There has been considerable recent work in the metaphysics and epistemology of self-awareness on what has come to be known as the elusiveness thesis (see, e.g., Shoemaker 1986, 1994; Cassam 1994, 1997; Bermúdez et al. 1995; Bermúdez 1998; Peacocke 1999). According to the elusiveness thesis, the self cannot be an object of introspective awareness. Alternatively put, when we find out about our own properties through introspection we are not acquainted with any object whose properties they are. In this respect, then, our introspective awareness of ourselves is fundamentally different from our awareness of ordinary physical objects and other psychological subjects. This general idea is developed in different ways depending on how the key notions of self and introspection are understood.
It is natural to think that this distinctive feature of introspection has epistemological consequences, and in fact the elusiveness thesis sits easily with a particular way of thinking about the epistemological issue of privileged access. We clearly have a form of privileged access to our own mental states. This privileged access is a function of certain ways we have of finding out about our own mental states, and it is standard to think that it applies only to introspective ways of finding out about our own mental states. Pursuing this line of thought, it is hard not to be struck by the thought that privileged access and the elusiveness thesis might be two sides of the same coin. Perhaps it is (at least partly) because we do not encounter the self as an object in introspection that the knowledge of the self that we gain through introspection is epistemically privileged—the epistemic privilege derives from the fact that our introspective self-awareness is awareness of a fundamentally different type from our awareness of ordinary objects and other psychological subjects.
This idea that there is a close connection between privileged access and the elusiveness thesis will be the subject of this essay. I will not be challenging the general thesis that a significant degree of epistemic privilege attaches to introspective self-awareness—although I shall have some things to say about how that epistemic privilege should be understood. My principal question is whether there are any forms of self-awareness that are epistemically privileged yet nonintrospective. In discussing the general issue of privileged access, philosophers have tended to concentrate on the various types of privileged access that might accrue to introspective awareness of psychological properties (see, e.g., Alston 1971). But there are equally important questions concerning the degree of privileged access accruing to the nonintrospective ways in which we can find out about our physical properties. These nonintrospective forms of self-awareness include proprioceptive bodily awareness and the perception of the self that occurs as an inextricable part of ordinary perceptual awareness. We shall be looking in some detail at the type of privileged access afforded by these nonintrospective modes of self-awareness. We will see that they require a fundamentally different type of explanation from ordinary introspective awareness—a type of explanation in which the elusiveness thesis has no part to play.
Section I surveys different formulations of the elusiveness thesis, focusing on the recent formulation by Sydney Shoemaker linking the elusiveness thesis with the phenomenon of immunity to error through misidentification (relative to the first-person pronoun). In section II, I explore proprioceptive bodily awareness, suggesting that it offers a prima facie reason for prising the phenomenon of immunity to error through misidentification apart from the elusiveness thesis. Somatic proprioception offers both a protoperceptual acquaintance with the embodied self and an identification-free source of information about one’s bodily properties. Section III looks at the nonintrospective forms of self-awareness that are an inevitable concomitant of exercising the ordinary perceptual systems. Perceptual awareness of the embodied self can give rise to judgments that are immune to error through misidentification and are even more obviously examples of perceptual awareness of the embodied self than somatic proprioception. In section IV we turn to the key question raised by earlier sections. What explains the immunity to error through misidentification of these different types of self-awareness? I shall offer a different explanation for each type.
The origins of the elusiveness thesis are usually taken to lie in a famous passage from Hume’s Treatise of Human Nature:
For my part, when I enter most intimately into what I call myself, I always stumble on some particular perception or other, of heat or cold, light or shade, love or hatred, pain or pleasure. I can never catch myself at any time without a perception, and can never observe anything but the perception. When my perceptions are remov’d for any time, as by sound sleep; so long am I insensible of myself, and may truly be said not to exist. And were all my perceptions remov’d by death, and cou’d I neither think, nor feel, nor see, nor love, nor hate, after the dissolution of my body, I shou’d be entirely annihilated, nor do I conceive what is farther requisite to make me a perfect non-entity. If anyone upon serious and unprejudic’d reflexion, thinks he has a different notion of himself, I must confess I can reason no longer with him. All I can allow him is, that he may be in the right as well as I, and that we are essentially different in this particular. He may, perhaps, perceive something simple and continu’d which he calls himself, tho’ I am certain there is no such principle in me. (Hume 1739–1740/1978, 252)
The elusiveness thesis, as we find it in Hume, is closely linked with Hume’s peculiar form of irrealism about the self. The self, as far as Hume is concerned, is nothing over and above a “bundle of perceptions”—and the elusiveness thesis is an important part of his argument for that radical view.
