Philos Stud (2013) 166:625–636 DOI 10.1007/s11098-013-0199-2
Responses to four critics Ernest Sosa
Published online: 12 September 2013 Springer Science+Business Media Dordrecht 2013
The agreement in these excellent sets of comments is of course welcome. Equally welcome is the disagreement, all of it interesting and instructive, despite some of it turning out to be verbal. In any case, my grateful thanks to all four commentators!
1 Response to David Henderson and Terry Horgan (H&H) 1. The comments of Henderson and Horgan reveal little substantive disagreement, or so I will argue. They do of course verbally disagree in alleging that my view is too intellectualist when it claims that knowing full well requires reflective knowledge, which in turn requires ascent to an apt second-order belief that the first-order belief is apt. True enough, according to my account, you know full well that p only if the correctness of your belief that p manifests your first-order competence to believe correctly on the question whether p, and you also believe correctly that you thus aptly believe that p, where the correctness of this second-order belief itself manifests your competence to believe correctly on the question whether you aptly believe that p. What exactly is my claim to which H&H raise their objection. My claim is that a first-order belief ascends to the higher level of knowing full well only when it is itself the object of a second-order belief that the first-order belief is apt.
E. Sosa (&) Rutgers University, New Brunswick, NJ, USA e-mail:
[email protected]
123
626
E. Sosa
To the contrary, H&H think one can know perfectly well that p provided one’s belief manifests mere ‘‘sensitivity’’ to when one can so believe without excessive risk. To require in addition a second-order belief is said to be implausibly intellectualistic. This alleged disagreement is only verbal, however, given my anti-intellectualist conception of a suitably broad category of ‘‘belief.’’ Although this broad conception figures large in my earlier writings, it figures not at all in the book under discussion, which helps explain H&H’s reaction. Here now is how I make the relevant distinctions and try to clarify what reflective knowledge amounts to, and how it comes in degrees. 2. Much of our most useful animal knowledge is constituted by ‘‘beliefs’’ that remain below the surface of consciousness, mostly beyond one’s ability to verbalize. Such beliefs are manifest rather in the appropriateness of our behavior relative to our perceived situations. This is the sort of implicit or tacit knowledge that guides us as we negotiate an obstacle path, as we recognize a friend through facial features, and so on. Provided such animal competences are reliable, we evince in trusting them not just belief (as does the bigot who considers a target group inferior) but also knowledge, animal knowledge. Such animal knowledge can also rise to a reflective level higher than just the minimum. This it does when accompanied, not only by the minimal reflective endorsement already noted, but also by a fuller perspective on the nature and reliability of the competence involved. This, moreover, is a matter of degree. With a fuller perspective on the nature and reliability of one’s object-level competences comes better reflective knowledge of the facts known on that object level. To know those facts full well requires a rich and coherent enough perspective that underwrites one’s grasp of the facts known. In order to know full well that p, one must know it reflectively: one must know that one knows. In addition, finally, the epistemic perspective in the best human knowledge (knowledge full well) must be rich and coherent enough, since it is such rich coherence that enables us to rise above the limitations of our modular faculties. 3. I thus recognize a broad sort of knowledge that can remain inarticulably implicit. Such knowledge is extensive and important, not only in quotidian contexts, but also in the lab and even in theoretical inference. My gladly allowing this suggests how extensive and substantive is the agreement between H&H and myself. 4. Extensive and substantive though our agreement is, it is not complete. Suppose we do allow such broad belief to constitute a subconscious second-order perspective that can make our knowledge reflective. H&H think that by being so flexible we collapse the distinction between animal and reflective knowledge. If animal knowledge is allowed to ascend to the reflective level so easily, in their view nothing deserving the title of knowledge can then be animal knowledge pure and simple, without being reflective. For it will always need to be reflective at least to that minimal degree. Here we do have a disagreement, and I will now argue for my side. (At least, I think we disagree, but even here the disagreement may be in large measure only verbal, as will soon emerge, in item 5.)
