This essay originally appeared in J. E. Tomberlin, ed., Philosophical Perspectives, 2, Epistemology (1988). Reprinted in William P. Alston, Epistemic Justification: Essays in the Theory of Knowledge, 1989.

The Deontological Conception of Epistemic Justification

William P. Alston

I. The Deontological Conception

The terms 'justified', 'justification', and their cognates are most naturally understood in what we may term a "deontological" way, as having to do with obligation, permission, requirement, blame, and the like. We may think of requirement, prohibition, and permission as the basic deontological terms, with obligation and duty as species of requirement, and with responsibility, blameworthiness, reproach, praiseworthiness, merit, being in the clear, and the like as normative consequences of an agent's situation with respect to what is required, prohibited, or permitted. More specifically, when we consider the justification of actions, something on which we have a firmer grip than the justification of beliefs, it is clear that to be justified in having done something is for that action not to be in violation of any relevant rules, regulations, laws, obligations, duties, or counsels, the ones that govern actions of that sort. It is a matter of the action's being permitted by the relevant system of principles.1 To say that the action was justified does not imply that it was required or obligatory, only that its negation was not required or obligatory. This holds true whether we are thinking of moral, legal, institutional, or prudential justification of actons. To say that Herman was (morally) justinea in refusing to take time out from writing his book to join in a peace march is to say that the relevant moral principles do not require him to march; it is not to say that he is morally obliged to stick to writing his book, though that may be true also. Likewise to say that Joan was legally justified in leaving the state is to say that her doing so contravened no law; it is not to say that any law required her to do so. Finally, consider my being justified in giving my epistemology class a take-home final rather than one to be taken in the classroom. Here we might be thinking of institutional justification, in which case the point would be that no regulations of my department, college, or university require a classroom final exam; but my being so justified does not imply that any regulations require a take-home exam. Or we might be thinking of pedagogical justification, in which case the point would b that sound pedagogical principles allow for a take-home exam for this kind of course, not that they require it; though, again, the latter might be true also.

The most natural way of construing the justification of beliefs is in parallel fashion. To say that S is justified in believing that p at time t is to say that the relevant rules or principles do not forbid S's believing that p at t. In believing that p at t, S is not in contravention of any relevant requirements. Again, it is not to say that S is required or obligated to believe that p at t, though that might also be true. With respect to beliefs we can again distinguish various modes of justification: moral, prudential, and epistemic. These may diverge. I may, for example, be morally justified in trusting my friend (believing that he is well intentioned toward me), and I may even be morally required to do so, even though, since all my evidence tends strongly against it, the belief is not epistemically justified. In this paper our concern is with epistemic justification. How is that distinguished from the other modes? The justification of anything, H, consists in H's being permitted by the relevant principles: epistemic, moral, or whatever. Thus the crucial question is: What distinguishes epistemic principles from moral principles? Well, the "epistemic point of view" is characterized by a concern with the twin goals of believing the true and not believing the false. To set this out properly we would have to go into the question of just how these goals are to be weighted relative to each other, and into a number of other thorny issues; but suffice it for now to say that epistemic principles for the assessment of belief will grade them in the light of these goals. Just how this is done depends on the conception of justification with which one is working. On a deontological conception of justification, the principles will forbid beliefs formed in such a way as to be likely to be false and either permit or require beliefs formed in such a way as to be likely to be true.2 Thus on the deontological conception of the epistemic justification of belief that is as close as possible to the standard conception of the justification of action, to be justified in believing that p at t is for one's belief that p at t not to be in violation of any epistemic principles, principles that permit only those beliefs that are sufficiently likely to be true.3 Let's say, for example, that beliefs in generalizations are permitted only if based on adequate inductive evidence, otherwise forbidden, and that a perceptual belief that p is permitted only if (a) it is formed on the basis of its perceptually seeming to one that p and (b) one does not have sufficient overriding reasons; otherwise it is forbidden. One will be justified in a belief of the specified sort if the relevant necessary conditions of permissibility are satisfied; otherwise the belief will be unjustified.

Since this is the natural way to use 'justification', it is not surprising that it is the one most often formulated by those who seek to be explicit about their epistemic concepts. Perhaps the most eminent contemporary deontologist is Roderick Chisholm.4 Because of the complexities of Chisholm's view, however, I shall take as my model deontologist Carl Ginet. He sets out the conception with admirable directness.

One is justified in being confident that p if and only if it is not the case that one ought not to be confident that p; one could not be justly reproached for being confident that p.5

Now this conception of epistemic justification is viable only if beliefs are sufficiently under voluntary control to render such concepts as requirement, permission, obligation, reproach, and blame applicable to them. By the time-honored principle that "Ought implies can", one can be obliged to do A only if one has an effective choice as to whether to do A.6 It is equally obvious that it makes no sense to speak of S's being permitted or forbidden to do A if S lacks an effective choice as to whether to do A.7 Therefore the most fundamental issue raised by a formulation like Ginet's is as to whether belief is under voluntary control. Only if it is can the question arise as to whether the epistemic justification of beliefs can be construed deontologically. As we shall see, there are various modes of voluntary control that have usually not been fully distinguished in the literature and that require separate treatment. I will be arguing in this paper that (a) we lack what I will call direct voluntary control over beliefs, (b) that we have only a rather weak degree of "long range" voluntary control over (only) some of our beliefs, and (c) that although our voluntary actions can influence our beliefs, the deontological notion of justification based on this indirect influence is not the sort of notion we need for the usual epistemological purposes to which the term 'justification' is put.

II. The Problem of Voluntary Control of Belief

There are many locutions that encourage us to think of believing as subject to requirement, prohibition, and permission. We say

"You shouldn't have supposed so readily that he wouldn't come through",
"You have no right to assume that",
"I had every right to think that she was honest",
"I ought to have given him the benefit of the doubt", and
"You shouldn't jump to conclusions".
We also often seem to suggest the voluntary control of belief:
"I finally decided that he was the man for the job",
"Make up your mind; is it coreopsis or isn't it?",
"I had to accept his testimony; I had no choice" (the suggestion being that in other cases one does have a choice).
And philosophers frequently fall in with this, speaking of a subject's being in a situation in which he has to decide whether to accept, reject, or "withhold" a proposition.8 All these turns of phrase, and many more, seem to imply that we frequently have the capacity to effectively decide or choose what we are to believe, and hence that we can be held responsible for the outcome of those decisions? It is natural to think of this capacity on the model of the maximally direct control we have over the motions of our limbs and other parts of our body, the voluntary movements of which constitute "basic actions", actions we perform "at will", just by an intention, volition, choice, or decision to do so, things we "just do", not "by" doing something else voluntarily. Let's call the kind of control we have over states of affairs we typically9 bring about by basic actions, "basic voluntary10 control". If we do have voluntary control of beliefs, we have the same sort of reason for supposing it to be basic control that we have for supposing ourselves to have basic control over the (typical) movements of our limbs, viz., that we are hard pressed to specify any voluntary action by doing which we get the limbs moved or the beliefs engendered. Hence it is not surprising that the basic voluntary control thesis has had distinguished proponents throughout the history of philosophy, for example, Augustine, Aquinas, Descartes, Kierkegaard, and Newman.11 Though distinctly out of favor today, it still has its defenders.12

Before critically examining the thesis, we must make some distinctions that are important for our entire discussion. First, note that although the above discussion is solely in terms of belief, we need to range also over propositional attitudes that are contrary to belief. Chisholm speaks in terms of a trichotomy of "believe", "reject", and "withhold" that p.13 Since rejecting p is identified with believing some contrary of p, it brings in no new kind of propositional attitude; but withholding p, believing neither it nor any contrary, does. The basic point to be noted here is that one has control over a given type of state only if one also has control over some field of incompatible alternatives. To have control over believing that p is to have control over whether one believes that p or not, that is, over whether one believes engenders instead some incompatible alternative.14 The power to choose A at will is the power to determine at will whether it shall be A or (some form of) not-A. Therefore, to be strictly accurate we should say that our problem concerns voluntary control over propositional attitudes. Although in the sequel the formulation will often be in terms of belief, it should be understood as having this more general bearing.

Second, something needs to be said about the relation between the control of actions and of states of affairs. Thus far we have been oscillating freely between the two. Now a belief, in the psychological sense that is being used here (as contrasted with the abstract sense of that which is believed), is a more or less long-lived state of the psyche, a modification of the wiring that can influence various actions and reactions of the subject so long as it persists. And the same holds for other propositional attitudes. Thus in speaking of voluntary control of beliefs, we have been speaking of the control of states. But couldn't we just as well speak of the voluntary control of, and responsibility for, the action of bringing about such states: accepting, rejecting, or withholding a proposition, forming a belief, or refraining from believing?15 The two loci of responsibility and control may seem strictly correlative, so that we can equally well focus on either. For one exercises voluntary control over a type of state, C, by voluntarily doing something to bring it about or inhibit it. And from the other side, every action can be thought of as the bringing about of a state of affairs. Whenever we are responsible for a state of affairs by virtue of having brought it about, we may just as well speak of being responsible for the action of bringing it about. There are the following reasons, however, for proceeding in terms of states.

First, in holding that beliefs are subject to deontological evaluation since under voluntary control, one need not restrict oneself to beliefs that were formed intentionally by a voluntary act. I can be blamed for believing that p in the absence of adequate evidence, even though the belief was formed quite automatically, not by voluntarily carrying out an intention to do so. Provided believing in general is under voluntary control, any belief can be assessed deontologically. It is enough that I could have adopted or withheld the proposition by a voluntary act, had I chosen to do so.

Another consideration that decisively favors the focus on states is that, as we shall see later, there is a way in which one can be responsible and blameworthy for a state of belief, or other state, even if one lacks the capacity to bring about such states intentionally.