Although the elusiveness thesis is usually credited to Hume, he is not in fact either the only or the earliest major figure in the history of philosophy to have maintained it. We find the elusiveness thesis clearly stated by Descartes who thought that we are never directly acquainted with the thinking “I.” This is a direct consequence of his view that the self is a substance—the self is the “thing which thinks,” in the famous phrase from the Meditations. It is part of Descartes’s conception of substance that we are never directly acquainted with substances, only with their properties, as he clearly states in the Principles:
However, we cannot initially become aware of a substance merely through its being an existing thing, since this alone does not of itself have any effect on us. We can, however, easily come to know a substance by one of its attributes, in virtue of the common notion that nothingness possesses no attributes, that is to say, no properties or qualities. Thus, if we perceive the presence of some attribute, we can infer that there must also be present an existing thing or substance to which it may be attributed. (Descartes 1984, 210)
So, for Descartes, just as for Hume, all I am directly acquainted with are thoughts. I am not directly acquainted in introspection (or in any other way) with the thinker of those thoughts. Nonetheless, unlike Hume, Descartes does not deny that there is any such thinker. The existence of a thinking self is, for him, guaranteed by the twin metaphysical facts, first, that thoughts are properties and, second, that properties cannot exist unless they are instantiated in something.
There is an interesting difference between Hume and Descartes with respect to the elusiveness thesis. It looks very much from the second part of the quoted passage as if Hume saw the elusiveness thesis as a contingent claim. There is, he seems to be saying, no incoherence in the thought that an introspecting subject might encounter a self lurking among his perceptions (“he may be in the right as well as I”). It is just that, as a matter of fact and, no matter how hard he tries, his attempts to introspect the self have not met with success. This apparent admission of contingency is compatible with Hume’s irrealism about the self, provided of course that one takes the nonexistence of a self over and above the bundle of perceptions to be itself a contingent matter. As far as Descartes is concerned, however, the elusiveness thesis holds necessarily rather than contingently. Given that the self is a substance, and given that we cannot enter into any sort of epistemic relation with substances qua substances, it follows as a matter of necessity that we cannot be directly acquainted with the thinking “I.”
It is hard, however, not to feel a certain dissatisfaction with the way in which both Descartes and Hume set up the framework that allows them to formulate their respective versions of the elusiveness thesis. The problem arises with the way in which they view the nature of introspective awareness, as if introspection were a matter of scanning the contents of the mind in much the way in which one might scan the contents of a room. Sydney Shoemaker (1986) has provided a formulation of the elusiveness thesis that speaks directly to this concern about the nature of introspective awareness. As Shoemaker understands the elusiveness thesis, it is as much a thesis about the nature of introspection as it is about the nature of the self. One of Shoemaker’s targets is the idea that introspection should be understood as a form of perception—the so-called act-object conception of introspection. According to this way of understanding introspection, introspection involves a subject standing in a protoperceptual relation to a mental particular (which might be a thought, a propositional attitude, an emotion, or a sensation). It is something like the act-object conception that lies behind Hume’s version of the elusiveness thesis, Hume’s claim being in effect that the mental particulars that he encounters when he introspects do not include a self. In Shoemaker’s version of the elusiveness thesis, in contrast, the central point is that introspection does not involve protoperceptual awareness of any mental particulars at all—and a fortiori no protoperceptual awareness of the self. Perception, as Shoemaker understands it, involves the pickup of what he calls identification information—that is, information that allows the perceiver to pick out one object among a range of objects in virtue of its perceived properties. But there is, he suggests, nothing analogous to such observationally based identification of an object in introspection—neither as applied to mental particulars nor as applied to the self.
Introspection gives rise to judgments that are, to use Shoemaker’s own phrase, immune to error through misidentification relative to the first person pronoun:
To say that a statement “a is ϕ” is subject to error through misidentification relative to the term “a” means that the following is possible: the speaker knows some particular thing to be ϕ, but makes the mistake of asserting “a is ϕ” because, and only because, he mistakenly thinks that the thing he knows to be ϕ is what “a” refers to. (Shoemaker 1984, 7–8)
Immunity to error is an epistemological notion relativized to the grounds on which the judgment or statement in question is made. It is a function of the particular form of warrant possessed by a type of judgment or statement, so that, for example, the same sentence can be employed to make two different statements, only one of which is immune to error, if the grounds on which the statement is made on the two occasions of utterance are appropriately different. A useful way of bringing out how the immunity to error through misidentification of a statement or judgment is a function of the grounds for that statement or judgment has been offered by Crispin Wright (1998). As Wright points out, a judgment that is subject to error through misidentification is one whose grounds will survive as grounds for an existential generalization if the misidentification is uncovered. So, suppose that I come to form the judgment “I’m looking scruffy” by seeing a scruffy person on a closed-circuit television display in a shop window and wrongly identifying that person as myself. If I discover the mistake then I must, of course, retract the original judgment—but the grounds on which I made that judgment will survive as grounds for the existential generalization “Someone is scruffy.” What characterizes judgments that are immune to error through misidentification, however, is that there can be no such retreat to an existential generalization.