123
Responses to four critics
627
5. Suppose there is indeed a range containing all beliefs, from the minimally reflective to the most fully reflective (measured, for example, by the quality of the meta-competence and the degree of awareness involved, which for humans would normally involve richness of coherent perspective). Compare our range of head-to-heel length, with the shortest of us at one end and the tallest at the other. Only those who are sufficiently taller than the average are tall, even though people less tall still have the property of head-to-heel length, and indeed of being so-many-units tall, and are normally taller than someone else, even when they are not at all tall. We may similarly view reflective knowledge as knowledge that is reflective enough (as measured, at least in part, by quality of meta-competence and degree of awareness). Compatibly with that, some degree of what (in sufficient measure) can make knowledge reflective may always be present even in the knowledge that is merely animal since insufficiently reflective. On this account, a bit of animal knowledge could involve reflection to some extent without being reflective knowledge. Compare the fact that someone short could have to some degree what (in sufficient measure) makes the tall tall, namely head-to-heel length. Someone short would have some degree of such length without having enough of it to make them tall. 6. However, the supposition that there is the sort of range posited can be challenged, based on how a super-blindsighter can just find a belief within, despite having been guided to form it by no risk assessment. To me it is intuitively quite plausible that such a belief can amount to knowledge of a sort. Can’t this be knowledge that is purely animal and entirely unreflective, based on no risk assessment whatsoever? Such blindsighter beliefs derive rather from subpersonal processes systematically reliable enough to yield a kind of knowledge. To the contrary it might be argued that the blindsighter’s inclination to believe is too modular, whereas belief proper requires the interplay of reasons characteristic of full-fledged human knowledge. However that may turn out, there is a more radical challenge to the supposed collapse of the distinction between two sorts of knowledge, the animal and the reflective, a more radical challenge that would block the alleged collapse. 7. Recall what happens when you take your yearly eye exam. In my experience, the technician seats me some distance away from a Snellen chart, and asks me to read the letters on line after line, starting with a huge A at the top to rows of smaller and smaller letters. At some point I start to guess and the technician finally stops the test and takes note of how reliably I could correctly read the letters how far down the chart, … or the like of that. Take line 10, the line that normal subjects can see at a maximum of 10 feet. Do you knowledgeably identify the letters as you view the 10 line? What if by that point you are only guessing, so you don’t really form any full-fledged belief on what they are. If we grant you a kind of knowledge in that case, there is a level of knowledge—subanimal knowledge, as it might be—that does not even require outright belief. You do make your affirmations to the tester in the endeavor to get it right, perhaps even in the endeavor to get it right reliably enough. Suppose there is a threshold of reliability requisite for knowledge. Sitting there in the dark as you view
123
628
E. Sosa
the Snellen chart, you do not intend to affirm correctly with reliability above that threshold (and only above that threshold), which means that, strictly speaking, you do not really intend to affirm correctly, even if you may try to do so. Let us distinguish between two senses of ‘‘affirming that p in the endeavor to get it right reliably enough.’’ In the case where you view the 10 line, you do affirm in the endeavor to get it right reliably enough, with reliability above the threshold, when you say ‘‘This is an E,’’ when you say it to the technician or perhaps when you first say it to yourself. You do after all focus intently, and you do consider carefully what you ought to affirm, etc. So, in that sense you do try to affirm as reliably as you can, and so you do try to affirm with reliability above rather than below the threshold. However, you are not at all sure that you will in fact attain a level of reliability above the threshold. So you are not at all confident that your affirmation will be a reliable enough: i.e., above the threshold. You are not sure enough that so affirming is a sufficient means to affirming reliably enough. Accordingly, you do not affirm as a sufficient means to affirming reliably enough. Thus, although you try to affirm correctly and even reliably enough, you do not properly intend to do so. Moreover, you do not try to accomplish the following: to affirm correctly regardless of how reliably you may do so. No, your attitude to performance that fails to attain its focused aim is that you prefer getting it right even if you do not get it right reliably enough. However, even if your preference is satisfied then by your merely getting it right (without doing so reliably enough), that will fulfill no intention of yours. By the time you get to the 10 line you may have no such intention. Strictly you do not even intend to affirm correctly. You hope