we began as follows: (1*) does disagreement with someone whom, were it not for the disagreement in question, one would regard as an epistemic peer, require substantial doxastic revision, and (2*) can there be reasonable disagreement between those whom, were it not for the disagreement in question, would regard one another as epistemic peers? It should now be clear that while I have defended the negative answer given by nonconformists to (1*), I do not thereby endorse their positive answer to (2*). That is to say, disagreement with someone who, were it not for the disagreement in question, one would regard as an epistemic peer, does not necessarily require doxastic revision, but it does not follow from this that there is reasonable disagreement.
Suppose that before evaluating a claim, you think that you and your friend are equally likely to evaluate it correctly. When you find out that your friend disagrees with your verdict, how likely should you think it that you are correct? The equal weight view says: 50%. (Elga 2007: 488)
It may be argued, however, that the disagreement in question is not actually changing one's beliefs about the probability that one is right in the above cases. For instance, consider my belief about the probability that I am right in my belief (p.314) about Estelle's presence in PERCEPTION. If you were to ask me what probability I would assign to my being right were I to disagree with Edwin on some unspecified perceptual issue, I would say 50%. But if you were to ask me what probability I would assign to my being right were I to disagree with him regarding, say, whether Estelle is sitting inches away from us, wouldn't I assign a much higher probability to my being right?35 Doesn't this, then, show that I am not downgrading the epistemic status of my peer on the basis of the disagreement itself, which is precisely what is at issue between the nonconformist and the conformist? For wouldn't the probability that I assign to my being right on the question whether Estelle is sitting inches away from us be the same at both t1 and t2?
No, and here is why. If I genuinely regard Edwin as my epistemic peer where perceptual matters are concerned, then even when asked in the abstract what probability I would assign to my being right were we to disagree regarding whether Estelle is inches from us, I would say 50%. It is only in the context of the actual disagreement itself—where I have a phenomenologically vivid experience of Estelle sitting inches from me and have relevant personal information about the normal functioning of my cognitive faculties—that I would offer a significantly lower probability. Given this, I downgrade Edwin's status as my peer precisely because of the disagreement in question, and thus the probability that I assign to my being right regarding whether Estelle is sitting inches from us differs between t1 and t2. This conclusion is clearly at odds with the view of disagreement advanced by the conformist.
At this point, then, there are three conclusions that I have reached: first, there are some cases where idealizing the conditions at issue renders the relevant disagreement either inexplicable or disconnected from the very disagreements that motivate the debate; second, there are some cases of ordinary disagreement where nonconformism delivers the intuitively correct result, at least with respect to the question whether I must revise my beliefs in the face of this disagreement; and, third, personal information can combine with the already high degree of justification held by a confident belief so as to provide precisely the symmetry breaker that is needed to explain why nonconformism seems to get it right in these cases.36 More will be said about these conclusions, but let us now consider whether any intuitive support can be garnered for conformism. (p.315)
One of the more intuitively compelling cases on behalf of conformism is provided by Christensen, which can be presented as follows:37
Now, the first point to notice is that, as we saw with the earlier cases, hyper‐idealizing the conditions in BILL CALCULATION renders the disagreement in question rather inexplicable. For instance, if evidential equality is required, it may be argued that I have the experience of going through a calculation that seems to support one‐fifth of the bill being $43, and Ramona has the experience of going through a calculation that seems to support one‐fifth of the bill being $45, so it is unclear how our evidence could be equal. Perhaps what is meant is merely that we have an equal amount of evidence supporting our different beliefs? But this surely cannot be what evidential equality requires, for there will then be countless cases where peer disagreement is explainable in terms of one party to the dispute being privy to a piece of crucial evidence that the other lacks. Moreover, full disclosure of the disagreement at issue will presumably include our sharing our calculations with one another, at least one of which is incorrect. So, after this full disclosure, what explains how idealized disagreement between epistemic peers persists in such a case?
BILL CALCULATION: While dining with four of my friends, we all agree to leave a 20% tip and to evenly split the cost of the bill. My friend, Ramona, and I rightly regard one another as peers where calculations are concerned—we frequently dine together and consistently arrive at the same figure when dividing up the amount owed. After the bill arrives and we each have a clear look at it, I assert with confidence that I have carefully calculated in my head that we each owe $43 and Ramona asserts with the same degree of confidence that she has carefully calculated in her head that we each owe $45.38
Given these considerations, BILL CALCULATION, like the earlier cases, seems best understood as involving ordinary disagreement. Thus, apart from the (p.316) disagreement at hand, Ramona and I both take ourselves to be epistemic peers with respect to the question under consideration—that is, we take ourselves to be in roughly a position of both evidential and cognitive equality regarding the amount each of us owes and to have fully disclosed our evidence to one another.39 What, then, is the rationally appropriate response to our disagreement? Christensen writes, ‘it seems quite clear that I should lower my confidence that my share is $43 and raise my confidence that it's $45. In fact, I think (though this is perhaps less obvious) that I should now accord these two hypotheses roughly equal credence’ (Christensen 2007: 193). While Christensen argues that Ramona and I should split the difference in the degrees of confidence in our respective beliefs, Feldman's view is that each of us should withhold belief about the amount owed; in both cases, however, the conformist requires substantial doxastic adjustment in the face of ordinary disagreement with peers.
