BELIEF IN THE FACE OF CONTROVERSY[i]
Hilary Kornblith
University of Massachusetts, Amherst
Disagreement is ubiquitous. One need not be especially opinionated, or especially well-informed, to be aware that, whatever one’s beliefs, there are a great many people who have beliefs which are contrary to one’s own. As I write these words, the United States is about to head to the polls for another election, and many of the issues which voters will face are ones on which the electorate is deeply divided. The electorate is divided about moral and political issues, and it is divided, as well, about the likely effects of various policies; it is also divided about simple matters of historical fact. Anyone who has views on religious questions of any sort, including agnostics and atheists, is well aware that there are many others who see matters differently. Famously, there is deep division of opinion about evolution, especially in the United States, but, to a growing extent, in other industrialized countries as well. At times, it can seem that there is hardly any subject on which one might have opinions on which there are not many others who disagree. What is the epistemological significance of such disagreement? To what extent, if any, need one modify one’s views in the face of disagreement?
Of course, not all disagreement presents even a prima facie epistemological challenge to one’s own beliefs. I believe that the capital of Maine is Augusta. In fact, I know full well that the capital of Maine is Augusta. In order to make certain of this, I just checked it on a map. I also know that many people believe that the capital of Maine is Portland. This doesn’t make me reconsider my opinion at all, nor should it. I have an explanation of why it is that so many people are mistaken: people often assume that the capital of a state with which they have little experience is its largest city. More than that, having checked my belief against a reliable source, I also know that those who disagree with me would come to recognize their error, and change their belief, when confronted with the evidence I have. There is not some deep disagreement here about what constitutes good evidence for believing that a city is the capital of a given state. So in this particular case, I have good reason to believe that I am in a superior epistemic position vis-à-vis the question of the capital of Maine, and the disagreement I have with others is epistemically unproblematic. I should go on believing as I do even in the face of this disagreement. In this case, at least, I know that I have better evidence on this question than do others who disagree with me.
There are other cases as well in which I may shrug off the disagreement of others, not because I have better evidence than they, but because I have better judgment. If in trying to help a young child with his arithmetic homework, I explain that he has made a simple error of addition–five plus seven is twelve, I try to remind him, not thirteen–I should be completely unperturbed if he should insist that, no, he is right; five plus seven, he tells me, really is thirteen. My arithmetical skills are not, of course, infallible, but surely in this case, it is perfectly reasonable for me to go on believing as I did before facing this disagreement. I know full well what the sum of five and seven is, and disagreement from an arithmetical neophyte should not cause me to reassess my views about arithmetic or about my mathematical abilities.
It is clear, however, that not all cases of disagreement are like either of these two cases, and so I cannot simply chalk up my disagreements with others to their inferior evidential situation or their inferior judgment. I certainly can’t insist that others’ evidence or judgment must be inferior to my own simply in virtue of their disagreement with me. And it is for this reason that the fact of widespread disagreement raises troublesome epistemological questions. There is a very wide range of topics on which we knowingly disagree with others, and yet we seem to lack any reason to think that their evidence or their judgment is inferior to our own. What should we do in such cases? What is it reasonable to believe in the face of such disagreement?
Disagreements within philosophy constitute a particularly interesting case of this kind of disagreement. Consider the debate between internalists and externalists about epistemic justification. I am a committed externalist. I have argued for this position at length and on numerous occasions. My arguments for the position are not merely a pose. I have not presented arguments for externalism merely to serve as a gadfly, provoking discussion. I sincerely believe that the arguments I have presented are good arguments, and I sincerely believe their conclusions. At the same time, I recognize, of course, that there are many philosophers who are equally committed internalists about justification, and that the arguments they offer are ones to which they are not only committed, but which they believe are good ones, and whose conclusions they sincerely believe.
It would be reassuring to believe that I have better evidence on this question than those who disagree with me, that I have thought about this issue longer than internalists, or that I am simply smarter than they are, my judgment superior to theirs. It would be reassuring to believe these things, but I don’t believe them; they are all manifestly untrue. So on the question of whether externalism or internalism is correct, I find that I have an opinion, but there are others who disagree with me who are, to adopt a useful term[ii], my epistemic peers: they are just as smart as I, just as well-informed, and have thought about the issue just as long as I and just as carefully. That my epistemic peers disagree with me on this question is surely relevant evidence that I ought to take into account. It is indirect evidence on the question of internalism vs. externalism, but it is important evidence nonetheless. And it surely seems that the proper way to respond to evidence of this sort is to suspend judgment, to suspend belief about the proper resolution of the debate between internalists and externalists.
As Richard Feldman has pointed out[iii], this is precisely what it seems we ought to do, and what we in fact typically do, in perceptual cases. Thus, to take his example, if you and I are each looking out of a particular window, in exactly the same direction, and I see someone standing in the middle of the quad and you see no one there at all, then we have a puzzle. If you and I have equally good vision and we are both in our right minds, then we should each be surprised by the other’s judgment. “What do you mean,” I will say, “that you don’t see anyone there?” “What do you mean,” you will say, “that you really see someone out there in plain view?” Once we become convinced, however, that each of us is playing no joke, that we are looking in the same direction, and that we are, to all appearances, functioning normally, we should suspend belief about what there is in the middle of the quad. Someone here is making a bad mistake, and that person may well have some serious problem, and we will surely both agree that all of this is true. But it would be completely unreasonable for me to conclude, for this reason, that you are the one with the problem. I have no more reason to think that you are making a mistake than that I am, and, for that very reason, I should suspend judgment, as should you. When we give proper weight to the judgment of our epistemic peers in perceptual cases, it seems, suspension of belief is required.
