Over the past thirty years or so, Roderick Chisholm has presented a series of ever more refined and penetrating accounts of the central notions of the theory of knowledge. Clearly we can do no better than to start by considering his seminal and exemplary work. Chisholm's central epistemological project, perhaps, has been the development of epistemic principles: noncontingent conditionals whose antecedents specify a relation between a person S and a proposition A, and whose consequents specify that A has a certain epistemic status for S—certainty, perhaps, or acceptability, or being beyond reasonable doubt, or being known. Chisholm's attention has been directed less explicitly to exploring the nature of warrant (or ‘positive epistemic status’, as he calls it) than to stating those epistemic principles; he has been less explicitly concerned with what warrant or positive epistemic status is than with the question under what conditions a belief has it. But of course he does propose a view as to what warrant is, and that is where I propose to focus our attention. What, according to Chisholm, is warrant or positive epistemic status? What is its nature?

The answer I propose to explore first is “classical Chisholmian internalism,” the answer he gives in his books Theory of Knowledge¹ (first and second editions) and The Foundations of Knowing² (hereafter TK and FK). The principal thing to see here, I think, is that Chisholm's answer stands squarely in the tradition of classical internalism. He concurs with the fundamental deontological intuition of the latter; he sees warrant or positive epistemic status as essentially connected with deontological epistemic justification, the condition of having satisfied one's epistemic duties or obligations. Of course his view differs, in interesting respects, from that of Descartes and Locke. For example, they limit knowledge to what is certain; he does not. Further, he writes as if warrant just is justification, in the sense that what makes the belief that P a case of knowledge for S is just the fact that believing P is an especially good way for S to fulfill his epistemic duty. Descartes and Locke, however, do not see duty fulfillment as essentially involved in knowledge. As they see it, knowledge is only of propositions that are certain for S—propositions that are self-evident to him for example. But (as they think) we have little or no voluntary control over whether we believe propositions that are certain for us; hence there is little scope for duty and its fulfillment with respect to believing these propositions. Still Chisholm's basic suggestion is that warrant or positive epistemic status is to be understood in terms of fulfillment of epistemic duty; and that idea displays obvious continuity with the thought of Descartes and Locke, for whom questions of epistemic duty and obligation are of central concern.

I. The Central Idea

Chisholm begins his inquiry by introducing an undefined technical locution: “p is more reasonable than q for S at t”; here the values for p and q will be such states as believing that all men are mortal and withholding the belief that all men are mortal—that is believing neither it nor its denial.³ When one believes or withholds something, then, Chisholm assumes, there is indeed something one believes or withholds. If I believe that Venus is about the same size as Earth, then there is something I believe, namely Venus is about the same size as Earth. I concur: suppose we call the things that can be thus believed and withheld propositions, ignoring, for the moment, the difficult questions as to what sorts of things propositions are (classes of possible worlds?⁴ items from some unusually large and powerful language? brain inscriptions? divine thoughts?), whether there are “singular propositions,” whether propositions have constituents, whether they sometimes have persons and other concreta as constituents, whether, if a proposition has constituents, it has them essentially, whether some propositions exist contingently, whether there is or could be a set or perhaps a proper class of all propositions, and so on. Given ‘is more reasonable than’ as an undefined term, Chisholm goes on to define a battery of “terms of epistemic appraisal” as he calls them: ‘certain’, ‘beyond reasonable doubt’, ‘evident’, ‘acceptable’, and so on. A proposition A is beyond reasonable doubt for a person, for example, if it is more reasonable for her to accept that proposition than to withhold it; and A has some presumption in its favor for her at t just if accepting it then is more reasonable than accepting its negation. The epistemological principles Chisholm presents are formulated by way of these terms of epistemic appraisal.

Now Chisholm introduces ‘is more reasonable than’ as an undefined locution; but of course he intends it to have a sense, and to have a sense close to the sense it has in English. In FK he says that “Epistemic reasonability could be understood in terms of the general requirement to try to have the largest possible set of logically independent beliefs that is such that the true beliefs outnumber the false beliefs. The principles of epistemic preferability are the principles one should follow if one is to fulfill this requirement” (p. 7). In his earlier TK he is a bit more explicit about intellectual requirements and explains reasonability in terms of a somewhat different requirement:

We may assume that every person is subject to a purely intellectual requirement: that of trying his best to bring it about that for any proposition p he considers, he accepts p if and only if p is true.

He adds that

One might say that this is the person's responsibility or duty qua intellectual being.… One way, then, of re-expressing the locution ‘p is more reasonable than q for S at t’ is to say this: ‘S is so situated at t that his intellectual requirement, his responsibility as an intellectual being, is better fulfilled by p than by q’. (p. 14)

I said Chisholm stands in the classical internalist tradition; here we see the first and deepest connection with that tradition. Reasonability, as Chisholm explains it, is a normative concept; more precisely it is deontological: it pertains to requirement, duty, or obligation. That deontological character need not be strictly moral or ethical; for perhaps an epistemic requirement is not a moral duty. Perhaps it is a sui generis form of obligation.⁵ But even if epistemic obligation is sui generis, it shares important elements off structure with moral obligation: there is supervenience, defeasibility, the application of the prima facie/all-things-considered distinction, the characteristic: relations among permission, obligation, and prohibition, and so on. And Chisholm's central claim here is that a certain requirement, or responsibility, or duty, or obligation lies at the basis of such epistemic notions as evidence, justification, positive epistemic status, and knowledge itself; for knowledge is defined in terms of reasonability together with truth⁶ and belief.