to do so, and you do intentionally try, but you do not intend to succeed. In fact that is not your aim, of course, when you take the eye exam. It is just part of the test, as we are now thinking of it, to take guesses once you move far enough down the chart, to just affirm in line with how the letters then seem to you, no matter how tenuously. You keep trying to answer correctly. That is what you are supposed to do. But you no longer intend to answer correctly, as opposed to intentionally trying while perhaps hoping for the best. (Really you need not even much hope, since the outcome is sure to make only minimal difference to the glasses you must in any case wear. So, here your attempt is fully blase´, since you care not at all one way or the other.) The tester will then determine your success rate, and that will in turn fix how good your vision is. So, your guesses are not affirmations of the sort required if you are thereby to know. They are not affirmations intended to get it right, and (of course) to do so reliably enough (and only reliably enough). You hope to get it right by so affirming, but you do not intend to get it right. In that context, you do not mean to affirm that way. You do affirm in the endeavor to get it right, of course. But that much you can do, while still in doubt as to whether by so affirming you will affirm reliably enough, and so you can affirm in the endeavor to get it right without properly intending to get it right by so affirming. So at that line you do not really judge. So, you do not judgmentally believe. Instead, you only guess. In fact, from the fact that you try to get it right reliably enough nothing follows about whether you also try to get it right regardless of how reliably. Indeed you
123
Responses to four critics
629
might even aim to get it right reliably enough, while also, compatibly, aiming to avoid getting it right but without doing so reliably enough. Note that guesses can attain extremely high levels of reliability. In our example, you would always get all the letters right at line 10, year after year. Your guesses would thus attain 100 % reliability! If you are like me, however, here’s your situation year after year. Before a recent web research, I had only a thin understanding of what is involved in a vision Snellen test. I had no idea that the lines are assigned numbers, such as 400, 200, 40, 20, and 10. And so, from year to year, over the course of many, many years, I have not kept track of how far down the chart the tester would let me go. Nor did I keep track of my success rate, and how it changed once I got to the lower lines. So, I did not gain any track-record reason to suppose that I was reliable up to and including the 10 line. So, I had no belief, nor even any opinion or thought, on that matter. Yet, suppose I was in fact extremely reliable, with a 100 % success rate, year after year, at the 10 line, despite my great hesitation once I got to that line. Most plausibly, I did surely enjoy a kind of knowledge about the contents of that line, a knowledge constituted by guesses, by affirmative thoughts and statements in the endeavor to get it right. However, the affirmations that constituted these guesses did not rise to the level of belief. They did not rise to the level of affirmations made as a sufficient means to affirming not only correctly but also reliably enough. Nor do I find it plausible that at least I was implicitly responding to cues that enabled me to be sensitive to risk. In guessing as part of the eye exam, risk is irrelevant. One just guesses uninhibitedly, in order to get it right, because it is by so acting that one cooperates in determining one’s acuity. 8. It would be very surprising if for no subject were there ever two lines at the higher of which the subject knew he could tell the letters while at the lower he still did not know that he could not tell the letters. The lines would then have to be spaced far enough apart that no-one ever was willing to merely guess at any line. No line would ever be a line where you were insufficiently sure whether you were reliable enough, although in the preceding line you were quite sure that you were reliable enough. Right after the last line where one judged confidently that one was reliable enough would always come a line where one judged confidently that one was not reliable enough. Suppose that is always so, although I know that in my own case it is not always so. Even if it were always so, given the customary spacing of the lines, we surely could just fill in more and more lines until it was no longer so for anyone in the population. That would enable much finer discriminations. Now people would not just be divided into those with 20/10 vision and those with 20/20, and nothing in between. Now we might have discriminations much finer than that. So, many might be able knowledgeably to tell the letters at line 20, while having to guess at line 19, since they would no longer be sure enough that they could still tell reliably enough. Suppose you are among these. You can tell the letters correctly at line 20, while knowing that you do so. But at line 19 you no longer know whether you tell the letters correctly. This of course is something you might come to know through track-record information. Initially, however, in the absence of such information about line 19, you do not know whether you tell the letters correctly.
