Introduction
Suppose that you and I are epistemic peers.Footnote 1 Of course, there may be some matters about which I have greater expertise, and vice versa, but in general we are equally rigorous, critically reflective and competent thinkers.Footnote 2 Now imagine that we disagree on some matter about which neither of us is particularly expert. Say that, on this matter, you believe that P, and I am convinced that not-P. Suppose further that this disagreement is genuine, such that our difference in opinion over the truth of P cannot be attributed to the fact that one of us lacks evidence that the other one has. Finally, suppose that we reveal to each other our doxastic attitudes towards P, and share our reasons, arguments and evidence. Assuming that neither of us has any independent reason to think that the other one is wrong in this case, that is, a reason which is independent of the disagreement at hand, what is the normative significance of our disagreement? As responsible epistemic agents, how, if at all, should we revise our respective beliefs? Once I discover that you believe that P should I, for instance, revise downward my confidence in my belief that not-P? Should you, likewise, revise downward your confidence in your belief that P? Perhaps, as some have suggested, absent any special reason to think that you are in a better position than I am to assess P, I ought to stick with my original doxastic commitment.Footnote 3 Or maybe, as yet others have argued, out of considerations of symmetry, we ought to give equal weight to each other’s views and both suspend judgment regarding the truth of P.Footnote 4
The arguments in favour of these otherwise divergent positions share an important assumption, which is that the normative significance of disagreement can be decided in a case like the one described above, that is, where there is a genuine disagreement with full disclosure between epistemic peers. In this paper, I argue against this assumption. A helpful way to understand the question of whether reasonable disagreement among peers is ever possible is to view it as a question about how to fulfill our duties as responsible epistemic agents, and the best way to answer this question, in turn, is to articulate the conditions for justified belief. As I see it, a theory of justification, as distinct from a theory of knowledge, sets out the standards for evaluating whether we have been suitably careful in forming our beliefs insofar as our overall aim is to have an accurate and consistent set of beliefs. In its barest form, justification involves having reasons to which one has access.Footnote 5 This internalist understanding of justification provides the rationale for a distinctly deontological conception of the notion, whereby to have a justified belief is to have done the best that one could do, epistemically speaking, insofar as from a first-person perspective one’s beliefs are supported by evidence. This internalist conception of justification gives us a handy interpretation of the question of the normative significance of disagreement between epistemic peers. Seen in this light, the question of whether I can responsibly maintain that not-P when someone whom I consider to be my epistemic peer maintains that P is just the question of whether, in such a case, I can justifiably believe that not-P while my peer justifiably believes that P.
In what follows, I argue, contra the evidentialist, that the answer to this question cannot be determined by looking at evidence alone. The scenario outlined above is missing an important variable; before we can evaluate the normative status of a belief, we need a further piece of information. In particular, in order to establish whether a subject S has a justified belief that P at time t we need to know the subject’s interest in P, and the corresponding level of risk she takes in accepting P as true. According to this interest-relative theory of justification, S has a justified belief that P at time t if and only if S’s evidence at time t supports P in proportion to S’s interest in P. On this view, when the risk a subject S takes at time t in accepting that P is true is high, the amount of evidence required in order for S to be justified in believing that P at time t goes up. The risk in question is thus best understood as an epistemic risk insofar as it arises for an individual qua knower, in the context of propositional assent as she considers, “What should I believe?”
In the first half of this paper I develop an argument in favour of this view which is motivated by what I take to be the fundamental problem with evidentialism, namely, that it fails to provide us with the tools necessary to apply its own epistemic norms. This argument gives us reason to accept an interest-relative theory of justification, independent of the issue of peer disagreement. That said, this theory has interesting consequences for the normative question about disagreement, which I will discuss in the second half of the paper. Because our epistemic evaluations of belief depend on more than evidence alone, a genuine disagreement with an epistemic peer in the case of full disclosure does not necessarily imply that one individual is being epistemically irresponsible, that is, has an unjustified belief; the difference in opinion might, instead, come down to a difference in epistemic risk.
