In “Mind-World Relations,” Ernie Sosa proposes a performance-theoretic account of three mind-world relations: action, perception, and knowledge (Sosa Reference Sosa2015). All three are species of success that is caused, in the right way, by competence – but what is it for success to be caused “in the right way” by competence? Sosa's answer is that success is caused “in the right way” by competence when success manifests competence. In each case we are able to distinguish the good case – e.g. in which you intentionally startle your boss, perceive the sun, or know that the cat is on the mat – from the bad case – e.g. in which you drop your tray and unintentionally startle your boss, enjoy a visual experience caused by a torch without perceiving, or have a justified true belief that the man who will get the job has ten coins in his pocket without knowing that the man who will get the job has ten coins in his pocket – by appeal to the fact that in the good case, but not in the bad case, competence is manifested.

Can we go further? Can we answer the question of what it is for competence to be manifested? A competence is a disposition to succeed, just as fragility is a disposition to break. But just as not all breakings because of fragility are manifestations of fragility, not all successes because of competence are manifestations of competence. Consider a normal case of fragility manifested: a fragile glass shatters upon impact with the floor. But compare an abnormal case of fragility not manifested: a fragile glass shatters upon impact with the floor because of the intervention of a fragility-hating zapper, who causes the glass to shatter just as it hits the floor. What distinguishes cases of manifestation from cases of non-manifestation? Sosa's answer is: unarticulated (and perhaps ineffable) community convention. Just as communities agree to conventions that draw a boundary between polite behavior and impolite behavior, by positing standards of etiquette, they agree to conventions that draw a boundary between cases of manifestation and cases of non-manifestation, by positing standards of normality. Fragility is manifested when a glass disposed to shatter in normal conditions shatters in those conditions because it is so disposed; fragility is not manifested when a glass disposed to shatter in normal conditions shatters in abnormal conditions because it is so disposed. And knowledge of what is normal and what is abnormal is just “part of the instrumentally determined commonsense that humans live by.”Footnote ¹

The idea that knowledge is (in some sense) normative is a persistent theme in contemporary epistemology. When we broaden our inquiry to include mind-world relations in general, we are reminded of the ideas of the normativity of content, of the normativity of psychology, of the normativity of propositional attitude attributions, of the normativity of the mental, and of related themes. Here I should like to ask: does Sosa's account imply that action, perception, and knowledge are normative? And in what sense?

Two ideas concerning knowledge can be bracketed. First, you might think that knowledge is normative because it is a norm: for example, you might think that a doctor ought not prescribe medicine unless she knows that it is safe. This provides a sense in which knowledge is normative – we might say that knowledge is “normative for prescribing medicine.” But Sosa's account is orthogonal to whether knowledge is a norm. Second, we can set aside any normativity arising exclusively from the factivity of knowledge. If you were wondering whether to believe that p, or to stop inquiring about whether p, or to treat the proposition that p as a reason for action, you would be enlightened to find out that someone knows that p, because this entails that p. But this seems true of factive states in general. Sosa's account straightforwardly implies the factivity of knowledge. But we can set this point aside.

In her influential discussion of normativity, Christine Korsgaard writes that certain concepts, including the concept of knowledge, “have a normative dimension” because “they tell us what to think, what to like, what to say, what to do, and what to be” (Reference Korsgaard1996: 9). Following Sosa's focus on metaphysical, rather than semantic or conceptual, analysis, we can say that some non-linguistic and non-conceptual thing is normative just in case it tells us what to think, what to like, what to say, what to do, or what to be. What could this mean? Consider suffering. Many would argue that suffering per se is bad, where this entails that we always have pro tanto reason to prevent suffering, or that we always ought to prevent suffering, other things being equal, or that we always ought to dislike suffering. This is just a rough sketch of the kinds of things that badness entails, but the idea that suffering per se is bad would provide a sense in which suffering is normative. Now consider knowledge. Many would argue that knowledge per se is good, where this entails that we always have pro tanto reason to produce knowledge, or that we always ought to produce knowledge, other things being equal, or that we always ought to like knowledge. Again, this is just a rough sketch of the kinds of things that goodness entails, but the idea that knowledge per se is good would provide a sense in which knowledge is normative. As well, the idea that action per se is good would provide a sense in which action is normative, and the idea that perception per se is good would provide a sense in which perception is normative.

