Wittgenstein on the Constitutive Uncertainty of the Mental

The idea that our recognition of others’ mental states is beset, not only by contingent but constitutional uncertainty is one to which Wittgenstein returns throughout his later work. And yet it remains an underexplored component of that work. The primary aim of this paper is to better understand what Wittgenstein means when he describes the mental as constitutively uncertain, and his conception of the kind of knowledge of others' mental lives consistent with it. The secondary aim is to connect Wittgenstein’s discussion of the constitutive uncertainty of the mental with two further components of his later thought – specifically, his remarks on aspect perception and on the pattern-like nature of the emotions.


Introduction
In RPP II, §657 Wittgenstein imagines an interlocutor insisting "But you can't recognise pain with certainty just from externals", to which Wittgenstein replies: The only way of recognising it is by externals, and the uncertainty is constitutional. It is not a shortcoming. (RPP II: §657) The idea that our recognition of others' mental states is beset, not only by contingent, but constitutive uncertainty is one to which Wittgenstein returns throughout his later work. And yet it remains an underexplored component of that work. 1 As a consequence, important questions regarding the phenomenon continue to press. What does Wittgenstein mean when he talks about a constitutional uncertainty in relation to mental states like pain? Does Wittgenstein conceive of the phenomenon as having merely epistemic import, or does it also have metaphysical implications? In those contexts in which there is constitutional uncertainty, might we nevertheless know what another person is thinking or feeling? If so, what is the character of such knowledge?
The primary aim of this paper is to answer these questions. The secondary aim is to connect Wittgenstein's discussion of the constitutive uncertainty of the mental with two further components of his later thought -specifically, his remarks on aspect perception and on the pattern-like nature of the emotions. The former aim is pursued in sections 2, 3 and 4, the latter in section 5.

Characterising constitutive uncertainty
We can distinguish at least two strands to Wittgenstein's views about the constitutive uncertainty of the mental, the first epistemic and the second metaphysical. The epistemic strand concerns an indeterminacy in the rules of evidence on the basis of which we sometimes attribute mental states to others. The metaphysical strand concerns an indeterminacy in the mental states themselves. In this section, I discuss these strands in turn.
In LW 1, Wittgenstein describes the uncertainty of whether someone else is in pain as "an (essential) trait" of the language game ( §877). In RPP I, he says: "The uncertainty of the ascription "He's got a pain" might be called a constitutional certainty" ( §141). And in Z §555, He says that our uncertainty about what another person is feeling "relates not to the particular case, but to the method, to the rules of evidence". In these passages, Wittgenstein's point appears to be primarily epistemic. His point, it seems, is that the rules of evidence on the basis of which we sometimes ascribe mental states to others contain some essential or ineliminable indeterminacy -that it is a constitutive feature of the rules of evidence for, say, irritation that in particular cases (though not necessarily in every case 2 ) they allow for uncertainty as to whether another person is irritated or, say, excited. 3 This feature is constitutive insofar as our uncertainty is to be explained in terms of the rules themselves, and not, for instance, in terms of our lacking evidence. Even if we were to take into consideration all the possible evidence for a person's being irritated, the rules of evidence may still fail to settle the question in particular cases. (To ease discussion, I refer to such contexts hereafter as epistemic indeterminacy contexts.) That this question would remain unsettled even in the face of all possible evidence is significant; it explains Wittgenstein's remark in RPP II, §657 that the uncertainty "is not a shortcoming". Wittgenstein's point is that there is a kind of uncertainty which attaches to third-person ascriptions of mental states which is not ultimately traceable to a deficiency in one's evidence, but, rather, to some indeterminacy in the rules of evidence for the relevant states. We can readily imagine a scenario in which my uncertainty about whether to describe someone else as, say, irritated or excited is due to a lack of evidence. Perhaps if I had seen the person's behaviour, as opposed to merely having had it described to me, I wouldn't be responding concerns our capacity to recognise pain in others quite generally, not the exercise of that capacity in a particular case. Read, however, in the context of remarks like Z, §556, PPF, xi, §330, PPF, xi, §353 and Z, §374, it is clear that this reading is mistaken. As these passages make clear, Wittgenstein's view is that we can be certain that another person is in pain, as well as about others' mental states considered generally. The point of RPP II, §657 is merely to draw our attention to a kind of uncertainty about others' pain which is constitutional in the sense outlined below. 3 In her 2007, Moyal-Sharrock argues that in fact, by the close of LW II, "Wittgenstein is no longer saying that uncertainty is a constitutional or essential trait of our psychological ascriptions" (p. 225). Part of Moyal-Sharrock's argument turns on the claim that Wittgenstein's considered view is that it can be objectively certain that another person is in a particular mental state -that third-person ascriptions of mental states can be certain in just the same way and to just the same extent as, for instance, the proposition that I am a human being or that 2 + 2 = 4. (For the distinction between subjective and objective certainty, see OC: §194.) I'm not convinced that Moyal-Sharrock is right on this exegetical point. But the more fundamental problem, it seems to me, is that the argument ignores the fact that constitutive uncertainty is a feature of the rules of evidence, and not of particular contexts. There is no reason why it cannot be true both that in a particular context it is objectively certain that another person is in a given mental state and that the rules of evidence for that state are constitutively uncertain. Even if it is objectively certain in this context, there may yet be some other context in which the rules of evidence fail to decide the question whether a person is in the relevant state and would do so even if we were to take all the possible evidence for their being in that state into consideration. uncertain. But the idea that all uncertainty about what other people are thinking, feeling or sensing is ultimately traceable to some poverty in one's evidence, Wittgenstein's thinks, is misguided. Sometimes our uncertainty is not a shortcoming; it is a consequence of an ineliminable feature of the rules of evidence themselves.
When Wittgenstein talks about rules of evidence for a psychological state, he has in mind rules like the following: (1) Rubbing one's cheek while groaning is evidence of toothache; (2) The fact that one has told the truth in the past about having some intention is evidence that one is telling the truth in the present about having that intention; (3) Swelling around the ankle is evidence of pain in the ankle; (1), (2) and (3) all correlate some occurrence or condition with a psychological state by asserting that the former is evidence for the latter. 4 (1) correlates the target state with present conscious behaviour, (2) with past conscious behaviour and (3) with the present condition of (a part of) one's body. I take it that rules of the first kind are of primary significance to Wittgenstein's discussion, just because they are most directly connected with the possibility of pretence; when one pretends it is typically by means of one's present behaviour.
