Performance normativity and here-and-now doxastic agency

Sosa famously argues that epistemic normativity is a species of “performance normativity,” comparing beliefs to archery shots. However, philosophers have traditionally conceived of beliefs as states, which means that they are not dynamic or telic like performances. A natural response to this tension is to argue that belief formation rather than belief itself is the proper target of epistemic normativity. This response is rejected here on grounds of the way it obscures the “here and now” exercise of cognitive agency that I view as central to any account of epistemic normativity and doxastic agency. Although the etiology of a belief can be relevant to its normative status, often so much more is relevant and more centrally so. This generates a dilemma for anyone following Sosa in pursuing the idea that epistemic normativity is a species of performance normativity.


Introduction
Genuine norms, many philosophers assume, are prescriptive in that they tell us primarily what we ought to do. Of course, some 'ought's tell us how things ought to be, but it is widely assumed that these 'ought's must connect somehow to what someone ought to do, in order to have genuine normative force. In epistemology we are interested in norms of belief. Indeed, if norms of action are the primary concern of ethics, then maybe norms of belief should be the primary concern of epistemology. But believing B Matthew Chrisman matthew.chrisman@ed.ac.uk is not a special kind of acting, so thinking of normative epistemology as the "ethics of belief" is problematic. Belief is a paradigmatic mental state; believing that p is a way of being, not something one counts as actively doing. So there is a tension at the heart of normative epistemology between the idea that norms are primarily prescriptive and the idea that epistemology should be primarily concerned with norms of belief.
There are several ways to resolve this tension. We could reject the first thought leading to it, arguing that not all norms are prescriptive, not even indirectly or implicitly so. 1 Or we could reject the second thought leading to the tension, arguing that believing is indeed a kind of doing rather than being. Less revisionary than both of these options, however, would be to argue that norms of belief are prescriptive but only indirectly or implicitly so. In saying that someone ought to believe that p, perhaps we are implicitly committing to norms governing something like the performance of forming a belief. That is something which is both active and whose normative status is clearly relevant to epistemology.
There is something attractive about this thought, but switching the proposed target of normative epistemic evaluation from belief to belief formation is surely too simplistic. When we think about people and what they ought to believe, of course we may sometimes be concerned (implicitly anyway) with what beliefs they ought to form, i.e. whether they ought to go from unopinionated about whether p to believing that p or believing that notp. But very often, we are concerned with something more complex than the formation of a belief. We are concerned with what someone ought to do cognitively given that they have various beliefs, memories, experiences, suspicions, hunches, etc. all of which stand in various relations of epistemological support and tension with one another. Perhaps one believes p but has an experience as if notp and one remembers that q, which one suspects makes p more likely than not, but that is not why one originally formed the belief that p. What should this person do cognitively? In this and many similar cases, if we say "She ought to believe that p," we do not seem to be making a claim about what belief she ought to form; she already has the belief that p.
In reality our doxastic lives are even more complex. The key point here, however, is not the psychology of our cognitive lives but that the norms governing these lives, assuming there are such norms, would seem to be much more about what we do here and now with the beliefs we have-maintain them, excise them, adjust our views about their epistemological relations of support and tension with other beliefs, memories, experiences, suspicions, hunches, etc.-than about the formation of those beliefs. 2 So, although the etiology of a belief may be relevant to its normative status, epistemic normativity seems to be at least as much if not more about here-and-now doxastic agency: what we should do with our various beliefs given that we have them, alongside myriad 1 Feldman (2001) defends the idea that epistemic 'ought's are "role-oughts" describing what counts as fulfilling some role well (though not necessarily prescribing one to try to do this). Kornblith (2001) criticizes this view but defends a related idea that epistemic 'ought's outline ideals, which need not be prescribed. These are two ways of working out the idea that epistemic 'ought's are evaluative and not prescriptive. 2 Boyle (2011, p. 6) makes a similar claim against those who would locate doxastic agency in the various processes that lead up to belief. This is also related to Moran (2001) influential discussion of the 'deliberative perspective' where, in considering what one believes and why, one rarely reflects on what causal-historical process led one to form a belief. experiences, memories, suspicions, hunches, etc. If this is right, then the attractively less revisionary way of resolving the tension at the heart of normative epistemology mentioned above would seem to leave out a lot of what is important about our doxastic lives and their governance by epistemological norms.
