1. Introduction
In a series of articles and books, Ernest Sosa has developed a sophisticated view of the nature of knowledge and epistemic normativity. Very roughly, Sosa's account depends on the idea that having a belief is a performance with a constitutive aim (or aims) and can be assessed in the same ways that any performance with an aim can be assessed. A belief counts as knowledge when the performance counts as a good performance (in the relevant sense).
Although I do not share this view, or even one in the nearby vicinity, I do have a great deal of sympathy for it. Indeed, as I argue in a forthcoming paper, I think there are very few accounts in the literature that show promise for explaining how it is that we are epistemically justified in reasoning in some ways rather than others, especially in cases where we do not have any prior justification to believe that those ways of reasoning are good ones.Footnote 1 There are not many accounts that explain where the “normative oomph” (to use a technical term) could come from. Sosa's account, or one like it, has the potential to explain this normativity. So I find the central ideas underlying Sosa's account deeply attractive.
What I'd like to do in this paper is to explain why I nevertheless do not accept Sosa's account of knowledge and epistemic normativity. My discussion will have two main parts. In the first part, I will identify two places where I think Sosa's account requires supplementation with additional theoretical resources if it is going to provide an account of epistemic normativity. Supplementing the account in these ways strengthens the view. But it also begins to suggest that we should take a different general approach to explaining epistemic normativity. This is where my biggest-picture concerns about Sosa's view lie.
In the second part of the paper, I will turn to a specific part of Sosa's account, a part that he has recently been emphasizing.Footnote 2 In particular, Sosa argues that there is a kind of belief – judgmental belief – that doesn't merely aim at truth but also aims at “aptness”, that this kind of belief is centrally important to our mental lives, and that this kind of belief is (in some sense) a higher grade of belief than belief that merely aims at truth. I will present a few concerns about this part of Sosa's picture. My objections are not conclusive, merely suggestive. But they provide reasons to be wary of the idea that there is an important kind of belief that aims at aptness.
But before I get to all of that, let me first present the necessary background on Sosa's views.Footnote 3 That is the task of the next section.
2. Background
2.1. Sosa's general picture
Let me start by sketching Sosa's general picture of performance and achievement. For Sosa, a performance is a state, action, or process that has a constitutive aim or aims. Aims come in two varieties: they may be functional (and arise from the biological, psychological, or sociological function of the performance), or they may be intentional (and arise from the performer's conscious or unconscious intentions).Footnote 4 If the constitutive aim of a performance is intentional, the performance counts as an endeavor to attain the relevant aim or aims. Endeavors form an important class of performances, since they are closely tied to the agency of the performer.
Performances can be evaluated as performances in at least three ways. (Sosa counts these as telic evaluations.) A performance is accurate if it succeeds in attaining its aim. A performance is adroit if it was skillfully performed – that is, if it is the exercise of an at least minimally reliable competence to succeed in attaining its aim.Footnote 5 Finally, a performance is apt if it is accurate because it is adroit. Or, more precisely, a performance is apt just in case its success in attaining its aim sufficiently manifests the performer's competence to succeed in attaining the aim.
An accurate performance is better, in some sense, than an inaccurate one. But an apt performance is better still. This is because in an apt performance, the performer's success in attaining the aim is due to skill rather than (a problematic kind of) luck. In an apt performance, success in attaining the aim is a genuine achievement of the performer. The success is something that is attributable or creditable to the performer.
Performances can have multiple constitutive aims. One important way this can happen is when they have two aims – a basic aim and the correlative aim of aptly attaining the basic aim.Footnote 6 Such performances are called “full performances”.
Full performances can themselves be accurate, adroit, and/or apt. The best kind of performance is the fully apt performance – a full performance that aptly attains its basic aim and also aptly attains its correlated aim of aptly attaining its basic aim. (Sosa also requires that in a fully apt performance, the performer's aptness in attaining the basic aim of the performance is guided by an apt awareness that the performer would likely enough attain the basic aim if the performer tried to do so. But I'll ignore this wrinkle in what follows.)
2.2. The case of epistemology
That is Sosa's picture of performances and achievement. Sosa goes on to apply this general picture to the particular case of epistemology. For Sosa, beliefs are performances. They are states with constitutive aims. There are two kinds of beliefs: functional beliefs and judgmental beliefs. Both have the aim of representing the world correctly. Functional beliefs are implicit, non-intentional, and are not under the direct control of the agent. Judgmental beliefs are intentional, often consciously intentional, and are under the direct control of the agent. (There is also a third, hybrid kind of belief that is under the indirect control of the agent, but that will not be relevant in what follows.)