In what sorts of contexts might rules of evidence thus understood give rise to uncertainty which is not traceable to a poverty in one's evidence? The following two cases are examples of such contexts: Someone gives expression to their pain. I am convinced that they feel the sensation as strongly as their expression suggests. You, meanwhile, are convinced that they are hamming it up. Even in the face of all the possible evidence, we are unable to resolve our disagreement.
Someone who has a habit of lying about their intentions declares an intention to ф. Even in the face of all the possible evidence, the rules of evidence fail to settle the question whether their declaration is sincere.
There are, of course, differences between the two cases. The first, but not the second, involves disagreement about the psychological state of the target individual. But this difference is superficial. We can readily adjust the second case so that it too involves disagreement: I am convinced that this person really does have an intention to ф, while you are convinced that they do not. Third-party disagreement is not a necessary or constitutive feature of epistemic indeterminacy contexts. It is conceivable that the rules of evidence may fail in a particular context to decide the question whether a person is in a given state despite there being no extant disagreement in that context about whether they are in that state; there may be general uncertainty. But typically, epistemic indeterminacy contexts in which there is no extant disagreement can readily be modified, consistent with their remaining epistemic indeterminacy contexts, so that there is disagreement. I say typically because there are some epistemic indeterminacy contexts in which disagreement is not easy to imagine. Consider the following example: Someone groans while under anaesthetic or in their sleep. We are asked "Is this person in pain?" But the evidence in insufficiently determinate to give an answer one way or the other (cf. LW II: p. 57).
We may find it difficult to imagine one's being certain either way in a case like this one.
There is further difference between this third case and the first two. In the first two cases, the explanation for our disagreement or uncertainty is the presence of competing evidence -evidence for this person's hamming it up or for that person's having an intention to ф. Compare the third case. Our uncertainty in this case does not seem to be explained in terms of our having competing evidence. Rather, we are uncertain, it seems, because the scenario is just so unlike the usual one. Because it's so unusual, behaviour which would otherwise qualify unambiguously as evidence that the person manifesting the behaviour is in pain -namely, groaning -is rendered ambiguous: it is not clear whether it is evidence of pain or whether it isn't.
There are, then, at least two ways in which the rules of evidence might leave it indeterminate whether another person is in a particular mental state. First, there might be an irresolvable conflict between the rules of evidence for mutually exclusive psychological states, as in the first and second cases. Second, because of the unusual nature of the circumstances in which an instance of behaviour occurs, the rules might leave it unclear whether that behaviour constitutes evidence for a given state, as in the third case.
Thus far, our concern has been with epistemic indeterminacyindeterminacy in the rules of evidence for particular mental states. But there is a second, metaphysical strand to Wittgenstein's views about the constitutive uncertainty of the mental. We may find ourselves tempted by the view that in every case, there is some fact of the matter with respect to whether another person is in some mental state, whether or not we are in a position to know that fact. In other words, we might think that even though it may be epistemically indeterminate, the question whether another person is in some mental state is never metaphysically indeterminate.
The view that it is always metaphysically determinate whether another person is in a particular state is one which Wittgenstein rejects: [I]t is misleading to think of the real irritation as a facial expression of an inner face, so to speak, such that this facial expression is defined completely clearly, and that it is only the outer face that makes it uncertain whether the soul really has this expression. (LW II: p. 70) In some cases, Wittgenstein thinks, our uncertainty about whether another person is in a particular state is due at base, not to their behaviour, but to there simply being no fact of the matter as to whether they are in that state.

Epistemic and metaphysical indeterminacy
What is the relation between the epistemic and metaphysical strands in Wittgenstein's discussion? On one straightforward interpretation, epistemic and metaphysical indeterminacy are, for Wittgenstein, exactly co-occurrent, such that: (i) in any case in which it is epistemically indeterminate whether S is in M, it is metaphysically indeterminate whether S is in M; and (ii) in any case in which it is metaphysically indeterminate whether S is in M, it is epistemically indeterminate whether S is in M. I leave (ii) aside in the remainder of this paper. It's not implausible that Wittgenstein endorsed (ii), but I don't wish to defend it or its attribution to Wittgenstein. My focus is on (i). I'll defend the claim that Wittgenstein rejects (i) and that, consequently, epistemic and metaphysical indeterminacy are not, on Wittgenstein's view, exactly co-occurrent. The defence of this claim will reveal Wittgenstein's positive account of our knowledge of our minds in epistemic indeterminacy contexts.
The attribution of (i) to Wittgenstein may seem to be supported by certain of his remarks regarding the relation between an inner process or event and its outward expression. Consider, for instance: An "inner process" stands in need of outward criteria. (PI: §580) This remark is bound up with Wittgenstein's sustained critique of a particular conception of the relation between the inner -the realm of sensations, thoughts, intentions, and so on (LW 1: §956) -and the outer. The conception to which Wittgenstein stands opposed is one according to which the realm of the "inner" bears only causal relations to its expression. On this conception, the "inner" is essentially hidden behind a veil of behaviour, to which it is only contingently related. Third-person psychological ascriptions are inferences, from observable behaviour, to the "inner" processes, events and states which issue in that behaviour, in much the same way that one's judgments about the state of, say, a clock's inner mechanism might be inferences from its face. On Wittgenstein's positive conception of the relation, the inner bears constitutive relations to the outer. The relation between, say, a thought and its expression is not merely causal, but logical (LW II: p. 63) or conceptual (LW II: p. 62). Part of what it is to have a particular sensation, say, is to be disposed, under normal conditions, to express it in particular ways.