Evidentialism
The interest-relative theory of justification which I develop here can be seen as a modified version of evidentialism, and we can motivate it by looking at what is wrong with standard versions of evidentialism, for example Feldman’s and Conee’s.Footnote 6 Evidentialism holds that evidence is the sole mark of a justified belief. On this view, S’s belief that P at time t is justified if and only if S’s evidence at t supports P and S believes that P on the basis of said evidence. More generally, the evidentialist holds that the right doxastic attitude regarding P (belief, disbelief, or suspension of judgment) is the one that fits the evidence, where evidence is understood as all the information relevant to the truth of P that a person has to go on, at the time.
Before looking at what is wrong with this view, let us first see its implications for the epistemology of disagreement. Because evidence is the sole mark of a justified belief and a matter of objective fact, given any proposition P and evidence E, there will be only one justified doxastic attitude about P licensed by E. Either the evidence supports P, or it supports not-P, or it fails to come down on either side. This is what Richard Feldman calls the ‘uniqueness thesis,’ the view that no body of evidence justifies more than one doxastic attitude (believing, disbelieving, suspension of judgment) on any given question. As he put it, “in cases of disagreement the evidence supports (at most) one side of the dispute and the party of the disagreement with the belief that is objectively supported by the evidence is justified and the other person is not”.Footnote 7 An important implication of this thesis is that there is no possibility for reasonable disagreement among epistemic peers, at least not after they have discussed the issue thoroughly and shared their evidence, reasons and arguments—a stage in a disagreement which Feldman calls “full disclosure.”Footnote 8
To see why this is so, take the case of epistemic peers Pro and Con. Suppose that Pro and Con have a disagreement on a subject about which neither is particularly expert, and let us say that this disagreement is genuine, such that, on the basis of common evidence, Pro affirms the proposition P, which Con denies. Suppose further that the evidence in fact supports P. Now, imagine Pro’s and Con’s epistemic situation as they consider their beliefs prior to learning what the other one believes. In isolation, according to the evidentialist, Pro’s belief is justified and Con’s is not, as their common evidence uniquely supports P. But the normative status of their respective beliefs changes once they learn about their disagreement. Because Con is Pro’s epistemic peer, Pro possesses no discernible epistemic advantage over Con. As such, absent any independent reason to think that Con might be wrong in this case—that is, a reason which is independent of the disagreement (say, for example, that Con is typically biased when it comes to matters relating to P)—the fact that Con believes not-P now gives Pro a good reason to think that she might be wrong about the truth of P, and vice versa. And what this means is that Pro ought to revise downward her confidence in her belief that P, and Con ought to do the same vis-à-vis not-P. In fact, according to the evidentialist, because Pro and Con are roughly equal with respect to intellectual rigour, background knowledge and reasoning powers, they ought to give equal weight to the opinion of the other as they would to their own opinion.Footnote 9 On this view, thus, the only justifiable doxastic state for them both is to suspend judgment regarding P, and this is the case even though P remains true.
The Problem with Evidentialism
At first glance, there is something compelling about the evidentialist principle of ‘equal weight,’ but as it turns out what is most attractive about this principle is also its fatal flaw, which is the presumption of symmetry that underlies it. Chipping away at this assumption will help to expose the fundamental problem with an evidentialist theory of justification, which is that it fails to provide us with the tools necessary to apply its theory’s own epistemic norms. I will explain precisely why this is the case shortly.
On the one hand, there is something perfectly reasonable about the idea that, if you and I are intellectual peers who have been privy to the same evidence on a matter about which neither of us is expert, then, in the case of a known disagreement, there should be no reason for me to privilege my conclusion over yours, and vice versa. The suggestion that, in such a case, all knowers be treated alike has the force of the unassailable principle of equality behind it, and thus carries a good deal of intuitive appeal. On the other hand, what makes this idea so persuasive is the thought that so long as these conditions are met, there is nothing that distinguishes you and me normatively. It is precisely this presumption that I want to challenge. It gets its purchase off the evidentialist claim that evidence is the sole determining factor for a justified belief, which means that two individuals with identical evidence regarding the truth of P are alike, justificationally.