I must admit to sharing the conformist's intuitions in BILL CALCULATION. Given that I argued on the side of nonconformism with respect to PERCEPTION, ELEMENTARY MATH, and DIRECTIONS, but admit that conformism provides the correct result here, I now want to consider whether there is a principled explanation of this difference.
The first point to notice about the two types of cases is the degree of confidence with which the beliefs in question are held. In all of the cases supporting nonconformism, I am extremely confident in the truth of the beliefs that I hold—I am, for instance, extraordinarily sure that Estelle is sitting next to me at the table, that 2 + 2 = 4, and that My Thai is on Michigan Avenue. In an ordinary case of disagreement, it would take a great deal more for me to adjust my doxastic states with respect to these beliefs than one person disagreeing with me, even one for whom I had, until that moment, good reason to believe was an epistemic peer. In contrast, when I divide in my head the amount of a large bill owed by five of us, I may be confident in my calculation, but I never come anywhere near the confidence that I have in my belief that a friend is currently sitting inches from me, or that 2 + 2 = 4, or that a restaurant that I have been frequently going to for the past fifteen years is on a street as unforgettable as Michigan Avenue. (p.317)
To test the relevance of the degree of confidence with which beliefs are held, consider the following:
While some doxastic revision may still be appropriate here, I do not at all have the same conformist intuitions that I have in BILL CALCULATION. And I suspect that if the case is modified to include more and more rounds of calculation done by me on paper, or perhaps even on a calculator, we may ultimately end up altogether eliciting nonconformist intuitions, even if Ramona is busy doing the same number of calculations. For in a wide range of cases, the amount of doxastic revision rationally required by ordinary disagreement seems to diminish as the degree of confidence with which the belief in question is held increases.
MODIFIED BILL CALCULATION: While dining with four of my friends, we all agree to leave a 20% tip and to evenly split the cost of the bill. My friend, Ramona, and I rightly regard one another as peers where calculations are concerned—we frequently dine together and consistently arrive at the same figure when dividing up the amount owed—and we both have a clear view of the bill. After repeatedly going through the division of the amount of the bill on paper, I assert with a high degree of confidence that we each owe $43. Ramona, who was also busy performing the division on paper, asserts with the same degree of confidence that we each owe $45.
But, clearly confidence, even when it is extraordinary, cannot be the only relevant feature here. For this would have the consequence that the hyper‐dogmatist—who is supremely confident in all of her beliefs—is never rationally required to adjust her doxastic states in the face of ordinary disagreement with epistemic peers. This brings us to the second point about the difference between those cases that support nonconformism and those that support conformism: what PERCEPTION, ELEMENTARY MATH, and DIRECTIONS have that BILL CALCULATION lacks is a highly confident belief that is very well‐justified epistemically. In the first three cases, I am in optimal epistemic conditions relative to the beliefs in question: I see my room‐mate of eight years sitting inches away from me in excellent lighting, I fully grasp with cognitive faculties that are not deficient in any way the truth of 2 + 2 = 4, and I vividly remember My Thai being on Michigan Avenue, where this is the most famous street in Chicago and the restaurant is one that I have walked to countless times in the past fifteen years. Given these optimal epistemic conditions, the probability of my being wrong in any of these three beliefs is extremely low, and thus my very confident beliefs are clearly highly justified.
In BILL CALCULATION, however, the situation is quite different: I arrive at a belief after dividing in my head the amount of a large bill owed by five of us. Several features of this case render the likelihood of my being wrong somewhat high. For instance, the bill is rather large—if it were instead a $25 bill that I was dividing equally by five, Ramona asserting that we each owe $6 rather than $5 would fail to clearly elicit conformist intuitions. I am also dividing the bill by five—if I were dividing a $218 bill by two, my believing that we each owe $109 may not intuitively require doxastic adjustment in the (p.318) face of an epistemic peer's disagreement. Moreover, I am doing the calculation in my head—if I were working the math out on paper, we may be less inclined to say that I should withhold belief given Ramona's disagreement with me. Finally, I perform the calculation only once—if I had repeatedly done the math and consistently arrived at the belief that we each owe $43, conformism no longer seems to be the clear response to the situation. What these considerations reveal is that the likelihood of my being wrong in BILL CALCULATION is significant, and thus neither my level of confidence nor my level of justification is very high with respect to the belief in question. To my mind, the absence of a highly justified confident belief is what explains the clear conformist intuitions in this case.
