Doxastic Self-Control

Transcription

1 1 Doxastic Self-Control I. The thesis that we have direct voluntary control over the content of our beliefs is held by some, but not many. Let us assume that we do not have such control, and that our conscious belief acquisition is hostage to what we take to be true. Is there still a sense in which we can be autonomous with respect to our beliefs? To be autonomous is to be self-governed, as opposed to being determined by forces extrinsic to the self. If a person is autonomous with respect to an attitude or an action, that attitude or action is correctly attributed to the person herself rather than to subpersonal or alien mechanisms. Doxastic autonomy would require that one does not passively find oneself believing certain things; autonomous beliefs must be in some sense self-determined. But selfdetermination and person-level activity are agential notions, and would seem to demand that an autonomous thinker exert control over what she believes. The possibility of doxastic autonomy is thus apparently in tension with the assumption that we have no such direct voluntary control. A further source of doubt as to the possibility of doxastic autonomy lies in skepticism about the coherence of epistemic akrasia. If it is possible to be autonomous in one's beliefs, it must also be possible to fail to exercise that capacity. And a paradigmatic way to fail to be autonomous is to be akratic, allowing oneself to be determined by extrinsic forces when in some sense one could have been self-determined. Reflecting on ways in which a person can be weak with respect to her beliefs may therefore illuminate what it is not to be weak. However, a number of philosophers have argued that there is no such thing as epistemic akrasia. Epistemic akrasia is generally defined as a matter of consciously believing that p while at the same time believing that it is unreasonable to believe that p. 1 It is a matter of some debate whether one can be in a state so described, but many illustrious thinkers have maintained that one cannot. For, as Smith and Pettit explain, " beliefs that run counter to what fact and evidence require will not allow those requirements to remain visible because the offending beliefs themselves give you your sense of what is and your sense of what appears to be." 2 In a related vein, Shah and Velleman claim that the deliberative question whether to believe that p inevitably gives way to the factual question whether p, because the answer to the latter question will determine the answer to the former. 3 If this is right, then there appears to be no 1 Unreasonable here is meant to be neutral between different views on what justifies a belief. 2 Michael Smith and Philip Pettit, "Freedom in Belief and Desire," Journal of Philosophy 93:9, P See also David Owens, "Epistemic Akrasia," Monist 83:5, 2002; Jonathan Adler, "Akratic Believing?" Philosophical Studies 110:1, 2002; and Susan Hurley, Natural Reasons, Oxford University Press, It is possible to interpret G.E. Moore as endorsing this view. 3 Nishi Shah and J. David Velleman, Doxastic Deliberation, Philosophical Review 114:4, 2005, p. 499.

2 independent standpoint from which one can either autonomously conform to the rational requirements on belief or self-consciously fail in this regard. But in philosophical work on agency, there are in fact two distinct phenomena that go by the label "weakness of the will." The first, more widely discussed type of weakness is what Davidson called 'incontinence': acting intentionally against one's own best judgment. 4 This is a conflict between an intention to act and a judgment concerning what one ought to do, held consciously and simultaneously. The aforementioned debate over the possibility of epistemic akrasia has been characterized as the structural analog of this type of synchronic weakness. However, there is a second type of diachronic weakness that consists in failing to maintain an intention formed at some previous time in the face of a later, temporary shift in preference or evaluative judgment. This latter instance constitutes weakness not because the intentional action conflicts with a concurrent judgment about what is best, for at the time one acts the action does accord with one's preferences. Rather, it is a failure to exercise the psychological capacity to resist changing one's mind even when it wrongly strikes one in the moment that one ought to do so. We might call this capacity "selfcontrol." 