123
630
E. Sosa
And yet the reliability of your guessing at line 19 might closely approximate or even equal your reliability at line 20. Isn’t it plausible that you would ‘‘know’’ in some very basic way, even despite the absence of belief? This I find more compelling than its correlate in the lore of Norman the clairvoyant, of Truetemp with his implanted temperature sensor, and of the student who loses confidence and nevertheless answers a test question correctly. Indeed it seems entirely possible that many of us in fact attain such knowledge year after year. In some sense we then know what the letters are despite the fact that we do not so much as believe what we nevertheless know. On this basic level, reliable enough guessing can plausibly yield knowledge. This knowledge would be unguided by second-order risk assessment, and indeed would not even be constituted by belief. From there, moreover, all it would take to arrive at belief-constituted pure animal knowledge is an overconfident eye-exam subject. Perhaps we could deny the overconfident subject judgmental knowledge despite granting a kind of animal knowledge to the reliable guesser, and despite the lack of any difference in reliability. Perhaps we would be reluctant to reward such epistemic misbehavior. But then this would not be so plausible about the functional (as opposed to judgmental) belief that we might enjoy at least briefly as we view those letters. (No doubt we could even find a case where the overconfident eye-exam subject was far more reliable than the guessing knower.) In conclusion, there can be pure animal knowledge with no admixture of riskassessing reflection, no matter how implicit. A very basic sort of knowledge can be found even below the level of animal knowledge, moreover, since it is constituted by mere guesses rather than beliefs. And such knowledge plausibly survives the overconfidence of a subject who does manage to believe (at least functionally) what others in his position only guess at sans belief. Even leaving those subjects aside, moreover, a bit of knowledge can have too little of the reflective backing required to make it a case of reflective knowledge. Instead it would properly be regarded as falling short of some required threshold of reflection.
2 Response to Jack Lyons 1. Lyons has concerns similar to those voiced by H&H, along with some distinctive points. He thinks I have three different conceptions of reflective knowledge. But, partly in line with the above, it will emerge that this is a misreading. For me reflective knowledge is just apt belief aptly noted. And to know full well is not merely to know thus reflectively, since it also requires that one’s first order belief be guided to aptness through one’s risk-assessing second-order apt belief. Note that this guidance is a further factor beyond mere reflection, beyond a knowledge that is not just animal but also reflective. In order to know full well one must be guided to apt belief on the first order by what one aptly believes on the second order as to the risk involved in so believing on the first order. That distinction emerges clearly through the example of a huntress, Diana, who takes a shot in the knowledge that the risk is appropriate, but guided not by that knowledge but only by a coin toss. Her shot would be not just animally apt, but also
123
Responses to four critics
631
reflectively apt, since all that is required for the latter is that her shot be not only animally apt but also accompanied by an apt meta-belief that the first-order risk is appropriate and the first-order shot apt. In order for her to succeed fully well, however, it is required not just that her shot be accompanied by such a second-order apt belief. It is required also that in opting to shoot, and in shooting as she does, she be guided by her second-order apt belief that the shot would run appropriate risk and be apt. She cannot just let a coin toss determine whether to take the shot. 2. However, the main issue on which we do seem to have a deep and substantial disagreement is perhaps the main issue that divides process reliabilism from virtue epistemology. Lyons raises the issue when he suggests that there is nothing more to be said for reflective knowledge, or for knowing full well, than for a knowledge that is, say, perceptual, and also buttressed by good testimony. Reflection adds nothing of a different sort than does testimony, nothing importantly distinctive. In both cases what is added is at best additional evidence, which may give one’s belief a more reliable basis, so that it is formed and sustained through a more reliable process. If the belief is to gain anything epistemically, this is how it must do so. There is nothing special that derives from the cross-level support constitutive of reflective ascent, nor even when one knows full well. All we can hope to derive epistemically from the cross-level support is more evidence and reliability, and this we can get equally well through the same-level support of good testimony. That reaction strikes me as deeply wrongheaded, for reasons I will now lay out. 3. An act that derives from some sort of agency will be a performance