To see what is wrong with this idea, first recall that the evidentialist restricts herself to two factors when evaluating the normative status of belief: a doxastic state—belief, disbelief, or suspension of judgment—and the evidence in favour of it. What this means is that doxastic attitudes are, in an important sense, detachable from the subjects who hold them; they are essentially interchangeable. If a proposition P is supported by evidence E, then, so long as S believes that P on the basis of E, S’s belief that P is justified, regardless of how important or trivial P is to S. If P is justified for you on the basis of E, in other words, then it will be justified equally for me or for anyone considering P on the basis of E.
Now, Conee and Feldman see this as a happy consequence of their view, but we ought to pause and appreciate precisely what is entailed by it. On this view, whether I have a keener interest than you in the truth of P is irrelevant, for instance, and thus so is the fact that I adjust my confidence in P appropriately, given my interests, as well as the care I demonstrate when I judge that my evidence regarding P is sufficient, or when I decide instead that I need to pursue a matter further. Indeed, on this view, any particular non-evidential facts about me, qua subject, are irrelevant. In this way evidentialism overlooks the importance of the subjective perspective when it comes to evaluating one’s evidence as reason for belief.
This may sound like a peculiar criticism against an internalist theory of justification like Conee’s and Feldman’s, which holds that only mental states and our phenomenal experiences can serve as evidence.Footnote 10 But it is important to understand that evidentialism privileges a certain kind of mental state experience—i.e., an evidential one—at the expense of others. The sort of mental state experiences that are relevant for evaluating a doxastic attitude toward a proposition P are just those experiences which are not uniquely mine, but rather those which are shared by any individual who is likewise considering evidence E.
The problem with this view is not that, in failing to capture properly the subjective perspective evidentialism fails to allow for a rich sense of our deontological notions of epistemic blameworthiness and praiseworthiness. That is surely true, but since Conee’s and Feldman’s evidentialism is not motivated by deontological concerns, this is hardly a fair criticism.Footnote 11 More to the point, the problem is that in failing to capture properly the subjective standpoint, evidentialism leaves us hamstrung in trying to apply its own epistemic norms. I suspect that we fail to notice this because of our habitual and mechanical-like tendency to fill in the assumptions needed, in any given case, to make the relevant judgments. However, without those assumptions, that is, by considering evidence and evidence alone, strict evidentialism leaves us unable to evaluate the justificatory status of a subject’s beliefs.
The evidentialist makes a faulty inference. He moves from the correct notion that evidence is the only relevant consideration when it comes to discerning the truth of P to the flawed one that evidence is the only relevant factor in determining whether one is justified in believing that P. And because evidence is the only pertinent factor when it comes to justification, on this view, there is no mechanism that we can use to determine how much evidence is sufficient from case-to-case. The evidentialist is hamstrung in applying the norm of evidence since evidence alone can never tell us whether a subject has enough evidence for justified belief.
The task of any theory of justification is to articulate the conditions under which a subject’s belief that P is justified, and these conditions are given by the epistemic norms that govern belief. It has been argued that these norms tell us that beliefs must cohere with other beliefs, or that beliefs must be drawn from foundational beliefs, or that beliefs must be acquired using a reliable belief-forming mechanism. But whichever the epistemic norm of favour, a theory of justification has to provide us with the tools needed to assess how well supported our own beliefs are, upon reflection, just as we ought to be able to evaluate the justificatory status of someone else’s beliefs.Footnote 12 If a theory of justification can’t enable us to do this, then it is failing at its most basic task. Again, this is precisely the problem with strict versions of evidentialism.
When we decide whether an individual’s belief is justified, we consider her reasons for belief in light of how important we take the proposition in question to be for her. We weigh the subject’s evidence in favour of the truth of P, in other words, in light of the subjective context in which her belief that P is held, that is, in light of her relative interest in the matter. Indeed, my point here is that we must do this, for without this information, we cannot determine whether a belief is living up to the standard set by the epistemic norm in question—in this case, the norm of evidence.