This provides the resources for answering a question that some readers may have at this point: why is there a symmetry breaker in the cases supporting nonconformism, but not in BILL CALCULATION? Here is the answer: personal information, when it combines with the already high degree of justification possessed by a confident belief—such as that enjoyed by my belief that 2 + 2 = 4—is sufficient for breaking the prior epistemic symmetry between you and me when we disagree. But the cases that clearly elicit conformist intuitions can be different from those that intuitively justify nonconformism in various ways. First, the nature of the disagreement in the latter cases indicates that at least one party to the dispute is seriously cognitively malfunctioning. When there is a disagreement regarding whether 2 + 2 = 4 or a friend is sitting at the table, for instance, it is unlikely that this can be explained by appealing to ordinary errors. Second, as noted above, the confidence enjoyed by the beliefs in the cases supporting conformism may not be very high. In BILL CALCULATION, for instance, it is plausible to think that I regard it as quite possible that I am wrong. Third, even if I tend to think very well of my abilities and thereby have an unusually high degree of confidence in my belief, this high degree is surely not justified given the substantial fallibility in this situation. Given these differences, the positive support provided by one's personal information will not be adequate, when combined with the relatively low degree of justified confidence, to render ordinary disagreement with a peer epistemically benign in cases supporting nonconformism.
We are now in a position to draw some conclusions about the epistemic significance of disagreement. First, the cases where nonconformism clearly provides the correct result are ones where there is a symmetry breaker between one's epistemic peer and oneself that is provided by the presence of personal information combining with a highly justified confident belief. More precisely: (p.319)
Second, the cases where conformism clearly provides the correct result are ones where there is a relatively low degree of justified confidence such that the positive support provided by personal information is insufficient for breaking the epistemic symmetry between one and one's epistemic peer. In particular:
No Doxastic Revision Required: In a case of ordinary disagreement between A and B, if A's belief that p enjoys a very high degree of justified confidence40, then A is permitted to rationally retain her same degree of belief that p if and only if A has a relevant symmetry breaker.
There will, then, be many cases that fall on the spectrum between no doxastic revision required, and substantial doxastic revision being necessary. If, say, A's belief that p enjoys a moderately high degree of justified confidence, then merely some doxastic revision may be required in the face of ordinary disagreement with an epistemic peer. For instance, rather than withholding belief or splitting the difference in degree of belief, A may be required to only somewhat reduce the degree to which she believes that p. On the other hand, if A's belief that p enjoys a moderately low degree of justified confidence, then more doxastic revision may be required in the face of ordinary disagreement, but perhaps still not as much as withholding or splitting the difference in degree of belief.41
Substantial Doxastic Revision Required: In a case of ordinary disagreement between A and B, if A's belief that p enjoys a relatively low degree of justified confidence, then A is rationally required to substantially revise the degree to which she holds her belief that p.
Third, in spite of my siding with nonconformists in some cases and conformists in others, it is a mistake to regard my view as merely a hybrid of the two. I do not agree with the nonconformist that peer disagreement is entirely without epistemic power, even when I support the intuitive results of such a view; nor do I agree with the conformist that peer disagreement possesses substantial epistemic power, even when I support the intuitive results of this view. In this sense, I reject both nonconformism—because the absence of doxastic revision in the face of peer disagreement is never justified merely by virtue of the fact that beliefs are either mine or are the product of correct reasoning—and conformism—because substantial doxastic revision in the face of peer disagreement is never justified merely by virtue of equal weight being given to my own beliefs and to those held by my epistemic peers. Instead, my view holds that peer disagreement's epistemic power, or lack thereof, depends on the degree of justified confidence with which (p.320) the belief in question is held combined with the presence or absence of relevant personal information. It is for this reason that my view is justificationist rather than simply a blend of nonconformism and conformism.
What notion of justification is at work here? While a detailed response to this question lies beyond the scope of this chapter, I should say at least a few words about how I am understanding this concept. To this end, let us briefly return to PERCEPTION which, it may be recalled, involves Edwin and me disagreeing over whether Estelle is present at the dining room table in our apartment. Now, let us suppose that Edwin's denial that Estelle is present is the result of hallucinating which, in turn, is caused by the fact that Edwin has been unknowingly drugged by a friend. The drug in question produces no signs discernible to the subject of its effects, and Edwin has no independent reason to doubt his own cognitive capacities.42 While I am happy to grant that, from a purely subjective point of view, Edwin is just as reasonable in his belief regarding Estelle's presence as I am in mine, I would not agree that our beliefs are even close to being equally justified. In particular, I am enough of an externalist about justification to require that the process or faculty responsible for the production of the belief in question be reliable or otherwise appropriately truth‐conducive.43 Thus, in the face of disagreement with one another, Edwin and I would both be entitled to hold our conflicting beliefs with the same degree of credence only to the extent that such beliefs are produced by processes that are equally reliable or truth‐conducive.44 Given that one belief is the result of a veridical perceptual experience, and the other the result of a hallucination, my view clearly grants radically different justificatory statuses to them.