with some sort of aim, whether this be conscious and intentional or subconscious and teleological. Since my case will be stronger for conscious and intentional judgments than for subconscious teleological functionings, let me focus on the former, though I believe a similar case can be made also for the latter. What is intentional belief? How is it structured? We focus on affirmation, and the corresponding disposition to affirm, in the endeavor to answer a given question correctly. Consider how important these are for a collaborative social species. They seem essentially required for collaborative deliberation and for information sharing. Take collaborative deliberation, right up to the most complex, as in a nation’s governance; also, information sharing, crucial as it is in a great many contexts, prominently in scientific inquiry. Such affirmation is largely conscious and intentional. If you add a column of figures in your head, for example, you may seem to get a certain result. But if the problem is complex enough, you may hesitate to affirm accordingly. You may first take out pencil and paper, or a calculator. Eventually, given coincidence of results, you may consider the evidence strong enough, which leads you to assent. You decide when to assent, you wait until the evidence is strong enough. We focus on such intentional, judgmental belief. How is it structured? Judgmental belief is definable as a certain sort of disposition to affirm. What sort of disposition? For a start let us take judgment that p to be a certain sort of affirmation in the endeavor to get it right on whether p. Judgmental belief can then be understood as a certain sort of disposition to judge in the endeavor to get it right on whether p, if one so endeavors.
123
632
E. Sosa
Before us so far is a partial account of judgment: as a ‘‘certain sort’’ of affirmation in the endeavor to get it right on whether p. Judgmental belief could then be understood as a corresponding disposition: to judge in the endeavor to get it right on whether p, if one so endeavors.1 4. Judgment is thus in my view essentially a form of action, of affirmation as a means to getting it right on a given question. What defines a normativity that is specifically epistemic, then, is that it is normativity proper to a domain of such epistemic action, with its proper aims, including prominently that of getting it right on questions taken up. When we act, when we endeavor to attain an end, there is a difference between lucky attainment of that end and competent attainment of it. Even competent attainment falls short, moreover, unless the attainment manifests the competence and is not just lucky. If we return to the case of Diana the huntress, we can see how her shot is assessable on two levels. She might shoot despite incorrectly thinking that the risk is far too high, whether this thought is competent or, worse, is not competent. Her shot is poorly selected, especially if her risk assessment is not only incorrect but also incompetent. And this is bound to reflect on the shot itself, which is now a poorly selected shot, and hence falls short at least in that respect. This is its own respect of evaluation of Diana’s shot, a dimension within which we assess how well selected her shots are. The accuracy of her shot, its success on that lower level, might nevertheless manifest her sublime skill as an archer. She was just wrong in thinking that her skill was not up to succeeding on that occasion. Despite how excellently apt her shot is, however, it still falls short because it is woefully selected. So, in a clear sense, I submit, it succeeds through a kind of luck that detracts from the credit that Diana earns even through her wonderfully apt shot. 5. That is a distinctive way in which a normal perceiver will excel beyond a blindsighter no matter how adept. A belief of the (over)confident blindsighter might enjoy exceedingly high reliability without being guided in the slightest by secondorder risk assessment. At some point, the blindsighter may just find himself believing, in the way we normally just find beliefs lodged in our memory banks long after the evidence in their favor has lapsed from memory. And the process that leads to the blindsighter’s belief may be super-reliable. Yet the belief will still fall short in being poorly selected. (And being unselected will be a special case of being poorly selected, if selection is called for in that case.) 6. Lyons properly wonders how, in what way, reflective knowledge is superior to mere animal knowledge, and one might similarly wonder how knowing full well is better than would have been the corresponding mere animal knowledge. I answer that knowing full well is an apt attainment of the truth through a well-selected judgment or belief, so that it does not succeed through a kind of ‘‘blind’’ luck. (And reflective knowledge is better than mere animal knowledge in that it attains at least
1
The importance of these phenomena for social beings should be obvious. In forthcoming work (Epistemic Agency book ms) I offer a fuller account of judgment and judgmental belief, and consider further theoretical benefits of recognizing them.