A true belief is true no matter how important it is, and yet how much evidence we need to support our beliefs varies from case to case. What accounts for this difference? Evidentialism, along with most contemporary theories of justification, does not require conclusive evidence for justification, so how much evidence is needed in any given case will be a matter of degree. But how do we determine this degree? Looking to the evidence itself will be of no use, since evidence alone cannot tell us how much evidence is sufficient for justification in any given case.Footnote 13 Rather, this evaluation will hinge on the relative importance of the belief in question for the subject, which is determined by the subject’s interest in a proposition. Again, this aspect of our normative assessment of belief is rarely articulated; it is part of the everydayness of normative assessments, the very fabric of epistemic evaluation. But that should not mislead us into thinking it plays an insignificant role, as we will see in the examples below. Our hopes and desires, our likes and dislikes, our interest in a matter—none of these mental state experiences are evidential and thus none count in favour of the truth of a proposition P, but, as the examples below illustrate, they function as tacit, background assumptions, and play an ineliminable role in the evaluation of the justificatory status of belief, both from a first-person and a third-person perspective.
Richard Feldman has dismissed this kind of consideration as irrelevant to the epistemic evaluation of belief, suggesting instead that one ought “to be guided by the evidence that one does have,”Footnote 14 but this suggestion misses the point. The advice is hollow. Evidence, on its own, is, strictly speaking, unable to guide us. Evidence cannot tell us when some matter is trivial to us or worth inquiring further over. And because evidence cannot tell us how much evidence is needed for justified belief, evidentialism, with its single-minded focus on evidence in assessing justification, lacks the tools necessary to apply its own epistemic norms. To see whether a subject has the appropriate amount of evidence for justified belief, we need to look at a subject’s interest in a proposition P, and the corresponding risk she takes in accepting P as true.
Interests and Epistemic Risk
By evaluating an individual’s evidence in favour of a proposition relative to how important that proposition is for her, the interest-relative theory of justification has the means to solve the problem facing strict versions of evidentialism. By providing us with a variable that we can employ to judge the amount of evidence needed in any particular case, this modified version of evidentialism has the tools necessary to apply evidentialism’s own epistemic norms, that is, the norm of evidence. According to the interest-relative theory of justification, S has a justified belief that P at time t if and only if S’s evidence at time t supports P in proportion to S’s interest in P. To say that a person has an interest in a proposition is just to say that the proposition in question matters to her.Footnote 15
Jason Stanley has recently articulated the concept of interest-relativity with a focus on practical interests,Footnote 16 but I prefer to think of interests as those factors that influence our actions or our general well-being, for in addition to having a practical stake in the truth of a proposition, I might care about its truth purely as a matter of emotional investment. I use the notion of ‘interests’ broadly, in other words, so as to include our practical interest in a proposition, that is, when it bears on an action that we might undertake, as well as our emotional interest, that is, when it bears on our happiness or on our general well-being.Footnote 17
The interest-relative theory of justification does a good job of filling the gap left by strict evidentialism and this gives us a good reason to accept it, but it does more than that. It offers a simple and plausible explanation of what’s going on in a particular kind of case of peer disagreement, namely, in those cases where the peers have a markedly different level of interest in the matter at hand. These kinds of cases expose a commonly shared intuition, which is that when the epistemic risk one takes in accepting a proposition is high for an individual, the evidential burden required for justification goes up.Footnote 18 The idea that certain subjective facts about an individual play an ineliminable role in the evaluation of the normative status of her beliefs goes a long way to explaining this intuition, and the interest-relative theory of justification makes room for it by allowing for a case wherein two individuals with the same evidence but conflicting beliefs can both be justified. And what this means is that the uniqueness thesis is false; two individuals with the same evidence and identical beliefs are not necessarily identical, justificationally.
Three Examples
To see how this works, let’s take an example where the epistemic risk one takes is raised for an individual because of practical reasons.
Tenure case: Suppose that you and I are, in general, epistemic peers, and, in particular, equally attentive colleagues, and that we are both present at a Department meeting in which the University’s Tenure and Promotion guidelines, which have been recently revised, are discussed. At the end of the meeting, at time t, you conclude that the number of refereed journal articles needed in order to get tenure has increased by one, whereas I believe that the number of articles has remained unchanged. At time t1, as we exit the meeting together, we discover that we disagree over how many articles are needed to get tenure. The question is: how, if it all, should we revise our original judgments?