5 My aim here is to take up the largely ignored question of whether there is a second type of epistemic akrasia corresponding to the lack of agential selfcontrol. I will argue that we do have the capacity for doxastic-self-control, understood as the ability rightly to refuse to reconsider a belief, and that we sometimes fail to exercise this capacity when we could have. This capacity underwrites an important sense in which we may be autonomous in our epistemic lives. I will begin by attempting to characterize in more detail why we have need of a capacity for doxastic self-control, and how otherwise insightful work on belief, responsibility, and first-personal authority has not placed sufficient emphasis on this diachronic dimension of belief. I will then flesh out the account by discussing the relationship between doxastic self-control and rational selfgovernance. I will first discuss whether doxastic self-control is a rational capacity or merely a form of self-manipulation. We can shed light on this question by comparing the maintenance of our beliefs over time to the exercise of self-control in our practical lives. I will argue that the refusal to reconsider an intention in the face of temptation must always involve some degree of self-manipulation, but that the refusal to reconsider a belief may be clear-eyed and fully rational (though it may not be). Surprisingly, it turns out that in this sense we have more rational control over our beliefs than our intentions. I will then take up the question of how the exercise of doxastic self-control constitutes an element of self-governance. I will argue that the rational form of self-control with respect to belief consists in self-consciously deferring to an earlier epistemic perspective in 4 Donald Davidson, "How is Weakness of the Will Possible?" reprinted in Essays on Actions and Events, Clarendon Press: Oxford, Richard Holton has lately done much to characterize this second type of weakness and to distinguish it from Davidson's notion of incontinence. See Willing, Wanting, Waiting, Oxford University Press,

3 favor of one's current perspective, and that this is possible only if the earlier perspective is conceived of as one's own. This in turn makes sense only if a thinker values having a continuous epistemic perspective or 'self' that persists over time and to which beliefs can be attributed. But this kind of diachronic continuity is precisely what is required if one is to be autonomous in one's beliefs, and thus a condition any thinker who aspires to doxastic autonomy will meet. II. Recent work on first-personal authority and the ethics of belief has highlighted the significance of holding a belief on the basis of reasons that are transparent to the believer. Not all beliefs are acquired via deliberation or accessible to the believer through rational reflection, but the claim is that a special authority or responsibility attaches to those that are. For example, according to Richard Moran s influential conception of first-personal authority, a thinker authors her beliefs to the extent that she is in a position to avow them as the conclusion of deliberation. 6 To avow a belief in this way is to take a stance on what is true and thereby to take responsibility for the belief, making oneself answerable to challenges as to the strength of one's reasons. Pamela Hieronymi has likewise proposed that a thinker is answerable for her belief that p just in case she settled for herself whether p, for only then can she correctly be asked for and evaluated on her epistemic reasons for settling the question in this way. 7 The general thought is that even if we cannot believe at will, some beliefs embody a thinker's answer to the question of what her epistemic reasons support and so are attributable to the rational agent herself. In contrast, beliefs that are not sensitive to her take on her reasons and thus discoverable only on the basis of evidence have a secondary, subpersonal status. I think such observations insightfully characterize an essential element of autonomy with respect to belief. However, they do not suffice for a complete account. The mere fact that a belief embodies a thinker s assessment of her epistemic reasons at a given time does not ensure that she is autonomous with respect to that belief. The problem is that this synchronic condition is silent about the cross-temporal aspects of investigating the question of what to believe, forming a judgment, and treating that judgment as settling the question for oneself over time. A thinker may fail to be self-governing in the process of forming her beliefs or in their maintenance over time in ways that thwart doxastic autonomy even if those beliefs do at some point reflect her take on her reasons. For instance, she might have been led by non-epistemic factors to consider only a subset of the relevant evidence, and thus influenced by extrinsic forces to judge on the basis of deliberation that p. It is a well-known result in social psychology, for example, that one can manipulate the belief deliberatively self-ascribed by a subject just by framing the question in slightly different ways. 3 6 Authority and Estrangement, Princeton University Press, "Responsibility for Believing," Synthese 161:3, 2008.