The final preliminary note is this. Our issue does not concern free will or freedom of action, at least not in any sense in which that goes beyond one's action's being under the control of the will. On a "libertarian" conception of free will this is not sufficient; it is required in addition that both A and not-A be causally possible, given all the causal influences on the agent. And other requirements may be imposed concerning agency. A libertarian will, no doubt, maintain that if deontological concepts are to apply to believings in the same way as to overt actions, then all of his conditions for freedom will have to apply to believings as well. In this essay, however, I shall only be concerned with the issue of whether believings are under voluntary control. If, as I shall argue, this condition is not satisfied for believings, that will be sufficient to show that they are not free in the libertarian sense as well.

III. Basic Voluntary Control

Let's turn now to a critical examination of the basic control thesis, the thesis that one can take up at will whatever prepositional attitude one chooses. Those who have attacked this view are divided between those who hold that believing at will is logically impossible and those who hold that it is only psychologically impossible, a capacity that we in fact lack though one we conceivably could have had.16 I cannot see any sufficient reasons for the stronger claim, and so I shall merely contend that we are not so constituted as to be able to take up propositional attitudes at will. My argument for this, if it can be called that, simply consists in asking you to consider whether you have any such powers. Can you, at this moment, start to believe that the United Sjates is still a colony of Great Britain, just by deciding to do so? If you find it too incredible that you should be sufficiently motivated to try to believe this, suppose that someone offers you $500,000,000 to believe it, and you are much more interested in the money than in believing the truth. Could you do what it takes to get that reward? Remember that we are speaking about believing at will. No doubt, there are things you could do that would increase the probability of your coming to believe this, but that will be discussed later. Can you switch propositional attitudes toward that proposition just by deciding to do so? It seems clear to me that I have no such power. Volitions, decisions, or choosings don't hook up with anything in the way of propositional attitude inauguration, just as they don't hook up with the secretion of gastric juices or cell metabolism. There could conceivably be individual differences in this regard. Some people can move their ears at will, while most of us cannot. However, I very much doubt that any human beings are endowed with the power of taking on propositional attitudes at will. The temptation to suppose otherwise may stem from conflating that power with others that we undoubtedly do have but that are clearly distinct.17 If I were to set out to bring myself into a state of belief that p, just by an act of will, I might assert that p with an expression of conviction, or dwell favorably on the idea that p, or imagine a sentence expressing p emblazoned in the heavens with an angelic chorus in the background intoning the Kyrie of Mozart's Coronation Mass. All this I can do at will, but none of this amounts to taking on a belief that p. It is all show, an elaborate pretence of believing. Having gone through all this, my doxastic attitudes will remain just as they were before; or if there is some change, it will be a result of these gyrations.

We should not suppose that our inability to believe at will is restricted to propositions that are obviously false. The inability also extends, at least, to those that are obviously true. A few pages back we made the point that voluntary control attaches to contrary pairs, or to more complex arrays of alternatives. If the sphere of my effective voluntary control does not extend both to A and to not-A, then it attaches to neither. If I don't have the power to choose between A and not-A, then we are without sufficient reason to say that I did A at will, rather than just doing A, accompanied by a volition. It is even more obvious, if possible, that responsibility, obligation, and their kindred attach to doing A only if the agent has an effective choice between doing and not doing A. If I would still have done A whatever I willed, chose, or preferred, I can hardly be blamed for doing it.

Thus, even if I willingly, or not unwillingly, form, for instance, perceptual beliefs in the way I do, it by no means follows that I form those beliefs at will, or that I have voluntary control over such belief formation, or that I can be held responsible or blameworthy for doing so. It would have to be true that I have effective voluntary control over whether I do or do not believe that the tree has leaves on it when I see a tree with leaves on it just before me in broad daylight with my eyesight working perfectly. And it is perfectly clear that in this situation I have no power at all to refrain from that belief. And so with everything else that seems perfectly obvious to us. We have just as little voluntary control over ordinary belief formed by introspection, memory, and simple uncontroversial inferences.

The discussion to this point will suggest to the voluntarist that he can still make a stand on propositions that do not seem clearly true or false and hold that there one (often) has the capacity to adopt whatever propositional attitude one chooses. In religion, philosophy, and high level scientific matters it is often the case that, so far as one can see, the relevant arguments do not definitively settle the matter one way or the other. I engage in prolonged study of the mind-body problem or of the existence of God. I carefully examine arguments for and against various positions. It seems to me that none of the positions have decisively proved their case, even though there are weighty considerations that can be urged in support of each. There are serious difficulties with all the competing positions, though, so far as I can see, more than one contender is left in the field in each case. So what am I to do? I could just abandon the quest. But alternatively I could, so it seems, simply decide to adopt one of the positions and/or decide to reject one or more of the contenders. Is that not what I must do if I am to make any judgment on the matter? And isn't that what typically happens? I decide to embrace theism or epiphenomenalism, and forthwith it is embraced.

There are also practical situations in which we are confronted with incompatible answers to a certain question, none of which we see to be clearly true or false. Here we often do not have the luxury of leaving the field; since we must act in one way rather than another, we are forced to form, and act on, some belief about the matter. It would be a good idea for me to plant these flowers today if and only if it will rain tomorrow. But it is not at all clear to me whether tomorrow will be rainy. I must either plant the flowers today or not, and it would surely be unwise to simply ignore the matter, thereby in effect acting uncritically on the assumption that it will not rain tomorrow. Hence the better part of wisdom would be to make some judgment on the matter, the best that I can. On a larger scale, a field commander in wartime is often faced with questions about the current disposition of enemy forces. But often the information at his disposal does not tell him just what that disposition is. In such a situation is it not clear that, weighing available indications as best he can, he simply decides to make a certain judgment on the matter and act on that? What else can he do?

Before responding to these claims I should point out that even if they were correct, it would still not follow that a deontological conception of justification is adequate for epistemology. For the voluntarist has already abandoned vast stretches of the territory. He has given up all propositions that seem clearly true or false, and these constitute the bulk of our beliefs. Controversial and difficult issues force themselves on our attention, especially if we are intellectuals, just because we spend so much of our time trying to resolve them. But if we survey the whole range of our cognitive operations, they will appear as a few straws floating on a vast sea of items about none of which we entertain the slightest doubt. Consider the vast number of perceptual beliefs we form about our environment as we move about in it throughout our waking hours, most of them short-lived and many of them unconscious. By comparison the controversial beliefs we have in religion, politics, philosophy, and the conduct of our affairs are negligible in number, however significant they may be individually. Hence if only the uncertain beliefs are under voluntary control, that will not enable us to form a generally applicable deontological concept of epistemic justification.18

To return to our philosopher, gardener, and military commander, I would suggest that in each case the situation is better construed in some way other than as initiating a belief at will.19 The most obvious suggestion is that although in these cases the supporting considerations are seen as less conclusive, here too the belief follows automatically, without intervention by the will, from the way things seem at the moment to the subject. In the cases of (subjective) certainty belief is determined by that sense of certainty, or, alternatively, by what leads to it, the sensory experience or whatever; in the cases of (subjective) uncertainty belief is still determined by what plays an analogous role, the sense that one alternative is more likely than the others, or by what leads to that. Thus when our philosopher or religious seeker "decides" to embrace theism or the identity theory, what has happened is that at that moment this position seems more likely to be true, seems to have weighter considerations in its favor, than any envisaged alternative. Hence S is, at that moment, no more able to accept atheism or epiphenomenalism instead, than he would be if theism or the identity theory seemed obviously and indubitably true. This can be verified by considering our capacities in a situation in which the above conditions are not satisfied; theism and atheism, or the various contenders on the mind-body issue, really seem equally likely to be true, equally well or ill supported. If that were strictly the case (and perhaps it seldom is), then could S adopt, for example, theism, just by choosing to do so? When I contemplate that possibility, it seems to me that I would be as little able to adopt theism at will as I would be if it seemed obviously true or obviously false. Here, like Buridan's ass, I am confronted with (subjectively) perfectly equivalent alternatives. If it were a choice between actions, such as that confronting the ass, I need not perish through indecision. I could arbitrarily make a choice, as we often do in a cafeteria line when two alternative salads look equally tempting. (Some people negotiate this more quickly than others.) But doxastic choice is another matter. How could I simply choose to believe one rather than the other when they seem exactly on a par with respect to the likelihood of truth, especially when that subjective probability is rather low? To do so would be to choose a belief in the face of the lack of any significant inclination to suppose it to be true. It seems clear to me that this is not within our power.20

The above account in terms of comparative subjective probability might be correct for all our cases, theoretical and practical. Thus the military commander might adopt the supposition about the disposition of enemy forces that seems to him at the moment best supported by the reports at his disposal. But I believe that there are cases, both theoretical and practical, in which the upshot is not triggered by some differential subjective probability of the alternatives. I have already argued that in those cases the upshot cannot be the formation of a belief, whether at will or otherwise. But then what? Here is one possibility. What S is doing is to resolve to act as if p is true, adopt it as a basis for action. This is often a correct description of situations like the military commander's. He may well have said to himself: "I don't know what the disposition of enemy forces is; I don't even have enough information to make an educated guess. But I have to proceed on some basis or other, so I'll just assume that it is H and make my plans accordingly." This is not to form the belief that the disposition is H; it is not to accept the proposition that the disposition is H, except as a basis for action. It would simply be incorrect to describe the commander as believing that the disposition of enemy forces is H, or having any other belief about the matter. He is simply proceeding on a certain assumption, concerning the truth of which he has no belief at all. One may also make an assumption for theoretical purposes, in order to see how it "pans out", in the hope that one will thereby obtain some additional reasons for supposing it to be true or false. Thus a scientist can adopt "as a working hypothesis" the proposition that the atomic nucleus is positively charged, draw various consequences from it, and seek to test those consequences. The scientist need not form the belief that the atomic nucleus is positively charged in order to carry out this operation: typically he would be doing this because he didn't know what to believe about the matter. Likewise a philosopher might take materialism as a working hypothesis to see how it works out in application to various problems. There may also be blends of the theoretical and the practical. One may adopt belief in God, or some more robust set of religious doctrines, as a guide to life, setting out to try to live in accordance with them, seeking to act and feel one's way into the religious community, in order to determine how the doctrines work out in the living of them, both in terms of how satisfactory and fulfilling a life they enable one to live and in terms of what evidence for or against them one acquires. Where the "acceptance" of a proposition in the absence of a significant subjective probability is not the adoption of a working hypothesis, there are other alternatives. (1) S may be seeking, for whatever reason, to bring himself into a position of believing p; and S or others may confuse this activity, which can be undertaken voluntarily, with believing or judging the proposition to be true. (2) As noted earlier, S may assert that p, overtly or covertly, perhaps repeatedly and in a firm tone, and this, which can be done voluntarily, may be confused with a "judgment" that p, of the sort that inaugurates a state of belief. (3) S may align herself, objectively and/or subjectively, with some group that is committed to certain doctrines -- a church, a political party, a movement, a group of thinkers -- and this, which can be done voluntarily, may be confused with coming to believe those doctrines. I am convinced that the analysis of a wide variety of supposed cases of believing at will in the absence of significant subjective probability would reveal that in each case forming a belief that p has been confused with something else. Thus I think that there is a strong case for the proposition that no one ever acquires a belief at will. But even if I am wrong about that, the above considerations do at least show that it is of relatively rare occurrence, and that it certainly cannot be used as the basis for a generally applicable deontological concept of epistemic justification.