Reflecting on Wittgenstein's critique of the picture of the "inner" as essentially hidden, one might be inclined to conclude that epistemic indeterminacy entails metaphysical indeterminacy in just the way maintained by (i) above. One might be inclined to this conclusion by way of something like the following train of reasoning. If the rules of evidence for a particular psychological state are constitutively uncertain, then there are contexts in which those rules fail to decide the question whether another person is in that state, and would do so even in the face of all the possible evidence. But if the rules fail to decide the question in a given context, then in that context we cannot know whether this person is in that state. Now, if there was nevertheless some fact of the matter about whether this person was in the relevant state, then it would be a context in which that fact was hidden in precisely the sense that Wittgenstein finds objectionable -essentially hidden. But the moral of Wittgenstein's critique of the "inner"/outer picture is that there are no such contexts. Wittgenstein's view is not that the picture of the "inner" as essentially hidden is true of some appropriately circumscribed set of cases, but, rather, that it is a bad picture generally. So Wittgenstein's view must be that in those contexts where the rules of evidence leave it essentially indeterminate whether another is in a particular psychological state, there simply is no fact of the matter about whether they are in that state; there is no state which is essentially hidden in the offending sense.
It is clear, however, that we ought not to ascribe (i) to Wittgenstein on the basis of the line of reasoning just sketched. That reasoning rests on the assumption that in those contexts in which the rules of evidence fail to decide the question whether another person is in a particular psychological state, we cannot know that they are in that state. But this is an assumption which Wittgenstein seems to reject. It is true that Wittgenstein often characterises epistemic indeterminacy contexts as contexts in which one is uncertain about what it is that another person is thinking or feeling, or in which such uncertainty would not be unwarranted or out of place. But nothing that he says implies that we cannot know what another's psychological state is in such contexts. In fact, his view seems to be that we can know. 5 Towards the end of PPF, xi, there is a series of remarks focused on the difference between the kind of certainty which attaches to third-person ascriptions of mental states and the kind which attaches to basic mathematical judgments (see, for example, PPF, xi: § §330, 332, 341) and to judgments of colour (see, for example, PPF, xi: § §346, 351, 352). One salient difference, according to Wittgenstein, is that, while there is general agreement with respect to judgments of the second and third kind, "there is no such agreement" in the case of judgments of the first kind (PPF, xi: §352). In the following two remarks, Wittgenstein considers cases in which there is marked disagreement about what another person is thinking or feeling. In §353 he suggests that in some such cases we can nevertheless be certain about what another person is thinking or feeling: I am sure, sure, that he is not pretending; but some third person is not. Can I always convince him? And if not, is there some mistake in his reasoning or observations? (PPF, xi: §353; cf. Z: §374) (Wittgenstein would answer his two questions in the negative. Compare colour judgments.) And in §354 he seems to imply that in some such contexts one might know another's psychological condition: "You don't understand a thing!"-this is what one says when someone doubts what we recognise to be clearly genuine-but we cannot prove anything. (PPF, xi: §354, emphasis added) That Wittgenstein would be happy to talk of knowing in place of recognising here is suggested by the passage immediately following, §355, in which Wittgenstein not only speaks explicitly in terms of knowledge but tells us something about its character: it is acquired through "experience": Is there such a thing as "expert judgment" about the genuineness of expressions of feeling? -Here too, there are those with "better" and those with "worse" judgment.
In general, predictions arising from judgments of those with better knowledge of people will be more correct.
Can one learn this knowledge? Yes; some can learn it. Not, however, by taking a course of study in it, but through "experience". -Can someone else be a man's teacher in this? Certainly. From time to time he gives him the right tip.-This is what "learning" and "teaching" are like here.
-What one acquires here is not a technique; one learns correct judgments. There are also rules, but they do not form a system, and only experienced people can apply them rightly. Unlike calculating rules. (PPF,xi: §355) These three successive passages, then, yield three positive claims. In at least some cases involving marked disagreement: first, I can be certain that another person is in a particular mental state; second, I can recognise or know that they are in that state; and third, I can acquire this knowledge through "experience". 6 In addition to these positive claims, these three passages yield at least three negative claims about such cases. They are contexts in which: first, although we can be certain, we can't necessarily convince someone else; second, we cannot prove anything; and third, although the knowledge involves rules, they do not form a system and only experienced people can apply them correctly.
What is the relation between the sorts of cases which Wittgenstein is focusing on here -namely, those cases involving marked disagreement in which knowledge of others' mental states is nevertheless available -and the sorts of cases which have been our focus hitherto -namely, epistemic indeterminacy contexts? It would seem that the two classes of cases are not co-extensive, on Wittgenstein's view. We have already acknowledged epistemic indeterminacy contexts in which disagreement is not readily imaginable. (Recall the case involving the individual who groans while under anaesthetic or asleep.) Moreover, there surely are some cases in which it is both epistemically and metaphysically indeterminate whether S is in M. 7 Presumably, in epistemic indeterminacy contexts of this sort, the relevant knowledge is not available, for there is no relevant fact to know. Still, there is significant overlap between the two classes of cases on Wittgenstein's view. As earlier stated, although third-party disagreement is not a necessary or constitutive feature of epistemic indeterminacy contexts, in the majority of such contexts either disagreement will be extant or it can be readily imagined. In at least some of these contexts, it would seem that knowledge of others' mental states is nevertheless available to those who are suitably experienced.
Consider, for instance, the case, introduced above as an example of an epistemic indeterminacy context, in which I am convinced that this person feels the sensation as strongly as their expression suggests, while you are convinced that they are hamming it up. Here we have an example of an epistemic indeterminacy context involving marked disagreement. And yet we can readily imagine someone who knows this person well, or who has requisite experience discriminating cases in which the expression of a sensation is overplayed from cases in which it isn't, being in a position to immediately recognise that, indeed, in this case this person is hamming it up. In other words, this seems exactly the sort of case in which the capacity for expert judgment, which is Wittgenstein's concern in §355, might serve to reveal another's psychological condition.
This same point holds true of the second example of an epistemic indeterminacy context given above. Someone who has a habit of lying about their intentions declares an intention to ф. As noted earlier, we can readily modify the case, consistent with it's remaining an epistemic indeterminacy context, so that it involves marked disagreement. And, as before, we can imagine someone who, for instance, knows the individual in question particularly well, being in a position to immediately recognise this person's declaration as in fact sincere.
None of this is to deny the point, acknowledged above, that there are epistemic indeterminacy contexts which are not examples of the sorts of cases Wittgenstein is focused on in § §353-355. In what follows, however, my concern is with those epistemic indeterminacy contexts which are simultaneously examples of such cases. Hereafter, when I talk about epistemic indeterminacy contexts, I'm talking specifically about these contexts.