The elusiveness of the self (to introspection) is, for Shoemaker, constitutively linked with the immunity to error of judgments based on introspective self-awareness. The argument is subtle. There is no point, he thinks, in even considering that the self might be encountered in introspection unless the accessibility of the self could perform some cognitive function, and the self could only be encountered in introspection as the owner or the bearer of introspectable mental states (rather than what Hume seems to have been looking for, which is some sort of disembodied entity over and above those mental states). But what cognitive function could be served by the introspectability of the owner of mental states, other than to allow that owner to be identified? And this is the crux of the matter, since the whole point of the immunity to error of the deliverances of introspection is that no such identification is needed. Once we realize that judgments based on introspection are immune to error through misidentification, the thought that there might be a perceptual encounter with the self in introspection swiftly reveals itself to be incoherent—simply because there is no gap between knowing that an introspectively accessible property is instantiated and knowing that it is instantiated in oneself.1
I will be discussing Shoemaker’s version of the elusiveness thesis in the remainder of this essay. It has considerable plausibility. It is hard to see how one might dispute Shoemaker’s claim that introspective awareness of psychological properties does not involve identification information. We do not come to the belief that we believe that p by encountering an object whose properties allow us to identify it, first, as a belief that p, and then, second, as our own belief that p. The claim as applied to introspective awareness of the self is, if anything, even more obvious.
We can note the obvious relevance to the issue of privileged access of Shoemaker’s version of the elusiveness thesis and the correlative conception of introspection. If introspection does not involve any form of observationally based identification, and if (moreover) it does not involve any sort of perceptual or proto-perceptual encounter with the self, then this provides a clear asymmetry between introspection and ordinary perceptual awareness of nonbodily objects—and it is precisely such an asymmetry that it is the burden of the doctrine of privileged access to establish. Privileged access is usually understood in terms of a certain class of beliefs about oneself being immune to a certain type of error, with the type of privileged access being a function of the type of immunity identified—immunity to correction (incorrigibility), immunity to doubt (indubitability), immunity to error (infallibility), immunity to ignorance (self-intimation), and so forth.2 These types of immunity are invariably discussed with reference to introspective beliefs about one’s own psychological states. Shoemaker’s version of the elusiveness thesis not only adds a further type of immunity to the list (immunity to error through misidentification), but in fact the type of immunity that it adds arguably has a foundational role to play relative to the other types, in at least the following sense. Any belief that has any of the other types of immunity will be immune to error through misidentification, whereas the converse does not hold—a belief can be immune to error through misidentification without being either incorrigible, indubitable, infallible, or self-intimating. Moreover, it may well be that there are in fact no other types of immunity—that incorrigibility, infallibility, self-intimation, and so on all turn out to be philosophers’ fictions. In that case, immunity to error through misidentification would be foundational by default.
All this strongly suggests that the elusiveness thesis holds considerable promise for explaining why there should be any type of privileged access at all. It is hard to see how we could have privileged access to certain types of information about ourselves unless we had ways of acquiring those types of information that do not lay us open to certain possibilities of error attendant on ordinary perceptual awareness—and the elusiveness thesis seems to go a considerable way toward explaining why introspective awareness should be different in this way from ordinary perceptual awareness.3
In this essay I shall suggest that the situation is considerably more complex than this attractive picture suggests. One complexity emerges because somatic proprioception and bodily awareness do provide (or so I shall argue) a form of protoperceptual acquaintance with the embodied self, while at the same time being a source of identification-free awareness of one’s own physical properties. So we need to separate out the elusiveness thesis from the immunity to error through misidentification thesis. The claim about the self does not follow immediately from the claim about immunity to error through misidentification.
Further complexity comes from a point that has been almost completely neglected both in discussions of the elusiveness thesis and in discussions of privileged access. The elusiveness thesis, as we have discussed it so far, is a claim about the nature of introspective self-awareness. It says, in effect, that introspective self-awareness does not involve observation-based identification either of the self or of anything else. The contrast it stresses is between introspective self-awareness, on the one hand, and ordinary perceptual awareness of physical objects and other psychological subjects, on the other hand. But even if we accept that this contrast exists, and that it should be characterized in broadly the terms offered by Shoemaker, a further question arises. There are ways of finding out about one’s physical and psychological properties that are not introspective (even when introspection is construed broadly, as some authors have suggested, to include somatic proprioception and bodily awareness). How, if at all, do these fit into the two-way distinction between introspection and ordinary perceptual awareness? Should they all be construed on the model of ordinary perceptual awareness? Or should some of them be assimilated to introspective self-awareness? Many philosophers have assumed that when we find out about ourselves in nonintrospective ways, we are merely applying the same methods and techniques of acquiring information that we employ when we find out about the physical properties of ordinary objects in the world or the physical and psychological properties of other embodied subjects. I shall be suggesting that this idea is misplaced—and hence that the elusiveness thesis can at best be only a part of an explanation of privileged access. We have nonintrospective modes of self-awareness that involve forms of privileged access to our own properties.
