This is an age of increasingly widespread knowledge, of great diversity in belief, and of sharp differences in judgment. The wisest course for those of us aware of this pattern is humility about our own correctitude and tolerance of disagreement from civilized disputants. The contemporary explosion of information – including misinformation – makes intellectual responsibility more needed than ever. But we also need a mean between a dogmatic self-assurance and a disabling fear of conviction. A central question for this paper is what standards should guide an intellectually responsible person in seeking this mean.

A prior question is what constitutes intellectual responsibility. Like the related notion of intellectual virtue, this concept needs explication. One problem arises because beliefs are neither actions nor under direct voluntary control. How to understand responsibilities that fit neither description is a challenge.Footnote ¹ Another problem is how to distinguish between intellectual responsibility for what we believe and moral responsibility for conduct aimed at maintaining good evidential standards. The former is chiefly a kind of liability to criticism, and its focus is typically present beliefs. The latter is a kind of obligation, and its focus is typically prospective beliefs. A third problem is how to square intellectual responsibility in retaining our views with the realization that someone we respect disagrees with us. A central case is our believing propositions which we know – or have good reason to think – are disbelieved by people whose intellectual credentials concerning the relevant matter are as good as our own.

To hold that intellectual responsibility is not reducible to moral responsibility does not commit one to holding that there is no ethics of belief, in the sense of a set of ethical standards governing actions having a certain kind of connection with belief. These actions include critically reflecting on certain propositions one either believes or is considering. Such belief-related actions can greatly influence cognition. They may, for instance, strengthen conviction by revealing new evidence for a belief one holds. They may also weaken conviction by contrary revelations. Reflection, then, like acquisition of new evidence, is a way by which we may exercise some degree of indirect control of our beliefs and other cognitions. This paper will first lay out some main elements needed for a sound ethics of belief and, on that basis, propose a wide-ranging group of principles for rationally responding to disagreements we have with those who deserve our intellectual respect.

1. Intellectual Responsibility

I begin with some assumptions that are unlikely to be controversial. First, consider doxastic control – roughly, control over whether or not we come to believe some proposition, p, continue to believe it, or cease to believe it. There are two possible forms: direct and indirect control. There are also two possible cases of each, negative and positive. The former distinction turns on the difference between what is achievable at will (directly, in a sense) and what is achievable only indirectly: by doing something else. For both of these kinds of control, direct and indirect, there is the positive power to generate or sustain a belief and the negative power to resist forming a belief or to eliminate a belief we hold. Suppose, for instance, that I discover that I have been caused to believe p by manipulation of my brain and I can think of no ground for the belief (or its negation). I should try to eliminate this apparently groundless belief. This holds whether or not I am aware of the manipulation: if, on careful consideration, p does not seem plausible to me either in virtue of its content or on the basis of some ground I have for it, say a memory impression, I should try to suspend judgment on p.

Many of us, upon finding only an inadequate basis (or what we take to be such) for something we believe, do cease to believe it, and this response is commonly “automatic”. Where it is not automatic, however, we can act to bring it about, say by concentrating on why our basis for believing it is evidentially inadequate. Many of us also automatically resist believing something – or at least our cognitive system resists belief-formation – when there is (or we take it that there is) inadequate ground for it; and some of us, especially those who are constitutionally skeptical, can, apparently at will, resist believing certain kinds of propositions even when there is good ground for them, or even strong evidence that we nonetheless consider inconclusive.

I assume here that we do not (and arguably cannot) have direct control of belief. There is little reason to doubt, however, that through what I shall call evidential conduct, we can have some degree of indirect control of belief formation and belief elimination. What is controversial is how this point applies to different kinds of intellectually and ethically significant cases. In what follows, a number of applications of the point will be described and appraised.

For understanding the ethics of belief, it is important to see that indirect control of belief formation – even just indirect negative control – suffices to ground some measure of doxastic responsibility, which is an important kind of intellectual responsibility. Here I use this term in the broad sense of liability to criticism either on account of what one believes (which may be an obviously absurd proposition) or on account of the basis or causal element(s) on which one believes it. The first case might be called content responsibility; the second – which we might call genetic responsibility – divides into (at least) responsibility for believing on a certain set of grounds, for instance evidentially inadequate grounds, and responsibility for believing as a result of some non-evidential cause, such as wishful thinking, that does not constitute a ground at all.

Take a different example. Imagine that, with the aim of coming to believe something I now do not believe, I expose myself to influential friends who I know will cause me to believe p without my having (or their providing) any ground for it. Then my believing it, as ultimately a result of my subjecting myself to their influence for this purpose, is such that I have (genetic) doxastic responsibility for my believing p. Given that we should try to avoid such groundless formation of our beliefs, I am prima facie culpable on that count, even if p is true and evidence for it exists.Footnote ² My doxastic responsibility here is my liability to criticism; my culpability consists in meriting criticism both for something I have done (exposing myself to the doxastically influential friends) and for holding a belief without any suitable ground for it. The point has two sides: I am in a state that reflects a kind of wrong-doing, the kind of conduct that led to acquiring the belief; and second, I hold a belief that is epistemically ill-grounded.