As David Christensen has pointed out[iv], the same seems to be true in mathematical cases. To take his example, suppose that you and I go out to a restaurant with a number of friends. After a large meal, the check comes and we agree to split the bill evenly. You and I are each quite good at mental arithmetic. I take a look at the bill and figure out what each person owes, and I put my share in the middle of the table. You look at the bill and figure out what each person owes and put your share in the middle of the table, and then we notice that we have put in different amounts. We are each well aware of the other’s mathematical abilities, and we are each convinced of the other’s honesty. At least one of us has made a mistake. It would surely be unreasonable for me to conclude that, since one of us has made a mistake, it must be you. The reasonable thing to do in this situation, and surely what most people in fact do in this situation, is suspend belief. We each go back to the bill and try to recompute the proper share.
The dispute about the bill at the restaurant and the dispute about the person in the quad are importantly different from the philosophical dispute about internalism and externalism. In the restaurant case, our disagreement will, in actual practice, quickly be resolved. One of us has made a simple arithmetical error. We are each quite good at mental arithmetic, and we will quickly determine who has erred when we recalculate. We suspend judgment when we first note our disagreement, but our judgment is suspended only briefly. After quickly recalculating, we will figure out which of us was right about the bill, and we will comfortably agree about its proper division. In the perceptual case as well, we are often able to resolve our dispute in short order. We look more carefully; we call in a third person; we go out into the quad. The dispute is, often enough, only momentary, and our puzzlement is resolved. (“Oh, now I see,” I say, “it was only a shadow.”) While not all such disputes are resolved, of course, they leave few lingering doubts about perception or perceptual belief. The occasional disagreements which arise on such matters and which resist easy resolution do not leave us suspending belief about perceptual matters generally, nor should they.
But philosophical matters are importantly different. If you and I disagree about internalism and externalism, and we’ve each, as I’ve supposed, thought about this issue for many years, then a quick run-through of the arguments will not resolve the issue between us; this is, in no way, like the restaurant case. By the same token, we cannot profitably, as in the perceptual case, do anything analogous to looking more carefully (since we’ve already scrutinized the arguments with great care), call in a third party (since we know that there are many third parties on each side), or head out into the quad for a closer look. If we set aside our opinions about internalism and externalism as a result of our dispute, then we are likely to be suspending judgment on this issue for quite some time. When we disagree about our share of the bill at the restaurant, or about what is going on in the quad, we recognize that these are isolated disagreements against the background of extraordinarily broad agreement about arithmetic and perceptual matters generally, and for this very reason, there is no threat of skepticism here about mathematics or the physical world. Even if our disagreement should prove to be unusually difficult to resolve, for whatever reason, it is an isolated disagreement, one which does not threaten to force us to suspend judgment very widely. But if the right thing to do in the internalism and externalism case is to suspend judgment, then it seems that we will be suspending judgment on philosophical matters generally, and not just for a moment. And since most of us have very deep philosophical commitments on a great many matters, this would involve substantial revisions in our corpus of beliefs, revisions most of us are not eager to undertake. A broad skepticism about philosophical matters threatens.
I will assume, for the purposes of this paper, that the proper resolution of the disagreements in the perceptual and mathematical cases is, in fact, as I have described: in the face of disagreement from epistemic peers, given no special reason to believe that either party is mistaken, one must, if one is to be epistemically rational, suspend belief. I believe the case for this has been well made by Feldman[v], Christensen[vi] and Adam Elga[vii], and in related work by Roger White[viii]. Supposing that this is correct about such mundane matters, however, how should we respond in the case of philosophical opinion? Are we to suspend judgment there as well? If so, does this force us to a broad skepticism about philosophical matters? And if we are rationally forced to such a skepticism about philosophical matters, how broadly does this skepticism generalize? These are the questions I wish to address in this paper.
I
When I find that others disagree with me on a certain question, this gives me, ceteris paribus, reason to be less confident than I was that I am right. In the cases we have been looking at, disagreement from epistemic peers gives one reason to suspend judgment entirely on the disputed question. It is clear, moreover, that the mere possibility that someone might disagree with me does not have the same epistemic significance. Indeed, if it did, then since there might always be people who disagree with one on any question at all, treating merely possible disagreement as on a par with actual disagreement would result in total skepticism. The worries generated by problems of disagreement, however, broad as they are, are not of this sort. So there seems to be an important asymmetry between actual disagreement and merely possible disagreement.
As Thomas Kelly urges[ix], however, one should not overstate the difference here. “Suppose,” as Kelly suggests, “that there would be considerable disagreement with respect to some issue, but all of the would-be dissenters have been put to death by an evil and intolerable tyrant.” In such a case, disagreements which do not in fact occur, disagreements which are not actual, are no less epistemically significant than actually occurring disagreements. “The significance of actual disagreement,” Kelly concludes, “need be no more intellectually threatening than certain kinds of merely possible disagreement.” How then should we determine the role which actual and merely possible disagreement should play in determining what to believe? Kelly argues that considerations of disagreement, in the end, drop out of account.