4 Subjects asked to complete the question "I date my partner because I " report significantly more love for their partners and claim it is likely they will marry them in far greater numbers than a second group asked to complete the question "I date my partner in order to." Plausibly, this is because the first group is led to attend to a different subset of reasons than the second group. 8 Whether or not we think such avowals are epistemically authoritative at the time they are made, they are a failure to be autonomous in the investigation leading to the deliberative avowal of the belief. Alternatively, one might have been autonomous in gathering evidence, reflecting, and forming a belief, only to reconsider that belief in the next instant as a result of wishful thinking, anxiety, lack of confidence in one s own judgment, and so forth. Suppose I inquire as to who will win the Presidential election. I reflect carefully on what I take to be the relevant evidence, judging that a statistical approach to weighting and amalgamating polling data will yield a reliable prediction. On the basis of this approach to interpreting the data, I conclude my inquiry by forming the judgment that Obama will win. The role of this judgment is to settle the question for me of who will win, and the fact that I formed it by reflecting on the supporting evidence puts me in position to avow the belief and to take responsibility for it. But again and again, out of ill-founded insecurity in my ability to assess the evidence and misplaced fear that I am mistaken, I re-open the question and deliberate anew. My counterpart who reaches the same judgment but who desires that Romney win the election also reopens the question and seeks out further, misleading evidence in the form of faulty campaign polls. I waste a great deal of psychic energy; my counterpart ends up changing his mind in favor of a false and unsupported belief. These are failures to be autonomous in the maintenance of a belief. The general point is that whether or not a belief is expressive of a person's influence as opposed to external manipulation cannot be decided by comparing that belief with a concurrent judgment about the reasons supporting it. It is in part a diachronic matter, to which both the genesis of the belief and its maintenance over time are relevant. Of course, in the election example, the emotional influences distorting my and my counterpart's judgment about our reasons are in one sense internal rather than external. But they are external to the processes by which a fully rational thinker assesses evidence, forms beliefs, and maintains them over time, and thus an impediment to self-determination. The problem for less than fully rational thinkers such as ourselves is that the influence of these non-epistemic factors can be very difficult to identify from the first-person perspective. They impede us from having the beliefs we ought to have not in the way that obstacles prevent one from reaching one's destination; rather, they pervade and distort our very ability to perceive, gather, and weigh 4 8 Krista Lawlor discusses this experiment in her critical discussion of Moran, "Elusive Reasons: a Problem for First- Personal Authority," Philosophical Psychology 16:4, The experiment was conducted by Seligman et. al. (Journal of Personality and Social Psychology 38, 1980).

5 evidence correctly. And if one deliberates about what to believe while in the grips of such influences, one will often come to the wrong answer. This predicament is well known to those who theorize about practical rationality. We planning agents frequently form intentions for the future that are meant to persist over time until the circumstances of action arise. The function of these attitudes is to settle at an earlier time the question of what the agent is to do at a later time. But it is an all-too-familiar problem that the preferences and evaluative judgments of agents like ourselves can undergo a temporary shift when we experience temptation: in the face of fine wine, comfortable beds, tortuous deadlines, and the like. A self-reflective agent may anticipate that she would regret indulging in the wine, oversleeping, and failing to make her deadline, and so take her desire to do these things to constitute an insufficient reason for action. Nevertheless, were she to consider the question of what to do when she is tempted by the wine and the comfortable bed, she would decide in favor of indulgence over meeting her deadline. On the assumption that the preferences and evaluative judgments that are more stable over time have a better claim to speak for the agent herself than those under the influence of temptation, this would constitute a failure of agential diachronic self-governance. To overcome this problem on the practical side, as Richard Holton has recently emphasized, we often resolve in a cool hour to resist such temptations at a later one. 9 On his view, a resolution is a composite of a first-order intention to perform a certain action (such as to decline to drink at the party) and a secondorder intention not to let the first-order intention be obstructed. They are formed in advance to combat contrary inclinations one fears one will be subject to at the time of action. But what is difficult to understand is how such resolutions could be effective. Setting aside "Ulysses" cases where one maintains one's resolution by depriving oneself of agency altogether, strength of will seems to require that one see a previously-formed resolution as settling the question of what to do even as one is simultaneously aware that one's preferences no longer support that course of action. Trickier still is that sometimes one rationally ought to change one's mind; perhaps the conditions turn out not to be as one had thought, or perhaps one's reasons have changed in the interim. It follows that strength of will cannot be a matter of refusing in every instance to re-open a practical question once one has made a decision, as Nietzsche advised. One must retain sufficient openmindedness to distinguish good reasons to reconsider from bad. My suggestion is that we have similar need of self-control in treating our past judgments as settling the question for us of what to believe in the face of temptation to reconsider. I will focus for the remainder of the paper on autonomy in the maintenance of a belief once it is formed, leaving aside questions about self-governance in the initial investigation. To see the parallel with practical resolutions, it will be helpful to reflect on why we even have need of an attitude that has the function of settling epistemic questions. It is not a matter of necessity that any psychology that aims at truth-tracking will treat 5 9 Holton (2008).