It should also be noted that the level of justified confidence operative in the antecedents of the NDRR and the SDRR is determined prior to the disagreement in question, but the disagreement itself can affect whether there is ultimately a relevant symmetry breaker. This point can be illustrated by using the language of defeaters. An instance of peer disagreement regarding the question whether p may be interpreted as providing me with a defeater for my belief that p. However, if my belief that p has a very high degree of justified confidence, then the personal information that I possess about myself and lack about my interlocutor can become salient and thereby provide me with a defeater‐defeater for my original belief. This then enables me to rationally retain my belief that p after the disagreement with the same degree of credence. If, however, my belief that p does not have a high degree of justified confidence prior to the disagreement, then the force of the disagreement can do the defeating work in question, thereby requiring substantial doxastic revision on my part. (p.321)
Fourth, as should be clear, I reject Uniformity. My view maintains that ordinary disagreement with a peer sometimes permits nonconformist results and, at other times, conformist results, depending on whether a symmetry breaker is provided in conjunction with the degree of justified confidence enjoyed by the belief in question. I take this to be a significant virtue of the present account, as it not only accommodates the apparently conflicting intuitions in the literature, but also provides an explanation of them.
Finally, unlike responses offered by conformists to counterexamples such as PERCEPTION and DIRECTIONS, my view provides an explanation of why no doxastic revision is required in such cases that is both principled and plausible.45 To see this, consider the following slightly modified case from Christensen (2007):
By way of response to this sort of case, Elga offers the following:
EXTREME BILL CALCULATION: While dining with four of my friends, we all agree to leave a 20% tip and to evenly split the cost of the bill. My friend, Mia, and I rightly regard one another as peers where calculations are concerned—we frequently dine together and consistently arrive at the same figure when dividing up the amount owed. After the bill arrives and we each have a clear look at it, I assert with confidence that I have carefully calculated in my head that we each owe $43. In response, Mia asserts with the same degree of confidence that she has carefully calculated in her head that we each owe $450, which is more than the total cost of the bill.
(p.322) So, according to Elga, conformists need not appeal to justification in order to accommodate the intuition that no doxastic revision is required in such a case, for there is a symmetry breaker in EXTREME BILL CALCULATION that can do the work all on its own; namely, that I find Mia's answer utterly insane. Given this, similar remarks could perhaps be extended to cover the counterexamples from Section 1, such as PERCEPTION and DIRECTIONS—I possess a symmetry breaker in both cases because I find the responses offered by Edwin and Jack utterly insane.
according to the equal weight view, your probability that you are right should equal your prior probability that you would be right, conditional on what you later learn about the circumstances of the disagreement. And one circumstance of the split‐the‐check disagreement is that you are extremely confident that your advisor's answer is wrong—much more confident than you are that your answer is right. Indeed, her answer strikes you as obviously insane. So in order to apply the equal weight view, we must determine your prior probability that you would be right, conditional on these circumstances arising.
To do so, think of your state of mind before doing the calculation. We have assumed that, conditional on the two of you disagreeing, you think that your advisor is just as likely as you to be right. But it is also natural to assume that, conditional on the two of you disagreeing and your finding her answer utterly insane, you think that you are much more likely to be right. If so, then when that circumstance arises the equal weight view instructs you to favor your own answer. That is the intuitively correct verdict about the case.
There are, however, several problems with this response. First, if the mere fact that one finds the view of one's opponent utterly insane suffices to provide a symmetry breaker in cases of peer disagreement, then not only is the very spirit of the conformist position put in serious jeopardy, but the door is also opened for bias and dogmatism to justify one in rarely engaging in doxastic revision. For isn't it the case that the most impassioned sexists typically find arguments on behalf of women's rights laughable, the most devout racists find the views of proponents of equality crazy, and the most committed homophobes find the idea of granting homosexuals even the most basic dignities perverse. But even putting aside such extreme versions of bias and dogmatism, isn't it quite common for those engaged in deep, impassioned debates to find the views of their opponents completely insane? Opponents of abortion, for instance, frequently argue that the position of pro‐choice advocates is incomprehensible, those against the war in Iraq are often at a complete loss trying to grasp the view of its supporters, and it is not at all uncommon for critics of Andy Warhol's art to have difficulty understanding the minds of its admirers. Yet given what Elga tells us in the passage quoted above, it looks as though doxastic revision would not be required in such cases since finding the view of one's opponent ‘utterly insane’ provides the necessary symmetry breaker. Surely, this is a most unwelcome result for an account such as Elga's that pitches itself as an equal weight view.
Once the conformist attempts to avoid this problem, however, it becomes clear that all of the work supporting nonconformist intuitions cannot be done without appealing to the notion of justified confidence, which brings us to the second problem with Elga's response: the possession of symmetry breakers makes sense in the cases under consideration only against the background of beliefs that have a very high degree of justified confidence. To see this, notice that in the cases at issue—PERCEPTION, DIRECTIONS, and EXTREME BILL CALCULATION—the beliefs opposing mine are not, in and of themselves, utterly insane.47 By itself, it is not utterly insane for Edwin to believe that Estelle (p.323) is not at the lunch table with us—this is insane only against the background of my very highly justified belief that Estelle is sitting inches away from us at the lunch table. In and of itself, it is not evidently insane for Jack to believe that My Thai is on State Street—this is extremely doubtful only when viewed against my very highly justified belief that it is on Michigan Avenue. And viewed in isolation, it is not evidently insane for my friend to believe that we each owe $450 when we are splitting our restaurant bill evenly among five of us—this is insane only against the background of my extremely highly justified belief that the amount owed by each of us when we evenly split the bill must be lower than the total bill. Thus, we can see that the symmetry breakers that Elga claims are doing all of the work are in fact efficacious only when there are highly justified beliefs underwriting them, thereby providing further support for my justificationist account of disagreement.