Whether we find the possibility of disagreement intellectually threatening, I suggest, will and should ultimately depend on our considered judgments about how rational the merely possible dissenters might be in so dissenting. And our assessment of whether rational dissent is possible with respect to some question (or our assessment of the extent to which such dissent might be rational) will depend in turn on our assessment of the strength of the evidence and arguments that might be put forward on behalf of such dissent. But if this is correct, then the extent to which merely possible dissent should be seen as intellectually threatening effectively reduces to questions about the strength of the reasons that might be put forward on behalf of such dissent. Now there might be cases in which we judge that the arguments and evidence that could be brought forth on behalf of a hypothetical dissent are truly formidable, and this might justifiably make us doubt our own beliefs. But in that case, the reasons that we have for skepticism are provided by the state of the evidence itself, and our own judgments about the probative force of that evidence. The role of disagreement, whether possible or actual, ultimately proves superfluous or inessential with respect to the case for such skepticism. [181-2]
Now Kelly clearly has in mind cases in which disagreements actually arise, and arguments are offered on behalf of competing views, or, alternatively, although no actual disagreement has arisen, arguments against a particular view are available. And in these cases, Kelly urges that we simply look at the arguments for or against the views in question, and let the fact of disagreement, when there is one, drop out of consideration. Whether such a disagreement actually arises, or instead, is merely possible, is of no epistemic weight at all, according to Kelly.
It is important to note, however, that in many cases where disagreements arise, arguments of the sort Kelly has in mind are not in play. Thus, remember Christensen’s restaurant case, where you and I reach different conclusions about the fair division of the check. Neither of us offers reasons for our conclusion about how the bill should be split, and yet we are each faced with a significant challenge to our respective beliefs. When I suggest that a fair division requires one payment and you suggest, instead, that it requires another, the problem is already set. Given that we each are aware of the fact that we are both highly reliable in figuring out such answers, the problem is set precisely by the fact of disagreement, and, at this point in the conversation between us, we cannot explain the epistemic problem we each face without pointing to that disagreement itself.[x] So at least in cases such as this, the role of the disagreement itself is anything but superfluous, contrary to Kelly. Perceptual cases are often much the same.
Kelly is right, of course, that my belief will not be threatened when you disagree with me if I know that your contrary opinion is unreasonable. This does not mean, however, that I must know the basis for your belief before it can threaten my own. As in the restaurant case, and perceptual cases as well, I may know that your beliefs on this sort of matter are generally reliable, without knowing anything specific about the basis of your belief; I need not, in particular, know any of your reasons at all. In cases of this sort, the fact of disagreement, in light of my background knowledge of your reliability, is an ineliminable part of my reason for suspending belief.
What should we say, however, about the kinds of cases Kelly clearly has in mind, cases where each party to the disagreement has actually offered arguments for his or her view–at least when there is an actual disagreement–or, alternatively, when there is no actual disagreement, and yet arguments are available for each of two conflicting views? This is precisely the kind of case which arises in philosophy. In the internalism and externalism case, there are arguments which have been widely noted on both sides of the issue, and the participants to the debate, whatever position they may hold, are aware of these arguments. Similarly, as Kelly points out[xi], there are issues in philosophy where arguments are available on either side of an issue, and yet one side of the argument has no actually existing adherents. Kelly’s view, in both sorts of case, is that the fact of disagreement, or the lack of disagreement, should drop out of rational consideration. We should consider the arguments, and let it go at that.
Now there is no question that this accords well with standard philosophical practice. We simply don’t tend to see arguments of the following sort in the philosophical journals.
Smith has offered an argument that p, and I must admit that the argument appears to be sound. But Smith has failed to take account of important evidence that p is false: I am one of Smith’s epistemic peers, and although I grant that Smith’s argument appears sound, I find that I am not convinced by it. As a matter of fact, I believe that not p. So there you have it: Smith believes that p; I believe that not p. It’s a tie. Given this evidence, we should just both suspend belief.
Nor does one see probabilistic variants of this kind of reasoning, in which the distribution of opinion within the field is offered as evidence for the degree of confidence which should be assigned to a disputed claim. Standard practice is to do as Kelly recommends: ignore the distribution of opinion and focus on the arguments.
Kelly’s suggestion that we focus on arguments in such cases is not meant as merely pragmatic advice. He is not suggesting that focusing on arguments in these cases will advance the cause of inquiry, or that it is more likely to help get one’s work published in respectable journals, or that it will be likely to advance one’s professional standing. His point, of course, is that the arguments are where the relevant evidence is to be found. That there is an actual disagreement, or that the preponderance of opinion falls on one side or another of the disagreement, is, on Kelly’s view, simply irrelevant to the question at issue.
So let us imagine following Kelly’s prescription. I’ve offered arguments for externalism, and when I examine them, I find them thoroughly convincing. You’ve offered arguments for internalism, and when I examine them, I don’t find them convincing. Now I know that the situation here is perfectly symmetrical: you are convinced by your arguments, and you are not convinced by mine, but these facts, Kelly tells us, are ones which I should not take into consideration in evaluating what to believe; they are simply irrelevant here. I should focus on the arguments and forget about who believes what.
But why should I do this? I know that you have thought about this issue as long as I have; that you are as familiar with the arguments on both sides of this issue as I am; and I know that you are just as smart as I am. We are epistemic peers. So, just as in the restaurant case, I have good reason to regard you as reliable, or, at a minimum, as reliable as I am. And if this can serve as a reason for taking the fact that your opinion is different from mine as reason for doubt when I know nothing of your reasons, it is hard to see why it should be that it cannot equally serve as reason for doubt when I do know your reasons.
After all, once our reasons are on the table, the dispute between us comes down to our abilities to assess the cogency of complicated arguments.[xii] And we agree that we are each quite good at that; we’re epistemic peers. But we disagree about the cogency of the arguments at issue. So the disagreement about internalism and externalism has now been replaced by a different disagreement, a disagreement about the cogency of various arguments. How are we to resolve this dispute? On Kelly’s view, we would have to give reasons here as well, and it is these reasons, presented in the form of arguments, on which the rational resolution of the dispute must turn.
Now no one, I think, will disagree with the suggestion that formulating such explicit arguments may be a useful thing, and it may serve to advance our inquiry. We should each do our best to formulate the arguments which we find most revealing, as well as the counterarguments which, we each believe, show the errors in our opponent’s view. But these points are pragmatic ones, points about what we should do in order to advance inquiry. And Kelly’s claim is not one about such pragmatic concerns, but one about what it is reasonable to believe right now in the face of disagreement.