Many find it obvious that I ought not to discount your opinion based simply on the fact that we disagree, that, all things being equal, the fact that a conclusion is reached in your head (and not mine) is immaterial, so that in such circumstances we both should become agnostics until further evidence becomes available. But suppose that we tweak the example a bit and note that you are a full professor who has come to the meeting more out of a sense of duty than concern, and that your view on the general fairness of the tenure document is not impacted by whether the number of articles required for tenure has been increased by one. I, on the other hand, am coming up for review in the next year, and this change would seriously impact my chances of getting tenure. Once we learn that I have a keener interest than you in the subject at hand the answer to the question of how, if at all, we should revise our original judgments, changes. And it changes not because this new information implies that I am more or less reliable than you, in this instance, or that I am biased in favour of a particular view, but rather because our varying interests in the subject means that we have varying evidential burdens in order to meet the epistemic standards set by a theory of justification. And this means that you and I are not in normatively symmetrical positions, not as we consider the matter in isolation, and not after there has been full disclosure between us, and this has two important implications. In the first place, the fact that, at time t, you affirm a proposition that I deny on the basis of common evidence does not imply that one of us has an unjustified belief. Second, the normative asymmetry runs its course through the disagreement: the different evidential burdens at time t remain present at time t1, as we discover that we disagree over the number of articles are needed to get tenure.
To be clear, while which belief the evidence favours might be an objective matter of fact, how much favouring is required in order for a belief to be justified depends on one’s interests. Returning to the example, let’s imagine that at the meeting in question, the source of information regarding the number of papers required for tenure was a casual remark by one of our colleagues. Given your relative indifference to the matter, the say-so of (yet) another epistemic peer would be sufficient to justify your belief that the number of papers required for tenure has gone up by one. And note that this is so even if your belief turns out to be false, that is, even if it turns out that you heard wrong, that, for whatever reason, your listening skills were less than reliable on that particular day, or that our colleague was joking and you failed to pick up on the humour in her comment. In either of these cases, your belief, though false, would still be justified, as you would have cleared the evidential standard set by your own interest in the matter.
Now imagine my doxastic situation at time t. I am equally as attentive as you, and I too heard our colleague comment that the number of papers required for tenure had been increased by one, but my immediate doxastic response was disbelief. Maybe I assumed that our colleague had been joking, having a little fun at my expense, or perhaps I figured that she must have been mistaken, otherwise, I would have reasoned, I most certainly would have heard about this change prior to the meeting. Either way, at time t, I reject proposition P, and rightly so, for while the say-so of a colleague provides some positive epistemic support in favour of P, given my vested interest in the matter, it is not enough to justify belief. Thus, at time t, I justifiably reject that P, while you justifiably believe that P.
Now consider our respective doxastic situations at time t1, after we learn of our disagreement. Once you discover that I believe that not-P your first thought, all things being equal, is to downgrade your confidence in your belief that P. But then you learn that my reasons for disbelief have to do with my extra evidential burden, given my practical stake in the matter, and this has some influence on how much weight you give to my opinion. After some consideration you decide, reasonably, to discount my view and stick to your guns. The fact that you believe that P makes me think twice about my initial rejection of it. That you took to heart the comment by our colleague suggests to me that I might have been too quick to dismiss it, and as a result I downgrade my confidence in my initial belief and suspend judgment regarding P until I can investigate the matter fuller. Thus, at time t1, after full disclosure, we have moved closer, doxastically speaking, but you and I still disagree over whether P is true, and again, both our doxastic states are justified.
What this example shows is that reasonable disagreement among epistemic peers, after full disclosure, is possible. Practical considerations cannot make a proposition true or false, nor can they count as evidence in favour of a proposition, but, contra the evidentialist, they can impact the current epistemic status of a belief insofar as they raise or lower the normative requirements for a justified belief.Footnote 19 So while the evidentialist is right that evidence is the only relevant consideration when it comes to discerning the truth of P, that does not mean that evidence regarding the truth of P is the only relevant consideration in determining whether one is justified in believing that P.
On the interest-relative theory of justification developed here, having a justified belief requires that we be responsive to evidence, but responding to evidence in the right way requires that we meet the evidential standard set by our relative interest in a matter. What this example shows is that, because interests—and thus evidential standards—are not uniform across people, there is no simple algorithm that we can employ in the case of a disagreement with an epistemic peer. Thus, the answer to the question of how much weight we give to the views of our epistemic peers in cases of genuine disagreement is that it depends. When our peer’s interest in a matter is roughly equal to ours, then the principle of equal weight is appropriate; but when our interests vary, as in the example above, then so do the normative requirements for justified belief.