So here Chisholm endorses a fundamental intuition of the classical internalist tradition: there are epistemic duties; and justification, being justified in one's beliefs, is the state of forming and holding beliefs in accord with those duties. Here Chisholm also endorses, at least by implication, the First Internalist Motif (see p. 19). On the view he presents, it is sufficient for my beliefs’ having positive epistemic status for me—even the highest levels of that quantity—that I do my epistemic duty, fulfill my epistemic obligation. But then whether my beliefs have positive epistemic status for me is up to me, within my control. I need only do my duty, to achieve that condition, and whether I do my duty is up to me.

The suggestions made in FK and TK as to what our intellectual requirement is differ in a more than superficial way; neither, furthermore, is exactly right. If the second suggestion were correct, one's duty as a being capable of belief might be satisfied by trying not to consider any propositions at all, or trying to consider only propositions that are obviously true or obviously false; if the first were correct, I could satisfy my epistemic duty by trying to restrict my attention as much as I could to propositions of simple arithmetic, considering and believing as many propositions of the form n + 1 > n as possible. Obviously something must be said about other epistemic values here: the importance of considering important propositions, of having beliefs on certain crucial topics, of avoiding unnecessary clutter and frivolous dilettantism, perhaps of having a coherent system of beliefs, and so on. The basic idea, however, is that our epistemic duty or requirement is to try to achieve and maintain a certain condition—call it ‘epistemic excellence’—which may be hard to specify in detail, but consists fundamentally in standing in an appropriate relation to truth. This is; a duty I have “qua intellectual being”—that is, just by virtue of being the sort of creature that is capable of grasping and believing (or withholding) propositions. We must pay a price for our exalted status as intellectual beings; with ability comes responsibility. And the idea, presumably, is that all intellectual beings have this responsibility: angels, devils, Alpha Centaurians, what have you—all are subject to this requirement or obligation. The obligation in question, furthermore, unlike, say, the obligation to repay the $10,000 you have unwisely borrowed from your children, is not one that can be discharged in a moment; it is a long-standing (indeed permanent) requirement to try to bring about and maintain a certain relatively complex and many-sided state of affairs. This state of affairs, whatever exactly it is, is such that trying to bring it about involves many actions over a considerable period of time.

The FK version is relatively unspecific, saying only that ‘_____is more reasonable than… for S at t’ can be understood in terms of “a general requirement attaching to intellectual beings”; the TK version is more specific, specifying that ‘_____is more reasonable than… for S at t if and only if S is so situated at t that his intellectual requirement is better fulfilled by_____than by…’ So suppose we consider the latter formulation. What stands out, here, is that Chisholm states the intellectual duty or obligation or requirement as one of trying to bring about a certain state of affairs. One's duty as an intellectual being is not that of bringing it about that one has a large set of beliefs, most of which are true; it is instead that of trying to bring about this state of affairs. My requirement is not to succeed in maintaining epistemic excellence; my requirement is only to try to do so. Presumably the reason is that it may not be within my power to succeed. Perhaps I don't know how to achieve epistemic excellence: or perhaps even though I do know how, I simply can't manage it. So my duty is simply to try to bring about this state of affairs. And of course here the First Motif is once more evident. Achieving justification, achieving a state in which my beliefs have positive epistemic status for me, is clearly within my control and up to me. All I have to do is try to achieve epistemic excellence; and of course whether or not I try to achieve that state is within my control and up to me.⁷

But how can I try to bring about such a state of affairs? In the general case, what one does to try to bring about some state of affairs is to take some action one thinks will bring about or contribute to the bringing about of that state of affairs. A university president, for example, has a duty to try to see to it that his university flourishes; he discharges this responsibility by taking actions that he thinks will contribute to its flourishing. Second, Chisholm's explanation of ‘is more reasonable than’ presupposes that my requirement to try to bring about a certain state of affairs can be better fulfilled by my doing one thing A than by my doing another thing B. How would that go? Well, clearly if I believe that A would tend to produce the state of affairs in question but do not think B would, then I can better fulfill that requirement by A than by B. But even if I think both A and B would contribute to producing that state of affairs, it could still be that I could better fulfill the requirement by A than by B. Perhaps I think A would contribute much more to the realization of the intended state of affairs; or perhaps I think it is much more probable that A would contribute to that realization than B; or perhaps I believe more strongly that A would than that B would.