Intellectual responsibility is not a monolithic notion. It is multidimensional. I have so far considered two basic kinds of intellectual responsibility as liability to criticism, content and genetic responsibility for belief. But ‘intellectual responsibility’ can also be used prospectively, to express future-directed intellectual obligations. There are at least five dimensions in which a rational, intellectually responsible person – roughly, a person of intellectual virtue – tends to monitor belief-formation and cultivate good patterns of cognitive response to experience. These five dimensions of responsibility also bear on how we should responsibly treat rational disagreement with our views. Let us take the five in turn.

1. Seeking evidence. We should in many instances seek, in a balanced way, evidence and counterevidence regarding propositions that we either believe or are considering as candidates for belief. This is not something we need to do, or even can do, for every belief. But some beliefs are especially important. Indeed, disagreement with someone we respect – or indeed even with a well-described hypothetical view we respect – is an occasion for such a search. Some of our beliefs are also credibly challenged by actual non-personal situations, such as certain discoveries of disconfirming evidence; and at least for important beliefs, credible challenges should be taken seriously.Footnote ³ Some categories of people, moreover – for instance philosophers – tend to have a greater obligation to seek evidence and counterevidence than is usual.

2. Reflective equilibrium. We should seek a reflective equilibrium – a kind of integrated outlook that contains minimal internal tensions – both in our overall view of the world and in important aspects of that view. In part for this reason, we should periodically review our outlook on important matters. Some of these matters are ethical, some political, some personal, and some, of course, intellectual.

3. Focus on grounds of belief. Even apart from direct voluntary control over our grounds, identifying and focusing on grounds with the aim of clearly assessing them is salutary. Focusing on grounds may result in automatic cognitive adjustments that enable our intellectual virtue to manifest itself. We may, for example, see by reflection that a ground is not clear (as with a highly vague proposition) or not supportive of p to the extent we have presupposed. It is useful in any case to identify our actual grounds for believing. We can then explain our views better and are less likely to miss some basis of belief (evidential or causal or both) or to rationalize by seeking reasons that we find plausible but use mainly to camouflage the basis on which we actually believe and on which the justifiedness of our belief in fact rests. We may then adduce considerations that we do not even accept or, even when we do accept them, on which our belief is not based. We should know what ground we are really standing on, lest we unwittingly stand on shaky foundations.

4. Interpersonal comparison. With grounds for believing, as with beliefs themselves, interpersonal comparison is often desirable. Particularly on realist assumptions, this is a good idea. We share the same world. But, even given our collective efforts to find truth, it is often elusive. Someone else may have evidence we lack or may be less biased about evidence we share. And suppose we agree with someone else on the truth of p. Our agreement is likely to be more stable, and our communication better, if we also agree on at least some of the grounds for p and, if possible, on how strong they are.

5. Achieving proportionality in conviction. As natural as it may be for rational persons to have a degree of conviction appropriate to their grounds for belief, the degree of their conviction may be disproportional to the normative strength of our grounds. Just as one may believe p with unjustified confidence, as where one is justified in believing it only in a circumspect and tentative way, one may believe it with undue diffidence. Since such epistemic facts about ourselves may well be determinable, we may be justly criticized if we do nothing to rectify these disproportions. A parallel point holds for the case in which a ground is in some way underestimated and merits not reduced influence, but an enhanced role.

In many cases of the kinds we have been considering, the intellectually virtuous may form beliefs about the status of their grounds, or of their beliefs, or of the relation between the former and the latter. Much intellectual virtue does not require forming such higher-order beliefs. But the self-scrutiny that goes with identifying and appraising grounds for one's beliefs encourages and sometimes automatically produces higher-order beliefs. A good measure of unselfconscious intellectual virtue seems possible without higher-order beliefs; but for most people who are self-critical, forming them is natural. Some people may be blessed with an ability to assess grounds and propositions and to respond appropriately without forming beliefs about their own beliefs; but forming such beliefs and beliefs about our grounds is common, and in a world in which our beliefs are often challenged we usually cannot help forming higher-order beliefs and beliefs about our grounds for holding them. In many cases, not forming some such beliefs is reprehensible.