123
Responses to four critics
633
one crucial element required for knowing full well, namely a competent and apt perspective that enables properly guided selection on the first order.)
3 Response to Pascal Engel Engel’s commentary poses several problems for my account of knowledge. He summarizes by objecting against my teleological picture based on the following two considerations (and I quote from his abstract): … [a] beliefs are not performances, and … [b] epistemic reasons or beliefs cannot be balanced against practical reasons. He then goes on to specify in detail how he thinks my formulations are vulnerable to the problems he raises. Here I will not attempt a point-by-point response. I do have a more detailed response to the interesting and challenging problems that he raises. But that complex response is to be found in work in progress on epistemic agency and related matters. In that forthcoming work I move beyond the earlier account. The new account is not fundamentally or radically different from the earlier account, but it does offer a new and substantially different view of judgment and of judgmental belief. As before, I still distinguish judgmental belief—which involves a free act of judgment, and a disposition to judge—from functional belief—which need involve no conscious deliberation or intention to affirm. Nevertheless, despite the specified difference between those two forms of belief, there is a deep affinity between them, which makes both of them cases of performance with an aim, even if the aim is consciously intentional in one case but teleological and subconscious in the other. Anyhow, in what follows I would like to give at least a fuller (though still brief) indication of how my approach proposes to deal with those, and other, problems. 3.1 The nature and varieties of belief: credence versus judgmental belief To judge that p is to affirm that p sufficiently in the endeavor to answer both correctly and reliably (reliably enough and only reliably enough) the question whether p. a.
b.
Thinking that p is a more general category. Affirming that p is more general yet. Our focus is here judgment (an act of affirmation) and judgmental belief (a disposition to perform that act when one faces the corresponding question and endeavors to answer it correctly as well as reliably enough). Not just any affirmation or disposition to affirm counts as such belief. Affirming can be done in the endeavor to lie, to comfort someone, or to reduce dissonance. One might affirm or be disposed to affirm in the endeavor to feel good about oneself, or to be a better spouse, or to gain confidence for athletic competition, and so on. None of these would thereby constitute the sort of belief of interest to us. These are instead forms of ‘‘make belief,’’ where the subject does not really
123
634
E. Sosa
believe if she affirms or is disposed to affirm essentially and sufficiently for pragmatic purposes, and not sufficiently through epistemic motivation. Makebelief is mock-belief. c. If someone affirms ‘‘essentially’’ for pragmatic purposes, then the extent to which her affirmation is due to the endeavor to hit the mark of truth is not by itself sufficient to yield the affirmation, in the absence of those operative pragmatic purposes. By contrast, really thinking that p is affirming or being disposed to affirm that p in the endeavor to affirm correctly (with truth), where the extent to which the affirmation is due to that endeavor does suffice by itself to yield the affirmation. You really believe (judgmentally) all and only what you are so disposed to affirm, when your endeavor is epistemically pure and disinterested, while aiming to attain truth reliably enough on the question at hand. d. Someone affirms sufficiently for certain purposes or through certain motivation only if such motivation is at that juncture for that subject sufficient to yield the affirmation, even in the absence of other operative purposes. The affirmation might conceivably be overdetermined by epistemic rationale on one side, and pragmatic rationale on the other. Really judging that p is affirming or being disposed to affirm that p in the endeavor to affirm correctly (with truth) and reliably enough, where such epistemic motivation is sufficient to yield the affirmation, even if the subject also has sufficient pragmatic rationale for doing so. You really believe (judgmentally) all and only what you are so disposed to affirm, when your endeavor is epistemically pure and disinterested, sufficiently epistemically motivated, while aiming to attain truth reliably enough on the question at hand. e. However, even someone who is not sufficiently epistemically motivated to affirm that p, when she faces the question whether p, could still satisfy to an enormous extent the functional profile of belief that p. Someone like that may functionally believe that p without judgmentally so believing. Someone deeply biased against a certain group, for example, may repudiate the bias upon considering consciously whether the target group is or is not inferior. And the like seems true of well-known delusions such as the disturbing Capgras. Does the subject really believe that his ostensible wife is just an impostor? No, he still believes her to be his wife, since he still interacts with her as he would with his wife. At least, he massively satisfies the relevant functional profile for that belief. Even without functionally believing his co-resident to be an impostor, nevertheless, he does seem judgmentally to believe it, in line with what he does affirm seriously to others and even perhaps to himself. 3.2 Does belief have dedicated reasons? a.