Now how does this apply in the case in question, the case of my being obliged to try to achieve epistemic excellence? First, I must have some idea as to what to do to achieve it. I need not have a correct idea as to what to do, but if I haven't any idea at all, nor even any idea how to find out what to do, then I cannot even try to do so. I have no idea how to make a computer out of wood and string, nor do I know whether it can be done. If you order me to do it, I can't even try to comply. (I could idly tie a couple of two-by-fours together and claim I was trying, but I would not be speaking the truth.) Similarly in the present case; I cannot try to achieve epistemic excellence unless I have some beliefs as to how to bring about that state of affairs. Second, if I do have beliefs as to how such a project should go, then (all else being equal) I can best satisfy my obligation to try to bring about the state of affairs by acting on those beliefs. Suppose it is my duty to try to be in Boston at a certain time; then the way for me to fulfill this duty is to act on my beliefs as to what I can do to get there then. This is true even if some of these beliefs are false. Suppose I believe I can get there on time only by flying through Detroit; then the way for me to try to get there on time is to fly through Detroit—even if, as it turns out, I am wrong, and in fact can get to Boston on time only by flying through Chicago.

We must make one qualification here. It may be that I am in some way blameworthy or properly subject to reproach in holding this false belief; perhaps I should have known better. Perhaps I hold the belief in question only because of laziness, or carelessness, or inattention. Under these conditions, acting on my belief might not be the best way to satisfy my obligation. You reproach me for not trying to get to Boston on time; I reply (by way of attempted self-exculpation) that I tried my best to get there then, but failed simply because of a false belief about plane schedules; you point out that I didn't even bother to make inquiries and arrived at my belief by relying on vague memories or just guessing. Then I may still be scored for flouting my obligation to try my best to get to Boston on time; I did not fulfill that duty.

Take another example. I have an obligation to try to bring up my children properly; I am obliged to try to bring it about that they enter adulthood in spiritual, moral, mental, and physical health. I can't even try to do this if I have no ideas at all about what to do—no ideas, for example, about what sorts of training and discipline are likely to produce the desired result. And if I do have beliefs about how to achieve the end in question, then the best way for me to fulfill this obligation is to act on these beliefs. Suppose it turns out that my beliefs are false, and acting on them wholly counterproductive: that doesn't show that I didn't try my best and doesn't show that I didn't satisfy my obligation thus to try. Perhaps I believed on good authority that it is an excellent idea to put your small children on a low-fat diet in order to help prevent cholesterol buildup later in life. Suppose, as it happens, I was wrong; low-fat diets are dangerous for young children.⁸ If, later on, my children fail to display normal growth patterns, I am not properly held responsible for failing to try to bring them up in such a way as to promote their health. I did the best I could but was misinformed; I was mistaken but not blameworthy. And clearly if I do believe that fatty food is unhealthy, then I am so situated that I can better fulfill my obligation to try to bring them up in a healthy manner by withholding it from them than by feeding them large quantities of it. The qualification, again, would apply where I hold this belief but do so carelessly, or with what Aquinas calls ‘undue levity’, or where I should have known better. In such cases, perhaps, I fail to try hard enough; in such cases I can be blamed for failing to fulfill that obligation. But if I (nonculpably) believe that the best way to achieve the goal in question is to take a certain line of action, then I have prima facie fulfilled my obligation to try to bring about this state of affairs by taking that line of action.

So my obligation is to try to achieve epistemic excellence, and the way to do so is to act on what I believe is the way to achieve that state. But then the Second Internalist Motif and its corollaries (see pp. 19–22) are also satisfied. The ground of warrant, that which confers warrant upon a proposition for me, is the aptness for fulfilling epistemic duty of my accepting that proposition. But in the typical case the degree of that aptness will depend upon the degree to which the proposition in question seems to me to be true; and that is something to which I have a sort of guaranteed access. Thus C2 (see p. 21) is satisfied. But so is C3 (p. 22). For what is my ratio cognoscendi here; how do I tell whether a given proposition is such that accepting it will contribute substantially to the fulfillment of duty to try to achieve epistemic excellence? Well, one way would be by determining how much or how strongly it seems to me to be true. The more strongly the proposition in question seems to me to be true, the more apt accepting it is for fulfillment of my epistemic duty. Ratio essendi and ratio cognoscendi, therefore, coincide. And of course the third motif is also exemplified: for trying to do something or other is surely in an important and recognizable (if hard to define) sense internal.

II. Classical Chisholmian Internalism Rejected

I propose to argue two points with respect to the main contours of classical Chisholmian internalism. First, the epistemic principles Chisholm offers do not in fact fit with his official claim that warrant is a matter of fulfilling epistemic obligation; if warrant is what he officially says it is, then most of his epistemic principles are clearly false. This suggests a certain tension in Chisholm's thought between what he officially says warrant is, and how he actually thinks about it. Second, I shall argue that positive epistemic status cannot be thus explained. My claim is that the fulfillment of epistemic obligation or requirement is nowhere nearly sufficient for warrant; it is also (though perhaps a bit more dubiously) unnecessary for it. Taking ‘justification’ in its original and most natural sense, what I shall argue is that justification is wholly insufficient for warrant; it is also not necessary for it.