When practical stakes raise the normative standards, individuals must exert more cognitive labour in order to meet the evidential burden for justified belief, and this is true not only when a proposition has some influence on a course of action which we undertake, but also when a proposition matters to us for emotional reasons, that is, when it bears on our general well-being. To illustrate this claim, I want to use one of Feldman’s examples. Feldman uses this example to demonstrate his view that no body of evidence justifies more than one doxastic attitude (believing, disbelieving, suspension of judgment) on any given matter—i.e., the uniqueness thesis. While the example, in its original form, does seem to support this thesis, we can see that with just a bit of tweaking it can be used instead to support an interest-relative theory of justification:
George and Gracie: Suppose you have two friends, George and Gracie, who each own a particular model and colour of car. You see a car of that kind arrive at your house but cannot see the driver. You have a good reason to think that George has just arrived, since George has a car of that kind. However, you have an equally good reason to think that Gracie has arrived (and suppose—unrealistically—that you are certain that no one else with that kind of car would arrive at your house).Footnote 20
Now, some might argue that it is equally reasonable to suppose that Gracie has arrived, or George (but, of course, not both), such that if you believe that it is Gracie and your spouse believes that it is George, this would be a case of reasonable disagreement—more specifically, a mutually recognized reasonable disagreement. But Feldman uses this example to draw a different conclusion, which is that, in cases like this one where the evidence is split, suspension of judgment is the only justifiable doxastic state. He argues that unlike action, with belief one always has the option of suspending judgment, even if one has a lot riding on the matter, even if, in other words, it matters a lot to us whether it’s George and not Gracie in the driveway.Footnote 21 According to Feldman, the option of suspending judgment is always a live one, even when the stakes are high.
I think this is partly right. It is true that, with respect to any particular proposition, the range of doxastic possibilities open to believers is not limited according to whether the stakes of believing are high or low. And Feldman is right that suspension of judgment is always an option. However, his claim that it is the only justifiable option is persuasive, in this example, because the evidence in this case is evenly split. But imagine that we tweak the example a bit. Suppose that there is a modest level of positive epistemic support in favour of P, the proposition that it is George in the driveway—say that you and your spouse both hear a man’s voice outside your front door. And suppose further that while you are indifferent to the truth of P, it matters a lot to your spouse, and not because it bears on any action that she may undertake, but rather because she finds George intolerably obnoxious, and she will be terribly disappointed if it is him and not Gracie popping in for a social visit. Thus, on the basis of your shared evidence, at time t, you believe that P, while your spouse suspends judgment regarding P, at least until she can go to the window and double check.
Your spouse’s hope that it is Gracie who has just arrived and not George does not make it more likely that it is Gracie in your driveway; that would be wishful thinking on her part, which is when our interest in a matter inappropriately lowers or wipes out our evidential burden. This is precisely the opposite of what is demanded by the interest-relative theory of justification, according to which our interest in a proposition, our wanting it to be true (or false), raises the evidential burden insofar as it raises the normative requirements for justified belief. In the George and Gracie example, because of the variation in epistemic risk, you and your spouse have different evidential burdens, and this is true even after there has been full disclosure between you, that is, even after you learn of your disagreement at time t1. A modest tip in the evidential balance in favour of P ought to be enough proof for you, but not for her, to have a justified belief that P.
Notice that, in this example, the increased evidential burden carries no practical consequences for your spouse. The risk she takes in accepting P as true is raised simply because she dislikes George, and not because it bears on any action that she might undertake. And note that the epistemic risk she takes, in this case, is not increased by much; your spouse’s well-being is effected by the truth of the proposition, to be sure, but not substantially. That said, we could imagine any number of factors that would further raise her interest and thus the epistemic risk that she takes in believing that P. Say, for instance, that your spouse would not only be disappointed by P but that it would put her in a terribly cranky mood, which would put a damper on your day as well. In this case, you too might have an increased interest in P, but more to the point, if your spouse knew that her bad mood would ruin your day as well, that might reasonably give her an even greater interest in P. But there is a limit on the impact of others’ beliefs on our interests. Because justification depends on our beliefs meeting the evidential standard set by our own interest in a matter, the fact that you are impacted by your spouse’s belief that P raises her interest in P only to the degree that she is aware of and cares about your feelings on the matter.