It should be clear that these normative points do not presuppose our having direct control, even direct negative control, either over our beliefs or over whether they actually rest on grounds we have for them. But we surely have some indirect control over our grounds-responsiveness: our tendency to respond cognitively and with the right level of conviction to grounds for belief and the right sensitivity to other relevant grounds accessible to us. Most people can enhance their grounds-responsiveness; and intellectual virtue commonly requires regular efforts in that direction.Footnote ⁴

2. Intellectual Responsibility, Moral Obligation, and Creditworthiness

Once intellectual responsibility is seen in the light so far shed on it, there should be no doubt that it is not simply derivative from moral responsibility. One might be intellectually reproachable for being uncritical in forming a belief even on a minor matter and where neither that nor the belief itself has any significant connection with any moral responsibility. Even in a solitary world I could be intellectually criticizable for cognitive error. This is not to say that there are culpable intellectual failings that cannot have a connection with moral responsibility; the point here is that the necessary connection required by the derivability claim does not hold.

Nothing said so far requires denying that intellectual responsibility applies directly only to behavior, whether covert, as with putting relevant evidence out of mind, or overt, as with silencing someone who is arguing against us. The point is that what is criticizable in such behavior may be intellectual rather than moral, aesthetic, prudential, or otherwise based on a non-epistemic perspective. The responsibility, in being directed toward action, is still broadly practical. There is no need to deny that overall responsibility, for an intellectually culpable act is greater where, because of the act, some further action – especially wrongful action – might be produced or sustained or even simply strengthened in the intensity of its motivation. That responsibility can be simply intellectual does not imply that there cannot be a coalition of intellectual and moral forces converging in support of either criticism or approval of our holding a belief given our grounds for it. Let us proceed to specific questions about the relation of belief to acts. For brevity I will concentrate on justification, though other normative notions may figure in our discussion.

Although intellectual responsibility does not reduce to any other kind, it is certainly connected with other kinds. The most important case here is responsibility for action on which the belief(s) in question bear. If, from either the moral or the prudential point of view, an action is important, there may be practical reasons, whether moral or prudential or some of each kind, to have especially good justification for a belief that is or may be a basis for the action. Other things equal, there is a proportionality here: the greater the importance of the action, the more reason there is to have adequately supportive grounds for a belief on which it is based.

So far, I've not emphasized something implicit: that justification for believing something (which does not entail actually believing it), i.e. propositional justification, and, by contrast, justified believing of something, i.e. doxastic justification, are subject to different standards. The latter implies the former but not conversely. Moreover, it is only upon beliefs and other causally efficacious elements that we act. This is a main reason why there is typically a need for justified belief as a basis of morally justified action, or even for action that is simply rational.

The distinction between doxastic and propositional justification is also connected with intellectual virtue and hence with living up to a sound ethics of belief in general. Virtue requires tending not just to do the right kinds of things but also having a settled tendency to do them for reasons of the right kinds. Here as elsewhere, practical reason is significantly analogous to theoretical reason, and the comparison is clarifying. Some of the questions we are raising for beliefs can also be raised for actions, and the answers should be parallel.Footnote ⁵

Kant's ethics is a case in point. In the Groundwork he held that an action not performed ‘from duty’ has no moral worth and that when we have a motive of inclination supporting an action that is also a duty, we cannot tell whether we acted from (a motive of) duty.Footnote ⁶ He apparently does not address the question whether, at will, we can prevent a motive of the wrong kind from serving as a motivating basis of action, and it is reasonable to take him to be committed, if not to claiming that we cannot, at will, unharness such motives, then to holding that we are unable to know that unharnessing has occurred even if it has.

This paper generally supports the epistemological counterpart of Kant's position on our power over the connection between our actions and our grounds for them, but here I must leave to the reader the formulation of the implications for practical reason of what emerges concerning theoretical reason.Footnote ⁷ I take it, however, that we should consider preservation of the parallel to practical reason to be a constraint in this sense: if the practical analogue of a view about our power over grounds of belief is not plausible, this is some reason to doubt that view for the domain of theoretical reason.

One clear parallel to be preserved is this: just as an action that takes time, such as assisting someone in preparing a meal, can be performed for one set of reasons at one time and for another set at a different time, a belief possessed over time can be held on one set of grounds at one time and on another set at a different time. The role of a given set of grounds, moreover, can intensify or diminish as the action (or activity) proceeds. Much the same holds in the cognitive realm. One might believe p initially on the basis of wishful thinking and later find evidence for it that becomes a basis on which one holds the belief. Similarly, one could have evidence for p but continue to believe it on a non-evidential basis. The justifiability of my belief that p on the basis of evidence I have does not entail the justifiedness of my holding it. It sometimes happens, moreover, that we initially believe p on an inadequate basis, such as unsubstantiated testimony from a person with influence over us, yet find good evidence for the proposition later and believe it mainly or wholly on that basis from then on. In what follows, I make the simplifying assumption that we are speaking of the intellectual responsibility of persons at a given time.