In the end, I suspect this issue of here-and-now agency should be viewed as a challenge to develop a more sophisticated and plausible account of the cognitive doings connected to believing rather than as the source of a deep objection to the less revisionary way of addressing the tension between thinking of norms as prescriptive and thinking of epistemology as focused on norms of belief. 3 The task of this short paper is not to address this challenge in any more detail 4 but rather to develop a line of criticism of a prominent account of epistemic normativity that I think obscures the challenge in an unfortunate way. This account is the treatment of epistemic normativity as a species of "performance normativity," worked out in such exquisite and seductive detail in many of Sosa's recent papers and books. 5 I believe Sosa is mistaken to conceive of epistemic normativity as a species of performance normativity, as this leads him to embrace the mistaken idea that belief itself is a performance rather than a state. I'll briefly explain this worry below. However, there is a natural and simple rejoinder on which I want to focus here: shift focus from belief itself to the formation of belief, which is surely more performance-like. I think there is something right about this move, but below I will argue that resting our account of epistemic normativity on the performance norms that apply to belief-formation obscures the here-and-now agency that I suggested above is so central to our doxastic lives (and so a proper centerpiece of any account of epistemic normativity).

Against performance normativity applied to belief itself
Sosa puts the core of his view about epistemic normativity as performance normativity writing, Belief is a kind of performance, which attains one level of success if it is true (or accurate), a second level if it is competent (or adroit), and a third if its truth manifests the believer's competence (i.e., if it is apt). Knowledge on one level (the animal level) is apt belief. The epistemic normativity constitutive of such knowledge is thus a kind of performance normativity (2011, p. 2).
And famously he uses the analogy of archery to illustrate the idea The archer's shot is a good example. The shot aims to hit the target, and its success can be judged by whether it does so or not, by its accuracy. However accurate it may be, there is a further dimension of evaluation: namely, how skillful a shot it is, how much skill it manifests, how adroit it is. A shot might hit the bull's-eye, however, and might even manifest great skill, while failing utterly, as a shot, on a further dimension. Consider a shot diverted by a gust of wind initially, so that it would miss the target altogether but for a second gust that puts it back on track to hit the bull's-eye. This shot is both accurate and adroit, yet it is not accurate because adroit, so as to manifest the archer's skill and competence. It thus fails on a third dimension of evaluation, besides those of accuracy and adroitness: it fails to be apt (2011, p. 5).
The problem with this is in its assumptions about the metaphysics of mind. Belief is a mental state and states are non-dynamic and atelic. That is, they don't involve change, and they don't involve direction towards some end. Archery shots are by contrast paradigmatic performances, being both dynamic and telic in involving some change and being directed at some end. How do we know this? Well, insofar as language is any clue, we can note that the verb 'believe' is a state verb, passing various semantic tests for being non-dynamic, whereas the verb phrase 'shoot an arrow' is a performance verb, passing various semantic tests for being dynamic and telic. Moreover-now kicking away the ladder of language-this doesn't appear to be some quirk of English that is more or less irrelevant to the underlying phenomena crucial to philosophical theory. There is a rich and productive philosophical tradition of treating beliefs as paradigmatic mental states, contrasting them with mental actions such as considering and inferring; this provides a powerful tool for explaining the stability of our minds over time and (in conjunction with desires also conceived as mental states) predicting and rationalizing agents' dispositions to behave in various ways. 6 Sosa's response to this is to suggest that he is just using the term 'performance' in a very loose sense. He writes. "…in our sense a 'performance' is just any state or action or process that has a constitutive aim" (Sosa 2015, p. 67). But how could a belief conceived of as state, which by definition involves no change or telos, have a "constitutive aim" and so therefore also be a "performance" in this loose sense? 6 I have made this argument in much more detail in Chrisman (2012Chrisman ( , 2016. It would take too much space to recapitulate all of the arguments of those papers here. But the basic thought is simple: performance normative evaluations (especially Sosa's notions of adroitness and aptness) apply to things in virtue of their direction at some end. Beliefs are states and so in virtue of their metaphysics not in themselves directed at some end. So it is wrong to think that performance normative evaluations apply directly to beliefs themselves. Or even more simply: belief is not a performance, so if epistemic norms are primarily norms of belief they are not performance norms. The goal of this paper is not to relitigate those arguments but to consider one of the most natural responses to it on Sosa's behalf, a response which has broader relevance to our understanding of epistemic normativity.