In a bit more detail, a judgment is a kind of affirmation – a “saying” that may be either public or in the privacy of one's own mind. A judgment that p is an affirmation that p in the endeavor of getting it right on the question of whether p, and in the endeavor of aptly getting it right on the question of whether p. That is, a judgment is a full performance, with a basic aim of affirming correctly on the question of whether p, and a correlated aim of doing so aptly.
A judgmental belief that p is a disposition to judge that p, in the endeavor of aptly getting it right whether p, if one should so endeavor. Judgmental beliefs form an important class of beliefs because they are closely tied to agency and because they most fully manifest our rational nature.Footnote 7 As before, a judgmental belief that p may be accurate, adroit, and/or apt. The best kind of judgmental belief is one that is a fully apt performance – it aptly attains the basic aim of getting it right whether p and it aptly attains the correlated aim of aptly getting it right whether p.
On Sosa's view, knowledge is identified with apt belief. A belief that p that aptly attains the aim of getting it right whether p counts as knowledge that p. A belief that p that aptly attains the aim of aptly getting it right that p counts as knowledge full well that p. On this view, a thinker knows that p when the thinker's success in attaining a true belief that p is creditable to the thinker. A thinker has knowledge full well that p when the thinker's success in attaining knowledge that p is creditable to the thinker. In a slogan, knowledge is creditable true belief. Knowledge full well is creditable knowledge.Footnote 8
This picture of the nature of knowledge and of epistemic normativity is very elegant.
2.3. Other parts of our mental life
Sosa also applies the general picture of performance and achievement to other parts of our mental life. On his view, actions are apt intentions. Perceptions are apt perceptual experiences.
Additional applications are possible, too. For instance, one might extend the view to emotional states and to desires and other conative states. One might view emotional states as performances with aims. Perhaps fear has the aim of correctly representing danger, anger has the aim of correctly representing norm violations, and sadness has the aim of correctly representing loss. One might view desires and other conative states as performances with aims. Perhaps the aim of an instrumental or realizer desire has to do with the final desire that it serves. Perhaps (much more controversially) the aim of a final desire has to do with whether what is desired contributes to well-being. If so, emotional states and desires may be accurate, adroit, and/or apt (or not).
The prospect of these further applications is an advantage of Sosa's view, since emotions and desires (and many other kinds of mental states) can be normatively assessed – such-and-such a fear is reasonable, such-and-such a desire is irrational, and so forth. If this entire range of normative assessments can be unified under a general theory of performance normativity, that would be a strong count in favor of Sosa's view.
3. Missing pieces?
Let me now turn to the first main part of the paper. In this part, I'm going to argue that Sosa's picture of epistemic normativity as a kind of performance normativity is incomplete. There are places where additional theoretical resources are necessary. At the very least, there are places where more needs to be said. In what follows, I'll focus on two such places.
3.1. Evaluating the aims
The first place where I think more needs to be said concerns the aims of belief and inquiry. On Sosa's view, performances have constitutive aims. Whether a performance has telic value depends on whether it attains its aim (or aims), and whether it does so adroitly, aptly, and the like.Footnote 9 But the aims are not themselves evaluated, at least not on telic grounds.Footnote 10 The aims of a performance can be evaluated in other ways – for instance, on moral or aesthetic grounds – but the kind of normativity involved in such evaluations is not performance normativity.
For example, as Sosa points out, an assassin's shot may be an apt shot, and even a fully apt shot, and thus have telic value. When a belief aptly (or fully aptly) represents whether p, it is valuable in exactly the same sense. The value of the shot or of the belief stems from its apt attainment of its aim (and perhaps from its apt attainment of its correlative aim of aptly attaining its basic aim).
Sosa views this feature of his account as one of its strengths.Footnote 11 It is meant to be an advantage of the account that epistemic normativity turns out to be a species of a familiar and more general kind of normativity – performance normativity. It is also meant to be an advantage of the account that epistemic normativity does not require the primary epistemic aim – having true beliefs – to have any kind of intrinsic or final value. Beliefs can be epistemically assessed while remaining neutral on this issue. The primary epistemic aim of having true beliefs can be taken as a given, and not itself assessed.
I am sympathetic to the claim that epistemic normativity is a species of a more general kind of normativity. After all, we can assess many other kinds of mental states for rationality besides beliefs, disbeliefs, and credences. But the claim that all there is to epistemic normativity is performance normativity seems to miss something important about epistemic normativity.