6 questions about contingent truths as settled. If we had unlimited cognitive resources and no need to act, we would have no reason to treat any genuine epistemic question as closed; we could simply keep a running tab of the probabilities of the truth of various hypotheses. But as we are, with significant constraints on our capacity to keep track of multiple investigations at once and a need to act, we do have reason to close our inquiries and settle on an answer. Thinkers like us would otherwise often be paralyzed and would waste valuable cognitive resources. Like in the practical case, the capacity to treat a question as settled must not systematically interfere with our ability to update on important new information or to correct past mistakes; one must not categorically refuse to reconsider. A diachronic believer must adjudicate the delicate balance between continuing to treat an investigation as closed and re-opening deliberation in the appropriate circumstances. But also like in the practical case, we face a variety of temptations to reconsider the question of what to believe that are not legitimate epistemic reasons. If an epistemic question is re-opened out of wishful thinking, anxiety, insecurity, or other such influences, this is a failure of a belief to play its proper settling function. The capacity for doxastic self-control is what enables a rational agent to treat a belief as settling the question for her unless and until she encounters sufficient reasons of the right kind to reconsider. Let me conclude this section by distinguishing the capacity I am referring to with the label 'doxastic self-control' from other, similar phenomena. John Heil has an influential discussion of the problem of 'doxastic incontinence', but he restricts it to cases in which one believes that p while one knowing or supposing one ought (epistemically) not believe that p. 10 Likewise, Al Mele has characterized a capacity he calls doxastic self-control, but he also has in mind the ability to conform one s beliefs to explicit higher-level judgments. 11 My interest is distinct from these notions, in that the kind of doxastic incontinence that is my target is not concurrent with a conflicting belief regarding what one ought to believe. In the cases I am thinking of, the thinker's confidence that her evidence sufficiently supports p significantly decreases as a result of the influence of nonepistemic, paradigmatically emotional factors. The vividness of some subset of the evidence increases while the rest is dulled; the soundness of one's past doxastic deliberation is groundlessly called into question; one is tempted by distaste for the conclusion to grasp at further considerations that constitute weak or misleading evidence. In such circumstances, the thinker no longer believes that she ought to believe that p. She justifiably had this higher-order judgment at an earlier time but abandoned it without sufficient epistemic reason. Doxastic self-control as I am employing the term is the capacity to continue to treat one's belief as settling the question even when such illicit influences have temporarily caused one to doubt the evidence supporting it "Doxastic Incontinence," Mind v. 93, p Irrationality, Oxford University Press (1987). Pgs

7 7 III. Is doxastic self-control a rational capacity? To answer this question, we must distinguish between two ways in which self-control, whether over intention or over belief, can be exercised. It is uncontroversial that we can manipulate ourselves into sticking with an intention or belief by depriving ourselves of the ability to reflect and reconsider. This is what I referred to earlier as "Ulysses" cases. A thinker can put herself to sleep, drug herself, bind herself, or otherwise bring it about that she is simply unable to change her mind. The much more interesting and difficult question is whether it is possible to exercise self-control in a clear-eyed, rational way, without depriving oneself of the ability to reflect. I will call this rational self-control. There are many conceptions of rationality, and thus many things one might mean by 'rational'. As I intend the phrase in this context, a choice is rational if it is adequately supported by the right kind of reasons, the agent is in a position to grasp this fact, and makes the choice because she grasps this fact. With respect to self-control, the choice is to refuse to reconsider a question one has already settled for oneself, and it is rational if it meets these conditions. My claim is that the exercise of self-control over our actions in the face of temptation cannot be fully rational in this sense, whereas doxastic self-control can be. Surprisingly, there is therefore a sense in which we exercise more rational control over our beliefs than over our intentions. My argument for this depends on the relatively benign assumption that an agent's