Third, as suggested above, even if utter insanity could be understood without smuggling a high degree of justification through the back door, this response on behalf of the conformist would fail to generalize to include all of the problematic cases. For there is nothing about Jack's response in DIRECTIONS that is utterly insane in any reasonable sense. Moreover, it can certainly be stipulated that in all of the relevant cases, both parties to the disagreement find one another's answer equally insane. Given this, there wouldn't even be the possibility of the ‘utter insanity’ symmetry breaker that Elga has in mind, and yet doxastic revision is still not intuitively required on my part in such cases.
Similar considerations apply to Christensen's response (2007) to cases such as PERCEPTION, DIRECTIONS, and EXTREME BILL CALCULATION. According to Christensen, I do not have to engage in doxastic revision in the face of peer disagreement so long as the best explanation I can offer of the disagreement lies with an error made by my interlocutor rather than with one made by me. For instance, in EXTREME BILL CALCULATION, Christensen claims that ‘The real ground for thinking that my friend made the error . . . derives from the fact I have evidence that my assessment of the disputed proposition is supported by an extremely reliable kind of reasoning, but I have no basis for supposing the same about my friend's contrary assessment’ (2007: 20). But there are two readings of this. On a subjective reading, so long as the best explanation from my subjective point view of the disagreement in question lies with an error made by my peer, no doxastic revision is required. This reading, however, opens the door to rampant bias and dogmatism justifying an extreme version of nonconformism. For so long as one is egocentric or biased enough, one can always convince oneself that the best explanation of a given instance of peer disagreement lies (p.324) with errors made by one's interlocutors. In BILL CALCULATION, for instance, I could be so dogmatic about my calculations that it just seems obvious to me that the best explanation of my disagreement with Ramona lies with reasoning errors on her part. This is obviously an unwelcome result. Given this, an objective reading of Christensen's clause may be more promising, according to which no doxastic revision is needed so long as the best explanation from an epistemic point of view of the disagreement in question lies with an error made by my interlocutor. But this reading leads to my justificationist account of disagreement's epistemic significance. For instance, what makes Edwin's error the best epistemic explanation of our disagreeing over Estelle's presence is the extraordinary degree of justified confidence I have in my belief that she is sitting inches from us at the table, which thereby enables the personal information that I have about myself to serve as a symmetry breaker. If both this level of justified confidence and personal information were removed, it would be utterly mysterious what would make Edwin's error the best epistemic explanation of our disagreement over Estelle's presence. Thus, on this reading of Christensen's clause, the best epistemic explanation of the disagreement in question just bottoms out in a combination of justified confidence and the presence or absence of relevant personal information.
There is one final problem with the conformist response that I should like to mention, one that afflicts the views of both Elga and Christensen. Recall that conformists explicitly require that the reason for preferring one's own belief in a peer dispute be independent of the disagreement in question, where this is meant to include both independence from the belief and from the reasoning grounding this belief. Both of the claims offered above, however, are at odds with this requirement. With respect to Elga's response, the charge of utter insanity, as mentioned above, makes sense only when features of the actual disagreement are taken into account. For instance, Edwin's denial of Estelle's presence in the room is insane to me only against the background of the high degree of justified confidence possessed by my belief that she is so present. But then breaking the symmetry between Edwin and me by appealing to the utter insanity of his answer is not independent of the grounds for my belief. Regarding Christensen's response, I know that my belief is the result of an extremely reliable process precisely because of the nature of the grounds of this belief, for example, my vivid phenomenological experience of Estelle. Once again, then, breaking the symmetry between Edwin and me is not independent of the grounds for my belief. Thus, in an attempt to accommodate cases of extreme disagreement, conformists undermine a central thesis of their view.
Unlike rival views, then, my justificationist account has the resources to explain why no doxastic revision is required in cases such as PERCEPTION, DIRECTIONS, and EXTREME BILL CALCULATION in both a principled and plausible fashion while appealing to just the original resources of the view. (p.325)
In this chapter, I have defended a justificationist view of the epistemic significance of ordinary disagreement among epistemic peers. Such an account provides a plausible explanation for why nonconformism delivers the intuitively correct result in some cases of peer disagreement while conformism provides the right response in others. These are significant virtues that the present view has over its rivals, as it reveals that my justificationist view is a fully generalizable account of disagreement's epistemic significance that provides principled and unifying explanations of intuitions that would otherwise appear to be in conflict.48
(1) Of course, I am not really biased in my holding this belief!