Correctly representing the world is not any old aim. It is not like the aim of committing a perfect murder or the aim of constructing the third-largest ball of twine in the local metro area. The former aim is abhorrent and the latter aim is pointless (absent other considerations). But the aim of correctly representing the world has something going for it. It is a good aim to have.Footnote 12
This is not to say, of course, that thinkers ought to aim to believe every truth and disbelieve every falsehood. As Gilbert Harman has long emphasized, one should not clutter up one's mind with trivialities.Footnote 13 One should not engage in long and complex investigations aimed at answering uninteresting questions.Footnote 14 But the aim of representing the world correctly is superior to the aim of representing the world incorrectly (or representing the world in the maximally humorous way, or …). And the kind of superiority at issue is plausibly a kind of epistemic or rational superiority (and not a kind of moral, aesthetic, or prudential superiority).Footnote 15 The claim that all epistemic normativity is performance normativity cannot accommodate this point, and so requires supplementation.
In response to this concern, Sosa could argue that performance normativity can accommodate (or explain away) the intuition that correctly representing the world has value. In arguing for this, Sosa could point to several different factors. First, Sosa could argue that when a performance has an aim, the performer must rationally prefer the attainment of the aim to its frustration. So, from a first-person point of view, it will seem good to a thinker that his belief is true.Footnote 16 Second, Sosa could argue that when we assess a performance third-personally, we assess the performance in part in terms of whether it attains its aim. So, from a third-person point of view, too, it will seem good that some other person's belief is true. Third, Sosa could argue that when we assess a performance in some domain, we assess the performance in terms of the normative standards operative in the domain. The primary normative standard operative in the epistemic domain is truth. So, it is natural to think that a belief is good when it is true. Finally, Sosa could argue that we all share the aim of having correct representations of the world. When we assess beliefs, it is natural to assess them with regard to these shared standards. So it is again natural to think that a belief is good when it is true.
The trouble with this response is that these four considerations do not distinguish between aiming at truth and having some valueless or negatively valued aim. For instance, consider the aim of performing a successful assassination. First, from a first-person point of view, an agent performing an assassination will rationally prefer the attainment of the aim to its frustration. Second, from the third-person point of view, when someone performs an assassination, that performance will be assessed in part in terms of whether it attains its aim. Third, the primary normative standard in the domain of assassinations is success in assassinating. Finally, we can imagine a community of assassins who all share the aim of carrying out successful assassinations. Members of the community will naturally assess one another with regard to their shared standards. In short, the four considerations apply just as well to neutral or bad aims as to good aims. Insofar as we think that aiming at true representations is good in a way that aiming at successful assassinations or enormous balls of twine are not, these considerations cannot account for the difference.
There is another potential response that Sosa could provide. In my description of Sosa's views, I have so far suppressed a distinction that Sosa makes between two parts of epistemology – the theory of knowledge and intellectual ethics. The theory of knowledge is “focused on the nature, conditions, and extent of human knowledge” while intellectual ethics “concerns evaluation and norms pertinent to intellectual matters generally, with sensitivity to the full span of intellectual values.”Footnote 17 For Sosa, knowledge (and related evaluative notions) is to be understood in terms of telic value. By contrast, intellectual ethics involves epistemic assessments that should be understood in some other way. The response, then, is that we should understand the value of correctly representing the world as part of intellectual ethics and not as part of the theory of knowledge.
There are two difficulties facing this response – one specific to Sosa's account of intellectual ethics and one more general. On Sosa's account, intellectual ethics primarily concerns the assessment of whether thinkers manifest two epistemic virtues (or the corresponding vices).Footnote 18 The first virtue is that of managing one's competences and position in the world so as to enable one to make apt judgments and have apt beliefs. The second virtue is that of properly guiding one's attention and aims. As Sosa writes, “Not all matters are worthy of our attention; not all aims are worth pursuing. That varies with the practical circumstances, with the attitudes, background, and situation of the agent.”Footnote 19 The second virtue, then, is the virtue of attending to worthy matters and aiming at what is worth pursuing. It is this part of intellectual ethics that may appear to hold promise for capturing the goodness of the aim of correctly representing the world.