Sosa's idea of course is to appeal to the venerable and vexed metaphor of belief aiming at the truth. A lot of ink has been spilled trying to cash out this metaphor, and I'll try not to spill too much more here, except to grant that there is considerable explanatory value in thinking of belief as constitutively governed by some standard of correctness. Something of the form: a mental state doesn't count as a belief unless it is such that it is correct iff X . And it's quite plausible to think X = true. Because of this, it's plausible to think that one doesn't count as forming a belief unless this formation is somehow constrained or governed by a standard of correctness for belief involving truth. 7 Is that enough to warrant thinking of belief itself as a performance? I think not. We can of course stipulate a technical use of the word 'performance' to cover anything subject to a constitutive standard of correctness. But that does not warrant applying categories like adroitly done and apt to it. For the technical stipulation cannot change the fact that believing itself is not a doing, i.e. something dynamic. It is a way of being, i.e. something stative. So, trivially, believing itself is not something one can do with or without competence (i.e. adroitly), which of course means that believing itself cannot involve a change from merely aiming at the target to achieving the target, perhaps aptly. Yet this is precisely the assumption in the metaphysics of mind that Sosa needs to make sense of an account of epistemic normativity as a species of performance normativity. Compare: the filled-in crossword-i.e. the ink on the page-might be said to be correct iff it matches the canonical solution, and that's part of what constitutes it as an answer to a crossword puzzle. However, that doesn't make the ink on the page a performance in any intuitive sense. It's the act of filling in the crossword-i.e. the putting the ink on the page-which is the relevant performance. This is what we evaluate in terms of concepts such as adroitly done and apt. Similarly, belief itself cannot be adroitly done or an apt performance.
Maybe our minds are usefully conceived of on the model of intellectual archers shooting for the truth, but it involves a category mistake to think of states of belief themselves as the shots. 8

A problem with the refined performance view
As mentioned above, the obvious response to my objection is to grant that beliefs themselves are states but to insist that they are states that we acquire or form in some way. 9 And belief formation is surely dynamic and telic. It involves a change from an unopinionated state to an opinionated state, and this change is directed at some telos. In short, someone sympathetic to Sosa might respond to my argument so far by suggesting that, even if belief itself is not a performance, belief formation surely is a performance. And this means that belief formation is properly evaluated with categories such as "adroitly done" and "apt." 10 Exactly right. We might use the term Refined Performance View to refer to the idea that epistemic normativity is indeed a species of performance normativity but one attaching to the performance of belief formation rather than belief itself. To be clear, I don't think this is Sosa's view, but an improvement on it, albeit one for which I'm going to attempt to raise problems. As dynamic doings, performances of belief formation could be construed as the proper target of normative epistemology. With this view, we also move epistemic normativity more clearly back into the realm of the prescriptive. Epistemic 'ought's are again about what someone ought to do (form beliefs) rather than how they ought to be.
Here is where the idea of here-and-now doxastic agency is relevant. The idea we're trying to make sense of is that epistemic normativity is a species of performance normativity. That is to say, when we normatively evaluate beliefs-e.g. as part of evaluating whether someone knows or understands something, or as part of outlining the aims of reflection and inquiry-we are evaluating performances. Which performances? Sosa claimed it is beliefs themselves which are the performances, but beliefs aren't performances, they're states. According to the Refined Performance view the focus should instead be on performances of belief formation.
Like I said above, I think this is sometimes right. When we're evaluating a piece of explicit deliberation about a topic on which the agent doesn't already have a belief, it is natural to assume that the question of what the agent ought here-and-now to believe is essentially the same as the question of what belief the agent ought here-and-now to form. For these cases the Refined Performance View seems right, and that's surely part of its seductiveness as a general view of epistemic normativity.
Most of our beliefs, however, are ones that were formed long ago or via complex cognitive processes well beyond our ken. Nevertheless, we are quite happy to subject those beliefs to normative evaluation. For example, I believe my parents were born in Eastern Kentucky. When and how did I form that belief? Did I do so adroitly? I have no idea, and neither do you. And I don't think the etiology of this belief is really important to the questions of whether I should now hold it, or whether I am rational in doing so, or how my credences in other propositions should interact with this belief.
Footnote 9 continued and believing. Maybe judgment is a way of forming a belief, or maybe it's the event which begins an episode of having a belief. If the former, some of what I say below is relevant, but if the latter, the situation is less clear. These are complex issues that would take me too far afield of main line of argument I want to develop here. The idea of focusing epistemic norms on judgments rather than beliefs criticized in Boyle (2011) and discussed further in Chrisman (2016).
After all, so much has happened since I formed that belief; I have acquired lots of new evidence. I have forgotten experiences that I had immediately prior to forming the belief. And all of this is potentially relevant to the questions of whether I am believing as I should when I believe this, and whether I am maintaining my system of beliefs more generally as I should in connection to this belief. How I originally formed the belief seems much less important to these questions.