IV. Nonbasic Immediate Voluntary Control

The demise of basic control, however, is by no means the end of "voluntarism", as we may term the thesis that one has voluntary control of propositional attitudes. Many deontologists, after disavowing any commitment to what they usually call "direct voluntary control of belief" and what we have called "basic voluntary control", proceed to insist that beliefs are subject to what they term "indirect voluntary control".21 All of them use the term 'indirect control' in an undiscriminating fashion to cover any sort of control that is not "direct", that is, basic. As a result they fail to distinguish between the three sorts of nonbasic control I shall be distinguishing.22 Some of their examples fit one of my three categories, some another. The ensuing discussion will show important differences between these three modes of control.

To get into this, let's first note that we take many familiar nonbasic overt actions to be voluntary (and their upshots to be under voluntary control) in a way that is sufficient for their being required, permitted, and prohibited. Consider opening a door, informing someone that p, and turning on a light. To succeed in any of these requires inore than a volition on the part of the agent; in each case I must perform one or more bodily movements and these movements must have certain consequences, causal or conventional, in order that I can be said to have performed the nonbasic action in question. In order for it to be true that I opened a certain door, I must pull it, push it, kick it, or put some other part of my body into suitable contact with it (assuming that I lack powers of telekinesis), and this must result in the door's coming to be open. In order to inform H that p, I must produce various sounds, marks, or other perceptible products, and either these products must fall under linguistic rules in such a way as to constitute a vehicle for asserting that p (if we are thinking of informing as an illocutionary act), or H, upon perceiving these products, must be led to form the belief that p (if we are thinking of informing as a perlocutionary act). Hence actions like these are not immediately consequent on a volition and so are not strictly done "at will". Nevertheless I might be blamed for my failure to turn on the light when it was my obligation to do so. The point is that in many cases we take the extra conditions of success for granted. We suppose that if the agent will just voluntarily exert herself, the act will be done. Here we might say that the action, and its upshot, is under the "immediate voluntary control" of the agent (more strictly, nonbasic immediate voluntary control), even though more than an act of will is required of the agent. I call this "immediate"23 control since the agent is able to carry out the intention "right away", in one uninterrupted intentional act, without having to return to the attempt a number of times after having been occupied with other matters.24 I will use the term "direct control" for both basic and immediate control. It is clear that if beliefs were under one's immediate control, that would suffice to render them susceptible to deontological evaluation.

But are beliefs always, or ever, within our immediate voluntary control? Our discussion of this will be largely a rerun of the discussion of basic control, with some added twists. As in the earlier discussion we can first exempt most of our doxastic situations from serious consideration. With respect to almost all normal perceptual, introspective, and memory propositions, it is absurd to think that one has any such control over whether one accepts, rejects, or withholds the proposition. When I look out my window and see rain falling, water dripping off the leaves of trees, and cars passing by, I no more have immediate control over whether I accept those propositions than I have basic control. I form the belief that rain is falling willy-nilly. There is no way I can inhibit this belief. At least there is no way I can do so on the spot, in carrying out an uninterrupted intention to do so. How would I do so? What button would I push? I could try asserting the contrary in a confident tone of voice. I could rehearse some skeptical arguments. I could invoke the Vedantic doctrine of maya. I could grit my teeth and command myself to withhold the proposition. But unless I am a very unusual person, none of these will have the least effect. It seems clear that nothing any normal human being can do during the uninterrupted operation of an intention to reject the proposition that it is raining (in the above situation) will have any chance at all to succeed. And the same can be said for inferential beliefs in which it is quite clear to one that the conclusion is correct. Since cases in which it seems perfectly clear to the subject what is the case constitute an enormously large proportion (I would say almost all) of propositions that are either the object of a definite attitude or considered as a candidate for such, the considerations of this paragraph show that immediate voluntary control cannot be a basis for the application of deontological concepts to most of our propositional attitudes.

But what about situations in which it is not clear whether a proposition is true or false? This is where voluntarists tend to take their stand. After all, they say, that is what inquiry is for, to resolve matters when it is not clear what the correct answer is. One certainly has voluntary control over whether to keep looking for evidence or reasons, and voluntary control over where to look, what steps to take, and so on. Since one has control over those matters, that amounts to what I have called immediate voluntary control over one's propositional attitudes.

If self-control is what is essential to activity, some of our beliefs, our believings, would seem to be acts. When a man deliberates and comes finally to a conclusion, his decision is as much within his control as is any other deed we attribute to him. If his conclusion was unreasonable, a conclusion he should not have accepted, we plead with him: "But you needn't have supposed that so-and-so was true. Why didn't you take account of these other facts?" We assume that his decision is one he could have avoided and that, had he only chosen to do so, he could have made a more reasonable inference. Or, if his conclusion is not the result of a deliberate inference, we may say, "But if you had only stopped to think", implying that, had he chosen, he could have stopped to think. We suppose, as we do whenever we apply our ethical or moral predicates, that there was something else the agent could have done instead.25

To be sure, the mere fact that one often looks for evidence to decide an unresolved issue does not show that one has immediate control, or any other sort of control, over one's propositional attitudes. That also depends on the incidence of success in these enterprises. And sometimes one finds decisive evidence and sometimes one doesn't. But let's ignore this complexity and just consider whether there is a case for immediate control of propositional attitudes in the successful cases.

No, there is not, and primarily for the following reason. These claims ignore the difference between doing A in order to bring about E, for some definite E, and doing A so that some effect within a certain range will ensue. In order that the "looking for more evidence" phenomenon would show that we have immediate voluntary control over propositional attitudes in basically the way we do over the positions of doors and light switches, it would have to be the case that the search for evidence was undertaken with the intention of taking up a certain particular attitude toward a particular proposition. For only in that case would the outcome show that we have exercised voluntary control over what propositional attitude we take up. Suppose that I can't remember Al Kaline's lifetime batting average and I look it up in the baseball almanac. I read there the figure .320, and I thereby accept it. Does that demonstrate my voluntary control over my belief that Kaline's lifetime batting average was .320? Not at all. At most it shows that I have immediate voluntary control over whether I take up some propositional attitude toward some proposition ascribing a lifetime batting average to Kaline. This is not at all analogous to my exercising my capacity to get the door open whenever I choose to do so. Its nearest analogue in that area would be something like this. I am a servant and I am motivated to bring the door into whatever position my employer chooses. He has an elaborate electronic system that involves automatic control of many aspects of the household, including doors. Each morning he leaves detailed instructions on household operations in a computer. Doors can only be operated through the computer in accordance with his instructions. There is no way in which I can carry out an intention of my own to open or to close a door. All I can do is to actuate the relevant program and let things take their course. Since the employer's instructions will be carried out only if I actuate the program, I am responsible for the doors' assuming whatever position he specified, just as I was responsible for taking up some attitude or other toward some proposition within a given range. But I most emphatically am not responsible for the front door's being open rather than closed, nor can I be said to have voluntary control over its specific position. Hence it would be idle to apply deontological concepts to me vis-a-vis the specific position of the door: to forbid me or require me to open it, or to blame or reproach me for its being open. I had no control over that; it was not subject to my will. And that's the way it is where the only voluntary control I have over my propositional attitudes is to enter onto an investigation that will eventuate in some propositional attitude or other, depending on what is uncovered. That would be no basis for holding me responsible for believing that p rather than rejecting or withholding it, no basis for requiring me or forbidding me to believe that p, or for reproaching me for doing so.

If Chisholm's claim is only that one can voluntarily put oneself in a position from which some doxastic attitude to p will be forthcoming (or perhaps that one can put oneself in a position such that a desirable doxastic attitude to p will be forthcoming), this capacity extends to all sorts of propositions, including those over which we obviously have no voluntary control. Consider propositions concerning what is visible. I have the power to open my eyes and look about me, thereby putting myself in a position, when conditions are favorable, to reliably form propositions about the visible environment. Again, with respect to past experiences I can "search my memory" for the details of my experiences of the middle of yesterday, thereby, usually, putting myself in an excellent position to reliably form beliefs about my experiences at that time. No one, I suppose, would take this to show that I have immediate voluntary control over what I believe about the visible environment or about my remembered experiences. And yet this is essentially the same sort of thing as the search for additional evidence, differing only in the type of belief-forming mechanism involved.

I suspect that deontologists like Chishom secretly suppose that the additional evidence, rather than "automatically" determining the doxastic attitude, simply puts the subject in a position to make an informed choice of an attitude. That is, despite their official position, they really locate the voluntary control in the moment of attitude formation rather than in the preliminary investigation, thus in effect taking the direct voluntary control position. But then, faced with the crashing implausibility of that position, they think to save the application of deontological concepts by pushing the voluntary control back to the preliminary search for decisive considerations. It is, then, their secret, unacknowledged clinging to the basic control thesis that prevents them from seeing that voluntary control of the investigation phase has no tendency to ground the deontological treatment of propositional attitudes. I must confess that I have no real textual evidence for this speculation, and that I am attracted to it by the fact that it explains an otherwise puzzling failure of acute philosophers to see the irrelevance (to this issue) of our voluntary control over the conduct of inquiry.