The trouble with this suggestion, however, is that Sosa suggests that which aims are worthy of pursuit depends on the particular attitudes of the agent. As I understand Sosa, intellectual ethics takes the basic aims of the agent as given, and considers whether additional aims are worthy of pursuit in terms of the basic aims. If an agent lacks the aim of correctly representing the world, or any aim that would be served by this aim, intellectual ethics will be silent on whether this aim is worth pursuing. Since an agent can, in principle, have any collection of aims whatsoever, intellectual ethics cannot capture the way in which the aim of correctly representing the world is a good aim to have for all agents.
The more general difficulty facing the proposed response is that intellectual ethics is intended to be divorced from the theory of knowledge. The assessments at issue in the theory of knowledge (accuracy, adroitness, aptness, full aptness) do not depend on the assessments at issue in intellectual ethics, or vice-versa. But it is an essential feature of accurate beliefs that they are superior to inaccurate beliefs. It is an essential feature of knowledge that it is superior to ignorance.Footnote 20 So it should fall out of the theory of knowledge that the aim of true beliefs is a good aim to have. This should not be left to an independent part of epistemic normativity.
In addition to the claim that aiming at true beliefs is a good aim to have, there is a still more radical claim that I'd like to endorse here. Consider a thinker who only minimally investigates the world around him. Suppose he doesn't try to predict or explain anything. Suppose he doesn't try to form beliefs about his immediate environment or figure out what's going to happen in his immediate future. Suppose, rather, that he spends his life sitting on a sofa, passively watching the play of light and color on a screen, and taking things as they come. (Suppose he is hooked up to an apparatus that keeps him fed, hydrated, and in good physical condition.) Such a figure strikes me as a tragic one, and not only on moral or prudential grounds. Rather, such a figure is rationally not as he ought to be.
If this is right, then there are norms that govern the endeavors that thinkers – or, at least, thinkers like us – rationally ought to engage in. It is plausible that irrespective of their goals and desires, thinkers rationally ought to engage in some amount of inquiry. They rationally ought to aim to explain the world around them and predict important phenomena, at least to some degree. Engaging in such cognitive projects are not rationally optional endeavors, open to us to adopt or not as we see fit. Rather, they are genuine requirements of rationality.Footnote 21
The existence of these rational requirements is connected to the fact that aiming at true beliefs is a good aim to have. Presumably, part of the reason that thinkers ought to engage in inquiry is that the products of inquiry have value, at least when the inquiry goes well. That is to say, having true representations of the world is objectively valuable. So the explanation of our rational obligations to inquire, to predict, and to explain is connected to the explanation of why having true representations of the world is valuable. A purely telic picture of epistemic normativity cannot deliver these explanations. This suggests that something needs to be added to the view.
3.2. Small steps and big leaps
The second place where more should be said concerns a different part of Sosa's view. Consider a thinker who knows (i) that it is snowing outside and (ii) that if it is snowing outside, the roads will be slippery, and on that basis directly infers (iii) that the roads will be slippery, using some form of Modus Ponens. Suppose the thinker has no defeaters for the conclusion, is in good cognitive shape, and the like. The conclusion belief is accurate (since the premises are true, and the premises entail the conclusion). Presumably, the thinker's belief in the conclusion is adroit, since it is the exercise of a reliable competence directed at the truth. And, presumably, the thinker's belief in the conclusion is apt, since it is accurate because it is adroit.
So far, so good.
Now compare a thinker who knows (i) that 13, 16, 17, and 5 are positive integers and (ii) that 5 > 2, and on that basis directly infers (iii) that 135 + 165 ≠ 175, using a rule of inference that we can call FLT: From the beliefs that x, y, z, and n are positive integers and that n > 2, infer that xn + yn ≠ zn. (FLT is thus a version of Fermat's Last Theorem in rule form.) Assume that the thinker does not antecedently know that Fermat's Last Theorem is true or that FLT is truth-preserving. Also assume that in using this rule, the thinker does not do any rapid mathematical calculations. The thinker does not somehow imaginatively survey all of the natural numbers or quickly prove Fermat's Last Theorem before moving from the premise to the conclusion. The thinker simply infers from the premises directly to the conclusion.Footnote 22
Suppose the thinker has no defeaters for the conclusion, is in good cognitive shape, and the like. The conclusion belief is accurate. On Sosa's view, the thinker's belief in the conclusion would seem to count as adroit, since applying the rule FLT to known premises is a competence – it involves a disposition to successfully determine whether xn + yn ≠ zn in a wide range of cases. The thinker's belief in the conclusion would also seem to count as apt, since the thinker's success in determining whether 135 + 165 ≠ 175 manifests the thinker's competence. So, on Sosa's view, the thinker would seem to count as knowing the conclusion of the inference.