There is still another important analogy between the practical and theoretical realms. Just as one can be morally reprehensible for intending to do something that is wrong even if one does not do it, one can be intellectually reprehensible for tending to believe something that, given one's overall evidence, one ought not to believe. This is common where biases or non-evidential influences affect us; but it may also occur where, for other reasons, we tend to believe under the influence of evidence that only seems sufficient to justify believing the proposition in question and is in fact inadequate. In both cases, one is inclining toward something reprehensible – action or belief – and, depending on what reflections one has engaged in or failed to engage in, may already merit criticism. Intellectual virtue, then, and the avoidance of intellectual error, depend on doing what one can to contribute to avoiding error and, in certain matters, to reaching true belief. We can be reprehensible for being on a path to a particular ill-grounded action or belief, even if we do not cross the threshold. Some of what one can and should do, then, concerns finding good grounds, resisting tempting ones that are inadequate, and bringing it about that what one believes tends to rest on good grounds.

3. The Challenge of Intellectual Diversity

That there is disagreement in moral matters is a prominent fact of contemporary life. Something similar holds for religious belief, especially if we consider the difference between theists and various agnostics and atheists. Given such substantial and often divisive disagreement on central questions in human life, it is important that we have a good account of how a rational, intellectually responsible person deals with disagreement.

It should be noted, however, that even if intellectual responsibility requires a rational cognitive response to certain kinds of disagreement, action that is called for by cognitions such as judgments of one's own obligation need not be supported only by those cognitions. Consider beneficent deeds. These may be supported by both religious beliefs and by moral convictions whose justifiability does not depend on a religious basis.Footnote ⁸ Thus, defeat of one's justification for one set of beliefs does not entail defeat of justification for an independent set. This is one reason why rational, intellectually responsible persons may – especially in the face of disagreement with people they find generally credible – seek more than one justification for actions as well as for beliefs themselves. Thus, suppose I encounter disagreement from a peer, in the epistemic sense of someone who, on the matter at hand, is comparably credible (in a sense to be clarified below). Even if this reduces my justification for an action-supporting belief below the threshold appropriate to its justifying the action in question, I may still have another set of beliefs that, brought to bear in ways I have yet to discover, can raise my justification to its previous level. Indeed, a morally responsible agent should foresee this possibility and seek to be supported by such justificatory plenitude, especially if the action would otherwise be seriously unjustified.

Although some ethical and religious beliefs may be unique to their holders, it is uncommon – particularly for general moral beliefs and especially for general religious beliefs – to be uniquely held as opposed to being objects of agreement in a socially constituted group. May intellectually responsible agents who encounter peer disagreement take comfort in the numbers on their side? The question here is that of the epistemic relevance of a kind of social support. One response to this question is the doxastic practice approach proposed by Alston.Footnote ⁹ That approach is consistent with the one offered here but quite different. My approach is to give weight to both social and evidential diversity. Two principles are suggested by reflection on these two cases.

The first principle can be seen by noting how we often enhance evidential diversity by using one sense to support a deliverance from another. Hearing a woody, cracking sound outside, I may want to confirm by sight that a branch has fallen in the wind. This diversity of evidence seems, other things equal, to yield even better justification than adding, to my own auditory evidence, another person's testimony confirming that person's hearing the same sound. The visual evidence adds more to my justification than simply, as it were, testimonially replaying my auditory evidence from outside.

To be sure, visual evidence in this kind of situation may be better in any case; but, supposing that is so, my point can be illustrated by equalizing the weight from each sense by making the auditory evidence clear and the visual evidence dimmed by fog. Adding the latter still seems more desirable than duplicating the former. An epistemic principle suggested here is

As my examples show, diversity may be of at least two kinds: intrasource, as where we listen a second time; and intersource, as where we add visual to auditory evidence. This principle is of course consistent with the possibility that one kind of ground is best or some kinds are better than others, whether in general or (more likely) for certain kinds of propositions.

I presuppose that testimony can provide evidence. It should be added that, even though testimonial evidential strength does not grow in direct proportion to the increase in the number of concurring attesters, it does tend to grow with an increase in their number, and the more so given differences in their points of view, which in turn imply an element of diversity. To make this point fully relevant to dealing with apparent disagreement from peers, we should add that epistemologically, discovering, without another person's communicating to one, that this person agrees, or disagrees, with one regarding p, tends to be as confirmatory or as disconfirmatory, as the person's actually attesting to p or to not-p, in one's presence. This suggests what I shall call

A number of clarifications are needed here. First, I take responding to the question whether p broadly: forming the belief that p suffices, and for this no questioning of p or expression of the belief that p is required, though in the typical cases the person will at least momentarily have the thought that p. Second, diversity tends to be more important than number, though there is no formula for comparing the weights of these two variables, and increments in either tend to be justificatorily less significant as the number of responders increases. Third, other things are not equal if – as often happens – there is an artificial, evidentially irrelevant influence on the responders, such as brain manipulation. Fourth, the greater the intellectual independence of the responders, the more bearing their disagreement has on the strength of one's grounds. Mere numbers of concurring voices do not enhance our justification.