So let us return to the disagreement you and I have about, as I’ve been supposing, internalism and externalism. You and I have opposing views, and we are each aware of the other’s view; we are also each aware of the other’s arguments. And the question at issue is not what we should do now to advance our inquiry; looking at arguments is, no doubt, a good strategy to pursue for that purpose. The question at issue is what, at this very moment, each of us is justified in believing in the face of all of the evidence we have. Kelly’s view is that the fact of our disagreement is not even relevant evidence here.[xiii]
In order to see how best to handle this sort of case, it will be useful to examine a different case in some detail. So let us look at what Kelly has to say about the Newcomb Problem.
II
Robert Nozick presented the Newcomb Problem, a problem in decision theory, in 1969.[xiv] There are two possible solutions to the Problem, the one-box solution and the two-box solution. The Problem arises because there is a highly intuitive argument for each of these incompatible solutions. As Kelly points out [182, n. 16], when Nozick first presented the Problem, he remarked[xv] that opinion within the community of decision theorists seemed to be roughly evenly divided. Currently, however, Kelly notes that “the by-now over three decades of sustained debate...has resulted in a significant shift in the original distribution of opinion in favor of Two-Boxing.” [182, n. 16] How should the distribution of opinion on this question affect our judgment about its proper resolution? Kelly’s answer, as we have seen, is that we must look to the arguments on either side, and ignore the distribution of opinion.
Why should we ignore the distribution of opinion on this case? Kelly asks us to imagine a student studying the Problem at the time it was introduced; in talking to others, she discovers that opinion is evenly divided on the question. There are other possible worlds, however, in which everyone she meets is a One-Boxer. Is it rational to believe different things in these two worlds? Not according to Kelly.
Should she take a different view about Newcomb’s Problem in the other, unanimous world than she does in our fragmented and divided world? Despite the fact that she has access to exactly the same arguments in both worlds? This seems extremely dubious–after all, can’t the student in the unanimous possible world simply look over at our own fragmented world, and realize that here she has epistemic peers who extol Two-Boxing? [183]
As Kelly notes, “whether there is any actual disagreement with respect to some question as opposed to merely possible disagreement might, in a particular case, be an extremely contingent and fragile matter.” [181] But surely this leads, he argues, to the conclusion that the distribution of opinion should carry no weight for us.
Now I think it is significant that Kelly has chosen an example from decision theory. Areas of formal investigation within philosophy, such as logic, probability and decision theory, are ones in which an extraordinary amount of progress has been made. There can be little doubt that there is as much reason for confidence in some of the results in these areas as there is in the empirical sciences. A look at the history of these fields, moreover, shows some interesting trends.[xvi] As in mathematics, intuitions about results and about arguments in these areas are extremely unreliable at the early stages of investigation. It is only as theory advances that we come to be able to understand the fundamental concepts at issue in these fields, the kinds of intuitions which are genuinely insightful, and the kinds of arguments which are genuinely sound. To take a single example, Stephen Stich notes[xvii] that as late as the end of the nineteenth century, one could find logic textbooks which defended the gambler’s fallacy as a legitimate statistical inference. Diversity of opinion is a commonplace in these areas early on in the development of theory; as theory advances, the opinions of experts in the field tend to converge. Even among experts, of course, convergence of opinion is no guarantee of truth, but one would have to be a radical skeptic about mathematics, logic, probability and decision theory to think that convergence of opinion is not, at this point in the history of these fields, evidence of truth. And at this point in the history of these fields, I think it is fair to say, radical skepticism is no longer a rational option.
Thus, as I see it, when we see that after three decades of work on the Newcomb Problem, there is an emerging consensus on a solution, this is itself, given the history of the field, strong evidence that the consensus is correct. What should we make of Kelly’s suggestion that we may always look over at other worlds in which the consensus is different, or in which there is no consensus? And what should we make of his suggestion that consensus may be “an extremely contingent and fragile matter”? I believe that these suggestions give a misleading view of the dynamics of consensus within formal fields in philosophy.
Let us consider, by way of contrast, views about the aesthetics of clothing. If I were to head off to buy a suit, I would find that there is some consensus in the fashion world about the proper width and shape of lapels. The consensus today is quite different from what it was five or ten years ago, and different still from what it was in the nineteen thirties. Someone might argue that consensus among fashion experts is evidence of truth: the lapels currently approved of are ones, we should believe, that are aesthetically correct; the fashion world is converging on the timelessly right answer here. Whatever one thinks of the kind of realism about aesthetic value which such a view presupposes, it is surely unreasonable to think that consensus here is any evidence of truth at all. The history and dynamics of consensus in the fashion world is strikingly different from what one sees in mathematics. In mathematics we see periods of disagreement followed, after intensive study, by a growing consensus; in the fashion world, we see a large consensus which simply changes from year to year. The second of these two patterns does not support any kind of confidence that this year’s consensus is more likely to get it right than last year’s consensus, or the consensus of the year before, or the year before that.
The consensus we find in the fashion world is one which we might accurately describe, to use Kelly’s words, as “an extremely contingent and fragile matter.” Although lapels of a certain width are currently in fashion, they might just as easily have been seen as evidence of bad taste. We need change little about the underlying dynamics of opinion about fashion if we are to imagine a world in which much broader or narrower lapels are approved of.