(2) Much will depend on which epistemic asymmetries are here relevant; I shall address this issue in detail later in this chapter.
(5) When I speak merely of ‘peer disagreement’ I mean disagreement between epistemic peers.
(10) In other words, the conformist claims that one cannot downgrade the epistemic status of one's peer merely because of the disagreement in question itself; one must have an independent reason for so downgrading. Thus, Adam Elga writes, ‘Suppose that . . . you and your friend are to judge the truth of a claim, based on the same batch of evidence . . . . Without some antecedent reason to think that you are a better evaluator, the disagreements between you and your friend are no evidence that she has made most of the mistakes’ (Elga 2007: 487, emphasis added).
(11) Full disclosure, according to Feldman, occurs when the parties to the disagreement ‘have thoroughly discussed the issues. They know each other's reasons and arguments, and that the other person has come to a competing conclusion after examining the same information’ (Feldman 2006: 220).
(12) In particular, he argues that when faced with peer disagreement, ‘(1) I should assess explanations for the disagreement in a way that's independent of my reasoning on the matter under dispute, and (2) to the extent that this sort of assessment provides reason for me to think that the explanation in terms of my own error is as good as that in terms of my friend's error, I should move my belief toward my friend's' (Christensen 2007: 199).
(13) Even more precisely, Elga writes: ‘Equal weight view: Upon finding out that an advisor disagrees, your probability that you are right should equal your prior conditional probability that you would be right. Prior to what? Prior to your thinking through the disputed issue, and finding out what the advisor thinks of it. Conditional on what? On whatever you have learned about the circumstances of the disagreement’ (Elga 2007: 490).
(14) A possible exception is van Inwagen (1996, 2010), who focuses specifically on disagreements that are ‘matters of interminable debate’ (1996: 141), such as those found in philosophy, politics, and religion. I should mention, however, that van Inwagen fails to offer any general arguments or principles to justify treating various kinds of disagreement differently.
(16) Christensen calls this ‘Cognitive Parity’.
(17) Adam Elga offers an alternative characterization, according to which you count someone as an epistemic peer if ‘you think that, conditional on a disagreement arising, the two of you are equally likely to be mistaken’ (2007: 487). Elga defends his ‘non‐standard’ usage of ‘epistemic peer’ on the following grounds:
Now, even if Elga's criticism of the standard use of ‘epistemic peer’ is correct, it would not apply to my characterization in the text since I specify that A and B are both evidential and cognitive equals relative to the question whether p. Thus, according to my use of ‘epistemic peer’, two people could not be evidential and cognitive equals with respect to the question whether p and yet deviate in their likelihood to be mistaken regarding this question.
On more standard usages, an epistemic peer is defined to be an equal with respect to such factors as ‘intelligence, perspicacity, honesty, thoroughness, and other relevant epistemic virtues' (Gutting 1982: 83), ‘familiarity with the evidence and arguments which bear on [the relevant] question’, and ‘general epistemic virtues such as intelligence, thoughtfulness, and freedom from bias' (Kelly 2005). In defense of my use, suppose that you think that conditional on the two of you disagreeing about a claim, your friend is more likely than you are to be mistaken. Then however intelligent, perspicacious, honest, thorough, well‐informed, and unbiased you may think your friend is, it would seem odd to count her as an epistemic peer with respect to that claim, at least on that occasion. (Elga 2007: 499 n. 21)
There are, however, independent reasons to question Elga's non‐standard use of this term. For on his account, two people could radically differ in both their evidential backgrounds and their cognitive abilities with respect to the question whether p, yet nonetheless turn out to be epistemic peers regarding this question. For instance, I may be a complete novice with respect to identifying birds of prey, and you may be an expert ornithologist. When I am sober and you are highly intoxicated, however, we may be equally likely to be mistaken about whether the bird flying overhead is an osprey. On Elga's account, then, you and I would be epistemic peers with respect to this question, but this strikes me as quite a counter‐intuitive result.
(18) It is unclear whether full disclosure for Feldman includes that A and B both know that all of the evidence relevant to the question whether p has been shared, but I will assume that this is the case at this point. Later in this chapter, I will consider the epistemic significance of less idealized forms of disagreement.
(19) More precisely, this condition (and the one to follow in my characterization of ordinary disagreement) should be expressed as follows:
This version of the condition makes clear that the relevant parties need not literally possess the concepts of evidential equality, cognitive equality, and full disclosure in order to participate in a disagreement; it is sufficient that such parties would regard one another as epistemic peers, were they in possession of these concepts. For ease of expression, however, I shall stick with the less cumbersome formulation found in (2), leaving the reader to interpret it along the lines found in (2*).
(2*) prior to recognizing that this is so, A and B would take themselves to be epistemic peers with respect to this question, were they presented with the relevant concepts and definitions involved in being such peers.
(20) Otherwise put, A and B disagree in an ordinary sense if and only if, relative to the question whether p, (1) A and B are aware that they hold differing doxastic attitudes, and (2) prior to recognizing that this is so, A and B are not aware of any relevant epistemic asymmetry between their situations.