reasons for action depend to some extent on that agent's preferences. This is consistent with a variety of views on the nature of reasons, including the truth of normative realism and reasons externalism, as long as it is the case that there is some value in satisfying one's desires. It is a benign assumption because views that assign no relevance to an agent's preferences in determining her reasons for action are highly implausible; they cannot account for what seems to be the obvious fact that to use an example of Mark Schroeder's if Ronnie likes dancing, the fact that there will be dancing at the party is a reason for him to go, whereas it is no reason for him to go if he does not like dancing. Our preferences must inform our reasons, even if they do not exhaustively determine them. Given this assumption, the problem notoriously posed by temptation is not only that it appears to the agent that she has reason to abandon her previouslyformed intention and re-open deliberation. In fact, she really does have such reason. It is part of the phenomenon of temptation that one's preferences and evaluative judgments undergo a temporary shift. We can identify such shifts as the effects of temptation rather than a genuine revision by the fact that they are transitory and something the agent is disposed to regret acting in light of. Nevertheless, during such a shift, the agent really does have reason to act on the basis of her new preferences, and her existing intention may no longer be most favored by her reasons. To be sure, she may not have sufficient reason to change her mind even taking her preference-shift into account; it will depend on the case. The crucial point, however, is that the kind of reason she will have to reconsider is

8 the right kind of reason: it is a reason that genuinely bears on what the agent should intend. Holton contends that we can avoid the negative impact of such temporary corruptions of judgment if we can implement a previously-formed intention without further reflection when temptation hits. The idea is that the ability to think less and simply follow through with resolutions is beneficial for mercurial agents like ourselves. This is clearly right if understood as the self-manipulative variety of self-control, e.g. via distracting oneself. But Holton also claims that one does not thereby make oneself irrational, or even arational; we are meant to be able to see ourselves as in rational control of our actions when implementing a prior resolution. This would require taking reconsideration to be a genuine option but refusing to do so while fully aware that the resolution is not currently supported by one s reasons and would be rational to revise if one did reconsider. This strikes me as a very difficult state of mind to consciously maintain, bordering on bad faith and unfit to be called 'rational'. Holton emphasizes that the general disposition to make and follow through on resolute choices can be advantageous to an agent from the perspective of her reasons for action over time, and we might call it rational in this strategic sense. But the fact remains that the particular refusal to reconsider often cannot be understood by the agent as supported by her reasons for action at the time control is exerted, because it is not. In contrast, the conditions that require the exercise of doxastic self-control do not in fact alter one's reasons for belief; it merely appears to the thinker that they have altered. Out of wishful thinking, anxiety, unreasonable insecurity, and other such factors, she comes to doubt that her act of concluding inquiry and treating the question of what to believe as settled was justified. It may in fact not have been, from an objective point of view. But irrespective of this fact, the emotional factors influencing her to call her previous judgment into question are not the right kind of reason to reconsider. Genuine reasons for reconsideration include the fact that one's previous deliberation was compromised by fatigue or intoxication, the existence of further relevant evidence one has not considered, the disagreement of equally informed and intelligent peers; and so forth. Doxastic self-control is appropriately exerted when none of these genuine defeaters has arisen. The difference is in principle something a thinker can develop the ability to recognize, perhaps through the acquisition of certain epistemic virtues (though I do not mean to suggest that this is an easy feat). The fact of the matter may be difficult to make out, but the appropriate exercise of doxastic self-control will be directly supported by one s epistemic reasons. The upshot is that self-control in our practical lives must always be manipulative to some extent, whereas doxastic self-control can be clear-eyed. Contrasting the two types of self-control therefore illuminates the respect in which practical self-control is a deficient form of autonomy, more akin to Ulysses s self-binding than Holton wishes to depict it. Doxastic self-control can be like this, but it need not be. It can be the self-conscious activity of resisting the 8