A. P1

Let us begin by considering two of the six “epistemic principles” Chisholm offers in FK. These principles, he says, are necessarily true if true at all (FK, p. 57); and they specify conditions under which a person and a proposition are so related that the latter has some degree of warrant—being certain, or acceptable, or evident, for example—for the former. According to the first of these principles,

P1 If the property of being F is self-presenting, then, for every x, if x has the property of being F and if x considers his having that property, then it is certain for x that he is then F. (FK, p. 12)

An instance of P1, says Chisholm would be:

P1a If the property of being sad is self-presenting, then, for every x, if x has the property of being sad and if x considers his having that property, then it is certain for x that he is then sad. (FK, p. 12)

To say that p is certain for S is to say that it is beyond reasonable doubt for him (is such that accepting it is more reasonable for him than withholding it) and hat furthermore there is no proposition q such that accepting q is more reasonable for him than accepting p (p. 8); so if p is certain for S, then it has the maximal degree of warrant for him. A proposition p is self-presenting for a person S if and only if necessarily, if S considers p and p is true, then S believes p (pp. 12–13); examples would be propositions of the sort S is appeared to redly, S is sad, and S believes that Albuquerque is in New Mexico. In the case in Question, therefore (the case of the person who is sad), there is no proposition q such that S can better fulfill his epistemic requirement by accepting q than by accepting the proposition that he is sad. So if S is sad and considers whether he is then the proposition that S is sad has, for S, the highest degree of positive epistemic status.

Now perhaps it is natural to attribute the impressive degree of warrant enjoyed by this proposition (and others like it) to the fact that it is indeed self-presenting. What is responsible for its exalted condition, you think, is just the fact that it is indeed self-presenting: the fact that it is necessarily such that if it is true and S considers it, then S does in fact believe it. It is impossible that this proposition be true and S consider it but fail to believe it. But here there is a problem. Suppose warrant or positive epistemic status is what Chisholm says it is: why should the fact that this proposition is self-presenting, in Chisholm's sense, confer maximal epistemic status upon it? Indeed, a proposition's being self-presenting seems to preclude its having such status. The suggestion is that if I feel sad, then believing that I feel sad is a good way of fulfilling my duty to try to achieve epistemic excellence; it is so good, in fact, that nothing else I could do would be a better way of fulfilling that duty.

But suppose I feel sad, at t: could it really be that believing that I feel sad is any way at all (let alone a maximally excellent way) of trying to fulfill my epistemic duty? In any case I can think of, if doing something A is a way in which I can try to bring about some state of affairs, A will be such that it is at least logically possible for me to consider it and carry it out, and also at least logically possible for me to consider it and fail to carry it out; there will be possible worlds in which I consider it and carry it out, but also worlds in which I consider it and do not carry it out. This is not so in the present case. For the proposition that I am sad is self-presenting; hence (according to Chisholm) it is not possible in the broadly logical sense that I should in fact be sad and consider whether I am, but refrain from believing that I am. But if that is how things stand, how could this proposition be such that at t I can better fulfill my epistemic duty—my duty to try to achieve epistemic excellence—by accepting it than by rejecting or withholding it? Perhaps you think I might be able to fulfill an epistemic duty (when I am sad) by, say, asking myself whether I am sad (I can consider whether to consider my phenomenal field and then either do it or not). You might add that if I do consider that question under those conditions, I will in fact believe that I am sad, so that my not believing that I am sad might show that I am not doing my epistemic duty. Perhaps so; but even if so, that does not show that I can satisfy any epistemic duty by believing that I am sad; what satisfies my duty, under that scenario, is my considering whether I am sad. If the fact is I am sad and I consider this proposition, then whether or not I accept it is simply not up to me; but then accepting this proposition cannot be a way in which I can fulfill my obligation to the truth, or, indeed, any obligation to try to bring about some state of affairs. Suppose I've just fallen off a cliff: could I be subject to an obligation to try to bring something about, which is such that I can better fulfill it by falling down rather than, say, by falling up or remaining suspended in midair? Hardly.

Chisholm's fundamental idea seems to be that the rational creature, the being capable of beliefs, considers the various propositions that come to his attention at a time t, deciding which to accept and which to withhold. If he is epistemically dutiful, he will make these decisions in the service of an effort to fulfill his duty to the truth, his duty to try to achieve and maintain epistemic excellence; and a proposition will have positive epistemic status (warrant) for him, at t, to the degree to which he can fulfill this obligation by accepting it. But if the proposition in question is self-presenting, then if it is true and he considers it, he already believes it. He cannot consider it, and then decide whether to accept it; if it is true and comes to his attention, then it is not up to him whether he believes it.