But now contrast this kind of case with cases of emerging consensus in formal areas of philosophy. Thus, consider the way in which a consensus was achieved following the publication of Gödel’s result on the incompleteness of arithmetic. We can easily imagine worlds in which Gödel did not discover his proof, and in which no proof of incompleteness was discovered until much later, or at all. We can certainly imagine worlds in which someone else discovered the famous proof, or in which it was discovered earlier. Some of these worlds are not very different from our own. What would, however, be very much different from the world we inhabit is one in which Gödel discovers his proof, the very proof which he presented in the actual world; he presents it to the mathematical community, and a consensus emerges that it is mistaken. Such a world would be one which is very different from our own. It would be a world in which the abilities of mathematicians are utterly different from those in the actual world, and, as a result, one whose history would look entirely different from the history of mathematics as it has actually unfolded. The fact that we can imagine a world in which the people who are called mathematicians are all incompetent, and that they reach consensus in much the way that consensus about fashion is reached in the actual world, says nothing whatsoever about how we should allow consensus among actual mathematicians to affect our opinions.
Consensus in the mathematical community, except in rare cases, is not a fragile thing at all. When there is consensus in this community, the consensus is typically very robust: one could not easily change the consensus, or undermine it, without making very substantial changes in the underlying facts (for example, by changing the facts about what arguments are available on either side of an issue, or by changing the facts about the basic competence of the members of the community). The same is true within the scientific community. The fashion community, of course, is a different matter entirely.
So what should our student do when she examines the arguments for each side in the case of the Newcomb Problem? If she wants to understand anything about the Problem at all, she needs to examine the arguments on either side with tremendous care and try to figure out which of the two sides seems to have the better argument. And if she wants to have any chance at all of advancing the community’s understanding of the issues, she will, again, need to examine the arguments. But if the question is, instead, what she is justified in believing when she knows that the community of experts is deeply divided on the issue, or, alternatively, what she is justified in believing when she knows that the community is unanimous in favor of one side rather than another, then the answer is that she should go with the community opinion.[xviii] And this is not simply true of students, who are not yet peers with the experts. It is true of the experts themselves as well.
Consider, once again, Christensen’s restaurant case. In the simple version of the case, already discussed, you and I come to different opinions about how the bill is to be fairly divided. But Christensen also presents another restaurant case, in which I am dining with 17 other people, each of whom is known to be as reliable at arithmetic as I, and each of whom has come to the same conclusion about the fair division of the bill, a conclusion different from my own. Clearly in this case, as Christensen argues, I should believe that it is I who have made the mistake. Things are no different if we move from dividing the bill at a restaurant to solving a problem in decision theory. If my views on a problem in decision theory are entirely at odds with the experts in the field, then even if I am myself such an expert–well known to hold a minority position on this issue–I would not be justified in continuing to hold the belief in the face of such opposition.[xix]
This is not, in many cases, how experts behave. Experts will often go on holding a view even when they know that they are in the minority. If we restrict ourselves to formal areas like mathematics, logic, probability and decision theory, however, or to the empirical sciences, then the long-run prospects for such dissenters is not very good. These are all areas in which there is a well established track record of reliable results issuing from the community, and although dissenters are sometimes proven right, these communities are sufficiently reliable, and have been so for long enough, that one would always be ill-advised to bet on the dissenter in the face of an overwhelming majority opinion. My point is simply that this is something which dissenters should realize as well. If the question at issue is thus whether one is justified in siding with the dissenter in the face of an overwhelming majority, the answer is that one is not, and this is true not only of bystanders who seek to inform their opinions by looking to the experts; it is true of the dissenter him- or herself.
Kelly’s suggestion then that we should look only to the arguments directly bearing on a disputed question, and simply ignore the distribution of opinion, cannot be sustained.
III
So let us return to the internalism/externalism dispute and the question, more generally, of what we should make of disagreement within philosophy outside of the most formal areas. I have been arguing that the distribution of opinion is, contrary to Kelly, importantly relevant to what we should think about disputed matters. In the case of the internalism/externalism dispute, I will suppose, what I believe to be in fact true, that there is not an overwhelming majority of opinion among the experts within epistemology as to which of these approaches is correct. If opinion is not evenly divided on this issue, there is certainly not an overwhelming majority of expert opinion on one side rather than the other. What is it reasonable then to believe about this issue given the current state of play? I believe that we are justified in withholding opinion on this issue, and that one would not be justified in believing either that internalism is correct or that externalism is correct.
Suppose, first, that one thinks that the philosophical community is, in relevant respects, like the mathematical or scientific communities. Suppose, that is, that one believes that individual philosophers are, by and large, quite reliable in their opinions in matters that touch on their areas of expertise, and that the community, overall, has shown a long history of steady progress on the issues it addresses. If one were to believe this (and be justified in believing it), then, as I’ve argued, the distribution of opinion is straightforwardly relevant to what one should believe, and when opinion is closely divided, one ought to suspend judgment. So in this case, suspension of judgment would be mandatory.
But surely it is not reasonable to believe that the philosophical community is like the mathematical or scientific communities in relevant respects. We don’t have a long history of steady progress on issues, and, as a result, the case for deferring to community opinion is thereby weakened. But this hardly strengthens the hand of those who would form an opinion one way or the other on matters such as the debate between internalists and externalists. When the community is composed of individuals each of whom are reasonably believed to be reliable, we must bow to majority opinion. But if we do not have reason to believe that the community is composed of individuals who are reliable, if the history of the field gives us no reason for confidence in the judgment of individual practitioners, then this, by itself, gives us reason to suspend judgment on questions which confront the field. If the history of the field shows no track record of success in addressing the issues it confronts, the only conclusion we can reasonably reach is that there is no basis for opinion here on anyone’s part at all. It certainly does not give one free reign to believe whatever one pleases.