Let me start with somatic proprioception. It has struck several authors that somatic proprioception is a source of information yielding judgments that are immune to error through misidentification (Evans 1982). If, for example, I form the judgment that my arms are folded on the basis of feedback from joint position sense, then there is no sense in which I can be mistaken about whose arms it is that are folded. And if I should find out that the judgment is not in fact warranted (perhaps because the relevant receptors have been artificially stimulated) my grounds for the judgment will not survive as grounds for the existential generalization that someone’s arms are folded. Accordingly there has been some discussion of whether somatic proprioception can count as a counterexample to the elusiveness thesis (Ayers 1991; Cassam 1995).
Recall that Shoemaker’s version of the elusiveness thesis, like Descartes’s version but unlike Hume’s, is based on an argument rather than an apparently contingent fact about the contents of the mind. Shoemaker argues, in effect, that there is no point postulating an introspective encounter with the self unless such an encounter serves a cognitive function. Since the only possible cognitive function would be to permit the identification of the states that are being introspected, and since there is no need for such identification (given the immunity to error through misidentification of introspection), the elusiveness thesis follows straightforwardly. The crucial point is that we do not need to be acquainted in introspection with the bearer of our psychological states in order to identify those psychological states as our own. What makes this argument seem so compelling is the fact that it is hard to see what else acquaintance with the self in introspection could achieve besides permitting the identification of the bearer of introspected psychological states.
Clearly, however, no such argument is likely to be effective in the case of somatic proprioception. It is true that the deliverances of somatic proprioception are immune to error through misidentification and, a fortiori, that there is no need to be acquainted with the bearer of those states (that is to say, with the embodied self) in order to identify them as one’s own.4 But, in contrast with ordinary psychological self-awareness, there is a range of cognitive functions that can be served by acquaintance with the embodied self. I have discussed these at some length in previous work (Bermúdez 1998; see also chapters 5 through 9 in this volume). Somatic proprioception provides a fundamental way of registering the distinction between self and nonself, and it does this in two basic ways. First, it provides an awareness of the limits of the body, primarily through the sense of touch. Second, it is one of the key ways by which one becomes aware that the body is responsive to one’s will. The feedback gained through kinesthesia, joint position sense, and the vestibular system explains how one is aware that the body is responding to motor commands. This yields a way of grasping the body as an object that is responsive to the will. In short, what somatic proprioception offers is an awareness of the body as a spatially extended and bounded physical object that is distinctive in being responsive to the will.
In the case of somatic proprioception, therefore, there is a clear cognitive function to be served by acquaintance with the embodied self, and the elusiveness thesis is correspondingly less plausible. But there is no need to rethink the original proposal that the deliverances of somatic proprioception are immune to error through misidentification. The cognitive functions served by acquaintance with the embodied self in somatic proprioception do not include permitting the identification of proprioperceived states as one’s own. The conclusion to draw from this is that the elusiveness thesis comes apart from the thesis about the identification-free nature of awareness of one’s own properties. In the case of somatic proprioception the inference from the identification-free nature of the associated self-awareness to the impossibility of acquaintance with the embodied self does not hold.
Nor, of course, does the fact that we are acquainted with the embodied self in somatic proprioception have any implications for the privileged access to our own physical properties that we derive as a function of the identification-free nature of proprioceptive self-awareness. This casts doubt on the earlier suggestion that the elusiveness thesis might play an important role in the explanation of privileged access. I suggested that privileged access might be a function of ways of acquiring information about one’s own properties that do not lay us open to certain possibilities of error attendant on ordinary perceptual awareness—and the elusiveness thesis seems to explain why the relevant types of awareness should be different in this way from ordinary perceptual awareness. That suggestion depends, of course, on there being a reciprocal relation between immunity to error through misidentification and the elusiveness thesis, so that no sources of information giving rise to beliefs that are immune to error through misidentification could be described as involving perceptual or protoperceptual acquaintance with the self. But it looks very much as if somatic proprioception is going to be a counterexample, because (as we shall see) a case can be made for assimilating proprioception to perceptual awareness.
Everything depends, of course, on how perceptual awareness is understood. There are ways of understanding perceptual awareness on which it follows straightforwardly that somatic proprioception could not possibly qualify—for example, if perceptual awareness is defined as involving the use of a dedicated sensory organ. We need a characterization as theoretically unladen as possible. Here, as elsewhere in this general area, Sydney Shoemaker has led the way. In his Royce lectures Shoemaker offers a general model of perceptual awareness in terms of the following three constraints (Shoemaker 1994).
The Object Constraint. While sense perception provides one with awareness of facts, this awareness of facts is a function of awareness of the objects involved in these facts in a sense experience distinct from the object of perception.
The Multiple Objects Constraint. “Ordinary modes of perception admit of our perceiving, successively or simultaneously, a multiplicity of different objects, all of which are on a par as nonfactual objects of perception” (Shoemaker 1986, 107).
The Tracking Constraint. Sense perception involves information that allows one to pick out the object of perception through its relational and nonrelational properties. Such information enables the “tracking” of the object over time, and its reidentification from one time to another. (Shoemaker calls this the Identification Constraint, but I have renamed it the Tracking Constraint to avoid prejudging issues to do with immunity to error through misidentification.)