So there is a deep difficulty with supposing that the sort of proposition I would express by saying ‘I feel sad’ or ‘I am appeared to redly’ achieves its high epistemic status by virtue of being self-presenting. As a matter of fact, if warrant is what Chisholm says it is, then its being self-presenting for me would preclude its having warrant for me; for propositions that are self-presenting, for me (as Descartes and Locke noted), are not such that whether I accept them is, in the appropriate way, under my control. And this suggests that, contrary to what Chisholm officially says, he does not really think of warrant as simply a matter of aptness for fulfilling epistemic duties. The sorts of propositions he thinks of as self-presenting do, no doubt, have impressive epistemic credentials; but the status they have is not that of being such that a person can best fulfill that epistemic requirement by accepting them. That's not the sort of epistemic status they have.

But now suppose Chisholm is wrong about the propositions he says are self-presenting—the sorts of propositions Chisholm thinks of as self-presenting—the ones typically expressed by such sentences as ‘I feel sad’ and ‘I am appeared to redly’. Suppose these propositions do in fact enjoy the high degree of warrant or positive epistemic status he says they do, but suppose they are not, in fact, self-presenting in his sense. Suppose it is or could be within my power to consider such a proposition as I am appeared to redly (when in fact I am) and then either accept or reject it. Suppose (by dint of enormous concentration and effort) I could train myself not to accept this proposition when I am thus appeared to and consider whether I am thus appeared to. Perhaps I could undergo some sort of strenuous regimen, so that at its conclusion it is within my power deliberately to refrain from believing that I am appeared to redly, despite the fact that I am thus appeared to. Then beliefs of this sort would not be disqualified for warrant—at any rate not for the previous reason. But would it then be correct to say that these propositions had maximal warrant (taking the latter to be what Chisholm officially says it is)? I think not. If P1 is correct, than it is necessarily the case that anyone who considers a proposition that is both true and of the sort in question is someone for whom that proposition has maximal warrant. Under those conditions, she and that proposition are so related that there is no proposition q such that she can better fulfill her epistemic requirement by accepting q than by accepting p. But suppose these propositions are not self-presenting for S; then clearly they are not related this Due to a brain lesion or the machinations of a malevolent Cartesian demon I might be deeply but nonculpably convinced, of the proposition⁹ I express by the words ‘I am appeared to redly’, that it is never true. Then the way for me to try to achieve epistemic excellence would be to try to reject that proposition, even when in fact I am appeared to redly.

Alternatively, a clever but unscrupulous epistemologist might offer me a subtly fallacious argument, convincing me both that this proposition—call it ‘P’—is false and that I ought not to accept it. His argument might go as follows:

1. P entails that there is at least one person.

2. Necessarily, if there are any persons, then they are composed of more than one but finitely many molecules.

3. For any person S and number n, if S is composed of n molecules, then it is possible that there be a person composed of n—1 molecules.

4. Anyone who carefully and reflectively considers (1)–(3) and believes them, and sees that (1)–(3) entail the denial of P, ought not to accept P.¹⁰

This argument may be mistaken, but that does not mean I couldn't be taken in by it. And if I were (through no fault of my own) taken in by it and convinced of its conclusion, then surely it would not be the case that I could fulfill my epistemic obligation—my obligation to try to achieve epistemic excellence—better by believing P than by withholding it, or believing its denial. As I argued (pp. 34–36), how I can best fulfill my obligation to try to bring about a certain state of affairs depends (among other things) upon what I believe as to how to bring about that state of affairs. If I believe that the best way to bring it about is to take a certain line of action, then (provided, anyway, that in acquiring that belief¹¹ I did not flout my duty to the truth) I can best satisfy that obligation by taking that line of action. I must do the best I can, according to my own lights. (Who else's?)

Accordingly, there are two relevant facts here. The first is that if a belief really is self-presenting, in Chisholm's sense, then whether I hold that belief on a given occasion is not up to me. But then I am not so situated with respect to that belief that either accepting or withholding it could be the best way for me to fulfill my duty to the truth—my duty to try to achieve epistemic excellence. And second, even if it were within my power to accept or withhold such a belief, it still would not follow that I can best satisfy this obligation by accepting it; I might believe that most such beliefs are false, or that my obligation to the truth, in these circumstances, requires that I refrain from accepting it. These two facts suggest that Chisholm himself isn't really thinking of certainty—that particular maximal form of warrant—as a matter of aptness for fulfilling my epistemic requirement. He isn't really thinking that a belief is certain for me if and only if I am so situated with respect to it that I can better fulfill my duty to the truth by accepting it than by accepting any other proposition. Finally, the beliefs Chisholm thinks of as self-presenting surely do have great warrant or positive epistemic status; but this status is not that of being (if true) overwhelmingly apt for the fulfillment of epistemic duty. In Warrant and Proper Function I shall try to say what sort of positive epistemic status or warrant they have, and why they have so much of it.