The sad truth, it seems, is that the history of philosophy does not look remotely like the history of science or mathematics when it comes to the dynamics of consensus among its most esteemed practitioners, and this has a striking bearing on the question of its epistemic credentials. One might try to carve out a recent piece of this history, and some particular subject-matter, where one believes that real progress is being made and that we are finally getting at the truth on some important issue. I must confess that, in my more optimistic moments, I find such a view tempting. But if we are to take any such view seriously, and subject it to real scrutiny, we would surely find that this view of the particular question at issue is itself a subject of real controversy among acknowledged experts in the field, and so it too must be seen, on careful consideration, as an issue on which we ought to suspend judgment. The field of philosophy in general, outside of the more formal areas which are most akin to mathematics, simply does not have anything like the epistemic standing of the empirical sciences. So much as we all find ourselves forming beliefs about disputed philosophical questions when we immerse ourselves in the arguments, we should acknowledge in quiet moments of reflection that these views we form are ones which are not epistemically justified. It would be as presumptuous to claim that we are justified here as it would in Christensen’s restaurant case when we find that our mathematically reliable dinner companion has reached a different conclusion about the division of the check.
It is not my view that the distribution of opinion is all the evidence one needs to determine what one should believe on any given question; such a view is completely untenable. It must be acknowledged, however, that the distribution of opinion among acknowledged experts must carry a tremendous amount of epistemic weight, and anyone who would reject a consensus among experts, or adopt a specific view in the face of deep division among experts, faces a very large hurdle. One must show that, in such a case, there is independent reason to believe that a substantial group of experts have gone wrong on the disputed matter.[xx] To take a purely imaginary case, one might be able to show that a large portion of some field of experts had been kidnapped and forced to express certain views at gunpoint, and, in such a case, one would be fully justified in ignoring the expressed consensus in the field. Here one’s evidence of the kidnapping would be entirely independent of the disputed question within the field, and once the expressed opinions of the kidnapped experts were thereby discounted, one might well be in a position to have confidence in one’s own judgment. But here, too, one would need to attend to the opinions of the remaining experts, if any. And I think it is perfectly clear that the high hurdle that bucking the consensus of expert opinion presents us with, even when we are among the experts ourselves, is one we are rarely if ever in a position to meet. I have no reason to believe, for example, that internalist epistemologists have been defending their views under threat from subversive kidnappers.[xxi]
I am thus forced to conclude, very reluctantly, that the opinions I hold on most philosophical matters–and I have a great many of them–are not epistemically justified. Given the current state of the field, no one’s opinions on these matters, it now seems to me, are epistemically justified.
More than this, this conclusion seems to generalize quite broadly. There are, for example, a great many moral and political issues, issues about which I have, in some cases, rather strong opinions, which are subjects of dispute among very intelligent, thoughtful and well-informed individuals. In some of these cases, we can explain away the disagreement of otherwise reasonable people by way of moves less desperate than the suggestion that our opponents have expressed their views only under threat by kidnappers. We do, after all, sometimes have good, independent reason to believe that someone who disagrees with us on a particular question is biased on that very question. We should, however, be wary of making this move too easily. The requirement of independence means that we cannot reject the opinions of our opponents as the product of bias merely on the basis of the view that their opinions are false–since they disagree with our own--and therefore must inevitably be due to some sort of biasing factor. What this means, in the end, is that this sort of maneuver will help us only rarely in dispensing with the challenge of disagreement from peers. And what follows from this, of course, is that a broad skepticism threatens.
This is not the sort of total skepticism of the Cartesian demon, but it would certainly force a radical revision in our body of beliefs.
IV
One might try to resist this conclusion, not, as Kelly does, by denying the relevance of the opinions of others, but, instead, by denying that giving others’ opinions their due weight in the interesting cases of disagreement–in philosophy, for example, and on moral and political questions–forces such a broad withholding of opinion. On such a view, idealized cases like Christensen’s restaurant example do genuinely show that one’s own opinions should be given no more deference than the opinions of one’s epistemic peers, but features of the more complex cases–somehow–allow one to avoid the unwanted and widespread change in our bodies of belief to which the general principle seems to lead. This, to be sure, is a consummation devoutly to be wished. It is, moreover, precisely the position defended by Adam Elga.[xxii]
As Elga points out, when you and I disagree about the fair division of the check at the restaurant, our disagreement is highly isolated. We don’t disagree about arithmetical questions in general. Indeed, I count you as my arithmetical peer precisely because we agree so broadly on arithmetical questions. More than that, it is not just that we agree. You and I might agree on some questions knowing full well that we are rank amateurs in an area where others are far better informed. But this is not the case in the restaurant example. We not only agree. We have good reason to believe that we are as well informed as just about anyone when it comes to these sorts of questions. We are fully justified in believing that we are both highly reliable in forming judgments about simple arithmetical issues. We simply disagree about an isolated question within arithmetic when we are each able to assess the other’s track record on arithmetical questions in general in a perfectly straightforward way.
This is not so, as Elga points out, in many other areas. Thus, Elga imagines two friends, Ann and Beth, who disagree about abortion. Their disagreement about abortion, if we are to make the case realistic, as Elga does, is not an isolated disagreement. Instead, there is a very wide range of related moral questions about which Ann and Beth disagree. But this is no minor complication, as Elga argues, for now the very basis each has for regarding the other as generally reliable on the kinds of questions at issue has been undermined. I may regard you as reliable about arithmetic even when we disagree about how to divide the check at the restaurant because we agree, it may be supposed, about every other arithmetical question we have ever jointly considered. But Ann and Beth cannot regard each other as reliable about moral matters generally precisely because their disagreement is so broad. We will only regard others as suitably reliable when they agree with us very broadly. It is for this very reason, Elga argues, that we may eat our cake and have it too: we may acknowledge that the opinions of our epistemic peers count just as heavily as our own antecedent opinions, and yet deny that in areas of moral and political dispute, for example, we must simply withhold opinion. Thus, Elga concludes:
...with respect to many controversial issues, the associates who one counts as peers tend to have views that are similar to one’s own. That is why–contrary to initial impressions–the equal weight view does not require one to suspend judgment on everything controversial.