9 wrong reasons for reconsideration while conforming to the right reasons. We thus arrive at the surprising conclusion that self-control with respect to our beliefs is a more genuine form of autonomy than that we exercise over our actions. IV. The final claim for which I will argue is that the rational exercise of doxastic self-control depends on the value of having a stable epistemic perspective that persists over time. A thinker who has no direct concern for diachronic continuity and cares solely about believing whatever the available evidence dictates at a time will not be able to exercise doxastic self-control except by manipulating herself. Rational self-control over one's beliefs is therefore bound up with the persistence of the self in a way that qualifies it as a genuine element of self-governance. The reasoning is as follows: (1) doxastic self-control is required only when one s current assessment of whether a belief is adequately supported by the available evidence conflicts with an assessment made at an earlier time; (2) it consists in deferring to the earlier assessment rather than the present one; (3) this is irrational unless one takes that earlier perspective to be metaphysically continuous with one s current perspective, and thus one s own; (4) the reason for taking the earlier perspective to be one's own stems from caring about having a diachronic self to which beliefs can be attributed. I will argue for these claims in turn. The first claim that doxastic self-control applies only to conflicts between a previously-formed belief and one s current assessment of the evidence supporting it essentially follows from my definition of the phenomenon. To be sure, the many ways in which our agency is exerted in inquiry before ever arriving at a belief seeking out evidence, directing our attention in various ways, withholding judgment or bringing the investigation to a close, and so forth are equally involved in the exercise of doxastic self-control. However, the particular interest here is in self-control, which I am understanding as applying to the maintenance of an already-existing attitude that can be attributed to the person in the face of distorting influences that are alien to that person. Doxastic selfcontrol will therefore come to bear only at the point at which an inquiry has been completed and a judgment made. More specifically, its exercise will only be required if at some point in time after a judgment has been made the thinker loses confidence that her belief is true and is tempted to re-open the question. The second claim is that doxastic self-control of the non-manipulative kind consists in deferring to a previous judgment rather than re-opening doxastic deliberation. There are several parts to this claim. First, non-reconsideration is a necessary but not sufficient condition for exerting doxastic self-control. Often we do not reconsider our beliefs simply because we remain confident in our grounds for them and see no reason to reconsider. This is not a manifestation of doxastic self-control. Manipulative self-control prevents reconsideration, but not via a self- 9

10 conscious choice. In contrast, rational self-control consists in self-consciously deferring to an earlier epistemic perspective. By 'deferring', I mean according the past judgment sufficient epistemic authority to trump one's current assessment. Importantly, this cannot simply be a matter of rationally reflecting at t 2 on the question of whether one's judgment that p at t 1 was correct. For this would involve reflecting on whether p, and that is precisely what is to be avoided when one is undergoing epistemic temptation. The authority accorded the earlier assessment must therefore be independent of one's current view on p. The third claim is that the rationality of deferring to a previous judgment in this way depends on taking this perspective to be one's own. This claim is less intuitive; why not simply treat an earlier judgment as if it were the testimony of another person? But this suggestion is problematic for essentially the same reason that many have thought the synchronic form of epistemic akrasia is impossible. To see why, let us reflect on what the rationale would be for accounting this testimony more weight than one s own current perspective. Suppose first that it rests on inductive reasoning. A thinker might reach the conclusion that when she comes to doubt a belief preserved in memory and is tempted to re-open the question, the preserved belief is more frequently correct than the new belief she arrives at, even if the preserved belief seemed to her quite dubious at the time. Reflection on the