The epistemic risk one takes is liable to escalate even further when we move from a compromised social visit with a disagreeable acquaintance to matters closer to the heart. Here is another example that illustrates this point, and which also helps to show the key difference between interest-relativity and wishful thinking, referred to above.
The biased parent: Suppose that you and I are, in general, epistemic peers, and that we are both moderately capable at judging artistic potential. Imagine that we are both in attendance at your daughter’s dance recital; we have equally good seats for the show, yet I observe her to be in the bottom rung of performers, whereas, as far as you can see, she’s the best in the class.
Examples like this one are often used to illustrate bias. Although on the face of it what we have here is a genuine disagreement among epistemic peers, we might think that I ought to ‘stick to my guns’ and discount your view, and not because it is at odds with mine, but because I have an independent reason to think you might be wrong in this case, i.e., the well-documented tendency of parents to form biases in favour of their own children. This is a classic case of wishful thinking insofar as you have allowed your interest in the matter to lower your evidential burden: you believe that your daughter is the best in her class because you want it to be true that your daughter is the best in her class. Your belief about your daughter’s talent is based on wishful thinking and not evidence, and thus is unjustified. The interest-relative theory demands that believers take precisely the opposite normative stance. Because you happen to know that it is important for your daughter to do well at dance, not just because you truly hope that she succeeds in all her endeavours but because she has indicated that she wants to be a dancer, your interest in the matter of her artistic potential as a dancer is high. According to the interest-relative theory of justification, this means that your evidential burden regarding the quality of her performance should go up, not down.
Let us modify the example a bit and suppose that you and I both judge your daughter to be in the bottom rung of performers. Whereas my belief that P, i.e., your daughter is not a good dancer, is justified, a more appropriate doxastic attitude for you to take, according to the interest-relative theory of justification, would be to suspend judgment, at least until you gather more evidence, say by asking a few other parents what they thought of the performance, talking to the dance teacher and engaging in other evidence-gathering activities. Your interest in the matter is greater than mine, and that lands us in normatively asymmetrical positions. Whereas I can justifiably form an opinion of your daughter’s artistic talent based on a single observation of her abilities, your evidential burden about her talent must meet a stiffer evidential standard than mine, and that is the case whether you observe her to be in the top rung of performers, or in the bottom.
Some Clarifications
What these examples help to illustrate is that evidence is not the only relevant factor when it comes to evaluating the normative status of belief, whether this evaluation comes from a first-person or a third-person perspective. The evidentialist is right that having a justified belief requires that we be responsive to evidence, and the evidentialist is also right that evidence is the only relevant factor when it comes to discerning the truth of P. However, the evidentialist is wrong to suppose that this means that evidence regarding the truth of P is the only relevant consideration in determining whether one is justified in believing that P. Here is where interests come into play; responding to evidence in the right way requires that we meet the evidential standard set by our relative interest in a matter. Thus, when evaluating doxastic attitudes, we must measure a subject’s evidence for her belief that P relative to the level of risk she takes in accepting P as true.
Again, the risk in question is best understood as epistemic insofar as it arises in the context of propositional assent. In each of the examples above, the possibility of an undesirable practical or emotional state of affairs—the chance that I might not get tenure, or that it is George at the door instead of Gracie, or that your daughter is not a talented dancer—is what raises our interest in a matter. But the risk arises for us qua knowers, in the context of propositional assent as we consider, “What should I believe?” Is it true that the standards for tenure have risen? Has George arrived for a visit? Is it true that your daughter is a bad dancer?
Our increased interest in a proposition raises the risk we take in accepting that proposition as true, and as such it raises the requirements of the normative standards set by a theory of justification like evidentialism. The above cases of disagreement between epistemic peers help to illustrate that when those standards are raised in one case and not another, then two individuals with shared evidence are not in normatively symmetrical positions. This illustrates that the uniqueness thesis is false.