(21) What about cases where the parties to the debate should take one another to be epistemic peers, but do not? Such cases may raise all sorts of interesting epistemological questions about egoism, dogmatism, irrationality, and the like. But these issues seem rather independent of the epistemic significance of disagreement itself, and thus lie outside the scope of this chapter.
(22) This may lead one to wonder whether there is a certain amount of talking past one another taking place in this debate.
(23) At the end of this section, I shall argue that this assumption is, in fact, incorrect.
(24) For instance, recall that Rosen says, ‘It should be obvious that reasonable people can disagree, even when confronted with a single body of evidence . . . it would appear to be a fact of epistemic life that a careful review of the evidence does not guarantee consensus, even among thoughtful and otherwise rational investigators' (Rosen 2001: 71–2).
(25) Recall, for instance, that Feldman writes, ‘in situations of full disclosure, where there are not evident asymmetries, the parties to the disagreement would be reasonable in suspending judgement about the matter at hand. There are, in other words, no reasonable disagreements after full disclosure’ (Feldman 2006: 235).
(26) It is of interest to note that proponents of conformism often appeal to analogies with non‐agential instruments in an attempt to show the absurdity of nonconformist results. For instance, Christensen provides the following example:
Feldman (2006: 234) makes a similar point appealing to different thermometers. The intuition we are invited to share by conformists here is that, without a reason independent of the disagreement in question for doing so, it is just as absurd to prefer my belief over my peer's as it is to prefer the time my watch says over that of my friend's. But notice: there are cases analogous to PERCEPTION involving non‐agential instruments. Consider the following:
I look at my watch, a one‐year‐old Acme that has worked fine so far, and see that it says 4:10. Simultaneously, however, my friend consults her watch—also a one‐year‐old Acme with a fine track record—and it reads 4:20. When she tells me this, it clearly gives me new evidence that her watch is fast: I should not trust her watch as much as I would have before finding out that it disagreed with mine. But just as clearly, I've just gotten new evidence that my watch is slow, and this should diminish my trust in it. In this case, it's obvious that the fact that one of the watches is on my wrist does not introduce an epistemically relevant asymmetry. (2007: 196)
It seems clear to me in WATCH that, despite not having a reason independent of the disagreement to prefer the time my watch says over that of Sonya's, I am rational in regarding hers as the inaccurate one. For it is simply not at all plausible to think that we could have eaten lunch, walked around the lake, gone shopping, and read for several hours at the bookstore café in an hour and fifteen minutes, nor is it likely that the sun would be setting in Chicago at 1:15 in the afternoon. Thus, I take it that conformists fail to garner the intuitive mileage that they hope to on behalf of their view from these sorts of non‐agential cases.
WATCH: Sonya and I eat lunch together at a restaurant at noon, take a long walk around the lake, shop at multiple stores, and read at the bookstore café for several hours. I look out the window, noticing that the sun is setting, and say, ‘It is 7:45 PM, so we should get going’, after which Sonya responds, ‘My watch says it is only 1:15 PM’.
(27) I should mention that Christensen (2007) and Elga (2007) consider a counterexample to conformism that bears some similarities to those found here and attempt to show how their respective views have the resources to rule out doxastic revision in such a case. In Section 3 of this chapter, I shall consider their responses in some detail and argue that they fail in various respects.
(28) I borrow this phrase from Nathan Christiansen.
(29) Perhaps this is why Kelly claims that ‘to uncritically assume that things are perfectly symmetrical with respect to all of the epistemically relevant considerations . . . is . . . to subtly beg the question in favor of the skeptical view’ (2005: 178–9).
(31) Henceforth, when I speak merely of ‘disagreement’, I mean ordinary disagreement.
(32) See, for instance, Sosa (this volume, ch. 14).
(34) It should be noted that there are two probabilities that need to be kept distinct: there is (i) the subjective probability that p is true, and (ii) the subjective probability that I am correct in my belief regarding p. Earlier in the chapter, the discussion focused on (i), but now I am emphasizing (ii). It should be clear, however, that (i) and (ii) are intimately related for the conformist. For instance, recall David Christensen's claim that, upon discovering that I disagree with an epistemic peer, ‘(1) I should assess explanations for the disagreement in a way that's independent of my reasoning on the matter under dispute, and (2) to the extent that this sort of assessment provides reason for me to think that the explanation in terms of my own error is as good as that in terms of my friend's error, I should move my belief toward my friend's' (Christensen 2007: 199). Here Christensen is saying that in cases of peer disagreement, to the extent that I am willing to assign a 50% probability both to my being correct in my belief regarding p and to my epistemic peer's being correct, I should split the difference with my epistemic peer in the degree to which I believe that p. Thus, the probabilities assigned with respect to (ii) directly determine the probabilities that should be assigned with respect to (i).
(35) I am grateful to Benjamin Almassi for this question.