In the remainder of this section, I'll consider briefly the first and second of the three negative claims enumerated above. In the following section, I'll consider the third negative claim, as well as Wittgenstein's positive conception of the knowledge which is available in epistemic indeterminacy contexts to those who are suitably experienced. In section 5, I connect this positive conception with Wittgenstein discussion, also in PPF, xi, of aspect perception and his views about the pattern-like nature of the emotions.
For Wittgenstein, the integrity of those concepts which figure in third-person ascriptions of mental states depends on there being simple cases (LW I: §967), cases in which we agree in applying those concepts immediately, without hesitation. If someone screams after having fallen into a fire, for instance, we wouldn't hesitate in describing that person as being in pain: If I see someone writhing in pain with evident cause I do not think, all the same, his feelings are hidden from me. (LW II: §22) In such cases, there is widespread agreement about the psychological states of others. That there is such agreement is foundational for the concept. But the further we move beyond these simple cases, the less resounding the agreement. At the opposite end of the spectrum, there is marked disagreement. What's more, the disagreement is not always resolvable. Compare disagreement about, for instance, the result of a calculation -"[S]uch disputes are rare and of short duration. They can be decided, as we say, 'with certainty'" (PPF, xi: §341)) -or about colour -"There is, in general, complete agreement in the colour statements of those who have been diagnosed normal. This characterises the concept of a colour statement" (PPF, xi: §352)). In both the calculating and colour cases, incorrigible dissent is typically regarded as a sign of irrationality, inattention or incomplete grasp of the concepts involved. Not so in the case of judgments about others' psychological states. In complex cases, one person may be certain while another is uncertain. Their inability to persuade the other does not necessarily indicate some failure of understanding, inattention or irrationality: Given the same evidence, one person can be completely convinced and another not be. We don't on account of this exclude either one from society, as being unaccountable and incapable of judgment. (RPP I: §685) This point is connected to Wittgenstein's characterisation of the evidence on which third-person psychological ascriptions are based as often imponderable (PPF, xi: §358) -that is, evidence which may be compelling to one, but which one cannot easily describe or articulate to others in such a way that they too find it compelling: "Imponderable evidence includes subtleties of glance, of gesture, of tone" (PPF, xi: §360). I may, for instance, be in a position to recognise a genuine look without being in a position to describe it (LW II: p. 61), or the difference between the genuine look and pretence (PPF, xi: §360). It is also bound up with Wittgenstein's insistence in PPF, xi, §354 that we cannot prove, for instance, that another person's expression is genuine. I may be convinced that another person is in a particular mental state and yet not be able to prove that they are, for there is not the sort of agreement around general principles which might support such a proof (LW II: p. 92). That I can't prove the correctness of my judgments about the psychological states of others is important, for it marks a distinction between the certainty involved in believing that another person is, say, in pain and believing a mathematical proposition: I can be as certain of someone else's feeling feelings as of any fact. But this does not make the sentences "He is very depressed", "25 x 25 = 625", and "I am 60 years old" into similar instruments. A natural explanation is that the certainty is of a different kind.-This seems to point to a psychological difference. But the difference is a logical one. (PPF,xi: §330) The kind of certainty is the kind of language-game. (PPF,xi: §332) In the case of 25 x 25 = 625, I can offer a proof. This marks the certainty as of a different kind. It is a certainty which can be backed by a demonstration of the truth of the claim believed. Not so in the case of my beliefs about the mental states of others.

Wittgenstein's positive conception of knowledge in epistemic indeterminacy contexts
In PPF, xi, §355 Wittgenstein says that although expert knowledge of others' mental states involves rules, they do not form a system and only experienced people can apply them correctly. In this section, I explore the significance of these negative remarks and offer an account of the kind of knowledge at issue.
The focus of PPF, xi, §355 is judgments concerning the sincerity of expressions of feeling or emotion. The rules mentioned in that passage are, accordingly, rules for distinguishing the genuine expression of emotion from mere pretence -rules like: (1) Usually, a smile "with the eyes" is a genuine expression of happiness; (2) A rushed apology is typically not a genuine expression of remorse; (3) Crying while "in character" is generally not a genuine expression of sadness. Unlike (1)-(3), (4)-(6) do not correlate an instance of behaviour with an inner state by asserting that the former is evidence of the latter. Rather, they assert of some behaviour that it is, or is not, typically genuinely expressive of some feeling. But this is not a substantive difference. Applications of rules of both kinds might form the basis for a judgment that the person exemplifying the relevant behaviour is in the relevant inner state. And, more significantly for our purposes, both kinds of rules are constitutively uncertain in the sense characterised above.
Certainly, (4)-(6) are sound guidelines. But they only purport to hold generally. Indeed, it is not difficult to imagine cases in which, for instance, a rushed apology is a genuine expression of remorse. If the person apologising is in a hurry, distracted, or nervous, the best explanation for their urgent apology is not obviously that it is insincere. Indeed, the answer to the question whether the apology is sincere might, in these circumstances, be unclear, even given all the possible evidence -just as the analogous question might in the case of (1)-(3). Like rules of evidence (and unlike, for instance, calculating rules), rules for distinguishing sincere from insincere expressions of feeling do not form a system, in that they do not suffice in every conceivable case to decide the question whether an expression is in fact genuine, even assuming that one is apprised of all the relevant information.
Of course, someone who is insensitive to the nuances of a particular case might nevertheless rush to judgment, applying the relevant rule in a way that yields an incorrect judgment. Even in the case where the person apologising is obviously nervous, for instance, they might conclude, on the basis of a blind application of (5), that the apology is in fact insincere. A more experienced person might rightly be ambivalent in such a case. Because (1)-(6) are not exceptionless, a subtlety of judgment not guaranteed by mere grasp of the rules' content is required if their application is to yield correct judgments. (Again, contrast calculating rules.) Wittgenstein's suggestion in §355 is apparently that such subtlety of judgment comes only as the result of experience with people.