I have argued at some length elsewhere (Bermúdez 1998, chap. 6) that somatic proprioception can be plausibly be taken to satisfy all three of these constraints, although not, of course, to the same extent as, say, vision.
In brief, the Object Constraint is met because somatic proprioception can provide sensory experiences of, say, limb movement or muscle stretch that yield factual information about the distribution of body parts and that are no less separate from the embodied self than ordinary visual experiences are from the objects they inform us about.5 The Multiple Objects Constraint is met because the sense of touch is a source of somatic proprioception and can be put to work to yield proprioceptive or exteroceptive information—information about the body or information about nonbodily objects in space. When the sense of touch is being used to yield exteroceptive information it obviously involves the possibility of perceiving a range of different objects, both simultaneously and successively. Even when the attention is fixed firmly on the proprioceptive dimension of tactile awareness, the exteroceptive dimension remains phenomenologically salient in background awareness (Martin 1995). And since it is uncontroversial that deploying the sense of touch exteroceptively permits both the simultaneous and successive perception of a range of distinct objects, it follows that the Multiple Objects Constraint is satisfied in all instances of tactile somatic proprioception, although, of course, in different ways.
In thinking about how the Tracking Constraint might be met, it is worth remembering that one can lose track of one’s body over time—or at least of various parts of one’s body. This happens, for example, when one absentmindedly walks home on automatic pilot instead of to the shops or taps one’s foot in time to a piece of music without noticing. The best reason for describing these as cases where one is losing track of what one’s body is doing is the feeling of surprise that comes when one notices what has been going on. What can confuse matters here is a failure to make the distinction stressed earlier between proprioceptive awareness and the operation of the proprioceptive information systems. In the two examples just given, the proprioceptive information systems are functioning as they must if actions like walking are even to be possible. But this doesn’t mean that the subject is keeping track of his body just because those systems are functioning, any more than does the fact that the proprioceptive information systems continue to function while one is asleep. What counts is the lack of the appropriate sort of proprioceptive awareness.
It would be foolish to argue that these three constraints are met in the case of somatic proprioception in anything like the same way in which they are, say, in visual perception. There is a continuum with ordinary visual perception at one end and ordinary psychological introspection at the other, and the question is where somatic proprioception falls on the continuum. My suggestion is simply that it falls somewhat closer to visual perception than to introspection—we might accordingly describe somatic proprioception as a form of protoperceptual awareness. Moreover, it is an obvious consequence of how this conclusion has been established that the embodied self is the object of proprioceptive protoperceptual awareness. This means, to return to the main thread of the argument, that our earlier suggestion about how privileged access might be explained must be rejected in the case of somatic proprioception. Somatic proprioception is a source of information for judgments that are immune to error through misidentification, but the explanation for this cannot be that somatic proprioception does not involve any form of perceptual acquaintance with the embodied self.
It is starting to look as if we are going to require different explanations of privileged access. Even if the explanation in terms of the elusiveness thesis will suffice for ordinary psychological introspection, it will clearly not do for somatic proprioception. We will return to these issues in section IV. In the next section, I will turn to other forms of self-awareness that bear a striking resemblance to those features of somatic proprioception that we have been stressing in this section. Namely, they provide information that is immune to error through misidentification, and hence a degree of privileged access, while involving perceptual awareness of the embodied self—and in fact a far more unequivocal example of perceptual awareness of the embodied self than occurs in somatic proprioception.
Let us say that through introspection and somatic proprioception we have certain ways of finding out about ourselves, and that the properties we can find out about through introspection include both physical and psychological properties. Two further points should be clear. The first is that there is a range of further ways of finding out about our physical and psychological properties. We can find out about our physical and psychological properties by exercising any of the five senses. We can also, of course, rely on testimony or memory. The second is that this range of ways of finding out about ourselves can be divided quite naturally into two categories, according to whether or not they involve anything that might be described as acquaintance with the embodied self. Suppose that I am trying to find out how long my hair is. I might just take a look in the mirror. Alternatively I might search my memory banks to find out whether I have recently had a haircut. The first of these ways of finding out involves precisely the sort of encounter with the embodied self that, according to the elusiveness thesis, is not possible in introspection. For the purposes of this essay I shall assume that the five world-directed senses exhaust the direct but nonintrospective ways of finding out about our physical and psychological properties, and that testimony and memory are canonical examples of indirect, nonintrospective modes of self-awareness.