This recognition of the value of intellectual independence does not, however, imply that the testimonial sources must be independent in the sense implying that none is influenced by the others. Even if one person's testimony or persuasive argumentation is the source of the view held by all the attesters, it may well be that each is sufficiently critical to make acceptance of that testimony or argument confirmatory. A crucial distinction here is between independent judgment regarding a range of propositions (including p), in the sense entailing the judgment's responding to evidentially independent grounds for the proposition(s) in question, and independence of judgment regarding that range of propositions, in the sense entailing a kind of critical capacity concerning this range. The latter does not entail the former.Footnote ¹⁰ You may believe p because I have assured you of it, but in accepting my testimony (or argument) you might still have been critically constrained by standards and cognitions whose permitting your acceptance gives additional weight to the overall case for p. These constraints may be quite rigorous. Like filters, they may vary greatly in what they allow to pass through, and even some very discriminating ones may allow certain things to pass through quickly. For this kind of reason, the existence of a belief may, at least for a recipient of testimony expressing it, lend more support to the proposition believed than is provided by the believer's grounds alone.

A third point is that some people would not express disagreement with one unless they were certain of p, or indeed had what they consider overwhelming evidence for it. If such people do express their disagreement, other things would not be equal, and their affirmation might outweigh two or more concurring parties. Similarly, if you know that certain people respect your judgment in such a way as not to disagree with you without having very strong evidence, that might give additional force to your simply knowing that they disagree (where they do not express the disagreement to you). It might give still greater force to their actually attesting to the contrary of what you believe. But an intellectually responsible person will tend to be concerned about peer disagreement whether it is expressed or not and will consider not just numbers of people who agree or disagree but also their credibility.

4. Some Guiding Principles for Responding to Peer Disagreement

We have now seen some of the dimensions of intellectual responsibility and some important standards that determine its extent. But in a world in which fast and detailed communications give great prominence to disagreements on important matters, philosophers must explore what principles should guide our response to the challenges to our beliefs which come from apparently well-grounded opposing judgments. Consider the difficult case of disagreement with a colleague one greatly respects. What if, regarding such a disagreement, I believe that the colleague is an epistemic peer: roughly, that (a) the colleague is as rational and as thoughtful as I (in the relevant matter, including the assessment of whether p is true), (b) has considered the same relevant evidence, and (c) has done so equally conscientiously?Footnote ¹¹ This should prevent my concluding, and rules out my reasonably concluding, that my colleague is irrational in holding a contrary view. It should prevent, and would render unjustified, dogmatism about my own correctness in believing p.Footnote ¹²

The description portrays a case of manifest disagreement: an opposing view has been expressed (orally or otherwise) to one of the disagreeing parties by another of them. There is also dispositional disagreement: an unexpressed disparity in belief or judgment, the kind of which one might say, having read two authors: ‘Though they've never discussed the topic, they surely disagree on major points.’ Awareness that someone disagrees with us dispositionally should, other things equal, have the same justificational significance for us as awareness of a manifest disagreement under the same conditions. It is an interesting question (that I cannot here pursue) whether the same holds of a hypothetical disagreement, the kind one has with an imagined opponent.

My description of peer disagreement is also applicable at a definite time: to synchronic as opposed to diachronic disagreements. Where disagreement persists over time, as it commonly does, the status of either party or the kind or amount of evidence may also change. Our basic concern should be synchronic disagreement, since we must understand that in order to understand how to evaluate changes that a person's justification may undergo over time.

Examination of apparent peer disagreements shows, however, that it takes much information, and may require extensive reflection, to be justified in believing (a), (b), and (c), i.e. that the disputant is as rational and as thoughtful as oneself in the relevant matter and has considered the same relevant evidence equally conscientiously. The breadth, complexity, and quantity of evidence needed about the other person are great, and error in assessing it is difficult to avoid. Moreover, there are several other variables relevant to one's justification in the face of apparent peer disagreement.

One variable is potentially influential background theories a disputant may hold, such as skeptical ones or ethical theories with myriad implications for the kinds of cases in dispute. If I hold a theory that predicts p, then (other things equal) I will be more likely than a neutral party to believe p given the same body of relevant evidence that supports it less than conclusively. Skepticism, by contrast, enhances the tendency to withhold belief given grounds that are not, in the person's eyes, conclusive. If we call the usual kinds of disagreement, in which one person believes p and the other a contrary of it (in the broad sense including its negation), direct (or contrarian) disagreement, the case in which one believes p and the other withholds it might be called indirect (or oblique) disagreement.