How does experience with people foster the capacity for judgment at issue here? One possibility is that it furnishes one with a more refined rule (or set of rules), one which includes a -perhaps exhaustive -list of exceptions. Instead of (4), perhaps the experienced person acquires (4)*: Smiling "with the eyes" is a genuine expression of happiness, except in circumstances C 1 , C 2 , C 3 … Wittgenstein's response to the suggestion that experience may furnish one with an exceptionless rule is likely to be two-pronged. First, he will question whether rules like (1)-(6) could really be rendered exceptionless merely by building exceptional circumstances into the rule. For regardless of the number of exceptions we build in, we can always, it seems, imagine a case in which none of those exceptions obtain, and yet we do not want to say that the target expression is, for instance, sincere. 8 Second, and more substantively, he will claim that the very idea that rules (1)-(6) could be rendered exceptionless mischaracterises the nature of the uncertainty that attends our judgments about what another person is thinking or feeling. 9 Granted, if we think of this uncertainty as explained in terms of the complexity of the psychological processes that give rise to those thoughts and feelings, then we will be inclined to think that an exceptionless rule, though extremely difficult to come to, is at least an in principle possibility. Once we understand those psychological processes in their full complexity, then our uncertainty will disappear. But, Wittgenstein will object, this gets things the wrong way around: It is not the relationship of the inner to the outer that explains the uncertainty of the evidence, but rather the other way around -this relationship is only a picture-like representation of this uncertainty. (LW II: p. 68) The inexactness of the rules on the basis of which we attribute thoughts and feelings to others is not a consequence of something else -the complexity of the psychological processes which give rise to those thoughts and feelings, or of the purely causal relation which mediates between thoughts and feelings and their expression (which, in any case, Wittgenstein disputes). It is, rather, a basic, ineliminable feature of the mental.
Might experience nevertheless furnish one with a more refined rule, even if the rule is not exceptionless? Wittgenstein may well agree with this point. But he will resist the idea that the only way in which we come to know what another person is thinking or feeling is by inferring on the basis of some more or less refined rule: "We see emotion."-As opposed to what?-We do not see facial contortions and make inferences from them (like a doctor framing a diagnosis) to joy, grief, boredom. We describe a face immediately as sad, radiant, bored, even when we are unable to give any other description of the features. (Z: §225) In general I do not surmise fear in him -I see it. I do not feel that I am deducing the probable existence of something inside from something outside; rather it is as if the human face were in a way translucent and that I were seeing it not in reflected light but rather in its own. (RPP II: §170) Often, our determinations as to what another person is thinking or feeling are not rule-based inferences or interpretations (Z: §218), but, rather, immediate judgments -the result of our seeing those thoughts and feelings expressed in the person's behaviour. Wittgenstein's thought, it seems, is that experience is significant insofar as it fosters the capacity to make such judgments: Ask yourself: How does a man learn to get an "eye" for something? And how can this eye be used? (PPF,xi: §361) We might say of an art critic that they have an "eye" for the style of a certain artist, in the sense that they are able reliably to identify artworks by that artist merely by looking. Someone with this capacity is in a position to determine when rules of thumb for identifying artworks by that artist (for example: such-and-such a brush technique is evidence that the artwork is by X; works in X's late period include this combination of colours in just these proportions -the analogues to (1) to (6)) are a reliable guide as to the provenance of a particular artwork and when they are not. They have the sensitivity of judgment requisite to discern those cases in which application of the rule yields a correct judgment from those in which it does not.
Similarly, we might say of someone that they have an "eye" for the genuine expression of emotion, in the sense that they are able reliably to identify genuine expressions of emotion just by looking. Like the art critic, someone with this capacity possesses the capacity to determine when rules (1)-(6) are a reliable guide to another's psychological state and when they are not. They have the sensitivity of judgment necessary to apply rules (1)-(6) correctly. 10 But what is the role of experience in the acquisition of such a capacity? Consider the art critic. How might she come by the capacity to reliably identify paintings just by looking? To begin, she may work exclusively with rules of thumb like those mentioned above. But such rules are unlikely to settle the question of the origin of an artwork in every case. Indeed, in some cases, they are likely to be misleading. How does the aspiring critic come reliably to determine when these rules are reliable guides and when they are not? The answer: by examining artwork after artwork after artwork, comparing those which belong to the artist to those which don't, scrutinising their features, noticing similarities and differences, and so on. On the basis of this body of experience, the critic may eventually be in a position to identify an artwork as belonging to a particular artist, not (or not simply) by inferring from some set of rules, but by seeing the artist's style expressed in the artwork.
Similarly, one might come to acquire the capacity to reliably identify genuine expressions of emotion just by looking, by interacting with a range of different people across a variety of different contexts, by seeing genuine expressions of happiness, seeing disingenuous expressions, noting similarities and differences, and so on. It is via such experience with people that one might acquire the capacity necessary to apply rules (1)-(6) correctlynamely, the capacity to see another's behaviour as expressive of their inner states. 11 This way of thinking about the significance of experience for our coming to know the minds of others makes sense of Wittgenstein's remark in §335 that the knowledge he is interested in does not consist in acquiring a technique, but in learning correct judgments. Someone who is able to identify others' mental states on the basis of (1)-(6) (or other, appropriately refined rules) has acquired a technique, understood as some procedure or protocol employable across a range of contexts, for determining how things are with other people psychologically. Not so in the case of someone who has acquired an "eye" for their genuine expression. Such a person has the ability to identify mental states in others in virtue of seeing certain behaviours as expressive of those states. Their judgment to the effect that another person is, for instance, in pain, is, when the result of the exercise of this ability, not an inferential judgment, but an immediate, broadly perceptual one. We might say of such a person, not that they have acquired a technique, but that they have learnt how to make correct perceptual judgments concerning the psychological conditions of other people. 12 Earlier, we said that if the rules of evidence for a particular mental state are constitutively uncertain, then there are contexts with respect to which those rules fail to decide the question whether a particular person is in that state, and would do so even if we were to take all the possible evidence for that person's being in that state into consideration. If it were Wittgenstein's view that one could come to know another's mind only by applying rules of evidence (or rules like (4)-(6)), then contexts in which the rules of evidence are constitutively uncertain would, it seems, be contexts in which one could not know the mind of another. But, as we have seen, that is not Wittgenstein's view. In epistemic indeterminacy contexts, one might come to know another's inner condition by seeing that condition expressed in their behaviour. 13 Let me summarise the discussion thus far. The question which prompted our investigation into Wittgenstein's positive conception of knowledge in epistemic indeterminacy contexts concerned Wittgenstein's conception of the relationship between epistemic and metaphysical indeterminacy. Specifically: does Wittgenstein think that epistemic and metaphysical indeterminacy are exactly cooccurrent? I claimed that he does not insofar as he rejects (i), according to which in any case in which it is epistemically indeterminate whether S is in M, it is metaphysically indeterminate whether S is in M. We are now in a position to appreciate the 12 Here, I am in agreement with a tradition in Wittgenstein scholarship that sees him as defending a broadly perceptual model of our knowledge of other minds. See, for instance, McDowell (1983) and Overgaard (2006) (although cf. Dain (2019)). 13 Of course, it is not only in indeterminacy contexts that Wittgenstein thinks we might come to know another's inner condition in this way. I take it that Wittgenstein thinks that we commonly know what others think or feel by seeing their behaviour as expressive of those thoughts and feelings. justification for this claim. As our discussion has revealed, Wittgenstein thinks that there are some epistemic indeterminacy contexts in which we can know another's mental state by exercising the sort of capacities just outlined. But if there are some epistemic indeterminacy contexts in which we can know that S is in M, then in those contexts there must be some fact of the matter as to whether S is in M. In other words, it cannot be metaphysically indeterminate whether S is in M. But it follows that there are some contexts in which it is epistemically indeterminate, though not metaphysically indeterminate, whether S is in M. So Wittgenstein rejects (i). So it is not Wittgenstein's view that epistemic and metaphysical indeterminacy are exactly co-occurrent.