We can, therefore, identify four fundamentally different types of self-awareness—namely, introspective self-awareness, proprioceptive self-awareness, direct nonintrospective self-awareness, and indirect nonintrospective self-awareness. Many philosophers have assumed that when we find out about ourselves in the third and fourth of these ways we are merely applying the same methods and techniques of acquiring information that we employ when we find out about the physical properties of ordinary objects in the world or the physical and psychological properties of other embodied subjects. The attractiveness of this idea in broad terms is easy to appreciate when we are dealing with forms of self-awareness that are nonintrospective and indirect. It is hard to see why there should be any difference between how I use testimony to find out about myself and how I use it to find out about anyone else. The idea also has a prima facie appeal when it comes to perceptual self-awareness, particularly when one considers the examples standardly given to illustrate the distinction between that and introspective self-awareness. These examples usually involve the use of mirrors or some other more or less nonstandard perceptual situation in which one observes someone performing some action or having a certain property and then identifies that person as oneself. In cases like these an act of recognition always takes place and there is a correlative possibility of misidentification.
Wittgenstein’s distinction between uses of “I” as subject and uses of “I” as object in the Blue Book offers a good illustration of the temptation to map the distinction between introspective and nonintrospective self-awareness on to the distinction between identification-free and identification-involving judgments. Wittgenstein identifies two different types of first-person judgment, corresponding to two different forms of self-awareness.
There are two different cases in the use of the word ‘I’ (or ‘my’) which I might call ‘the use as object’ and ‘the use as subject’. Examples of the first kind of use are these: ‘My arm is broken’, ‘I have grown six inches’, ‘I have a bump on my forehead’, ‘The wind blows my hair about’. Examples of the second kind are: ‘I see so-and-so’, ‘I try to lift my arm’, ‘I think it will rain’, ‘I have a toothache’. (Wittgenstein 1958, 66–67)
Judgments of the sort that would be verbally expressed by Wittgenstein’s “use as subject” of the first-person pronoun are based on introspective self-awareness, while judgments expressible through the “use as object” will be based on nonintrospective self-awareness (whether direct or indirect). As we can see in the continuation of the passage, Wittgenstein thinks what distinguishes the use as subject from the use as object is the fact that the form of self-awareness grounding the use as subject is identification-free. The passage continues:
One can point to the difference between these two categories by saying: The cases of the first category involve the recognition of a particular person, and there is in these cases the possibility of an error, or as I should rather put it: The possibility of an error has been provided for. … It is possible that, say in an accident, I should feel a pain in my arm, see a broken arm at my side, and think it is mine when really it is my neighbor’s. And I could, looking into a mirror, mistake a bump on his forehead for one on mine. On the other hand, there is no question of recognizing a person when I say I have toothache. To ask ‘are you sure that it’s you who have pains?’ would be nonsensical. (Wittgenstein 1958, 67)
In this passage, identification-free self-awareness seems to be confined to introspective awareness of one’s psychological properties, implying that any nonintrospective form of self-awareness will involve some sort of observation-based identification of a particular person as oneself. And in fact, this conclusion appears to have been tacitly accepted even by those who challenge the premise that identification-free self-awareness is confined to the domain of the psychological. When Gareth Evans suggested that we can, through somatic proprioception, have identification-free awareness of our physical properties (e.g., Evans 1982), he challenged the alleged entailment from immunity to error through misidentification to introspective awareness of the psychological. But Evans seems not to have challenged the alleged entailment from nonintrospective perceptual awareness to susceptibility to error through misidentification.
Yet this second alleged entailment cannot be accepted. Various central categories of nonintrospective direct self-awareness neither depend on observation-based identification nor are susceptible to error through misidentification.6 I might, for example, see where I am relative to a particular landmark in front of me and, on that basis, come to form a judgment such as “I am in front of the department store.” It is clear that in forming this judgment I do not identify myself as the person who is in front of the department store. Coming to find out in this way through perception that someone is in front of the department store just is coming to find out that I myself am in front of the department store. Nor is it possible for me to be mistaken about who it is who is in front of the department store. My judgment is immune to error through misidentification relative to the first-person pronoun. Of course, it is possible to imagine cases where I do form the judgment “I am in front of the department store” on the basis of ordinary perceptual awareness and yet that are not immune to error through misidentification. I might, for example, catch a glimpse in a shop window of a person with the department store behind him and then work out that that person is, in fact, me. But these sorts of cases can hardly be taken as standard.
Similar points hold for cases in which one becomes aware of one’s locomotive properties through ordinary perceptual awareness. It is clear that ordinary visual perception can yield not only information about whether one is moving or at rest, but also a fairly accurate indication of one’s speed and trajectory. This type of awareness of one’s physical properties is identification free and immune to error through misidentification relative to the first-person pronoun. I do not normally first discover that someone is moving at a certain speed in a certain direction and then work out that that person is myself—nor is it possible for me to be mistaken about who it is moving at that speed in that direction.
This point needs handling with some delicacy. Someone might object that this type of kinesthetic information does seem prone to errors of misidentification. To take what should be a familiar case, if I am sitting in a train and the train next to mine starts moving then I may well gain the false impression that I myself am moving. Why is this not a case in which I am subject to error through misidentification, given that I mistakenly identify myself as moving when I am not moving? To appreciate what is going on here, it is helpful to recall Wright’s reminder that the grounds for a judgment that is subject to error through misidentification will survive as grounds for a corresponding existential generalization if the identification turns out to be mistaken. Nothing like this occurs in the train case. When I discover that my train is in fact stationary, I cannot then retreat to the claim “Someone is moving, even if I am not.”7 The visual information on which I based my judgment provided an illusion of movement, and when my judgment is defeated so too is the judgment that there is any movement at all.