The interest-relative theory of justification helps bring to light something commonplace about the relationship between evidence and justification, which is that the amount of evidence needed in order to count as adequate support for a proposition is always a matter of degree, in cases of disagreement or otherwise. It is seldom the case that evidence comes down conclusively and irrefutably in favour of P or against P, or exactly in the middle. And, more to the point, what might look like favourable support to you, might carry a different character of conviction for me. The interest-relative theory of justification provides a simple and plausible explanation for this insofar as it tells us that the amount of evidence needed in any given case hinges on a subject’s interest in the matter at hand. Interests are thus a key component of our normative assessments and thus play an ineliminable role in the evaluation of the justificatory status of belief, and while this aspect of our epistemic evaluation is rarely articulated, without making assumptions about a subject’s interests we cannot determine whether her beliefs are living up to the standard set by the epistemic norm in question.
The examples above expose a commonly shared intuition, which is that when the epistemic risk one takes in accepting a proposition as true is high, the evidential burden required for justification goes up. An interest-relative theory of justification, with its emphasis on the relationship between one’s interests and one’s evidence, has the explanatory power to make clear why this is so. Again, a true belief is true no matter how important it is, so if getting at truth varies from case to case, that variation must be accounted for by looking at something other than evidence.
To bring this point home, imagine that you and I are watching television together, and as we are flipping channels, we see the same news flash about a pilots’ strike at British Airways (BA). Now, suppose further that I have a ticket booked on BA to take me to my best friend’s wedding in London the following weekend, a highly anticipated event to which I have been looking forward for months, whereas you are indifferent to the news. In this case, while you and I have the same evidence regarding P, i.e., there is a pilots’ strike at BA, that fact does not land us in normatively symmetrical positions. The modest positive epistemic support in favour of P is enough to justify your belief that P. However, because of my practical investment in the truth of the matter, the more fitting doxastic attitude for me, at least with the evidence I have in hand, is to suspend judgment about P.
This example is reminiscent of Fantl’s and McGrath’s Cases 1 and 2 of the Train to Foxboro, which aims to show the failure of strict versions of evidentialism. As Fantl and McGrath point out, if our intuitions are correct in these kinds of cases (and they think that they are), then justification cannot simply be a matter of the evidence one has, since the evidence in both cases is the same, yet the justified doxastic attitudes vary.Footnote 22
Conclusion
The interest-relative theory of justification that I have defended here can be seen as a modified version of evidentialism. It shares with evidentialism the idea that factors external to the mind do not make a difference, justificationally. Both theories are thus internalist, and both see evidence—evidence to which we have cognitive access—as the epistemic norm that beliefs must meet in order to be justified, but that is where the likeness ends. As I have argued here, strict versions of evidentialism suffer from a fundamental problem. By failing to take into account certain subjective facts about an individual, by focusing on evidence and evidence alone, evidentialism lacks the tools necessary to apply its own epistemic norms and as such leaves us hamstrung in trying to evaluate the epistemic status of belief. This gives us a good reason to accept an interest-relative theory of justification, independent of the issue of peer disagreement. That said, we have seen that this theory has interesting consequences for the normative question about disagreement. The examples discussed above show that, because our epistemic evaluations of belief depend on more than evidence alone, a genuine disagreement with an epistemic peer in the case of full disclosure does not necessarily imply that one individual has an unjustified belief. As we have seen, the difference in opinion might well come down to a difference in epistemic risk.
As I suggested at the start of the paper, one way of understanding the normative question about disagreement is to see it as a question about how to fulfill our duties as responsible epistemic agents. Seen in this light, the question of whether I can responsibly maintain that not-P when someone whom I consider to be my epistemic peer maintains that P, is just the question of whether, in such a case, I can justifiably believe that not-P while my peer justifiably believes that P. Because there is an ineliminable subjective component in this equation—namely, our interest in a proposition—it is possible that you and I could have conflicting, yet justified, doxastic attitudes about P. This opens the door to the possibility of reasonable disagreement amongst epistemic peers, even after full disclosure.
According to the interest-relative theory of justification, where there is a genuine disagreement among epistemic peers, it is possible for both sides to be justified, and this means that the uniqueness thesis is wrong. It also means that there is not only one maximally rational response to a given evidential situation;Footnote 23 rationality, on the interest-relative theory of justification, is more permissive than that.