Thus far I have been considering one way in which deontologists seek to defend a claim of immediate voluntary control over beliefs. We have seen that way to fail by irrelevance, since it has to do with voluntarily putting oneself in a position to form the most rational attitude, whatever that may be, rather than voluntarily taking up some specific attitude. However, there is no doubt but that people do sometimes set out to get themselves to believe that p, for some specific p. People try to convince themselves that X loves them, that Y will turn out all right, that the boss doesn't really have a negative attitude toward them, that the Red Sox will win the World Series, that materialism is true, or that God exists. Epistemologists don't like to cite such disreputable proceedings as a ground for the application of deontological concepts. To try to get oneself to believe that p, prior to being in a good position to tell whether p is true or not, is not a procedure to be commended from the epistemic standpoint. Nevertheless, these undertakings have to be considered in a comprehensive survey of possible modes of voluntary control. Proceeding in that spirit, the point to note here is that such goings-on provide no support for a supposition of immediate voluntary control over belief. For such enterprises can be successfully carried out only as long-term projects. If I, not currently believing that X loves me, were to set out to bring about that belief in one fell swoop, that is, during a period of activity uninterruptedly guided by the intention to produce that belief, then, unless I am markedly abnormal psychologically, I am doomed to failure. We just don't work that way. Again, I wouldn't know what button to push. My only hope of success would lie in bringing various influences to bear upon myself and shielding myself from others, in the hope of thereby eventually moving myself from disbelief to belief. This might include dwelling on those encounters in which X had acted lovingly toward me, shutting out evidences of indifference or dislike, encouraging romantic fantasies, and so forth. Thus this sort of enterprise belongs, rather, to the category of long-range voluntary control, a topic to which we now turn.

V. Long-range Voluntary Control

We have seen that we cannot plausibly be credited with either sort of direct control over our propositional attitudes. Taking up such an attitude can be neither a basic action like raising one's arm nor a non-basic action like flipping a switch. Hence the deontological treatment of belief can borrow no support from the applicability of deontological terms to actions like these. But the possibility still remains that we have more long-range voluntary control over belief. The considerations of the last paragraph encourage this supposition, at least for some cases. Before examining this possibility I will firm up the distinction between this type of control and the previous one.

In introducing the notion of "immediate control" I said that when one has this species of control over a type of state, C, one is able to bring about a C "right away, in one uninterrupted intentional act". When conditions are propitious, one can get a door open, get a light on, get one's shoes on, or tell Susie the mail has come, by doing various things under the direction of a single uninterrupted intention to bring about that state of affairs. One does not have to return to the attempt a number of times after having been occupied with other matters.26 And since one is not ordinarily capable of keeping an intention in an active state for more than a relatively short period of time, the sorts of actions over which one has immediate control must be capable of execution within a, short time after their inception.

Long-range control is simply the foil of immediate control. It is the capacity to bring about a state of affairs, C, by doing something (usually a number of different things) repeatedly over a considerable period of time, interrupted by activity directed at other goals. One has this sort of control, to a greater or lesser degree/over many things: one's weight, cholesterol concentration, blood pressure and disposition; the actions of one's spouse or one's department. One can, with some hope of success, set out on a long-range project to reduce one's weight, improve one's disposition, or get one's spouse to be more friendly to the neighbors. The degree of control varies markedly among these examples. I have, within limits, complete control over my weight; only sufficient motivation is required to achieve and maintain a certain weight. My ability to change my disposition or to change behavior patterns in my spouse is much less. But all these cases, and many more, illustrate the point that one can have long-range control over many things over which one lacks direct control. I cannot markedly reduce my weight right away, by the uninterrupted carrying out of an intention to do so, for instance, by taking a pill, running around the block, or saying "Abracadabra". But that doesn't nullify the fact that I have long-range control.

It does seem that we have some degree of long-range voluntary control over at least some of our beliefs. As just noted, people do set out on long-range projects to get themselves to believe a certain proposition, and sometimes they succeed in this. Devices employed include selective exposure to evidence, selective attention to supporting considerations, seeking the company of believers and avoiding nonbelievers, self-suggestion, and more bizarre methods like hypnotism. By such methods people sometimes induce themselves to believe in God, in materialism, in communism, in the proposition that they are loved by X, and so on. Why doesn't this constitute a kind of voluntary control that grounds deontological treatment?

Well, it would if, indeed, we do have sufficient control of this sort. Note that people could properly be held responsible for their attitudes toward propositions in a certain range only if those who set out to intentionally produce a certain attitude toward such a proposition, and made sufficient efforts, were frequently successful.27 For only if we are generally successful in bringing about a goal, G, when we try hard enough to do so, do we have effective control over whether G obtains. And if I don't have effective control over G, I can hardly be held to blame for its nonoccurrence. Even if I had done everything I could to produce it, I would have had little chance of success; so how could I rightly be blamed for its absence? (I might be blamed for not trying to produce it or for not trying hard enough, but that is another matter.) This is a generally applicable principle, by no means restricted to the doxastic sphere. If I am so constituted that the most I can do with respect to my irritability is to make it somewhat less likely that it will exceed a certain (rather high) average threshold, I can hardly be blamed for being irritable.

It is very dubious that we have reliable long-range control over any of our beliefs, even in the most favorable cases, such as beliefs about religious and philosophical matters and about personal relationships. Sometimes people succeed in getting themselves to believe (disbelieve) something. But I doubt that the success rate is substantial. To my knowledge there.are no statistics on this, but I would be very much surprised if attempts of this sort bore fruit in more than a very small proportion of the cases. In thinking about this, let's first set aside cases in which the attempt succeeds because the subject happens onto conclusive evidence that would have produced belief anyway without deliberate effort on his part to produce belief. These are irrelevant because the intention to believe that p played no effective role. Thus we are considering cases in which the subject is swimming against either a preponderance of contrary evidence or a lack of evidence either way. S is fighting very strong tendencies to believe when and only when something seems true to one. Some of these tendencies are probably innate and some engendered or reinforced by socialization; in any event they are deeply rooted and of great strength. To combat or circumvent them one must exercise considerable ingenuity in monitoring the input of information and in exposing oneself to nonrational influences. This is a tricky operation, requiring constant vigilance as well as considerable skill, and it would be very surprising if it were successful in a significant proportion of the cases. I am not suggesting that it is unusual for people to form and retain beliefs without adequate grounds, reasons, or justification. This is all too common. But in most such cases the proposition in question seems clearly true, however ill supported. The typical case of prejudice is not one in which S manages to believe something contrary to what seems to him to be the case or something concerning which he has no definite impression of truth or falsity. It is a case in which his socializat has led it to seem clearly true to him that, for example, blacks are innately inferior.

Thus a long-range control thesis does not provide much grounding for deontologism, even for the sorts of propositions people do sometimes try to get themselves to believe or disbelieve. Much less is there any such grounding for those propositions with respect to which people don't normally even try to manipulate their attitudes. We have already noted that most of our beliefs spring from doxastic tendencies that are too deeply rooted to permit of modification by deliberate effort. Most of the matters on which we form beliefs are such that the project of deliberately producing belief or disbelief is one that is never seriously envisaged, just because it is too obvious that there is no chance of success. Thus even if we were usually successful when we set out to produce a propositional attitude, the voluntary control thus manifested would not ground the application of deontological concepts to beliefs generally. So, once again, the most we could conceivably have (and I have argued that we do not in fact have even that) would fall short of a generally applicable deontological concept of justification.

VI. Indirect Voluntary Influence: A Different Deontological Conception of Epistemic Justification

Up to this point I have been examining the support for a deontological conception of epistemic justification provided by the treatment of propositional attitude formation on the model of intentional action. We have considered whether, or to what extent, it is in our power to carry out an intention to take up a certain propositional attitude, either at will (basic control), or while uninterruptedly guided by the intention to do so (immediate control), or as a complex long-term project (long-range control). We have seen that for most of our beliefs we have control of none of these sorts, and that for the others we have, at most, some spotty and unreliable control of the long-range sort. I conclude that we do not generally have the power to carry out an intention to take up a certain propositional attitude. Insofar as the conception of epistemic justification as believing as one is permitted to depends on that assumption, it must be rejected. The inauguration of propositional attitudes simply does not work like intentional action.

However, this is not necessarily the end of the line for the deontologist. He has another move. We can be held responsible for a state of affairs that results from our actions even if we did not produce that state of affairs intentionally, provided it is the case that something we did (didn't do) and should have not done (done) was a necessary condition (in the circumstances) of the realization of that state of affairs, that is to say, provided that state of affairs would not have obtained had we not done (done) something we should not have done (done). Suppose that, although I did not do anything with the intention of bringing about my cholesterol buildup, still I could have prevented it if I had done certain things I could and should have done, for instance, reduce fat intake. In that case I could still be held responsible for the condition; it could be my fault. This is a way in which deontological concepts can be applied to me, with respect to a certain state of affairs, even though that state of affairs did not result from my carrying out an intention to produce it.

This suggests that even if propositional attitudes are not under our effective voluntary control, we might still he held responsible for them, provided we could and should have prevented them; provided there is something we could and should have done such that if we had done it we would not have had the attitude in question. If this is the case, it could provide a basis for the application of deontological concepts to propositional attitudes, and, perhaps, for a deontological concept of epistemic justification, one that bypasses the above critique. Let's use the term 'indirect voluntary influence' for this kind of voluntary control, or better, "voluntary impact" we may have on our beliefs.

It may be helpful to display in outline form the various modes of voluntary control we have distinguished.