(36) What if the conformist were to respond that cases involving an asymmetry of access to personal information are ones where I in fact have reason, independent of the ordinary disagreement in question, to downgrade my opponent? (I am grateful to David Christensen for this type of response.) Given this, the conformist could grant the intuition that no doxastic revision is required in PERCEPTION, ELEMENTARY MATH, and DIRECTIONS, but then deny that it is a problem for her view. By way of response to this line of thought, notice that such a move has the consequence that there are virtually no actual instances of the kind of disagreement to which the conformist view purportedly applies. For in nearly all cases of ordinary disagreement, there is some sort of asymmetry of personal information. How often, for instance, do I know to the same extent as I do in my own case that my opponent is not distracted, sleep‐deprived, melancholy, and so on? Not very often (if ever). But then, given the response under consideration, it would turn out that there are virtually no cases of actual disagreement where equal weight should be given to both my view and my opponent's and, hence, it would turn out that there are virtually no cases where disagreeing peers are rationally required to doxastically conform. This, I take it, is an unwelcome result for conformism.
(37) A further defence of conformism frequently found in the literature appeals to the following:
Indeed, Kelly (forthcoming) argues that ‘a commitment to The Equal Weight View carries with it a commitment to The Uniqueness Thesis' (forthcoming: ms, p. 11). For proponents of this thesis, see White (2005), Feldman (2006, 2007), and Christensen (2007). While it lies outside the scope of this chapter to discuss The Uniqueness Thesis, let me say that none of the arguments I make here depends directly on its truth or falsity.
The Uniqueness Thesis: A body of evidence, E, justifies at most one doxastic attitude—i.e. believing, disbelieving, suspending judgement—toward any particular proposition.
(39) I am assuming that Ramona and I each asserting that we have carefully performed the relevant calculations in our heads suffices for us to have reason to believe that full disclosure has taken place. I should say, however, that the condition of full disclosure itself can be fleshed out in more or less idealized ways. For instance, at one end of the spectrum, it may be required that each of us provides all of the details of our respective calculations to one another. In this case, we again face problems making sense of the possibility of the disagreement. For if we have each gone step by step with one another through our fairly elementary calculations, what room is left for us to continue to disagree about the amount owed? At the other end of the spectrum, full disclosure may require nothing more than each of us asserting that we have arrived at the conclusion in question. But then this does not add anything significant to the initial disagreement. Thus, I think it is best to understand full disclosure as falling somewhere in the middle, which is what I have built into BILL CALCULATION.
(40) By a belief enjoying ‘a very high degree of justified confidence’ I mean a very confident belief that is highly justified.
(41) In my (forthcoming), I provide a far more detailed defence of my justificationist view that adds crucial further conditions to the above two principles. In particular, I argue that no doxastic revision is necessary if there is either a highly justified target or ‘protecting’ belief, and that substantial doxastic revision is required if there is neither a highly justified target nor ‘protecting’ belief. A's belief that p is protected by A's belief that q if and only if both A's belief that p and A's belief that q are members of a subset of A's beliefs, each of which is challenged by the same instance of ordinary disagreement with B, and where A's belief that q is highly justified and confidently held.
(42) I am grateful to Nikolaj Jang Pedersen for raising this sort of question to my account.
(44) Or, at least, roughly equal in terms of reliability or truth‐conduciveness.
(45) It should be noted that while Christensen (2007) and Elga (2007) both try to accommodate the intuitions elicited by cases such as PERCEPTION and DIRECTIONS within their conformist framework, Feldman (2006, 2007) instead tries to show that his view delivers the correct result in such situations: both parties should indeed withhold belief with respect to the question under dispute.
(46) In a similar spirit, Christensen (2007) argues that in order for Mia to have reached such an absurd answer, it is plausible for me to think that she must have failed to use a highly reliable, commonsense method of checking her results. Since I am confident that I did use such a reliable method, I do not have to engage in any doxastic revision in the face of our disagreement.
(47) The only possible exception to this claim is ELEMENTARY MATH, where it may be argued that denying that 2 + 2 equals 4 is insane in and of itself. I am inclined to think that this claim of insanity, too, makes sense only against the background of one's extraordinarily justified belief that 2 + 2 does equal 4. But even if one does not share my view here, the conformist is still left with the problem of explaining why doxastic revision is not intuitively required in PERCEPTION, DIRECTIONS, and EXTREME BILL CALCULATION in a way that does not appeal to the notion of justification.
(48) For helpful comments on earlier versions of this paper, I am grateful to Larry BonJour, David Christensen, Jeremy Fantl, Richard Feldman, Richard Fumerton, Peter Graham, Jon Kvanvig, Matt McGrath, Nikolaj Jang Pedersen, Blake Roeber, Andrew Rotondo, Ernie Sosa, Peter van Inwagen, audience members at the University of Washington, the Social Epistemology Conference at the University of Stirling, the Disagreement Conference at the University of Calgary, the 2007 meeting of the Eastern APA in Baltimore, Brown University, the University of St Andrews, and, especially, to Baron Reed.