This is an uncomfortable claim to endorse. Intuitively, there is an important difference between a belief inferred using Modus Ponens from known premises and a belief inferred using FLT from known premises. On the face of it, the former belief will typically count as knowledge and the latter belief will not. The view that knowledge is apt belief does not easily accommodate this intuitive difference. So we have an apparent counterexample to Sosa's view.Footnote 23
One way to bring out the intuitive problem is to note that thinkers who infer using Modus Ponens seem to be taking only a small step in their thinking. Thinkers who infer using FLT seem to be taking a giant leap in theirs. Such giant leaps are incompatible with the thinking yielding genuine knowledge. A different way to bring out the intuition – one more closely tied to performance normativity – is that a thinker who infers using Modus Ponens from known beliefs typically deserves credit for ending up with a true belief. A thinker who infers using FLT from known beliefs does not. Again, this suggests that a belief formed using FLT from known beliefs should not count as knowledge.
There are (at least) two ways that Sosa could respond to this objection. First, Sosa could argue that a thinker who uses FLT to conclude that 135 + 165 ≠ 175 has a defeater for this belief and so does not end up with knowledge. The idea would be that (in the absence of knowing that Fermat's Last Theorem is true or the like), an ordinary thinker who employs FLT, or a rule like it, would have good reason to think that they are likely relying on a mistaken rule. (Perhaps this is because they would know that anyone in our community who employed a complex mathematical rule and was not a mathematical prodigy was likely suffering from an unusual psychological condition uncorrelated with using a reliable rule.) So the epistemic status of beliefs formed on its basis would be defeated. Second, Sosa could claim that the thinker's belief that 135 + 165 ≠ 175 is in fact apt (and therefore counts as knowledge), but that it lacks some important higher-order epistemic property. For instance, perhaps it does not count as fully apt (and therefore is not knowledge full well). Our intuition that there is something wrong with the belief formed on the basis of FLT is really an intuition about full aptness, not aptness itself.
These lines of response are not very promising, however. True, an ordinary thinker in our society who relies upon FLT (without knowing the truth of Fermat's Last Theorem or the like) would presumably have evidence that they were likely using an invalid rule, and so would not count as knowing the conclusion. And, even if they did not have such evidence, we would view them as incredibly lucky to employ a valid rule, and would naturally resist thinking that the conclusion belief was fully apt. So considering only ordinary thinkers in our society, each of the two responses has some plausibility. But we can modify the counterexample to immunize it from each of these lines of response.
Consider a society of thinkers with a slightly different constitution to our own – they employ all the same basic epistemic rules that we do, with the addition of FLT. This is their innate constitution – they use FLT just as we use Modus Ponens (or whatever rules we use innately). Suppose these thinkers are skillful in applying FLT without error, apply the rule in a wide range of cases, and do so automatically. When people in their society inquire with the aim of figuring out whether various numerical identities obtain, they take FLT to be a good rule for doing so. And, in general, they treat FLT just as we treat Modus Ponens. This restores the symmetry between the two rules. On Sosa's view, then, members of this society would seem to aptly believe that (say) 135 + 165 ≠ 175, and indeed to have a fully apt belief that 135 + 165 ≠ 175. But it remains implausible that they know that 135 + 165 ≠ 175 or have knowledge full well that 135 + 165 ≠ 175.
The general problem here is due to Sosa's claim that all that matters for knowledge (and for related higher-order epistemic properties) is attaining the truth and being disposed to attain the truth (and having higher-order dispositions that also involve attaining the truth). Given this claim, there is no obvious way to break the symmetry between Modus Ponens and FLT.
This again suggests that something needs to be added to Sosa's view. Something needs to be said to explain how Modus Ponens and FLT are epistemically disanalogous, and how it is that using Modus Ponens is an epistemically reasonable or responsible thing to do but using FLT is not.Footnote 24 In Sosa's framework, the natural suggestion to make is that we should add restrictions on when a belief – or a performance more generally – counts as skillful or adroit. (Or perhaps we should add restrictions on what counts as a competence.) It is insufficient that the performance manifests a disposition to succeed reliably enough. But it is not clear to me how to impose a natural restriction that would be broadly in the spirit of Sosa's general picture of epistemic normativity.