However, given Sosa's account, neither action nor perception nor knowledge is plausibly understood as per se good. Action, on Sosa's view, is “apt intention” – someone acts when her Φing manifests her competence to Φ. But action, so understood, is not per se good – the goodness of actions depends on the content of their constitutive intentions. There is nothing good about my apt intention to own a saucer of mud, i.e. my acquisition of a saucer of mud, unless there is something good about my owning a saucer of mud. Actions aren't good just in virtue of being actions. And the goodness of perception is likewise conditional. Perception, on Sosa's view, is “apt perceptual experience,” where this “involves functional, teleological aimings, through the teleology of our perceptual systems.” Our perceptual systems achieve their aim when they function properly. (In this case proper functioning is a matter of natural history: for our perceptual systems to function properly is for our perceptual systems to do that which they evolved to do.) But proper functioning isn't per se good: nutritious meals are sometimes nasty; reproductive sex is sometimes no fun. So perceptions aren't good just in virtue of being perceptions.

What about knowledge? Here Sosa offers a disjunctive account. One sort of knowledge is “teleological” and “perception-like” and is the fulfillment of a teleological aim – but this sort of knowledge isn't plausibly understood as per se good, for the same reason that perception isn't plausibly understood as per se good. The other sort of knowledge is “judgmental” and “action-like” and is the fulfillment of an intention – but this sort of knowledge isn't plausibly understood as per se good, for the same reason that action isn't plausibly understood as per se good.Footnote ²

I think it is important that this conclusion – that action, perception, and knowledge aren't plausibly understood as per se good – is consistent with the idea that we have good reason to employ the concepts of action, perception, and knowledge. The boundary between cases of competence manifested (as in the good cases of action, perception, and knowledge) and cases of mere causation by competence (as in the corresponding bad cases) is down to convention, but Sosa suggests that we have good reason to draw this boundary where we do. The difference between competence manifested and mere causation by competence makes a difference when it comes to credit and discredit, praise and blame, approval and disapproval, trust and distrust, and this “has a large bearing on human flourishing, individually and collectively.” But a valuable distinction is not necessarily, or even typically, a distinction in value. We do well to distinguish hawks from handsaws, but neither hawks nor handsaws are therefore valuable, nor are hawks therefore better than handsaws (or handsaws therefore better than hawks). The utility of the distinction between competence manifested and mere causation by competence thus tells us nothing about the value of action, perception, and knowledge. Importantly, it does not tell us that knowledge is better than mere true belief, just as the utility of the distinction between shatterings that manifest fragility and shatterings that are merely caused by fragility does not tell us that the former are better than the latter.Footnote ³

Given Sosa's account, neither action nor perception nor knowledge is plausibly understood as per se good. Arguing that action, perception, and knowledge are per se good would have been a way to vindicate the idea that action, perception, and knowledge are normative. But Sosa's account does not jibe with such an argument.

We have focused on the normativity of three non-linguistic and non-conceptual mind-world relations: action, perception, and knowledge. What about the normativity of concepts (e.g. the concepts of action, perception, and knowledge) and the normativity of bits of thought and language (e.g. attributions of action, perception, and knowledge)? Consider knowledge attributions. Sosa compares the boundary between knowledge and non-knowledge to the boundary between polite and impolite behavior. To attribute politeness to some particular behavior is often (or perhaps always) to express endorsement, approval, praise, or some other such “pro-attitude” towards said behavior; this provides a sense in which politeness attributions are normative. Perhaps the same is true of knowledge attributions – perhaps they are normative in the same sense.Footnote ⁴ But this is neither entailed nor explained by Sosa's metaphysical account. Attributions of knowledge, for Sosa, are attributions of a particular kind of manifestation; attributions of manifestation, in general, are not normative (in the present sense). And it does not seem that action attributions, for example, are normative (in the present sense): just to call something an action is not to endorse or praise it. The unity of action, perception, and knowledge does not suggest the normativity of knowledge attributions.