One might try, in response to this move, to think of the dispute between Ann and Beth not as one about abortion alone, but as one about a cluster of related issues which includes the abortion question. Ann and Beth, one might argue, regard themselves as reliable on issues outside this cluster, and it is for this reason that they are troubled by their disagreement about the cluster itself. They should regard each other as reliable on the basis of their agreement on issues outside the cluster surrounding abortion, and this then forms the basis for taking the disagreement over the cluster of issues surrounding abortion as seriously as they do. Once one views the disagreement in this way, it seems, the parallel with the restaurant case is restored.
But Elga denies this.
Contrary to what the objection supposes, Ann does not consider Beth a peer about that cluster [of issues related to the abortion question]. In other words, setting aside her reasoning about the issues in the cluster, and setting aside Beth’s opinions about those issues, Ann does not think Beth would be as likely as her to get things right. That is because there is no fact of the matter about Ann’s opinion of Beth, once so many of Ann’s considerations have been set aside... Once so much has been set aside, there is no determinate fact about what opinion of Beth remains.
Thus, according to Elga, once again, we see that the basis for seeing our opponents on moral and political matters as generally reliable, when our disagreements are very broad[xxiii], is undermined. And once our judgment of the reliability of our opponents is thus undermined, he argues, we are entitled to go on believing as we did before the disagreement was encountered.
I do not believe, however, that Elga is right about this. First, it seems to me that Elga’s attempt to undermine Beth and Ann’s judgments about each other’s reliability here is unsuccessful. It is true that, in the imagined case, there is a large cluster of moral issues about which Ann and Beth disagree. But we need not exaggerate their disagreement. Thus, for example, Elga describes Ann and Beth as being “at opposite ends of the political spectrum.” So it would not be unfair to imagine Ann as, say, someone who characterize herself as a typical American pro-choice Democrat, and Beth as someone who characterizes herself as a typical American pro-life Republican. They disagree, to be sure, on a wide range of moral and political issues. But although Beth and Ann disagree about a great deal, their disagreement is not at all like their disagreement with, for example, Zena, a homicidal sociopath. Zena doesn’t just disagree with Ann or Beth about the cluster of moral issues surrounding the abortion question. She disagrees with them about virtually every moral question one might care to raise.
Now I think it is safe to say that neither Ann nor Beth will be much troubled by Zena’s disagreement with them, nor should they. And the reason why they should not is precisely that, when we subtract the moral issues on which Ann or Beth disagree with Zena from the totality of moral issues, there is virtually nothing left at all on which they might base a judgment that Zena is, but for their little disagreement, generally reliable about moral issues. Here we may reasonably say precisely what Elga says about the disagreement between Beth and Ann: once we set aside the issues on which Beth and Ann disagree with Zena, there simply is no basis for forming an opinion about Zena’s reliability on moral issues. When we set aside the areas of disagreement, after all, there are no moral issues that are left.
But this, it needs to be emphasized, is not at all the case with the disagreement between Beth and Ann. They disagree profoundly about an important range of moral issues, but neither regards the other as the moral equivalent of Zena. Indeed, it is because they do not regard each other as Zena’s moral equivalent that they are so engaged, and so disturbed, by each other’s opinions. Beth and Ann regard each other as basically decent, caring, thoughtful individuals whose opinions on a very wide range of moral matters, outside the sphere of issues most closely related to abortion, are trustworthy and insightful. While they rightfully dismiss Zena’s moral views out of hand, they are in respectful agreement about a very wide range of moral issues. And it is on this basis that they regard each other as moral epistemic peers, something they simply cannot do with Zena. Given that they do, justifiably, regard each other as moral epistemic peers, their grounds for withholding belief on the cluster of issues on which they disagree is thereby restored.
Note that the same is true when we consider disagreement about philosophical questions. I disagree with others about the proper resolution of the internalism/externalism debate in epistemology. But this is not like the restaurant case, where there is a disagreement about a single claim against the background of complete agreement about all other issues on the same general subject. Rather, once again, we have a case like the abortion issue. Those who I disagree with about internalism and externalism are philosophers with whom I disagree about a wide range of related issues. Does this then mean that I am no longer in a position to see these philosophers as my epistemic peers, as Elga suggests is the case with Beth and Ann on the abortion issue? Not at all. Even these disagreements, broad as they are, take place against a background of very broad agreement, agreement about the important issues in epistemology, about which positions are worth taking seriously, about what counts for and against various views, and so on. In short, I view internalists in epistemology in much the same way that Ann views Beth; I do not view internalists in the way that Ann views Zena. What this means, of course, is that there is ample room to view such philosophers as my epistemic peers, which is, in fact, precisely how I view them.
Disagreement on a wide range of related issues, as in the abortion debate or the internalism/externalism controversy, is not automatically a bar to reasonably viewing one’s opponents as one’s epistemic peers. And it is for this very reason that on many matters of great controversy, the only rational thing to do is suspend belief.