Although I have been discussing visual perception exclusively, these types of nonintrospective self-awareness also occur in other sensory modalities. As Gareth Evans pointed out, for example, audition is a powerful source of information about one’s location and is also one of the cues we employ to keep track of our movement. But vision is, of course, dominant and has been the most closely studied as a source of self-awareness (particularly by psychologists in the Gibsonian tradition, to be discussed further in section IV).
The important point, for present purposes, is that these cases of direct nonintrospective self-awareness provide further examples of how the particular type of privileged access associated with judgments that are immune to error through misidentification comes apart from the elusiveness thesis. All the judgments we have considered in this section are obviously immune to error through misidentification and equally obviously involve a direct perceptual acquaintance with the embodied self. On the scale running from psychological introspection at one end and ordinary perceptual awareness at the other, they fall somewhere between somatic proprioception and ordinary perceptual awareness. They provide, then, even clearer evidence that the elusiveness thesis cannot provide a satisfying explanation of the privileged access associated with immunity to error through misidentification.
We have seen that there are (at least) three different types of self-awareness possessing the privileged access characterized by immunity to error through misidentification. If one concentrates on the most obvious and familiar of these types—namely introspection of one’s own psychological properties—then it is natural to think that privileged access is a function of the fact that the type of self-awareness involved is not in any way perceptual. This nonperceptual nature of introspective self-awareness is closely linked with the fact that introspection does not involve any sort of encounter with the self as the bearer of these psychological properties, and so one might think that there is a plausible line of explanation running from the elusiveness thesis via the nonperceptual nature of introspective self-awareness to the privileged access associated with the immunity to error through misidentification of introspectively derived psychological self-ascriptions. The discussion in the previous two sections, however, shows that even if this line of explanation works for introspective judgments, there are important classes of epistemically privileged judgments about one’s physical and psychological properties on which it can get no purchase. We need to look further for a full understanding of the type of privileged access associated with immunity to error through misidentification—and different sources of information leading to judgments that are immune to error through misidentification will work in different ways to afford different types of privileged access.
Even in the case of psychological self-ascriptions one might wonder whether the proposed explanation could be fully satisfying. The fact that there is no introspective encounter with the bearer of psychological properties is not a full explanation of the nonperceptual nature of introspective self-awareness and hence, a fortiori, not of the privileged access yielded by such self-awareness. We might flesh out the sense in which psychological self-ascriptions based on introspective self-awareness are nonperceptual in the following way. A subject making such a self-ascriptive judgment moves from ϕ-ing that p to the self-ascriptive judgment “I am ϕ-ing that p.” This transition is based on an awareness of ϕ-ing that p but not on an awareness of himself ϕ-ing that p. The elusiveness thesis explains only a part of this. It explains why there is no awareness of himself ϕ-ing that p, but it does not explain how a subject can form the judgment that he is ϕ-ing that p in the absence of such an awareness.
So how does the subject make the transition? The simple answer (explored at greater length in chapter 3 above) is that the transition is available to subjects who have a basic mastery of what might be termed a simple theory of introspection (by analogy with the simple theory of perception and action discussed in Evans 1982 and Campbell 1994). In this case, the simple theory of introspection amounts to nothing more than some level of mastery of the a priori link between being introspectively aware of a thought and it being the case that one is thinking it—which is, of course, an appreciation of the fact that psychological self-ascriptions are identification free.8
Can we use a similar account to explain self-ascriptions based on somatic proprioception? The most obvious difference between proprioceptively based judgments about one’s physical properties and introspectively based judgments about one’s psychological properties is that it is not even a matter of natural necessity that one can only become aware of one’s own bodily properties through somatic proprioception. It is perfectly compatible with the existing laws of nature that one might be hooked up to someone else in such a way that one receives proprioceptive input from that person’s body as well as from one’s own and hence actually needs to identify which body a given proprioceptive input is coming from. This possibility does not cast doubt on the immunity to error through misidentification of somatic proprioception, given that, as things are, we are not wired up to other bodies and have no reason to think we might be. In fact, it is really the lack of any grounds for thinking that we might be wired up to other bodies that secures immunity to error through misidentification. That is what makes it inconceivable to ask whose body a given input is coming from, and what prevents the grounds for a defeated proprioceptively based self-ascription surviving as the grounds for an existential generalization. Nonetheless, the fact that there is such a possibility does show that this immunity is at best de facto rather than logical, to employ Shoemaker’s distinction.9 And, if the immunity to error is merely de facto then there can be no a priori link between being proprioceptively aware of some bodily property and that property being a property of one’s own body. Nonetheless, this is compatible with there being a simple theory of proprioception based on the de facto link between ownership and the objects of proprioception. One would expect this simple theory to be more complicated than that implicated in ordinary psychological introspection, given the points made earlier about the role of proprioception in underwriting an awareness of one’s body as a bounded physical object uniquely responsive to the will.10
Things get more complicated, however, when we turn to judgments based on the perception-based forms of self-awareness discussed in the previous section. We need a completely different account of the immunity to error through misidentification of these judgments. The proposal I will develop in the remainder of this essay is that their immunity to error through misidentification is a function of the distinctive way in which the self is represented in the content of perception. I will develop this proposal for visual perception, since vision is the dominant modality and has received the most detailed study. It will be recalled from the previous section that two types of visually grounded judgments are immune to error through misidentification in the required sense. The first type includes perceptually based judgments about one’s own location, such as “I am in front of the cathedral.” The second type includes judgments about the nature and direction of one’s movement such as “I am veering toward the central reservation” or “I am walking toward the door.”