Recall the persistence of the theme of the normativity of knowledge in contemporary epistemology. Imagine that a particular metaphysical analysis of knowledge entails that knowledge is not normative. (I have not argued that Sosa's account is an account of this kind, but I do suspect that it is.) The idea that knowledge is normative is a kind of truism in contemporary epistemology. You might think that any metaphysical analysis of knowledge is flawed if it fails to reveal a nature that entails and explains the normativity of knowledge – and so our imagined analysis is therefore flawed.

Truisms come and go. We should consider their rise and fall carefully through both idealist and materialist historical analysis. Questions of value were infrequently discussed in 20th century Anglophone epistemology. In this connection, two events warrant mention. First, consider the marginalization of ethics and aesthetics in the middle part of the century, arising, on the one hand, from the positivistic rejection of normative language as meaningless and, on the other, from the dangers of appearing political in a nervous Cold War climate. All this predicts for the survival of a distinctly non-normative discipline of epistemology. Second, consider the rapid growth of universities throughout the century, with the subsequent economic need for increased academic specialization, and the subsequent division of philosophers into “epistemologists,” “ethicists,” “metaphysicians,” etc. All this predicts for the need to develop the idea of “the epistemic” as a category distinct from the ethical and the aesthetic.

It may be useful here to ask why we may find ourselves unsatisfied with an analysis of action, perception, and knowledge that appeals to causation “in the right way,” and gives no further account of this. Why, in other words, do we think it a virtue to be able to give an account of causation “in the right way”? One reason, surely, is that we fear our appeal to “the right way” may mean that we have given no account at all of action, perception, and knowledge. When we say that action, perception, and knowledge are caused “in the right way,” we may just mean that they are caused in the ways that respectively cause action, perception, and knowledge. However, another reason may be that we detect the specter of unexplained normativity in the phrase “in the right way” – spectral, in virtue of our commitment to a philosophical naturalism that seeks to explain normativity in non-normative terms. From this perspective, you might think that it is an asset for any metaphysical analysis of knowledge that it is silent on the normativity of knowledge.

Reliabilism in the theory of knowledge was first motivated, for some, by the kind of naturalism just described.Footnote ⁵ Reliabilists aimed to analyze knowledge without appeal to such terms as “justification,” “evidence,” and “the right to be sure.” Success would have meant an analysis of knowledge in non-epistemic and non-normative terms, perhaps even in scientific terms. We might see Sosa's present account as an attempt to bridge the gap between the early reliabilist analyses – on which knowledge is obviously not normative – and the contemporary truism that knowledge is normative. It seems to me that this gap has yet to be bridged.

There is a sense of “normative” that we have not yet discussed. Consider Arnon Goldfinger. We might say that gold is normative for Goldfinger – it is what he cares about, and thus what animates and guides him in deciding what to do and how to live. This is all just another way of saying that Goldfinger loves gold. Now, given this sense of “normative,” we could say that knowledge is normative for us. The plausibility of this claim will depend, of course, on who we mean by “us.” With the scope of “us” suitably restricted – whether to epistemologists, philosophers, the curious, or the intellectually virtuous – it is plausible that knowledge is normative for us. Knowledge is what we care about, and thus what animates and guides us in deciding what to do and how to live. And this is just to say that we love knowledge. However, to say that something is normative for someone is just to describe how she regulates her conduct; it is not to prescribe anything to her. So the present sense of “normative” is fundamentally different from Korsgaard's sense (above). But, on the present sense of “normative,” it is perfectly compatible with Sosa's account that knowledge is normative for us – where this is just to say that we love knowledge. And we might say the same of action, perception, and other mind-world relations.