***
Neither of the issues I have raised in this section provides an objection to the main ideas underlying Sosa's view of knowledge and epistemic normativity. Rather, they each provide a place where more needs to be said to capture important epistemic phenomena. Additional theoretical resources need to be deployed to explain (i) the way in which epistemic aims are genuinely good aims, and can be rationally non-optional irrespective of the thinker's goals and desires; and (ii) the way in which some forms of reasoning are epistemically better than others, even when they are equally conducive to attaining the truth. So it is possible to address these issues not by rejecting central parts of Sosa's view, but by supplementing it further.
That said, my own view is that addressing these issues motivates moving away from a performance normativity-based view of the epistemic. If we say (as I would like to) that there are rationally required epistemic aims, that Modus Ponens is a better way of reasoning than FLT because it is useful or important or indispensable for achieving rationally required aims, and that this is what explains why using Modus Ponens is something thinkers ought to do (or are epistemically responsible in doing), the resulting picture of the epistemic looks very different from a view on which epistemic normativity is grounded in performance normativity. But that's a long story, one I cannot tell here.Footnote 25
4. Aiming at aptness
Let me now turn to the second main part of this paper, concerning the aim of aptness. Recall that on Sosa's view, judgmental beliefs form a particularly important class of beliefs because they are closely tied to agency and most fully manifest a thinker's rational nature. In that way, judgmental beliefs are a higher-grade kind of belief.Footnote 26
Judgmental beliefs differ from other beliefs because they have intentional rather than functional aims. Moreover, they don't merely constitutively aim at accuracy – at correctly representing the world – but also constitutively aim at aptness – at correctly representing the world in virtue of manifesting one's competence or skillfulness. These two features are what explain their tie to agency and rationality, and their status as high-grade beliefs.Footnote 27
What I would like to do now is to raise some difficulties for this part of Sosa's view. The worries I'm going to raise are not conclusive, but they are telling. They suggest that it is not the case that there is an important class of beliefs that intentionally aim at manifesting one's skillfulness or aptness.Footnote 28
One worry about this part of Sosa's view is that I, at least, am not introspectively aware of any intentions to form beliefs skillfully or aptly. When I engage in inquiry, even sophisticated kinds of inquiry – for instance, solving a mathematical problem, reasoning about some vexed philosophical issue, or making a difficult practical decision – I am focused on the first-order question at hand and not on my own skillfulness. My preference is to answer the question, not to be skillful in so answering it. Of course, I'd like to arrive at the correct answer if there is one. And I may prefer for the answer I arrive at to itself be elegant – for instance, I might be interested in identifying an elegant mathematical proof or an elegant way to articulate or develop a philosophical view. But even then, my focus is on the elegance of the output of my inquiry and not on the skillfulness or adroitness of the reasoning leading up to it. And this is so even for the kinds of inquiry that seem most closely tied to epistemic agency and rationality.
Now, it could be that I'm psychologically idiosyncratic and that other people are introspectively aware of an aim of aptness. But I suspect I'm not unique in this regard. Indeed, I expect that most people are not introspectively aware of an aim of aptness. If that's right, then this is evidence against the claim that there is an important class of beliefs with a consciously accessible aim of aptness.
Here's a second worry about Sosa's view: Suppose the epistemology oracleFootnote 29 (or, perhaps better, the epistemology pill pusher) offers me the choice between two pills, each of which will remove some of my beliefs and leave the rest in place, and otherwise make no changes. The oracle – who I'm rationally certain is completely trustworthy – assures me that the pills are safe and effective. The first pill will remove some of my false beliefs. The second will remove some of my non-apt beliefs. (For concreteness, assume that the pills remove 5% of the relevant class of beliefs, and will do so at random. If it matters, also assume that the pills’ effects are limited to beliefs that are the best candidates for being judgmental beliefs – sophisticated beliefs tied to rational agency.)
Which pill would I choose to take? The obvious answer is that I would choose the first pill, the one that removes false beliefs.Footnote 30 This suggests that if I do have an aim of aptness, it not as strong as my aim of accuracy. And I suspect most people would make the same choice.
We can consider variants of this story. Suppose the oracle instead offers me a pill that will remove some of my accurate but non-adroit or non-apt beliefs. (Again, assume the pill's action is limited to beliefs that are good candidates for being judgmental beliefs. In this scenario, it should also be specified that after I take or refuse to take the pill, I'll forget everything that just happened. So I won't be able to use the fact that one of my beliefs was removed as new evidence for its truth.)