Seeing aspects and patterns
The discussion thus far has taken for granted the notion of seeing an instance of behaviour as expressive of another's inner state. But how exactly are we to understand this notion? In this section, I defend two conclusions: first, seeing another's behaviour as expressive of their inner state bears instructive links to a more general phenomenon, which also features prominently in Wittgenstein's later philosophy of psychology -namely, the phenomenon of seeing an aspect; second, to see another's behaviour as expressive of their inner state is to see that behaviour as the continuation of a pattern "in the weave of life" (LW II: p. 42). I defend these conclusions in turn.
In PPF, xi, §113, Wittgenstein introduces the phenomenon of aspect dawning or noticing an aspect: I observe a face, and then suddenly notice its likeness to another. I see that it has not changed; and yet I see it differently. I call this experience "noticing an aspect".
Here, noticing an aspect is characterised in terms of an experiencethe experience of seeing that an image or object -in this case, a face -has not changed while nevertheless seeing it differently. This is the experience we might have when, for instance, we see Jastrow's duckrabbit, an image or "picture object" (xi: §119) which can be seen either as a rabbit or as a duck, now as a duck, now as a rabbit.
Several remarks later, in §118, Wittgenstein introduces a related phenomenon, the phenomenon of the "continuous seeing" of an aspect. The duck-rabbit image is ambiguous, but suppose that one has not noticed the ambiguity -that one sees the image only as a rabbit. Although one is seeing the image under an aspect, it would be infelicitous to describe such a case as involving the dawning or noticing of an aspect. Rather, the case is more accurately described as one in which one sees the image continuously under an aspect.
These remarks begin a detailed investigation into the topic of aspect perception more generally. A primary concern of the investigation is to resolve the sense of puzzlement which the initial statement of aspect dawning might engender. How could I see the picture differently if, at the same time, I see that it has not changed? Do I really see something different in each case, or do I see the same thing but interpret it differently? Wittgenstein's considered view is that we ought not to assimilate the phenomenon of aspect dawning -or of aspect perception considered more generally -either to the phenomenon of seeing, or to the phenomenon of interpreting. The pressure to so assimilate it has its base in the assumption that a visual phenomenon must ultimately be either one of seeing or one of interpreting. But that, Wittgenstein thinks, is an assumption which we are under no obligation to accept. Instead of shoehorning aspect perception into one or another category, we should treat it as a phenomenon in its own right -albeit one which bears resemblances to seeing, interpreting and a variety of other phenomena. 14 Aspect perception, as it interests Wittgenstein, is not exclusive to our perception of images. His discussion of the phenomenon takes in, not only puzzle pictures, but also drawings (xi: §180), music (xi: §178), inanimate objects (xi: §141) and -most importantly, for our purposes -living things and their behaviour. In PI, §420, for instance, Wittgenstein invites us to try to see the behaviour of other people as the behaviour of automata. Indeed, the very example Wittgenstein uses to introduce the notion of aspect dawning is one which involves our suddenly seeing a human face as resembling another. These examples suggest that aspect perception generally, and aspect dawning specifically, has application beyond mere pictures and images -specifically, to our perception of one another.
Indeed, consider the following structurally similar case to the case Wittgenstein imagines in §113. I am played a video of a person apologising. For whatever reason -perhaps because of my mood -I am unable to see this person's behaviour as anything other than pretence. It strikes me as bearing all the hallmarks of a false apology. Not wanting to allow my judgment to go unchallenged, however, I resolve (a day, a week, a month later) to look again. Now, under different circumstances, I see the person's behaviour differently, as genuinely expressive of remorse. 15 In this case, I have an experience analogous to that which §113 introduces as characteristic of the phenomenon of aspect dawning: I see that this person's behaviour has not changed, and yet I see it differently. And, not surprisingly, there arise just the same questions as those raised by the case which Wittgenstein discusses in §113. How could I see their behaviour differently if, at the same time, I see that it has not changed? Do I really see it differently, or do I just interpret what I see differently? We should give analogous responses to these questions. The phenomenon of seeing an instance of behaviour now as pretence, now as genuinely expressive of remorse, is at base neither the phenomenon of seeing differently, nor the phenomenon of interpreting differently. It is, if not an instance of the more general, standalone phenomenon of noticing an aspect, then very closely related to it.
We can and should say the analogous thing in the interpersonal case in which person A sees (cannot but see) this behaviour as pretence, while person B sees (cannot but see) it as genuinely expressive. We should say that A and B's disagreement is not traceable to a difference in what they see, nor to a difference in how they interpret what they see. Rather, their disagreement is like the disagreement of two people who see, say, an image (continuously) under different aspects. In both the intra and interpersonal cases described, then, it is instructive to think about our knowledge of others' mental states on the model of aspect perception.