B. P5a

Now suppose we consider Chisholm's P5, which I shall divide into two principles, P5a and P5b (and shall also state it in terms of propositions rather than properties):

P5a For every x, if (i) x perceptually takes there to be something that is F, and if (ii) his perceiving an F is epistemically in the clear for x, then it is beyond reasonable doubt for x that he perceives something that is F. P5b If conditions (i) and (ii) are fulfilled and furthermore x's perceiving something that is F is a member of a set of propositions which mutually support each other and each of which is beyond reasonable doubt for x, then it is evident for x that he perceives something that is F. (FK, p. 21)

I shall comment only on P5a. We have two locutions that require explanation: ‘perceptually takes there to be something that is F’ in the first couple of lines, and ‘epistemically in the clear for x’ in the third. Take the second first: a proposition is epistemically in the clear for x if and only if it is not disconfirmed by the conjunction of all those propositions that are such that it is more reasonable for S to accept them than to accept their negations (for details, see FK, pp. 18ff.). Second, ‘perceptually takes there to be something that is F’ is to be understood as follows: “The property of being F is a sensible property such that x is appeared to in such a way that he directly attributes to himself the property of being appeared to in that way by something that is F” (FK, p. 20). Sensible properties, says Chisholm, are such properties as “being red, round, yellow, putrid, rough (to the touch)” and the like. What this means, therefore, is that a person perceptually takes there to be something that is F if he is appeared to in a certain way Z, believes that there is something that appears to him in way Z, and believes that that thing has the property of being F. You look out of your window at your backyard; you are appeared to in the characteristic way that goes with perceiving a yellow playhouse. If, under those conditions, you take it that there is something that is appearing to you in that way, and that the thing appearing to you in that way is yellow, then you perceptually take there to be something that is yellow. Finally, recall that a proposition is beyond reasonable doubt for x just if believing that proposition is a pretty good way for x to fulfill his epistemic duty. (In the counterexample I offer here, therefore, I shall assume with Chisholm that a person has a considerable degree of control over what he believes.)

P5a I believe, is defective in more than one dimension. First, I shall argue that (as in the previous case) if positive epistemic status is what Chisholm says it is then P5a is not necessarily true: a person could satisfy the antecedent of P5a with respect to some proposition and nonetheless utterly fail to fulfill his epistemic obligation in believing that proposition, so that the proposition would not be beyond reasonable doubt for him. Suppose Paul, like the rest of us, is such that when he is appeared to redly (under normal conditions) he has an inclination to believe that he perceives something red. Paul, however, has read his Kant. He has a deep concern for the dignity and autonomy of free, rational creatures such as himself. It seems to him a bit undignified, possibly even faintly ridiculous, to be pushed around in this way by his passional or impulsive nature. He therefore solemnly decides to edit his cognitive nature, thus demonstrating his autonomy, his independence of natural tendencies, and his mastery over his lower nature. He undertakes a regimen to free himself from the tyranny of these impulses; by dint of long and arduous training, he develops the power to withhold the normal or ordinary belief in many perceptual situations. For example, when he is appeared to in that characteristic way one is appeared to upon perceiving a large, fully developed oak from 40 yards, he is able (at the cost of considerable effort) to withhold the belief that he is perceiving a large tree. (Sadly enough, he also develops an unfortunate tendency to patronize those more ordinary mortals who have not thus enhanced their rational autonomy.) Further, he trains himself to acquire unusual beliefs in ordinary perceptual situations. On occasions when he is aurally appeared to in the way one is appeared to upon hearing church bells, the belief he forms is that there is something that is appearing to him in that way, and that it is bright orange.

On a given occasion, therefore, he is appeared to in that church-bell way and, in accord with his training, forms the belief that the thing appearing to him in this way is orange. That he is thus perceiving a thing that is orange, we may suppose, is epistemically in the clear for him; it is neither confirmed nor disconfirmed by the conjunction of propositions beyond reasonable doubt for him. The antecedent of P5a is therefore fulfilled, for Paul. Yet surely it is not the case that he can then better fulfill his duty to the truth by accepting the proposition that he is perceiving something orange than by withholding that proposition. Forming beliefs in accord with this foolish and arrogant policy of his is no way to fulfill any duty at all. If positive epistemic status is fundamentally a matter of deontological justification, therefore, P5a is not necessarily true. Someone who instantiates its antecedent may still be failing to do his epistemic duty.

Perhaps you think this example unduly fanciful; if so, there are plenty that are less esoteric.¹² Suppose I form the belief all horses are white and form it in some epistemically culpable fashion. (Perhaps the Lone Ranger and Silver make a guest appearance in my hometown; I am very much impressed by their magnificent appearance and hastily generalize to the belief in question.) Later I hear a horse whinny and form the belief that I am being appeared to in that fashion by a thing that is white. Then I am not properly fulfilling my duty by forming the belief in question. Still, I satisfy the antecedent of P5a: I perceptually take there to be something that is white, and the proposition that I do so, we may suppose, is neither confirmed nor disconfirmed by the conjunction of propositions beyond reasonable doubt for me, so that it is epistemically in the clear for me.