Would I take this pill? The obvious answer is no. I would refuse. Indeed, I would strenuously object to taking such a pill, and would be willing to give up some things of value in order to avoid taking it. This again provides some evidence that adroitness and aptness are not aims of my beliefs, even for the beliefs that are the best candidates for being judgmental beliefs.
There is a reply that Sosa could make to these objections. He could argue that they are based on a mistaken picture of what is required to have an intentional aim of aptness.Footnote 31 First, Sosa could say that having such an aim does not require that it be consciously accessible. And second, Sosa could say that having such an aim does not require that it manifest itself in behavior the way that the two oracle cases presuppose that it does.
What, then, is required to have an intentional aim of aptness? There is a potential dilemma here. Having an intentional aim of aptness is supposed to help to explain why judgmental beliefs are closely tied to agency. If the requirements on having such an aim are very weak – for instance, if they are just a matter of being disposed in some circumstances to behave so as to make aptness more likely – it is difficult to see how aiming at aptness could be tied to agency and rationality.Footnote 32 There would not be anything distinguishing having an intentional aim of aptness from having some kind of non-agential and arational compulsion toward aptness. On the other hand, if the requirements on having such an aim are sufficiently strong, it is difficult to avoid the worries presented above. So what is needed is some principled middle ground.
Sosa suggests an account of the aim of aptness that seems intended to occupy this middle ground.Footnote 33 On Sosa's view, judgmental beliefs have a primary aim of accuracy and a secondary aim of aptness. Sosa presents a general account of primary and secondary intentional aims to explain how this could work. A performance has the primary aim of P and the secondary aim of S if: (i) the performer has an intention to attain P, (ii) S is a determinate of P – that is, one way to attain P is to attain S, and (iii) the performer prefers attaining the combination P & S to the combination P & not-S.Footnote 34 Applying this account to the case of interest, Sosa claims that when a thinker has a judgmental belief, (i) the thinker has the intention to form an accurate belief, (ii) aptness is a determinate of accuracy,Footnote 35 and finally, (iii) the thinker prefers an accurate and apt belief to an accurate and inapt belief.
This account, if correct, can answer the two difficulties raised above. The response to the first difficulty, concerning the lack of an introspectively accessible intention, is that when I intentionally aim at aptness, I need not have an intention with the specific content that my belief be apt; it suffices that I have an intention with the content that my belief be accurate. So it is unsurprising that I am not introspectively aware of an intention concerning aptness. The response to the second difficulty, concerning the epistemology oracle, is that what is needed to have an intentional aim of aptness is that I prefer accurate and apt beliefs over accurate and inapt beliefs. It is compatible with this that I prefer to lose inaccurate beliefs over inapt beliefs, and that I prefer to retain accurate but inapt beliefs over losing them. In other words, my preferences may be lexically ordered – accuracy comes first, and aptness second.
This account of an intentional aim of aptness faces difficulties. One general worry is that the account may be subject to the second horn of the dilemma above. If all that is required to have an intentional aim of aptness is that one intentionally aims at accuracy and that one prefers accuracy and aptness over accuracy and inaptness, it is difficult to see how intentionally aiming at aptness could be tied to agency and rationality. How and why is it that having such preferences explains our epistemic agency and manifests our rational natures?
One might try to add conditions to the account – for instance, perhaps to have an intentional aim of aptness, one has to also possess a plan to attain this aim which guides one's thinking. Such a condition is more plausibly connected to agency and rationality. But if we impose such a condition, the worry about introspective accessibility returns. If we have a plan for aptness, why is it that we have no introspective access to the aim of the plan?Footnote 36
We can also test Sosa's account of the intentional aim of aptness with another scenario involving the epistemology oracle. Suppose I'm equally curious about whether p is the case and whether (some unrelated) q is the case, but I only have the time and resources to investigate one of these two questions. Suppose the two questions are objectively just as interesting and pragmatically just as important as one another. Suppose I'm about to decide which of the two possible inquiries to launch when the epistemology oracle appears. The oracle – this time relying on her oracular powers rather than her stock of unusual pharmaceuticals – informs me that if I inquire into whether p, I'll end up with an accurate belief about whether p, but the resulting belief will be inapt. (The oracle may inform me that this is because I'll be in a Gettier scenario or even because my inquiry will be ineptly conducted.)Footnote 37 By contrast, if I inquire into whether q, the resulting belief about whether q will be both accurate and apt. Suppose, finally, I know that after I make my choice, I'll forget what the oracle told me. (Thus, I won't be able to leverage the oracle's pronouncements to gain an apt belief into whether p in an indirect way.) For what it's worth, my intuition about this case is that it would be perfectly fine for me to inquire into whether p and it would be perfectly fine for me to inquire into whether q. I don't have any clear preference for ending up with an accurate and apt belief over an accurate but inapt belief.