Let me answer two concerns which one might have about this proposal. The first is that it's not obviously felicitous to talk, as I do, about knowledge in connection with aspect perception. When one sees Jastrow's duck-rabbit as, for instance, a duck, it doesn't make sense for one to say on that basis that one knows it is a duck (or for others to say of one "She knows it's a duck"). Such talk of knowledge doesn't appear to have any application here. So why think that it has application in cases involving others' mental states?
The second concern has to do with the general applicability of my proposal. Our focus has been on cases in which, either there is reasonable disagreement about another's inner state, or such disagreement is readily imaginable. But what about a more everyday case? Suppose I happen to be watching as S strikes their thumb with a hammer. 16 I see immediately from S's grimace, their clutching their thumb and so on, that they are in pain. Reasonable disagreement about whether S really is in pain does not seem possible. If someone were to purport to see S's behaviour as pretence, for instance, I would take that as evidence that they have not grasped the concept pain, or that they were inattentive or deluded. In such a case, is it really plausible that my recognition of another's inner state is properly thought of on the model of aspect perception? I'll consider each of these concerns in turn.
In relation to the first concern, it's true that it doesn't make sense for one who sees the duck-rabbit as a duck to say on that basis that one knows it is a duck (or for others to say of one "She knows it's a duck"). Plausibly, this is because the picture is essentially ambiguous (and is acknowledged to be so) with respect to the relevant categories, the categories picture duck and picture rabbit: although one can see the picture as a duck or as a rabbit, it is in fact neither a picture of a duck nor of a rabbit. More generally, an item is essentially ambiguous with respect to a class of categories just in case: (i) the object can be seen (at different times) as an instance of the categories within the class; and (ii) the object is not in fact an instance of any category within the class. In cases where an item is essentially ambiguous with respect to, say, the categories A, B, and C, and is recognised to be so, it is infelicitous to talk of one's knowing that the item is, say, an instance of A (on the basis of seeing it as an instance of A, or otherwise). It is infelicitous because (it is recognised that) the object is not in fact an instance of A. This, I suggest, is the reason why it is infelicitous for one who sees the duck-rabbit as a duck to say on that basis that one knows it is a duck (or for others to say of one "She knows it's a duck").
Not every case of aspect perception is one in which the item seen under an aspect is essentially ambiguous in this sense. 17 Consider, for instance, a case in which, after looking in their direction for a while, I suddenly recognise an acquaintance in a crowd (PPF,xi: §144). In this case, the person at whom I am looking is not essentially ambiguous with respect to the relevant class of categories, the categories acquaintance and stranger: (i) I see the person now as a stranger, now as an acquaintance; but (ii) the person is in fact an acquaintance. There is, we might say, an independent fact to which my perception is answerable such that, in seeing this person now as a stranger, now an acquaintance, I may succeed in seeing, or fail to see, this person as they really are. Not so in the case of the duckrabbit. We may see the image now as a rabbit, now as a duck, but in seeing the image in one or another of these ways we are not succeeding in seeing, or failing to see, the image as it really is -for there is no way it really is. There is no independent fact to which one's ways of seeing the image are answerable. 18 Talk of knowledge 17 A reviewer questions this point. As evidence for the view that Wittgenstein thinks that seeing an aspect "necessarily involves a certain ambiguity or dissociation", they cite inter alia PPF, xi, § 203: "… I cannot try to see a conventional picture of a lion as a lion, any more than an F as that letter (though I may well try to see it as a gallows, for example)" (emphasis in original). On my view, this passage and others like it are not making the point that essential ambiguity in the sense described above is a precondition for aspect perception. Rather, their point is that expressions like "Now I see it as…" and "I am trying to see it as…" are out of place in cases where no alternative way of seeing the item in question is salient for one (as they typically aren't in the case where the item in question is a conventional picture of a lion). See the discussion of saliency further on. 18 This point presumably has its limits. We might think that even in the duck-rabbit case, our ways of seeing the image are in some sense answerable to facts about the image itself -for instance, to the fact that it can be seen as a duck or as a rabbit but not just as anything at all (not, for instance, as a horse).
But the fact that ways of seeing the duck-rabbit are answerable in this sense is perfectly consistent with: (i) it's being inappropriate to assert "I know that it's a duck" in does not seem out of place in the acquaintance case. It seems perfectly natural to say of someone (or of ourselves) that they know that this person is an acquaintance on the basis of seeing them as such. Plausibly, the reason it seems natural is that in this case the item perceived is not essentially ambiguous with respect to the relevant categories. This person is in fact an acquaintance and I can come to know that they are by seeing them as such.
In this respect at least, the expressive case clearly has more in common with the acquaintance case than it does with the duckrabbit case. Assuming that it is not metaphysically indeterminate whether another person is in pain, it is not essentially ambiguous whether this behaviour is genuinely expressive of pain or mere pretence. In seeing their behaviour as expressive of pain one may succeed in seeing, or fail to see, that behaviour as it really is. To this extent, it is not infelicitous to talk of knowledge in the expressive case.
In relation to the second concern, certainly, my seeing that S is in pain is not properly thought of on the model of aspect dawning, for it is not accompanied or constituted by the experience which is characteristic of that phenomenon -the experience of seeing that S's behaviour has not changed and yet seeing it differently. I cannot but see S's behaviour as expressive of pain. Might it nevertheless be properly thought of on the model of the continuous seeing of an aspect? The answer to this question depends, I think, on whether the continuous seeing of an aspect requires that other ways of seeing the image or object in question are salient, given the context. We can characterise the notion of saliency as follows: for a way of seeing an image or object to be salient given the context is for it to be readily available to at least some subset of those agents who grasp the relevant concepts and are attentive and rational. 19 In the context involving S and the hammer, alternative ways of seeing S's behaviour are decidedly not salient. I cannot see S's behaviour other than as the duck-rabbit case; and (ii) it's being the essential ambiguity of the image relative to the categories picture duck and picture rabbit which accounts for the inappropriateness. So the fact that ways of seeing the duck-rabbit are so answerable does not pose a problem for my account. 19 This is nothing more than a characterisation of saliency. It is not meant to amount to a set of necessary and sufficient conditions. expressive of pain, nor can others, who grasp the relevant concepts and are attentive and rational, see it otherwise. Compare the sorts of cases which are the focus of this paper -epistemic indeterminacy contexts. Even if I cannot see another's behaviour as anything other than, say, an expression of anger, so long as I am in an epistemic indeterminacy context, others, again, who grasp the relevant concepts and are attentive and rational, may readily see it differently. This fact does not, of course, mean that there really is no right way to see the behaviour, or that I ought to be any less assured of my own way of seeing it. Certain ways of seeing the behaviour in question will, and others will not, be revelatory of that behaviour as it really is. Still, the context is such that one may grasp the relevant concepts, be attentive and rational and yet not see the behaviour in the way that I see it. To this extent, alternative ways of seeing the behaviour are salient given the context.