  1. Direct control
    1. Basic control
    2. Nonbasic immediate control
  2. Long-range control
  3. Indirect influence

Now it does seem that we have voluntary control over many things that influence belief. These can be divided into (1) activities that bring influences to bear, or withhold influences from, a particular situation involving a particular candidate, or a particular field of candidates, for belief, and (2) activities that affect our general belief-forming habits or tendencies.28 There are many examples of (1). With respect to a particular issue, I have voluntary control over whether, and how long, I consider the matter, look for relevant evidence or reasons, reflect on a particular argument, seek input from other people, search my memory for analogous cases, and so on. Here we come back to the activities that people like Chisholrn wrongly classify as the intentional inauguration of a propositional attitude. Although the fact that it is within my power to either look for further evidence or not does not show that I have voluntary control over what attitude I take toward p, it does show that I have voluntary control over influences on that attitude. The second category includes such activities as training myself to be more critical of gossip, instilling in myself a stronger disposition to reflect carefully before making a judgment on highly controversial matters, talking myself into being less (more) subservient to authority, and practicing greater sensitivity to the condition of other people. It is within my power to do things like this or not, and when I do them with sufficient assiduity I make some difference to my propositional attitude tendencies, and thus indirectly to the formation of such attitudes.29

Actually, there would be no harm in including in the first category attempts to bring about a certain specific attitude, and the successful carrying out of such an attempt when and if that occurs. For these too would be things over which we have voluntary control that influence our propositional attitudes. The point of stressing other things is that, since our earlier discussions have provided reason for thinking that such attempts are rarely successful, I want to emphasize the point that even if we are never successful in carrying out an intention to believe, (reject, withhold) p, still there are many things over which we have voluntary control that do have a bearing on what propositional attitudes are engendered.

Thus it will sometimes be the case that had we performed (not performed) some voluntary actions A, B, . . . , we would have (not have) taken up some attitude we did not (did) take up. The only remaining question is as to whether deontological concepts apply to the sorts of activities we have been discussing. Is it ever the case that we ought or ought not to engage in some activity of searching for new evidence or refraining from doing so? Is it ever the case that we ought (ought not) to strive to make ourselves more (less) critical of gossip or more (less) sensitive to contrary evidence? Deontologists typically aver that we have intellectual obligations in such matters, obligations rooted in our basic intellectual obligation to seek the true and avoid the false, or, alternatively, rooted in our basic aim, need, or commitment to believe the true and avoid believing the false. Let's go along with our opponents on this point. I can do so with a clear conscience, since I am seeking to show that even if we admit this, and make the other concessions I have been making, a deontological conception of epistemic justification is not viable.

Thus it will sometimes be the case, when I believe that p, that I would not have done so had I done various things in the past that I could and should have done but failed to do. Suppose that I accept some idle gossip to the effect that Jim is trying to undermine Susie's position as chair of the department. It may be that had I been doing my duty by way of making myself more critical of gossip, and by way of checking into this particular matter, I would not have formed that belief or would not have retained it for so long. In that case I could be held responsible for believing this in the same way as that in which I can be held responsible for my cholesterol buildup. I can be properly blamed for it, even though I did not intentionally bring it about.

Note that this application of deontological concepts to beliefs is a derivative one. What is primarily required, permitted, and forbidden are the voluntary activities we ranged in two categories, various sorts of activities that influence belief. Deontological concepts are applied to beliefs only because of some relation these attitudes have to those primary targets of permission, and the like. This asymmetrical relation of dependence attaches to all those cases in which one is responsible for a state of affairs without being responsible for an action of intentionally bringing it about.

Now let's consider just what deontological terms can be applied to beliefs in this derivative way and how this application is to be understood. Remember that we are taking requirement, prohibition, and perrmission to be the basic,deontological concepts. When dealing with intentional actions, it is best to think first of general principles that lay down conditions under which an action of a certain sort is required, forbidden, or permitted, and then consider a particular action to have one of these statuses because it exemplifies some general principle. Thus if we take the forming of a belief to be an intentional action, we will envisage general principles that hold, for instance, that it is forbidden to believe that p in the absence of sufficient evidence. Then if I form a particular belief without sufficient grounds, that belief is forbidden, or, if you prefer, I have violated a prohibition in forming that belief. We can then apply other deontological terms like 'responsible', 'blame', and 'praise' on this basis. If one intentionally does something that falls under a principle of one of the above sorts, one is responsible for what one has done. If in doing it one has violated a requirement or a prohibition, one can rightly be blamed for it. If one has not violated any requirement or prohibition, one is justified in doing it.

But on the present way of looking at the matter, we can have no principles laying down conditions under which a belief is required, forbidden, or permitted, just because we lack sufficient voluntary control over belief formation. What the relevant principles will require, and so forth, are activities that are designed to influence factors that, in turn, will influence belief formation.30 Hence there is no basis for taking a particular belief to be required, prohibited, or forbidden. And so if we are to say, on the rationale given above, that one can be responsible and blameworthy for a belief, that will be the case even though the belief is not prohibited. If one is puzzled by this, the cure comes from realizing that responsibility and blame supervene on requirement, prohibition, and permission in two quite different ways. First, and most simply, one is to blame for doing something forbidden or for failing to something required. But second, one is also to blame for the obtaining of some fact if that fact would not have obtained if one had not behaved in some manner for which one is to blame in the first sense, that is, for doing something forbidden or failing to do something required.

So far, in discussing indirect influence, we have seen that one can be to blame for a certain propositional attitude provided one wouldn't have that attitude had one not failed to conform to some intellectual requirement or prohibition. But this formulation must be refined. On reflection it turns out to be too broad. There are certain ways in which dereliction of duty can contribute to belief formation without rendering the subject blameworthy for forming that belief. Suppose that I fail to carry out my obligation to spend a certain period in training myself to look for counterevidence. I use the time thus freed up to take a walk around the neighborhood. In the course of this stroll I see two dogs fighting, thereby acquiring the belief that they are fighting. There was a relevant intellectual obligation I didn't fulfill, which is such that if I had fulfilled it I wouldn't have acquired that belief. But if that is a perfectly normal perceptual belief, I am surely not to blame for having formed it.31

Here the dereliction of duty contributed to belief formation simply by facilitating access to the data. That is not the kind of contribution we had in mind. The sorts of cases we were thinking of were those most directly suggested by the two sorts of intellectual obligations we distinguished: (a) cases in which we acquire or retain the belief only because we are sheltered from adverse considerations in a way we wouldn't have been had we done what we should have done; (b) cases in which the belief was acquired by the activation of a habit we would not have possessed had we fulfilled our intellectual obligations. Thus we can avoid counterexamples like the above by the following reformulation:

(I) S is (intellectually) to blame for believing that p iff if S had fulfilled all her intellectual obligations, then S's belief forming habits would have changed, or S's access to relevant adverse considerations would have changed, in such a way that S would not have believed that p.

Another issue has to do with the "absoluteness" of the counterfactual involved in this formulation. (I) involves the flat requirement that S would not have believed that p under these conditions. But perhaps S is also blameworthy for believing that p if some weaker condition holds, for instance, that it would be much less likely that S would have believed that p had S fulfilled her intellectual obligations. Of course, the relation between this and (I) depends on one's account of counterfactuals. For present purposes we need not enter this forbidding swamp. I am shortly going to argue that the concept of epistemic justification that emerges from (I) is inadequate for epistemology, and that argument will not rest on taking the counterfactual to be stronger or weaker.

We can now move on to developing a deontological notion of epistemic justification that is based on the above. One point is obvious: when S is to blame for believing that p, that belief is not justified. But that will presumably cover only a tinVproportion of beliefs. What about the others? X^

One possibility would be to treat being justified as the mirror image of being unjustified. To justifiably believe that p, then, is for one's belief that p to be to one's credit. That is, one is justified in believing that p iff one wouldn't have believed that p unless one had fulfilled one's intellectual obligations in some way: by doing what is intellectually required of one or by refraining from doing what one is intellectually forbidden to do. This might seem to leave us with precious few justified beliefs. How many of our beliefs have intellectually dutiful deeds as an essential part of their ancestry? But before embracing that conclusion we should remember that part of the formulation that has to do with refraining from doing what is forbidden. The formulation does not imply that we are justified only where some positive act of duty is in the causal ancestry. One might argue that it is always open to us to engage in attempts to build up disreputable belief-forming tendencies, for example, wishful thinking; and that often we wouldn't have the perfectly respectable beliefs we do have if we had engaged in that enterprise with sufficient vigor. But even so, it still remains that no such counterfactual would hold for beliefs that are beyond the reach of voluntary endeavors, like typical perceptual, memory, and introspective beliefs. No amount of striving after wishful thinking would dislodge most of these. Hence they would fall outside the scope of justified belief on this construal.

There is a simple way to set up the justified-unjustified distinction on the basis of (I) without coming into so violent a conflict with our ordinary judgments. We can take any belief that is not unjustified by the above criterion to be justified. On this construal, 'unjustified' would be the term that "wears the trousers" (Austin), and 'justified' would simply be its negation. A belief is justified iff the subject is not intellectually to blame for holding it. This brings us back to the deontological conception of justification advocated by Ginet and others. Remember that we quoted Ginet as saying that one is justified provided that "one could not be justly reproached for being confident that p". The only difference is that whereas Ginet was thinking of blame as attaching to belief as something that is itself under voluntary control, we are thinking of it in the more complex, derivative way developed in this section.

VII. Critique of This New Conception

The upshot of the paper thus far is that the only viable deontological conception of justification is the one that identifies being justified in believing that p with not being intellectually to blame for believing that p, in a sense of 'to blame for' explicated in (I). To put this into a canonical formula:

(II) S is justified in believing that p iff it is not the case that if S had fulfilled all her intellectual obligations, then S's belief-forming habits would have changed, or S's access to relevant adverse considerations would have changed, in such a way that S would not have believed that p.

What follows the iff is, of course, the denial of the account of being to blame for a belief given by (I). "Of course", since on this conception, being justified in believing that p is just not being to blame for believing that p.