Another variable, not always sharply distinguishable from the first, is background beliefs. A belief may have a theory as its content or may be of a broadly theoretical kind, but not all background beliefs relevant here are of either kind. I include prejudicial beliefs that may not surface in the discussion or in the reflections of either party, say that poor people are more likely to steal than prosperous ones or that educated people are more reliable in moral matters than those with only eighth grade schooling. Loyalties may also non-evidentially influence our beliefs, as may other non-cognitive (or not wholly cognitive) elements. Someone with a strong loyalty to a person or cause may resist more than a neutral person believing propositions that seem to put the person or cause in an unfavorable light. The kinds of psychological influences in question may reduce or even wholly undermine the justifiedness of a person's belief.

Still another variable is conceptions, especially normative ones, that may affect assessment of an issue without even entering consciousness. Consider the effects of reading ethically significant books, such as some of Freud's, or powerful novels, such as Crime and Punishment. These may affect our conceptions of emotion and motivation, of causes of human behavior, of guilt, forgiveness, and many other elements prominent in human life. Our conception of, say, crime, will be connected with various of our beliefs; but conceptions of something may also affect our perspective – especially instances of seeing as – say seeing crime as angry or greedy, even if they yield only a disposition to believe it to have the relevant properties rather than an actual belief that it does.

Thus, two people can be equally rational and consider the same evidence for p equally conscientiously, but differ in the background cognitions and conceptions they bring to the assessment of that evidence. Such background elements may also include (as suggested above) religious convictions or theoretical commitments, say to skepticism or epistemological coherentism. Sometimes background factors of these kinds are included in ‘total evidence’. Then differences in them might imply that the disputants are not peers in the matter; but I use ‘evidence’ more specifically to designate considerations that objectively support the proposition in question. Their actual effect on belief may, however, sometimes be causal rather than justificatory.

In addition to the difficulty – especially in complex cases – of acquiring justification for believing someone else to have considered the same relevant evidence equally conscientiously, to be equally rational in the matter, and to be free of background cognitions and other psychological influences that reduce the person's overall justification regarding p, there are at least three further factors. One is that someone else's disbelieving p is itself a reason (even if not a strong one), for a person who rationally believes p, to doubt that the other is a full-scale epistemic peer in the matter.Footnote ¹³ The second factor is that we are better positioned to make a critical appraisal of our own evidence – at least where it is experiential, as with memory impressions and intuitions – and of our responses to it, for instance in assessing whether our belief that p is based on the evidence rather than on, say, wishful thinking, than of anyone else's evidence or responses to that evidence.Footnote ¹⁴ Other things equal, then, we are better justified in our assessment of our own basis for believing p and of our response to that basis than in our assessment of the basis of anyone else's believing it or of anyone else's response to that basis. Third, as we check and re-check our own grounds for a justified belief that p and our responses to them, we tend to increase our justification for believing p, at least where we retain that belief in the light of this effort. Indeed, even if, from a skeptical disposition, we do not retain it, our propositional justification regarding p may still rise, as where we come closer to having conclusive evidence but, being highly cautious regarding p, withhold belief from it.

These considerations provide some support for the modest conclusion that the very exercise of critically seeking to establish the epistemic parity of a disputant may give a rational person a justificatory advantage in the dispute. Perhaps we may conclude that other things equal, a rational conscientious attempt to establish the epistemic parity of a disputant tends to favor the conscientious inquirer who, on the basis of such an attempt, retains the belief that p. This conclusion, qualified as it is by a tendency element, does not entail that such conscientious inquirers should never withhold p instead. But where they should not withhold it, retention with humility might well be a commonly justified result of the kind of reflection in question. Such retention, moreover, should normally be accompanied by a sensitivity to, and in some cases a search for, further evidence concerning p.

At least two further qualifications are pertinent here. First, each of two disagreeing peers can go through the same kind of conscientious review of the apparent parity of the other, in which case the enhancement in justification may be equivalent on each side. Second, despite the sometimes very great value of such a review, repeating it should not be expected to increase one's justification for p indefinitely and will tend to do so decreasingly with each repetition. Given these qualifications, it should be stressed that the kind of increase in justification that such critical reflection may bring does not automatically warrant reduced humility and may indeed call for more of it.

The difficulty of determining whether a disputant is indeed an epistemic peer, together with the first of the three further factors noted – the justificatory weight of rationally believing p – bear on the degree of justification one should take the disputant to have for believing p. The second and third factors pertinent to justification bear on whether one's own and the disputant's beliefs are suitably responsive to (and based on) the evidential grounds that support the conflicting proposition. We should be seriously concerned with both aspects of justification: whether our grounds are adequately strong and whether our belief that p is properly responsive to them and not improperly influenced by non-evidential factors.