track record of deferring to the past belief would support a general strategy of continuing to do so, but only in the way that a clairvoyant might adopt the strategy of relying on beliefs that simply pop into his head without taking them to have any direct evidential support. That is, it would be a truth-conducive strategy, but the thinker would not be in a position to avow the relevant belief or defend it by appeal to reasons for thinking it true. This would collapse into self-manipulation, since one could not see one s belief as directly supported by one s reasons. A second possibility is that it simply follows from the nature of belief that we are justified in being epistemically conservative, such that an existing belief should be maintained until some possible defeater is encountered. 12 It would be irrelevant on this rationale that it is the thinker s own belief; the default conservatism would have the status of a norm governing the attitude of belief as such. Now, I am sympathetic to the idea that there is such a norm, but it would not solve the current problem. Even a norm of default epistemic conservatism must allow that it is sometimes permissible to re-visit an epistemic question even if no new defeaters arise, if for instance one s original deliberation was compromised. In the circumstances we are considering, the question has arisen already for the thinker of whether this is one of those cases. The default justification of an existing belief will not help to answer that question. Finally, suppose the thinker does inquire directly into whether the earlier epistemic perspective treated as that of another person should be considered more authoritative than her own. Here the problem is that she cannot rationally arrive at the conclusion that the answer is yes. It is of course possible to believe Thanks to John Bengson for this suggestion.

11 that one previously had better evidence than one currently does, e.g. because one has forgotten what the evidence is, and thus rationally to judge that the previous perspective is epistemically superior. But in that kind of case the thinker has no need of doxastic self-control, since she is not even tempted to reconsider her belief. The conditions of interest to us are precisely those in which if the thinker were to examine the credentials of that belief, her clouded judgment would lead her to reject their adequacy. Deferring nevertheless to that belief would be senseless a matter of judging that another epistemic perspective is or might be inferior and then deferring to it against one's own best judgment. It would be the type of epistemic akrasia many think is deeply problematic. Thus we arrive at claim (3): for doxastic self-control to be exercised rationally, the thinker must conceive of the past judgment as her own. It is only if the past belief is conceived of as one s own that the conclusion is not I judge that my own best judgment is flawed, but rather my best judgment consists in my earlier perspective rather than my current one. This allows the thinker to see the deferral to a past perspective as in keeping with her own best judgment. This brings us to claim (4) concerning the notion of a diachronic doxastic 'self'. As far I can see, it is not a necessary condition of being a believer that there be any kind of continuity in the content of one's beliefs over time, except as a byproduct of any diachronic continuity in one's assessment of the evidence. From the point of view of the constitutive norms on belief, there is thus no imperative to treat the beliefs of past time-slices as having any distinctive authority. However, my suggestion is that it is a condition on being autonomous with respect to belief. Given our original assumption that we lack direct voluntary control over our beliefs, there is simply no room for any merely synchronic fact to amount to doxastic self-determination. As I argued earlier, it is not enough for autonomy that one's beliefs conform to one's assessment of the reasons as formed through conscious deliberation, since that assessment might have been misleadingly constrained or distorted by extrinsic factors. It is only by taking into account the thinker's activity over time, either in re-opening the investigation and seeking out new evidence or maintaining her belief by deferring to her own previous judgment, that we can judge whether she was genuinely selfdetermining. A thinker who aspires to be autonomous will therefore have reason to conceive of her past judgments as expressive of a broadly continuous firstpersonal perspective, and to give these judgments a distinctive weight not held by the testimony of another person. Deferring to a past perspective is both an element of doxastic self-control and something we have reason to do from the point of view of diachronic autonomy in our beliefs. Sarah Paul University of Wisconsin Madison 11