It is widely thought that there are two forms of reason, the theoretical and the practical. This is supposed to go with two sorts of reasons: reasons for belief, and reasons for intention and action. Although belief that p, being a state in the world, will have consequences, and although this may seem to provide reasons
123
Responses to four critics
b.
c.
d.
e.
f.
635
for acquiring and sustaining that belief, or for avoiding it, these are widely thought to be inappropriate reasons for belief that a believer should not and perhaps cannot use as a basis for so believing. I would like to lay out some reasons to think that the two supposedly divergent domains—action and judgment—are much closer than they first seem to be. If the agential view of judgment and belief is on the right track, many of us are going wrong in supposing that judgment and judgmental belief are categorically different from action, and that reasons for belief must differ radically from reasons for action. On the agential view, reasons for occurrent judgment, and for other cognitive options, such as dispositional judgment, suspension of judgment, and others, are just special reasons for action, for the action that is constitutive of the pertinent cognitive option. Compare a means-end intentional action. Here it is important to distinguish between two things: (i) reasons why one might X in the endeavor to attain end E, as a means to attain that end, and (ii) given that one is endeavoring to attain E, reasons why one might X in that endeavor. What practical reasons might I have for flipping a certain switch in the endeavor to turn on the light? I may just want to see the contents of a room that is now dark. But here there is an interesting contrast. Given I am endeavoring to illuminate the room, there are dedicated reasons for flipping that switch. What dedicated reasons might favor flipping that switch, rather than flipping some other switch, or clapping my hands, etc.? Among such dedicated reasons would presumably be found reasons in favor of the proposition that flipping that switch will cause the light to go on (as well as that proposition itself, at least provided it is a true proposition). Compare dedicated reasons for judging that p, under our agential account of judgment. Suppose that judgment is affirmation in the endeavor to get it right on the question under judgment. In that case, we can distinguish dedicated reasons for so affirming, given that we are endeavoring to get it right on that question. If that is indeed our endeavor, then among our dedicated reasons would be found reasons in favor of the proposition that in affirming we will thereby attain our objective, namely getting it right on that question. Thus, our agential account of judgment (and judgmental belief) helps to explain why it is that there are dedicated reasons for judgment. If judgment is a meansend action, then, given one’s end, there will be distinctively appropriate reasons for or against adopting the means that one adopts. The distinctively appropriate reasons will be those that bear on how likely it is that one’s adopted means will lead to the given end. Practical reasons having to do with reputation, earned income, etc., are inappropriate reasons so long as they do nothing to support the relevant means-end connection. Again, why is it that reasons for judgment (and judgmental belief), must be truth-related? We can now answer: Because judgment is constitutively truth-aimed, and the distinctively appropriate reasons for adopting certain means given a certain end will be the reasons in favor of the proposition that the means will help attain the end.
123
636
E. Sosa
I have not given a detailed defense responsive to Engel’s critique; that defense must await the publication of my work now in progress. Nevertheless, I have offered considerations that do respond to the two main points highlighted in his abstract of his commentary, namely that… ‘‘[a] beliefs are not performances, and … [b] epistemic reasons or beliefs cannot be balanced against practical reasons.’’ I have tried to sketch features of my proposed framework that explain what is plausible in those two main points of the critique, but that I hope will also help to clarify what is not so plausible in them.
123