Still further, it is clear, I think, that in the sorts of situations P5a is designed to apply to, positive epistemic status is not (or is not merely) a matter of deontological justification. Return to Paul of the paragraph before last and alter his situation a bit. As in the previous case, he is such that when he is appeared to in one sense modality, he forms beliefs appropriate to another; only this time it is due not to some project born of pride and foolishness, but instead to a brain lesion or perhaps a whimsical Cartesian demon. When Paul is aurally appeared to in the church-bell fashion, therefore, he finds himself with a powerful, nearly ineluctable tendency or impulse to believe that there is something that is appearing to him in that fashion, and that that thing is orange. He does not know about this quirk in his epistemic equipment, and his lack of awareness is in no way due to dereliction of epistemic duty. Indeed, Paul is unusually dutiful, unusually concerned about fulfilling his epistemic duties; fulfilling these duties is the main passion of his life. And let's add that those around him suffer from a similar epistemic deficiency, They have all been manipulated in this way by demons or Alpha Centaurian cognitive scientists; or they all suffer from similar lesions due to radioactive fallout from a missile test. Then, surely, Paul is doing his epistemic duty in excelsis in believing as he does; but the proposition in question has little by way of positive epistemic status for him. He is deontologically justified, and more; for in working as hard as he does to achieve epistemic excellence, he performs works of epistemic supererogation. Using the term not as a synonym for ‘warrant’, but in a neutral way, we may say that there is a kind of positive epistemic status his beliefs have: they were formed in accord with a serious and determined effort to live dutifully.

Nevertheless there is also a kind of positive epistemic status this belief lacks for him, a kind crucial for knowledge: it lacks warrant. For that sort of positive epistemic status, it is not sufficient that one satisfy one's duty and do one's epistemic best. Paul can be ever so conscientious about his epistemic duties, and still be such that his beliefs have very little warrant. And the reason, fundamentally, is that even though he is doing his epistemic duty to the uttermost, his epistemic faculties are defective; this deprives the belief in question of any substantial degree of that quantity.

III. Justification versus Warrant

Suppose we briefly review our conclusions thus far. The discussion of Chisholm's principles displays a certain pattern: in both cases we noted first that if positive epistemic status is what Chisholm officially says it is, then the principle in question is wholly unsatisfactory. This is not, in these cases, a matter of fine tuning; it is not as if by further chisholming we could find a satisfactory set of principles substantially like the ones in question. The difficulties go deeper than that. What the classical Chisholm officially says is that warrant is a matter of a proposition's being so related to a person that he can better fulfill his duty—that of trying to bring it about that he is in the right relation to the truth—by accepting the proposition in question than by, for example, withholding it. But if this is what warrant is, then in a wide variety of cases—cases of the sort I mentioned above—Chisholm's principles yield wholly wrong results. In many of these cases, the way for a person to try his best to achieve epistemic excellence will depend, naturally enough, upon what he (nonculpably) believes about the way to achieve epistemic excellence. In particular, if a person is strongly (and nonculpably) convinced of the truth of a proposition—if that proposition seems obviously true to him—then (barring defeating conditions) the way for him to try to achieve epistemic excellence is to accept it; and the more obvious it seems to him, the more status of this sort it has for him. But Chisholm's principles don't anywhere nearly yield this result.¹³

This suggests that, despite his official statements on the nature of warrant or positive epistemic status, Chisholm does not or does not consistently think of it as a matter of aptness for fulfillment of epistemic duty. He implicitly thinks of it as also involving something else. And here he is surely correct. Epistemic dutifulness or deontological justification is attractive from an internalist perspective; but reflection on Chisholm's principles reveals that it cannot possibly be anywhere nearly sufficient for warrant. No degree of dutifulness, no amount of living up to one's obligations and satisfying one's responsibilities—in a word, no degree of justification—can be sufficient for warrant.

A couple of final examples. (If you are already convinced, please skip the next couple of pages.) Suppose I develop a rare sort of brain lesion that causes me to believe that I will be the next president of the United States. I have no evidence for the proposition, never having won or even run for public office; my only political experience was an unsuccessful bid for the vice-presidency of my sophomore class in college. Nevertheless, due to my cognitive dysfunction, the belief that I will be the next president seems to me obviously true—as obvious as the most obvious truths of elementary arithmetic. Now: am I so situated that I can better fulfill my obligation to the truth by withholding than by accepting this proposition? Can I better fulfill my obligation to try to bring it about that I am in the right relation to the truth by withholding than by accepting it? Surely not. That I will be the next president seems to me to be utterly and obviously true; I have no awareness at all that my cognitive faculties are playing me false here. So if I try to achieve epistemic excellence, I will count this proposition among the ones I accept. The way for me to try to achieve epistemic excellence in these circumstances, I should think, is for me to act on what I (nonculpably) believe about how best to achieve this end. But this proposition seems obviously true to me; so, naturally enough, I believe that the way to achieve epistemic excellence here is to accept it. We may add, if we like, that I am exceptionally dutiful, deeply concerned with my epistemic duty; I am eager to bring it about that I am in the right relation to the truth, and am trying my level best to do so. Then, surely, I am doing my epistemic duty in accepting the proposition in question; nevertheless that proposition has little by way of warrant for me. Even if by some mad chance I will in fact be the next president, I do not know that I will be.