The objection here is not that Sosa or a defender of the Refined Performance View is missing details about how the psychology of belief formation and maintenance works. The objection is that belief itself is not a performance, and the performance of forming a belief is not always or even usually centrally relevant to the assessment of whether one ought to have that belief.
If the goal of appealing to performance normativity is to explain the whole or even the central part of the structure of epistemic normativity, then the Refined Performance view fails to move far enough from Sosa's actual view. 11 Since belief states stick around, interacting in myriad ways with everything else going on in our minds, the performance involved in their formation is often going to be more-or-less irrelevant to assessing whether they are beliefs we ought to have. For this reason, I think the etiology of beliefs (how well or poorly they have been formed) will often (but not always) be irrelevant for assessing whether they count as instances of knowledge. So, although we can and do evaluate performances of belief formation for accuracy, adroitness, and aptness, that appears to be a special case of epistemic normativity rather than a reflection of the general structure of epistemic normativity.
What about all of the other cognitive performances that seem to be possible objects of epistemic normative evaluation-revising beliefs, reconsidering evidence, adjusting credences, refining one's appreciation of the inferential relations between beliefs, etc.-aren't these now viable candidates for an account of epistemic normativity as a species of performance normativity? Yes and no. Yes these are performances which we can evaluate with epistemic norms, and these norms might be usefully conceived as a species of performance norms. However, no, that still won't mean that epistemic normativity is a species of performance normativity. For there are other cognitive things we do that aren't performances because they aren't directed at some end; cognitive activities such as imagining, considering, maintaining, etc. are things we do but not performances in the sense of having an end pursuit of which could be adroit or not. Moreover, there is a difficult programmatic question about what makes these various cognitive doings epistemically relevant; and my favored view is that epistemic norms are fundamentally about what makes the difference between beliefs that are knowledge and beliefs that are not. I won't argue for this here, but if it is right, then Sosa was right to focus on belief itself in thinking about epistemic normativity. It's just that, since belief itself is not plausibly thought to be dynamic and telic, it is not plausible to think that the central epistemic norms applying to it are ones characteristic of the structure of performance normativity.

Conclusion
At the beginning of this paper, I highlighted a tension facing epistemologists thinking about epistemic normativity. On the one hand, norms are supposed to be prescriptive, to have something to do with things agents can actively do. On the other hand, epistemology is supposed to be about norms of belief; and beliefs are not standardly or plausibly thought to be things agents can actively do (but rather ways one might be). I suspect one of the attractions of Sosa's idea that epistemic normativity is a species of performance normativity is that beliefs conceived of as performances, where this is construed widely to include anything with a constitutive aim (including a state like belief), straddle the divide between things one can actively do and ways one might be. Beliefs needn't be actions, but as long as we still think of them as things that can achieve their aims and do so in better and worse ways, we'll be on the way to making sense of the way epistemic normativity is both prescriptive and primarily about doxastic states.
However, it is wrong to think of beliefs themselves as something actively achieving or failing to achieve an aim. They are the characteristic product of a kind of performance (belief formation), which does indeed have an aim. As such beliefs may sometimes be the result of an apt performance. But the importance of here-and-now doxastic agency to normative evaluations of our cognitive lives belies focus on these performances of belief formation as the whole of or even a central part of the target of normative epistemology.
Another way to put the argument of this paper is in terms of a dilemma faced by those who would like to appeal to the categories of performance normativity to explain the structure of epistemic normativity. On the one hand, if they focus on belief itself, then this isn't a performance and so the norms applying to it are not plausibly construed as species of norms applying to performances. On the other hand, if they focus on the performance of belief formation, then although this is certainly a performance relevant to epistemology it is not all or even a central part of the sorts of norms that apply to here-and now doxastic agency. Either way, epistemic normativity doesn't look like it can be entirely or even mainly a species of performance normativity.
So what about the tension I started with between thinking of norms as prescriptive and thinking of epistemic norms as applying to doxastic states? I think this should be addressed not by construing epistemic normativity as a species of performance normativity but rather by better investigating the indirect relation between the normative evaluation of beliefs and the various things we do as part of the activity of belief-system maintenance. 12 Activities, such as maintaining a marriage or sustaining a healthy diet, are characteristically active but atelic. So, although it won't make sense to evaluate them for aptness, it may still makes sense to think they can be done more or less well. It is here that I would look to develop alternative categories, alongside the categories of performance normativity, for understanding epistemic normativity. 13 Open Access This article is distributed under the terms of the Creative Commons Attribution 4.0 International License (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made.