It might seem that the threat peer disagreement poses to one's conviction comes only from the conflicting grounds that can be provided by the peer, but surely the fact that an apparent peer has responded to those grounds with a belief incompatible with ours is a further concern. That a rational person – the only kind in question, though the notion of intellectual parity may apply to certain irrational people – forms a belief that p on the basis of a conscientious consideration of relevant evidence is some reason to think that the evidence adequately supports p; and if the evidence is genuine and not misleading, it is some reason to think that p is true. This point is particularly important given that parity alone does not imply that the evidence each has is enough. But a rational person's believing on the basis of a body of evidence is some reason to think it is evidentially adequate. Evidential adequacy cannot be explicated here, but two points should add clarity. First, adequate evidence for p is a kind such that it is reasonable to believe p on the basis of it; second, such evidence is normally knowledge-sufficing, in the sense that a true belief held on that basis constitutes – apart from a fourth condition problem such as we find with accidentality and other ‘fourth condition cases’ – knowledge.

We can now see some of the reasons why intellectual responsibility carries a prima facie obligation to be justified in believing that one has adequate evidence for believing that p, especially when p is challenged by rational disagreement. The more nearly adequate the evidence is to determine whether or not p is true, the more evidentially (hence epistemically) serious peer disagreement is, on the plausible assumption that epistemic peers should tend to agree. As the strength of the evidence approaches that determinative level, one is at least closer to a dead-end with no solution in sight for consensus. Here the ethics of belief suggests that where one has justification for believing the evidence is insufficient to determine whether p, one should seek more evidence if the matter is important. The ‘should’ here is ethical at least where the question whether p is itself ethical.

It will be apparent, then, that on my view, even where we do have good reason to consider a disputant an epistemic peer, skepticism about justification for the proposition in question may not be warranted. This is not to say that one may simply ignore apparent peer disagreement. Indeed, in many cases it is intellectually best – and possibly required by intellectual responsibility – to go in the direction of skepticism. This may occur in many ways. Some deserve mention here.

One possibility is retention of one's belief while adopting second-order skepticism. There are two main cases, depending on whether the higher-order question concerns whether one knows that p or whether one is justified in believing p. I might say, for instance, ‘I see your reasons for thinking him guilty; but, though I realize I don't have conclusive evidence for it, I still believe he's innocent.’ Given a justified belief that one differs with a peer regarding p, it is plausible to hold that one is not justified in believing one knows that p (though of course one may know it). Second, it seems almost as plausible to hold that one is not justified in believing one is justified in believing p.

A quite different response, and one that often bespeaks humility, is reduction in one's confidence that p. Beliefs are held with different degrees of confidence; and even if (as I maintain) we cannot in general control our degree of confidence at will, we can (and should) acquire and maintain a critical intellectual disposition that makes it natural for us to gain or lose conviction in accordance with our sense of how good the overall grounds for p are.

The points that have emerged suggest a number of normative principles concerning disagreement with someone who is an epistemic peer with respect to p. Most are epistemic, but the first is mainly practical. It is

Other things are not equal, for instance, where one must flatly assert that p to prevent an ill-considered suicide. This cautionary principle is especially plausible if knowledge is the ‘norm of assertion’, in a sense implying that there is (typically) a normative impropriety in asserting that p without knowing p. The principle does not, however, depend on that view.Footnote ¹⁵

A plausible principle in the same range is suggested by the point famously stressed by Hume among others. Noting that ‘there are all imaginable degrees of assurance, from the highest certainty to the lowest level of moral assurance,’ Hume continues: ‘A wise man, therefore, proportions his belief to the evidence.’Footnote ¹⁶ Peer disagreements bear on how one should view one's evidence, and certainly the wise will seek a proportionality of conviction to their degree of evidential justification. A guiding principle here might be

High confidence does not imply a tendency to ascribe any particular probability to p, but in calling confidence high I imply that it is strong enough to preclude any significant doubt that p, and it may go as far as psychological certainty. The behavioral implications of this principle prominently include these. First, given justification for believing that a peer disagrees, we have a prima facie obligation to try to keep our confidence level appropriately responsive to our appraisal of the disagreement. Second, we have a prima facie obligation to act with appropriate caution: roughly, to see to it that we have sufficient ground for any act we intend or are seriously considering that is justified only if p is true.

These two principles bear mainly (though not exclusively) on cognition, and even the one regarding assertion concerns a narrow range of actions. In both cases, other things are not equal if one can reasonably attribute the disagreement to some of the factors cited above – such as prejudices or background theories – other than determinants of parity.

There is also a behavioral principle supported by the considerations brought out in this paper:

This principle concerns both (a) the intellectual responsibility not to act ‘cognitively’ toward others – say by psychological or neural manipulation – on an inadequately justified belief and (b) the prima facie moral wrongness of both cognitive and behavioral coercion on such a basis. Here ‘inadequacy’ of justification is consistent with a fairly high degree of it, such as grounds implying p's being more likely than not-p. What should count as inadequacy is in part a matter of how important the issue is and what further evidence is accessible, but the principle is a useful guide even if there is no way to be absolutely precise about this aspect of its rationale.