If the continuous seeing of an aspect does not require that other ways of seeing the image or object are salient, given the context, then there is no barrier to conceiving of my seeing that S is in pain on the model of continuous aspect perception. 20 If, however, it does so depend, then such a conception is probably not accurate. It is not my concern to settle the question whether the continuous seeing of an aspect does or does not so depend, and so to settle the question whether aspect perception provides a useful model in cases like the one involving S and the hammer. In the sorts of cases which are the focus of this paper -epistemic indeterminacy contexts -alternative ways of seeing the target behaviour are salient in just the sense outlined above. Even if seeing another's behaviour as expressive of their inner state is not properly thought of on the model of aspect perception in every case, I maintain that it is properly thought of in this way in contexts in which alternative ways of seeing another's behaviour are readily available to rational, attentive and conceptually competent onlookers.
I said that the phenomenon of seeing an instance of behaviour now as pretence, now as genuinely expressive of remorse, is if not an instance of the more general, standalone phenomenon of noticing an aspect, then at least very closely related to it. Does conceiving of the seeing of another's behaviour as genuinely expressive on the model of aspect perception thus rule out the possibility of our saying anything illuminating about what it is to see an instance of behaviour in this way? No. Wittgenstein himself makes a number of positive suggestions, particularly in the case of emotional states. For example, at several points in his later work, he suggests that we recognise others' actions and behaviour as expressions of emotional states, not by considering those actions and behaviour in isolation, but by locating them within a broader pattern of human goings-on: Someone smiles and his further reactions fit neither a genuine nor a simulated joy. We might say "I don't know my way around with him. It is neither the picture (pattern) of genuine nor of pretended joy." (LW II: p. 61) We judge an action according to its background within human life, and this background is not monochrome, but we might picture it as a very complicated filigree pattern, which, to be sure, we can't copy, but which we can recognise from the general impression it makes. (RPP II: §624) On Wittgenstein's view, there is an essential diachronic aspect to our making sense of another's behaviour. To see an instance of behaviour as genuinely expressive is to see it as the continuation of a pattern, as the extension of a strand in the "weave of life" (LW 1: §862). On that view, knowing another's mind is less a matter of careful scrutiny of the person in isolation from the broader context and more a matter of attunement to a motif or theme which runs through our lives together. Of course, none of this is in tension with, or an alternative to, the thought that to see another's behaviour as expressive of pain is an instance of aspect perception. Seeing for the first time a shape, say, as the continuation of a pattern of shapes is an instance of the dawning of an aspect. (One sees that the shape has not changed and yet one sees it differently.) So too seeing for the first time an instance of behaviour in this way.
One objection to Wittgenstein's view might be expressed as follows: I grant that seeing another's behaviour as expressive of grief, say, requires experience with past expressions of grief (and perhaps the expression of other emotions besides). One couldn't recognise some behaviour as expressive of grief unless one had identified, or been in the presence of the identification of, grief (or similar emotions) in the past. But it is one thing to claim that experience with past expressions of grief is required for one to see an instance of behaviour as expressive of grief. It is quite another to claim that to see some instance of behaviour as expressive of grief just is to see it as bearing a pattern-like relation to those past expressions. The former is an explanatory claim, the latter a constitutive one. What reason do we have to accept the constitutive claim?
This concern is buoyed by a particular conception of grief, and emotions more generally. On this conception, grief is the emotion that it is purely in virtue of its intrinsic qualities. Facts about, for instance, the way in which grief is typically expressed, about the way in which it interacts with other emotions and thoughts and sensations, about the circumstances under which we would be justified in saying of someone that they are experiencing grief, while nevertheless illuminating, are entirely incidental to the nature of grief itself. But this conception of the emotions is one which Wittgenstein rejects. Instead, he proposes that we think of emotions as themselves patterns in the weave: "Grief" describes a pattern which recurs, with different variations, in the weave of our life. If a man's bodily expression of sorrow and of joy alternated, say with the ticking of a clock, here we should not have the characteristic formation of the pattern of sorrow or of the pattern of joy" (PPF, i: §2) If grief itself is a pattern, then it is no surprise that its recognition involves the seeing of an instance of behaviour as a continuation of that pattern.

Conclusion
The concern of this paper has been to elucidate Wittgenstein's views about the constitutive uncertainty of the mental. We have seen that one strand of that view is that there are contexts with respect to which the rules of evidence for mental states fail to decide the question whether a particular person is in the relevant state, and would do so even if we were to take all the possible evidence for that person's being in that state into consideration. Still, we might know another's mind in such contexts. Such knowledge is the result, not of inferring from a rule akin to (1)-(6), but rather the exercise of a capacity to see another's behaviour as expressive of their psychological condition. Like the capacity to see an artwork as expressive of a particular artist's style, such a capacity requires for its acquisition experience of a particular kind. We also noted the connection between Wittgenstein's discussion of the constitutive uncertainty of the mental and two further components of his later philosophy: his discussion of the phenomenon of aspect-perception and his views about the pattern-like nature of the emotions.
On this final point, our discussion has revealed an important unity to the remarks in PPF, xi. Wittgenstein opens that section by introducing the phenomenon of aspect-dawning, and the theme of aspect perception more generally. He closes it with discussion of cases involving marked disagreement about the mental states of others, and of the kind of knowledge available to the experienced in such contexts -knowledge which, as we have seen, we should think of on the model of aspect perception. 21