A last example: perhaps you think that what goes in excelsis with satisfying duty is effort; perhaps (in a Kantian vein) you think that genuinely dutiful action demands acting contrary to inclination. Very well then, suppose this time, Paul (due to lesion, demon or Alpha Centaurian) is utterly and nonculpably convinced that his nature is deeply misleading. Like the rest of us, he has an inclination, upon being appeared to redly, to believe that there is something red lurking in the neighborhood; unlike the rest of us, he believes that this natural inclination is misleading and that on those occasions there really isn't anything that is thus appearing to him. He undertakes a strenuous regimen to overcome this inclination; after intense and protracted effort he succeeds: upon being appeared to redly he is able to withhold the belief that something red is appearing to him, and finally to form, on those occasions, the belief that nothing red is appearing to him. His devotion to duty costs him dearly. The enormous effort he expends takes its toll upon his health; he is subject to ridicule and disapprobation on the part of his fellows, who view his project as at best Quixotic; his wife protests his unusual behavior and finally leaves him for someone less epistemically nonstandard. Determined to do what is right, however, Paul heroically persists in doing what he is nonculpably convinced is his duty. It is obvious, I take it, that even though Paul is unusually dutiful in accepting, on a given occasion, the belief that nothing red is appearing to him, that belief has little by way of warrant for him. Epistemic duty fulfillment, even epistemic works of supererogation—these aren't anywhere nearly sufficient for warrant.

But neither are they necessary for it.¹⁴ Suppose there is the sort of epistemic duty Chisholm suggests: a duty to try to bring it about that I attain and maintain the condition of epistemic excellence; and suppose I know this, am dutiful but also a bit confused. I come nonculpably to believe that our Alpha Centaurian conquerors, for reasons opaque to us, thoroughly dislike my thinking that I am perceiving something that is red; I also believe that they are monitoring my beliefs and, if I form the belief that I see something red, will bring it about that I have a set of beliefs most of which are absurdly false, thus depriving me of any chance for epistemic excellence. I thereby acquire an epistemic duty to try to withhold the beliefs I naturally form when I am appeared to redly: such beliefs as that I see a red ball, or a red fire engine, or whatever. Of course I have the same epistemic inclinations everyone else has: when I am appeared to redly, I am powerfully inclined to believe that I see something that is red. By dint of heroic and unstinting effort, however, I am able to train myself to withhold the belief (on such occasions) that I see something red; naturally it takes enormous effort and requires great willpower. On a given morning I go for a walk in London; I am appeared to redly several times (postboxes, traffic signals, redcoats practicing for a reenactment of the American Revolution); each time I successfully resist the belief that I see something red, but only at the cost of prodigious effort. I become exhausted, and resentful. Finally I am appeared to redly in a particularly insistent and out-and-out fashion by a large red London bus. “Epistemic duty be hanged,” I mutter, and relax blissfully into the belief that I am now perceiving something red. Then this would be a belief that was unjustified for me; in accepting it I would be going contrary to epistemic duty; yet could it not constitute knowledge nonetheless?

The conclusion to be drawn, I think, is that justification properly so-called—deontological justification—is not so much as necessary for warrant. Justification is a fine thing, a valuable state of affairs—intrinsically as well as extrinsically; but it is neither necessary nor sufficient for warrant. Chisholm's powerful and powerfully developed versions of deontological internalism—classical Chisholmian internalism—must be rejected.¹⁵

In chapter 1, I argued that in the twentieth century received tradition in Anglo-American epistemology—a tradition going back at least to Locke—sees justification as essentially deontological but also as necessary and nearly sufficient (sufficient up to Gettier) for warrant. Although this is an attractive, indeed, a seductive approach to an understanding of warrant, it is nevertheless at bottom deeply flawed. No amount of dutifulness, epistemic or otherwise, is sufficient for warrant. My doing my duty in accepting a proposition (or the proposition's being apt, even maximally apt, for epistemic duty fulfillment, for me) isn't anywhere nearly sufficient for that proposition to have warrant or positive epistemic status for me. Knowledge does indeed contain a normative element; but the normativity is not that of deontology. Perhaps this incoherence in the received tradition is the most important thing to see here: the tension between the idea that justification is a deontological matter, a matter of fulfilling duties, being permitted or within one's rights, conforming to one's intellectual obligations, on the one hand; and, on the other, the idea that justification is necessary and sufficient (perhaps with a codicil to propitiate Gettier) for warrant.¹⁶ To put it another way, what we need to see clearly and first of all is the vast difference between justification and warrant. The lesson to be learned is that these two are not merely uneasy bedfellows; they are worlds apart. Classical deontologism is no better off than classical foundationalism. We have learned to acquiesce or rejoice in the demise of classical foundationalism; but the classical deontologism that lies at its root is no better off.