One way to strengthen this intuition is to consider a slight variant of the case. Suppose I have already decided on inquiring into whether p when the oracle appears to me. (Suppose I've written it down in my daily planner.) Does the oracle's information provide me with any reason at all to switch to the other inquiry? I maintain the answer is no. Indeed, I would be relieved to hear the oracle's pronouncements – it would mean I will arrive at the correct answer in my inquiry, which is all I really care about in the end.
If this is right, what this suggests is that I (at least) don't have an intentional aim of aptness, even on Sosa's understanding of what such an aim would come to. I don't prefer accuracy and aptness over accuracy and inaptness. Even for the best candidates of beliefs tied to agency and rationality, my sole aim is that of accuracy.Footnote 38
There is one last worry that I'd like to raise about the aim at aptness, however it is to be understood: Aiming at aptness is overly self-directed. In reasoning – even in the highest grade of reasoning – one's focus should be on the world and not on oneself. (I'm putting aside cases where one is inquiring into a question about oneself.) Something is wrong with a thinker who aims at being skillful instead of – or even in addition to – aiming at the truth.Footnote 39 Aiming at aptness is making things too much about oneself.Footnote 40
The issues here are a little delicate. Certainly, in good cases of reasoning, one's aim has something to do with oneself. When I try to determine what to believe on some topic, my aim is that I have true beliefs on the topic, not that the number of true beliefs in the world increases. Moreover, in reasoning, it is often the case that I should take into account my own reliability in my reasoning on the topic. If I know that I'm intoxicated or that I'm prone to forgetting to carry the one, I should downgrade my degree of confidence in the results of my reasoning and perhaps give up the belief. But the reason I should pay attention to the reliability of my reasoning is that my estimate of my reliability is used in determining what to think about the world. Even when I'm taking into account evidence about my reliability, my aim should be solely to have correct representations about the world.
One way to see the point is to think of examples of people who do intentionally aim at aptness. Examples of people who intentionally aim at aptness in some of their thinking can be found in works of fiction, in particular in the genre of detective stories.Footnote 41 Maybe the clearest example of this is Agatha Christie's detective Hercules Poirot. He doesn't merely want to solve the case, but to manifest his skillful reasoning in determining the guilty party. This is not portrayed as a positive trait. Numerous characters throughout Christie's series of books comment on how conceited he is. Indeed, Christie herself came to detest her creation, calling him an “egocentric little creep”.
This kind of character flaw is rampant throughout the genre.Footnote 42 Meeting Sherlock Holmes and seeing him show off his skills as a detective, Watson thinks to himself, “This fellow may be very clever … but he is certainly very conceited.”Footnote 43 The trouble, again, seems to be that Holmes is not focused only on figuring things out and solving his cases, but on his own skillfulness. His reasoning – impressive as it is – is too much about himself, and not focused enough on the matter at hand.Footnote 44
This kind of argument is, of course, not decisive. It is always possible to argue that Poirot and Holmes's flaw is not that they aim at their own skillfulness or aptness, but something else. For instance, perhaps the relevant character flaw is taking too much pleasure in one's skillfulness, or attending too much to one's skillfulness, or calling too much attention to one's skillfulness, or aiming at too high a level of skillfulness, or having skillfulness as a primary and not a secondary aim, or something like that. But it is striking that the characters in literature who are the clearest cases of people with the intentional aim of manifesting skill or aptness in their epistemic performances are presented as flawed characters. And their flaw clearly has something to do with their aim of manifesting their skill.
5. Conclusion
Let me take stock. I've argued for two main points. First, there are places in Sosa's picture of epistemic normativity where more needs to be said – namely, on the goodness of epistemic aims and on the ways in which equally accurate forms of reasoning may differ in their epistemic status. I've suggested – tentatively – that filling in the story in these places will give the resulting view a different complexion. Second, I've argued that it is not the case that there is an important class of beliefs that intentionally aim at aptness. Such beliefs, when they exist, are overly self-directed, and seem more closely connected to vices than epistemic virtues. If that's right, then the picture of judgmental beliefs as aimed at aptness and as closely tied to agency and rationality should somehow be modified.Footnote 45