V
It is worth thinking about these controversial matters from a somewhat broader perspective. Consensus and near-consensus on formal issues within philosophy is so weighty epistemically, as I’ve suggested, because there is a history within these formal areas of undeniable progress. Against that background, consensus among the experts is a formidable matter. By the same token, disagreement among experts in those fields must also carry great weight. The same is true, of course, in the empirical sciences. When we look at the track-record of less formal matters in philosophy, however, or on matters of public controversy having to do with moral and political issues, the case for a well established track record of progress is, to be sure, harder to make out. Without that background of longstanding progress, we must look at individual investigators quite differently. In the sciences, we have good reason to believe that individual experts are each highly reliable overall, and they should surely regard one another in this way. Their considered opinions should thereby be tempered when there is disagreement in the field. In philosophy, however, and the other areas of controversy we have been considering here, there is no such history of longstanding progress, and for that very reason we should not consider the experts in the field–including, of course, ourselves–to be highly reliable. The history of the field simply does not give any ground for credence in such a view. But it is then, for that very reason, that we must, in the end, withhold opinion on the issues under consideration.
I do not mean to suggest that we should stop thinking about these issues, or that thinking about them, and trying to work out tenable views, is not intellectually respectable. This is not my view at all. And given the nature of human belief, I very much doubt that philosophers will stop forming views about the subjects they think about for so long and with such care. When we stand back, however, and reflect on our practice and on the beliefs which that practice generates, it seems to me that the history of our field makes epistemic modesty the only rational position available. We may hope that in trying to work out the views we are most sympathetic to that, in the long run, we may somehow contribute to an approach to philosophy which will look more progressive than any we have thus far seen. But at the present time, we should all recognize that this is merely a hope, and that rational belief must be tempered by the facts about our current situation.
[i]I have received helpful comments from the audience at a presentation of this paper at the Free University, Amsterdam and at MIT. I also wish to thank David Christensen, Adam Elga and Tom Kelly for comments on a draft of the paper.
[ii]This term was introduced by Gary Gutting in his Religious Belief and Religious Skepticism, University of Notre Dame Press, 1982.
[iii]”Epistemological Puzzles about Disagreement,” in Stephen Hetherington, ed., Epistemology Futures, Oxford University Press, 2006, 223.
[iv]”Epistemology of Disagreement: The Good News,” Philosophical Review, forthcoming.
[v]Op. cit.
[vi]Op. cit.
[vii]”Reflection and Disagreement,” Noûs, forthcoming.
[viii]”Epistemic Permissiveness,” Philosophical Perspectives, 19(2005), 445-459.
[ix]”The Epistemic Significance of Disagreement,” Oxford Studies in Epistemology, 1(2005), 181. Quotations from Kelly in the remainder of this paragraph are from this page.
[x]There is, of course, an argument which each of us can give which explains our epistemic predicament, but any such argument makes essential reference to the fact of disagreement. These are not, of course, the kinds of arguments Kelly has in mind.
[xi]184-5.
[xii]Actually, I don’t think this is entirely correct, but I will assume it for the sake of argument in the text since it is most favorable to Kelly’s position. In fact, I believe, the reasons we are able to articulate often fail fully to do justice to the reasons for which we believe. It is these latter reasons, rather than the ones we can verbalize, which determine the epistemic status of our beliefs. I have discussed this further in “Distrusting Reason,” Midwest Studies in Philosophy, XXIII(1999), 181-196, and Knowledge and its Place in Nature, Oxford University Press, 2002, chapter 4.
[xiii]Kelly does, in addition, present an argument that even if we were to count the fact of disagreement as relevant evidence, this would not rationally force us to suspend belief in the cases described. [185-90] Nevertheless, his view is, as the quoted passage shows, that facts about disagreement are epistemically irrelevant to the issue under dispute.
[xiv]”Newcomb’s Problem and Two Principles of Rational Choice,” reprinted in Nozick, Socratic Puzzles, Harvard University Press, 1997, 44-73.
[xv]”Newcomb’s Problem and Two Principles of Rational Choice,” 48.
[xvi]See, for example, Ian Hacking, The Emergence of Probability, Cambridge University Press, 1975.
[xvii]The Fragmentation of Reason, MIT Press, 1990, 83.
[xviii]I am assuming here that our student has not come up with a new argument, for in such a case our student is familiar with an argument which the other side has not yet considered. There are interesting questions about the extent to which one should be confident that one is right even in such cases, but I will not consider them here.
[xix]Again, as in the previous note, I am assuming that I have not discovered some new argument in favor of my position which is as yet unknown to the other members of the community.
[xx]Christensen’s discussion of this point about independence is particularly illuminating.
[xxi]This bears on a question addressed by David Lewis in “Academic Appointments: Why Ignore the Advantage of being Right?,” reprinted in his Papers in Ethics and Social Philosophy, Cambridge University Press, 2000, 187-200. Lewis notes the common practice, not only in philosophy departments, of bracketing the truth of a candidate’s views in making judgments about whether the candidate should be offered a job. Why should we do this? Lewis’ answer is as follows:
We ignore the advantage of being right because we comply with a tacit treaty to do so. It is reasonable for all of us to think that this treaty, and therefore our present compliance that sustains it, serves the advancement of knowledge. [197]
While I don’t doubt that considerations of this sort may play a role, I suspect that this is not the major factor in explaining the practice, and that it is not the most important factor in explaining the legitimacy of the practice. Rather, in my view, it is a somewhat inchoate awareness of the weak epistemic standing of our opinions on disputed questions which both motivates and justifies the practice.
[xxii]Op. cit.
[xxiii]This is an important qualification, as Elga notes. On Elga’s view, disagreement on these matters with those who see things largely in the way we ourselves do can serve as an important check on our own opinions. As Elga notes, this, by itself, if taken to heart, would force important revisions in our bodies of belief. It would thus be wrong to paint Elga as some sort of quietist. He is merely trying to resist the very sweeping withholding of belief on controversial matters which I am defending here.