In the remainder of the paper I shall present reasons for denying that (II) gives us a concept of justification that is what we are, or should be, looking for in epistemology. I shall point out ways in which one can be justified according to (II), and yet not justified in any way that is crucial for epistemological concerns; and, conversely, that one can be justified in an epistemologically crucial way and yet not deontologically justified, as spelled out by (II). But first a terminological disclaimer. My linguistic intuitions tell me that 'justified' and its cognates are properly used only in a deontological sense. To be justified in doing or believing . . . something just is to not have violated any relevant rules, norms, or principles in so doing, believing. . . . If, as I believe, most epistemologists use 'justified' for some quite different notion, they are speaking infelicitously. However, this way of talking is so firmly entrenched that I shall go along with it, albeit with an uneasy linguistic conscience.

If I am to argue that (II) does not amount to real epistemic justification, I must proceed on the basis of some assumption as to what epistemic justification really is. I have no time to argue for any such assumption at the tag end of this paper.32 Hence I shall argue the point separately for two different conceptions, one more externalist, the other more internalist.33 I shall devote most of the time to my favorite conception, which is basically externalist with an internalist twist. Let me begin by briefly explaining that.

Start from the idea of forming a belief in such a way as to be in a good position to get a true belief. Call this, if you like, a "strong position" conception of justification.34 One is justified in believing that p only if that belief that p was formed in such a way as to make it at least very likely that the belief is true, or, as is sometimes said, only if it was formed in a "truth-conducive" way.35 Reliability theory is a natural way of further developing the notion: the belief is justified only if it was formed in a reliable fashion, one that can generally be relied on to produce true beliefs. Note that there is no guarantee that the subject will be aware of the crucial aspects of the mode of formation, much less of the fact that that mode is truth conducive; that is what makes this conception externalist. My internalist twist consists in also requiring that the belief be based on a "ground" that the subject can be aware of fairly readily. This twist does not negate the externalism. I do not also require that the subject be aware, or have the capacity to become aware, that the ground is an adequate one or that the belief was formed in a reliable or a truth-conducive way. My internalist qualification will play no role in what follows. The argument will depend solely on the requirement that the ground be in fact an adequate one (sufficiently indicative of the truth of the belief). I mention my internalist twist only to point out that the ensuing argument against (II) does not depend on embracing the most extreme form of externalism.

The first point to mention about (II) is that the concept does not apply at all to subjects that lack sufficient sophistication, reflectiveness, or freedom to be subject to intellectual requirements, prohibitions, and the like. This includes lower animals and very young children as well as the mentally defective. But I don't want to stress this consideration, since it can be plausibly argued that the notion of epistemic justification has no significant application to such subjects either. If I went along with the popular view that justification is necessary for knowledge, I would resist this claim, for it is clear to me that lower animals and very young children often know what is going on in their environment. But since I am prepared to recognize knowledge without justification, I am free to acknowledge that the notion of epistemic justification gets a foothold only where subjects are capable of evaluating their own doxastic states and those of others and responding to those evaluations appropriately. Hence the discussion will be restricted to normal adult humans, to whom deontological concepts are applicable.

Next I shall explore ways in which one may be deontologically justified in a belief without forming the belief in a truth-conducive way. But first, how we are to tell when one is free of blame in forming a belief? That depends on whether the belief stemmed, in the specified way, from any failure of obligations. But how are we to think of those obligations? I am not now asking about the content of our intellectual obligations. As for that, I shall simply draw on the illustrations given earlier. I am asking rather: how much is a person obliged to do along these lines in a particular situation? And the main point is that we must distinguish between "counsels of perfection" and what it is reasonable to expect of a person. With world enough and time we could require people to carry out an exhaustive investigation of each witness, search through all the relevant literature for considerations pertinent to each candidate for belief, check each calculation ten times, and so on. But we simply do not have time for all that. Even if we were exclusively devoted to the search for truth, we would not be able to do that for all the matters on which we need to form beliefs. And given that we have various other commitments and obligations, it is doubly impossible. Hence, abandoning counsels of perfection, let us say that one can properly be blamed for a belief only if that belief stems, in the specified way, from failures to do what could reasonably beexpected of one: simply 1 failing to do what would be ideally adequatels not enough.

In Essay 4, I present two examples of subjects who are deontologically justified but in a poor position to get the truth. One is a case of cultural isolation. S has lived all his life in an isolated primitive community where everyone unhesitatingly accepts the traditions of the tribe as authoritative. These have to do with alleged events distant in time and space, about which S and his fellows have no chance to gather independent evidence. S has never encountered anyone who questions the traditions, and these traditions play a key role in the communal life of the tribe. Under these conditions it seems clear to me that S is in no way to blame for forming beliefs on the basis of the traditions. He has not failed to do anything he could reasonably be expected to do. His beliefs about, for example, the origins of the tribe stem from what, so far as he can see, are the best grounds one could have for such beliefs. And yet, let us suppose, the traditions have not been formed in such a way as to be a reliable indication of their own truth. S is deontologically justified, but he is not believing in a truth-conducive way.

The first half of this judgment has been challenged by Matthias Steup, who takes a hard line with my tribesman.

No matter how grim the circumstances are, if an agent holds a belief contrary to evidence, it is within his power, given that he is a rational agent, to reflect upon his belief and thereby to find out that he had better withhold it, or even assent to its negation. Being a rational agent, I would say, involves the capacity to find out, with respect to any belief, whether or not it is being held on good grounds.36

Hence, contrary to my judgment, S is not free of intellectual blame and so is not deontologically justified.

I think that Steup is displaying an insensitivity to cultural differences. He supposes that there are standards recognized in all cultures that determine what is adequate evidence, or good enough grounds, for one or another kind of belief. That does not seem to me to be the case. There may very well be transcultural epistemic standards, such as consistency and reliability, but I see no reason to suppose that they are sufficient to settle all issues as to what counts as adequate reasons or grounds. On the contrary, the criteria for this vary significantly from one culture to another. The judgments of adequacy of grounds that are transmitted across generations will differ across cultures. Hence what can reasonably be expected of a subject with respect to, for instance, critical examination of beliefs and their bases will differ across cultures. We requjre adults in our culture to be critical of "tradition", but this is a relatively recent phenomenon, given the time humans have been on earth; it cannot be reasonably required of everyone in every society. Note that I am not saying that what is adequate evidence varies with the culture. I am no cultural relativist. On the contrary. My judgment that S's belief lacks adequate grounds was based on the supposition that there are objective standards for adequacy of grounds that hold whatever is accepted in one or another culture. But that is just the point. Deontological justification is sensitive to cultural differences because it depends on what can reasonably be expected of one, and that in turn depends on one's social inheritance and the influences to which one is exposed. But truth conducivity does not so depend. Hence they can diverge.

The other case I presented was a "cognitive deficiency" case. It concerned a college student who doesn't have what it takes to follow abstract philosophical exposition or reasoning. Having read parts of Book IV of Locke's Essay, he takes it that Locke's view is that everything is a matter of opinion. He is simply incapable of distinguishing between that view and Locke's view that one's knowledge is restricted to one's own ideas. There is nothing he could do, at least nothing that could reasonably be expected of him, given his other commitments and obligations, that would lead him to appreciate that difference. Hence he cannot be blamed for interpreting Locke as he does; he is doing the best he can. But surely this belief is outrageously ill grounded, based as it is on the student's dim-witted impressions of Locke.

Steup challenges this case by claiming that even if the student is incapable of attaining a better understanding of Locke, he could have done something that would have led him to withhold acceptance of the interpretation in question, viz., ask himself "Do I understand Locke's Essay well enough to be justified in assenting to this interpretation?".37 Now, as Steup intimates, I certainly don't want to depict the case in such a way that the student is incapable of asking himself this question. I do, however, want to construe it in such a way that asking the question would not lead him to withhold assent. The case I have in mind is one in which the student feels quite confident of his reading; this is definitely the way it strikes him, and he has no tendency to doubt it (at least not prior to seeing the grade he gets on the final exam). Certainly that scenario is a possible one, and it, too, illustrates the possibility of a gap between deontological justification and truth-conducive justification.

However, it may have been poor strategy to trot out this hapless student as one of only two cases, for it undoubtedly raises too many controversial issues. Moreover, it may give the impression that counterexamples based on cognitive deficiency are limited to such extreme and, we may hope, such unusual cases as this, whereas in fact they are all too common. We have such a case whenever one forms a belief, on poor grounds, on something beyond one's intellectual capacity; and this is surely a common occurrence. Just consider a person who forms the belief that socialism is contrary to Christianity, for the reasons that are often given for this view by the New Right, and who is intellectually incapable of figuring out how bad these reasons are.

However, cultural isolation and cognitive deficiency cases only scratch the surface. We have so far been considering cases that are either rather extreme in our culture or come from a very different culture. But there are other sorts of cases that are around us every day. I am thinking particularly of those in which we lack the time or resources to look into a matter in an epistemically ideal fashion. Consider the innumerable beliefs each of us forms on testimony or authority. Most of what we believe, beyond what we experience personally, comes from this source. Ideally we would check out each source to make sure that it is reliable before accepting the testimony. But who has time for that? We can do it in special cases where the matter is of particular importance, but no one could do it tor even a small percentage of the items proferred by others for our belief. Nor is it a real option to withhold belief save where we do run a check. That would leave our doxastic structure so impoverished we would not be able to function in our society. Practically everything we believe about science, history, geography, and current affairs is taken on authority. Moreover, even if we had the time to check up on each authority, in most cases we lack the resources for making an informed judgment. For the same reason that I cannot engage in astrophysics on my own, I am in no position to determine who is a competent authority in the field, except by taking the word of other alleged authorities. Thus in most cases in which I uncritically accept testimony I have done as much as could reasonably be expected of me. And now let us consider those cases in which the authority is incompetent or the witness is unreliable. There we are forming a belief on an objectively unreliable basis, though deontologically justified in doing so. One could hardly deny that this happens significantly often.