All of the principles so far formulated concern disagreement with a single apparent epistemic peer. Two qualifications should be added. One concerns an increase in the number of apparent peers. The other concerns a kind of cognitive difference – including indirect or oblique disagreement – that (if examined in the literature so far) has received less attention than disagreement in the sense entailing the parties' believing contrary propositions. These qualifications can be largely captured by two principles.

Given that the number of peers is important, one principle here should be something like

To explain what has to be equal is no easy matter. Suppose the others in question merely follow oneself or, on the other hand, deferentially follow a single disputant – though one would hope no such person could be one's epistemic peer. Then their additional evidential importance approaches zero (depending on what critical faculties they have that might still operate regarding the relevant evidence). The issue for the numbers principle is the right kind of independence. A second point is suggested by this: the more conscientious the response of the apparent peers to the relevant evidence, and the less the influence of non-evidential factors on their belief regarding p, the more evidential weight their conviction carries. Third, as the number of disagreeing parties rises, there is a tendency toward decreasing marginal importance, so that the addition of another person tends to matter less. Fourth, even apart from this, the principle does not imply that mere number of dissenting voices is significant for justification.

A further qualification pertinent to the numbers principle and others that concern peer disagreement is quite different. It concerns the distinction between disagreement and disparity: a justificationally relevant difference in cognitions regarding p manifested in (possibly among other things) either (a) difference in degree of conviction or (b) belief on one side and withholding on the other, as with indirect disagreement, or (c) presence, on one side of the dispute but not the other, of one or more higher-order beliefs regarding the status of p, or the status of the person's belief that p. It is important to see that the third condition does not entail the second, though it may explain it: I might withhold judgment on p without believing that my grounds are insufficient but might also withhold it because I have come to believe they are insufficient. In either case, someone who believes p and rationally takes me to be a peer regarding p should be responsive to the disparity between our cognitions. If there is a principle warranted here, it is something like this:

One might well ask why higher-order beliefs matter. They need not if they are unjustified, but in the case of a peer we should not lightly think that they are. If justified, they represent a wider perspective and a kind of double check on the status of p (or of a belief that p) in relation to the evidence. What of third-order and still higher-order beliefs of this kind? I suggest that, with each rise in level, there tends to be an incremental though decreasing rise in the significance of such higher-order beliefs. Thus, a third-order justified belief that one's belief that one is justified in believing p is itself justified may add something to one's justification for believing p, but (other things equal) not as much as the second-order belief does. Metaphorically, the higher we build from the justificatory foundations, the greater the risk of error.

Something should also be said about the rationale for some of the principles I have proposed. On the plausible assumption of evidential determinacy – the view that a given body of evidence supports a proposition to a determinate degreeFootnote ¹⁷ – the level of conviction on the part of rational peers regarding p should tend to be the same as in one's own case (since they have the same relevant evidence regarding it). Given this implication, a rational person who believes p is likely to be disturbed upon finding that a peer has zero conviction regarding it. If we are properly responding to the same relevant evidence and that evidence has a determinate justificatory bearing on p, we should tend to agree on whether p is true.

The view that evidence for a proposition does have a determinate justificatory bearing on it is, however, not essential in this paper. The view is even more plausible, however, if such determinacy does not require there being an exact numerical probability; but suppose it does. This apparently does not entail the narrow conclusion that only one cognitive attitude is justified by such a body of evidence. In addition to the possibility that a rational person may be justifiably unable to discover the relevant probability, there is the special case in which p and not-p each have a probability of ½. One rational response is that p is at least as likely as not-p; and a different rational response, intuitively less optimistic about p, is that not-p is at least as likely as p. To be sure, a proposition with this status in one's cognitive purview may not justify any significant action that proposition may call for, but this is a different and largely non-epistemological matter.

What the evidential determinacy assumption highlights is that the frequency of our disagreements when we are highly rational and work from the same relevant evidence apparently indicates the magnitude of non-evidential influences on cognition. This point leads to another perspective on disagreement. Suppose we add to the evidential determinacy assumption a consequentiality assumption: the plausible idea that a person's justification (doxastic as well as propositional) is determined by (consequential on) psychological facts about the person. Then, other things equal, the closer my peer is to being my psychological duplicate, the smaller the probability of any cognitive disparity between us. At the limit – qualitative psychological identity – the probability is zero. These considerations suggest

One reason for the ceteris paribus clause is that not all psychological similarities and differences are relevant to every case of justification. My recollections of paintings at the National Gallery may have no evidential bearing on my views in astronomy, so my peer's lacking such recollections is epistemically irrelevant to our disagreement on a matter of astronomy. But by and large, if someone has grown up in my own sociocultural group and been educated in all the same institutions and programs as I, then disagreement on certain topics should tend to concern me more than disagreement on those same topics on the part of someone of utterly different culture and education.