This same pattern is found outside the sphere of testimony. Consider perception. Sometimes peoples' eyes deceive them because of physiological or psychological malfunctioning, or because of abnormalities in the environment (cleverly constructed imitations, unusual conditions of the medium, and the like). Should we check for such abnormalities each time we are on the verge of forming a perceptual belief? Obviously we have no time for this, even if our perceptual belief-forming mechanisms were sufficiently under voluntary control. Hence, except where there are definite indications that things are off, we will not have failed in our intellectual obligations if we simply form perceptual beliefs unself-consciously and uncritically; and hence we are deontologically justified in doing so. But now consider cases in which our visual impressions are misleading, even though we are not aware of any indications of this. There one is forming beliefs on an unreliable basis, though deontologically justified in doing so.

Next consider irresistible beliefs and belief tendencies. If it is strictly impossible for me to alter a certain belief or tendency, then I can hardly be expected to do so. But some of these irresistible beliefs may be formed in an unreliable fashion. The most obvious examples concern strong emotional attachments that are, in practice, unshakeable. For many people their religious, or irreligious, beliefs have this statues, as do beliefs concerning one's country, one's close relations, or one's political party. Such beliefs are often not formed in a truth-conducive fashion. But the person cannot be blamed for having something she can't help having, and so we get our discrepancy once more.

Finally, consider timing problems. Suppose that I come to realize that it is incumbent on me to look more fully into matters relevant to basic religious issues: the existence of God, the conditions of salvation, the authority of Scriptures, and so on. I have deeply rooted beliefs on these matters; I am not going to throw them over just because I am reopening the questions, nor am I obliged to do so. And even if suspension of belief would be ideally required, it is not a real possibility for me until I see conclusive negative evidence. In any event, I enter onto my investigation. Let's say that the investigation reveals that my beliefs were ill founded all along. As soon as I see that, I cease to believe, either immediately or after some period of readjustment. But while the investigation is proceeding, something that might occupy many years, I am deontologically justified in continuing to hold the beliefs, for I am not obliged to give them up, even if I could, just because questions have been raised; and yet they are not held on truth-conducive grounds. Again our discrepancy. And again it would seem that such cases are quite frequent.

This completes my case for the possibility, and the actuality, of deontological justification without truth-conducive justification. Even if I am mistaken about the possibility or actuality of some of the above cases, I can safely ignore the possibility that I am mistaken about all. We may take it that our deontological formula, (II), fails to capture what we are looking for in epistemology under the rubric of 'justification', when we are looking for something in the neighborhood of "being in a favorable position in believing that p", favorable from the standpoint of the aim at believing the true and avoiding believing the false.

But we can have discrepancies in the opposite direction as well: believing on an adequately truth-conducive ground while not deontologically justified. This possibility will be realized where: (l) I form a belief that p on ground G; (2) G is in fact an adequate ground for that belief; (3) if I had reflectedcritically on this belief-forming proclivity, as I should have done, I would have found sufficient reasons to doubt its adequacy, and as a result this belief would not have been formed. Here is an example. Let's suppose that it is incumbent on me to look into the credentials of anyone on whose word I believe something of practical importance. An acquaintance, Broom, tells me that Robinson, whom we are considering for a position in my department, has just been made an offer by Princeton. The press of affairs and my instinctive confidence in Broom lead me to neglect my duty and accept Broom's report uncritically. If I had looked into the matter, I would have found strong evidence that Broom is untrustworthy in such matters. This evidence, however, would have been misleading, and in fact Broom is extremely scrupulous and reliable in reporting such things. Thus I formed the belief on an objeeuvely adequate ground, but had I done my intellectual duty I would have mistrusted the ground and hence not formed the belief. I was justified on truth-conductivity standards but not according to (II).

I have been seeking to show that the deontological conception of justification, the only one of that ilk we have found to be internally viable, fails to deliver what is expected of justification if those expectations include truth conducivity. But not all contemporary epistemologists go along with this; in particular, the most extreme internalists do not. Not that they sever justification altogether from the aim at attaining the true and avoiding the false. They hold, to put it into my terms, that for a belief to be justified it is necessary, not that its ground be in fact such as to render it likely that the belief is true, but that the subject be justified in supposing this, that the belief appear to be truth-conducive "from the subject's own perspective on the world".38 Although this kind of internalism is developed in various ways, by no means all of which exactly fit the formula given in the last sentence, we cannot go into all that in this paper. I shall work with the characterization just given.

I believe this view to be subject to an infimte regress of requirements of justification, and to other fatal difficulties.39 My present concern, however, is to point out divergencies between it and (II) and hence to show that the deontological conception runs afoul of both externalism and internalism. After the lengthy discussion just completed I can be briefer here. The general point is that even after one has done everything that is reasonably expected of one intellectually, it is by no means guaranteed that one is justified in supposing that the ground of one's belief is an adequate one. Let's glance at a few of the cases just presented in connection with my moderate externalism. First, the point with respect to irresistible beliefs is precisely the same. If a belief is irresistible, then no matter how intellectually virtuous I am, I will form that belief whether or not I am justified in supposing its ground to be an adequate one. Turning to resistible beliefs, let's note first that we often do not have time to look into whether the ground of the belief is an adequate one. In such cases (assuming the belief doesn't stem from some other failure to carry out intellectual obligations) one would be deontologically justified; but, assuming that a failure to consider the matter would prevent one from being justified in supposing one's ground to be adequate, one would not satisfy internalist requirements.40 Again, consider another version of lack of cognitive powers. It seems plausible to suppose that many cognitive subjects are simply incapable of engaging in a rational consideration of whether the grounds of their beliefs (at least many sorts of beliefs, for instance, those involving complex inductive grounds) are adequate ones. Even if they raise the question, they are not capable of coming to well-grounded conclusions. These people might have done everything that could reasonably be expected of them in an intellectual way, and yet, because of their inability to effectively submit the grounds of their beliefs to a critical assessment, would not be justified in supposing the ground of a certain belief to be sufficient. Again, deontological justification without internalist justification.

The internalist may reply that we have set our standards for higher level justification too high. It is only necessary that one have a justification for believing the ground of one's lower level belief to be an adequate one; and this does not require that one have actually formed any such higher level belief. That is, what should be required is that one be in possession of sufficient evidence for the proposition that the ground in question is adequate, whether or not one has actually made use of that evidence to support a belief in the adequacy of the ground. But we will still get divergencies. The irresistible beliefs will still pose problems. One can't be faulted for holding those beliefs, whether or not one is in possession of adequate evidence for a correlated higher level belief in adequacy. And again it seems that in cases of lack of time I could be deontologically justified in believing that p, whether or not I have sufficient evidence for supposing that ground of that belief to be adequate. Finally, in cases of deficient cognitive powers one could not be faulted for failure of intellectual obligations in believing that p, even if one lacks evidence for the belief in the adequacy of grounds.

Moreover, here too we get the possibility of internalist justification without deontological justification. Consider one who is justified in supposing the ground of her belief that the hostages are in Iran to be an adequate one. That is, the information and general principles at her disposal indicate her evidence to be sufficient. She is internalistically justified in this belief about the hostages. Yet had she engaged in further investigation, as she should have done, her internal perspective would have been enlarged and corrected in such a way that she would no longer be justified in this higher level belief. Her total knowledge and justified belief would then have indicated that her evidence for this proposition is not sufficient. Here we have internalist justification without deontological justification.

Thus the deontological conception embodied in (II) matches an internalist conception of justification no better than it matches an externalist one. I conclude that there is nothing to be said for the deon tological conception, the only one that is not vitiated by internal flaws, as a fundamental concept for epistemology. This is not to deny that it is an interesting and important concept. There are, no doubt, contexts in which it is highly relevant to consider whether a person has failed in any intellectual duty and what bearing this has on the fact that he now believes that p. We would want to consider this if, for example, we were engaged in training the person to be more intellectually responsible or to improve his belief-forming tendencies. But we have seen that these deontological issues are not central to the basic concerns of epistemology with truth and falsity, whether this is conceived externalistically as the formation of propositional attitudes in such a way as to maximize truth and minimize falsity, or internalistically as the formation of propositional attitudes in accordance with what is indicated by the subject's perspective as to the chances for maximizing truth and minimizing falsity. Deontological justification is not epistemic justification.

VIII. Conclusion

Let's draw the threads of this paper together. We have examined several forms of a deontological conception of epistemic justification in terms of freedom from blame in taking up a certain propositional attitude. All of these but one was seen to be untenable by reason of requiring a degree of control over our propositional attitudes that we do not enjoy. The only version that escapes this fate was seen to be not the sort of concept we need to play a central role in epistemology. Therefore, despite the connotations of the term, we are ill advised to think of epistemic justification in terms of freedom from blame for believing.41


1 Robert Audi has suggested that for me to have justifiably done A it is also necessary that I did it because it was permitted by the relevant system of principles (though this is not required, he says, for A's having been justified for me). This would be on the analogy of the distinction between S's being justified in believing that p, and the proposition that p's being justified for S. This may be right, but I am unable to go into the matter that fully here. The point I need for this paper is that permission, rather than requirement, by the relevant principles is necessary for the justification of action.

2 Principles having to do with the way a belief is formed is not the only possibility here. A deontologist might prefer to make the permissibility of a belief depend on what evidence or grounds the subject has for the belief, rather than on the grounds that were actually used as a basis in the formation of the belief. Again, she might prefer to make permissibility depend on whether the belief is formed in the right way or on the basis of an adequate ground, so far as the subject can tell, rather than on the actual chances of the beliefs being true. Since these differences are not germane to the issues of this paper, I have chosen to state the matter in terms of my position, according to which the actual adequacy of the basis on which a belief was formed is crucial for its justificatory status.

3 This formulation is itself subject to both internalist and externalist versions (in fact to several varieties of each), depending on whether the "likelihood" is objective or "within the subject's perspective". Moreover, as foreshadowed in the last sentence in the text, this version of a deontological conception is only the one that is closest to the usual concept of the justification of actions; in the course of this essay it will be found not to be viable and will be replaced by a deontological conception that is further from the action case.

4 See Roderick Chisholm. "Lewis' Ethics of Belief." in The Philosophy of C. I. Lewis, ed. P. A. Schilpp (La Salle, Ill.: Open Court, 1968); Theory of Knowledge, 2d ed. (Englewood Cliffs, N.J.: Prentice-Hall, 1977); "A Version of Foundationalism," in The Foundations of Knowing (Minneapolis: University of Minnesota Press, 1982).

5 Carl Ginet, Knowledge, Perception, and Memory (Dordrecht: D. Reidel, 1975), p. 28. Other epistemologists who explicitly endorse a deontological conception are Laurence Bonjour, The Structure of Empirical Knowledge (Cambridge: Harvard University Press, 1985), chap. 1; Paul Moser, Empirical Justification (Dordrecht: D. Reidel, 1985), chaps. 1, 4; Margery Naylor, "Epistemic Justification," American Philosophical Quarterly, 25, no. 1 (1988), 49-58; Alvin Plantinga, "Reason and Belief in God," in Faith and Rationality, ed. A. Plantinga and N. Wolterstorff (Notre Dame, Ind.: University of Notre Dame Press, 1983); Nicholas Wolterstorff, "Can Belief in God Be Rational If It Has No Foundations?", in ibid.; John Pollock. Contemporary Theories of Knowledge (Totowa, N.J.: Rowman & Littlefield, 1986), pp. 7-8. It should be noted that Plantinga makes it explicit that he would be just as happy with a conception in terms of "excellence" rather than in terms of freedom from blame. This indifference has since shifted to a definite preference for the latter. See his "Positive Epistemic Status and Proper Function," Philosophical Perspectives, 2 (1988). Alvin Goldman, though a reliabiliat, also advocates a deontological conception in Epistemology and Cognition (Cambridge: Harvard University Press, 1980), chap. 4.

6 Various exceptions to the principle have been noted recently. See, e.g., Michael Stocker, "'Ought' and 'Can'," Australasian Journal of Philosophy, 49, no. 3 (1971), 303-16. However, none of the exceptions involve kinds of actions that are not normally under voluntary control. Hence they have no tendency to show that one could be required or forbidden to believe while one generally lacks voluntary control over beliefs. The formulations in this paragraph should be taken as requiring that one normally have voluntary control over one's beliefs, not that one has voluntary control over particular belief in a particular situation.

7 As we have pointed out, for an action or whatever to be justified in a deontological sense is for it to be permitted, rather than required. Thus it is the necessity of an effective choice for something to be permitted that is crucial here.

8 See, e.g., Chisholm, Theory of Knowledge, pp. 14-15.

9 This qualification is needed because things we can bring about by a basic action, e.g., a movement of one's arm, we can also bring about by doing something else, as when one lifts one arm by moving the other arm.

10 I will often omit the modifier 'voluntary' in speaking of control. In this paper it is always to be tacitly understood.

11 Most of these people limit their voluntarism to cases in which it is not clear to the subject whether the belief is true or false. For an excellent account of the history of thought on this subject, with many specific references, see Louis Pojman, Religions Belief and the Will (London: Routledge & Kegan Paul, 1986).

12 See, e.g., Carl Ginet, "Contra Reliabilism," The Monist, 68, no. 2 (1985), and Jack Meiland, "What Ought We To Believe? The Ethics of Belief Revisited," American Philosophical Quarterly, 17 (1980). Neither of these maintains that belief is always under basic voluntary control

13 Theory of Knowledge, chap. 1.

14 To be sure, one might lack the power to determine which of a number of incompatible alternatives is realized, but one could not have the power to choose A at will without also having the power to determine at will that some contrary of A (at least not-A) is realized.

15 I understand "accepting" a proposition as an activity that gives rise to a belief. Therefore, unlike Keith Lehrer, "The Gettier Problem and the Analysis of Knowledge" (in Justification and Knowledge, ed. G. S. Pappas [Dordrecht: D. Reidel, 1979]), and others, I am not using the term in such a sense that one could accept a proposition without believing it; though, of course, the belief engendered by an acceptance may be more or jess long lived. I also recognize processes of belief formation that do not involve any activity of acceptance.

16 The best-known defense of the logical impossibility claim is by Bernard Williams, "Deciding to Believe." in Problems of the Self (Cambridge: Cambridge University Press, 1972). It has been criticized by, inter alia, Trudy Govier, "Belief, Values, and the Will," Dialogue, 15 (1976), and by Barbara Winters, "Believing at Will," Journal of Philosophy, 76 (1979).

17 It may also stem from misdiagnoses of a sort to be presented shortly.

18 Ginet disagrees. He holds that "we can interpret an ascription of unjustifiedness to a belief that the subject cannot help having as saying that, if the subject were able to help it, she ought not to hold the belief" (Knowledge, Perception, and Memory, p. 183). Thus we can extend a deontological concept of justification to irresistible beliefs by invoking a counterfactual. I have two comments to make on this move. (1) This renders epistemic justification quite different from the justification of action, where 'justified' and other deontological terms are withheld from actions the subject couldn't help performing. (2) Insofar as we can make a judgment as to what would be permitted or forbidden were a certain range of involuntary states within our voluntary control, it will turn out that the deontological evaluation is simply a misleading way of making evaluations that could be stated more straightforwardly and more candidly in other terms. Suppose that we judge that if we had voluntary control over the secretion of gastric juices, then we ought to secrete them in such a way as to be maximally conducive to health and a feeling of well being; e.g., we should not secrete them so as to produce hyperacidity. But since gastric juices are not within our voluntary control, this would seem to be just a misleading way of saving that a certain pattern of secretion is desirable or worthwhile. The deontological formulation is a wheel that moves nothing else in the machine.

19 Cf. Poiman, Religious Belief and the Will, chap. 13, for excellent diagnoses of putative cases of basic control of beliefs.

20 In maintaining that one cannot believe that p without its at least seeming to one that p is more probable than any envisaged alternative, I am not joining Richard Swinburne (Faith and Reason [Oxford: Clarendon Press, 1981], chap. 1) in supposing that to believe, that p is just to take p to be more probable than some alternative's).

21 See, e.g., Alvin Goldman, "The Internalist Conception of lustification," Midwest Studies in Philosophy, 5 (1980); Plantinga, "Reason and Belief in God"; Wolterstorff, "Can Belief in God Be Rational?"; Moser, Empirical Justification, chap. 4; Matthias Steup, "The Deontic Conception of Epistemic Justification." Philosophical Studies, 53, no. 1 (1988).

22 Even the extended treatment in Pojman, Religious Belief and the Will, fails to make any distinctions within "indirect control".

23 When the 'nonbasic' qualifier is omitted, 'immediate' is hardly a felicitous term for something that contrasts with basic control. Nevertheless, I shall, for the sake of concision, mostly speak in terms of 'immediate control'. The 'nonbasic' qualifier is to be understood.

24 This notion of doing something "right away" will serve to distinguish the present form of direct control from the next.

25 Chisholm, "Lewis' Ethics of Belief," p. 224.

26 I should make it explicit that I do not suppose that an intention must be conscious, much less focally conscious, during all the time it is playing a role in guiding behavior.

27 I am not saying that S could be held responsible for taking attitude A toward p only if S himself had in fact been successful in intentionally bringing about that attitude. The requirement is rather that p be the sort of proposition toward which people generally are usually successful in bringing about a certain attitude when they try hard enough to do so. If the more stringent requirement were adopted, for actions generally, it would prevent us from holding S responsible for a purely habitual action where he could have successfully carried out an intention to refrain from that action if he had had such an intention. For in that case the action was not in fact the carrying out of a specific intention to perform it. Nevertheless, provided the agent could have inhibited the action had be formed an intention to do, we would feel justified in holding him responsible for it.

28 See Wolterstorff, "Can Belief in God Be Rational?", for an excellent discussion of these modes of influence.

29 Note that the activities in this second category are even further removed from the intentional formation of a certain belief than those in the first, which themselves are clearly distinct from any such thing. The activities in the first category are concerned with a small number of alternatives for attitude formation. Though the activities are not undertaken with the aim of taking up one particular attitude from this field, they are directed to seeking out influences that will resolve this indeterminancy in some way or other. The activities in the second group, however, are directed much more generally to our tendencies of attitude formation on a wide variety of topics and in a wide variety of situations. But the most important point is that in neither case do the activities in question involve the carrying out of an intention to take up a particular propositional attitude.

30 Thus the closest we get to the principle mentioned above would be something like: "One should do what one can to see to it that one is so disposed as to believe that p only when that belief is based on adequate evidence". Hence the power to do things that influence belief formation can be thought of as, inter alia, a higher level capacity to get ourselves into, or make it more likely that we will be in, a condition that would be required of us if we had sufficient voluntary control over belief.

31 I am indebted to Emily Robertson for calling this problem to my attention.

32 For a defense of a particular conception of epistemic justification see Essay 9. For an account of some differences in concepts of epistemic justification, together with a brief defense of a chosen alternative, see Essay 4.

33 For an account of the varieties of internalism and externalism see Essay 8.

34 In Essay 4 this is called an "evaluative" conception, evaluative from the "epistemic point of view".

35 A more detailed account would look into the epistemic status of the belief at times subsequent to its formation as well. We must forgo that in this brief discussion.

36 Steup, "The Deontic Conception of Epistemic Justification," p. 78.

37 lbid., p. 80.

38 See, e.g., Richard Foley, The Theory of Epistemic Rationality (Cambridge: Harvard University Press, 1987), esp. chaps. 1-3; Bonjour, The Structure of Empirical Knowledge, chap. 1; Richard Fumerton, Metaphysical and Epistemological Problems of Perception (Dordrecht: D. Reidel, 1985), chap. 2; Keith Lehrer, "The Coherence Theory of Knowledge," Philosophical Topics, 14, no. 1 (1986).

39 For details see Essay 8.

40 Some epistemologists understand 'S is justified in believing that p' to mean something like 'If S were to believe that p, in S's present situation, that belief would be justified'. On this understanding the above requirement does not hold. However, I have throughout the paper been understanding 'S is justified in believing that p' as 'S justifiably believes that p'.

41 Thanks are due to Robert Audi and Jonathan Bennett for very useful comments on this paper. I have profited from discussions with Carl Ginet about these issues.