In discussing these two types of judgment I will be leaning in particular on the account of vision developed by the perceptual psychologist J. J. Gibson (particularly in Gibson 1966 and 1979). It is in Gibson’s work that we find the most developed account of how the embodied self features in the content of perception as a distinctive object. I shall be arguing that those features of the content of perception to which Gibson draws our attention play a crucial role in underwriting the immunity to error through misidentification of perceptually grounded first-person judgments.
Let me start, though, by getting clear on the nature of the problem. What exactly are we looking for in an account of the immunity to error through misidentification of the relevant class of judgments? Our earlier discussion of introspection and somatic proprioception provides a clue. What we sought there was an account of what warrants the immediate transition from one’s awareness of ϕ-ing to the judgment that one is oneself ϕ-ing. Judgments that are immune to error through misidentification are identification free in an epistemic sense. That is, the warrant for them does not depend on the warrant one might have for identifying a particular person as oneself. Nonetheless, there must be some warrant for what one might term the self-specifying aspect of the judgment. In the case of proprioceptive and introspective judgments, the warrant comes from mastery of the link (a priori in the case of introspection and de facto in the case of proprioception) between being aware of a particular property and that property being one’s own. These links exist because both introspection and proprioception are modes of acquiring information that yield information only about one’s own properties. Clearly, there are no such links in the case of perceptual awareness, since the bodily self is presented in perception as one of a range of physical objects. So what provides the warrant for the self-specifying aspect of the judgment?
The answer lies, I claim, in the distinctive way in which the embodied self appears in the content of visual perception. In the case of self-locating beliefs, the embodied self is fixed as a distinctive object at the origin of the field of view in such a way that perceptually derived information about the position of objects in the world is explicitly relational. The self-specifying aspect of a perceptually based self-locating judgment is warranted because it is grounded in what is given explicitly in the content of perception—and the immunity to error through misidentification derives from the fact that this relational information yielded by vision could not be information about the position of anyone but oneself. The same holds for perceptually based judgments about one’s movement. It is part of the content of perception that one is moving—one can normally tell with considerable accuracy the difference between moving relative to the environment and having the environment move relative to oneself. Again the information is explicitly relational—and relational in a way such that it could not possibly be information about anyone else’s movement.
It is important to realize that the embodied self is the originating point of perception in a much richer sense than simply being the geometrical origin of the field of view. This is so in two respects that Gibson has drawn to our attention. The first is that the embodied self is directly perceived as the boundary of the field of view. The field of view is oval, bounded by the shape of the head. The head serves to block out roughly half of the surrounding environment, but it does this in a way rather different from the way one nonbodily object blocks out another. The boundaries are vague and indefinite. They define the point of observation in a way that cannot in principle be shared by anyone else. Those facial features that occupy permanent structural positions within the field of view enhance this perspectival uniqueness. The nose, for example, is almost always present in visual experience and serves itself to mark an important boundary within the field of view. The nose is the leftmost thing that can be seen by the right eye and the rightmost thing that can be seen by the left eye. Moreover, of all the edges that the observer can see, the nose sweeps across the field of view when the observer moves his head at a faster rate than any other perceived surface. For many of us the eyebrows and cheekbones also occupy a permanent place framing the perceived environment. The second point that Gibson stresses is that even those body parts that do not occupy fixed positions in the field of view are nonetheless perceived in a manner quite unlike the way in which nonbodily physical objects are perceived. The bodily extremities protrude into the field of vision from below in a way that occludes the environment, and yet that differs from the way in which one nonbodily physical object in the field of vision might occlude another. A solid angle is an angle with its base at some perceived object and its apex at the eye. Every object is capable of presenting a range of solid angles, depending on its size and distance from the perceiver. The farther away the object, and/or the smaller it is, the smaller the solid angle. But the solid angles subtended by occluding body parts cannot be reduced below a certain minimum—the minimum corresponding to the maximum distance the body part in question can travel from the point of observation. Perceived body parts are, in Gibson’s provocative phrase, “subjective objects” in the content of visual perception.
We see, then, that the self is presented in perception as an object of a distinctive and peculiar type in virtue of certain features of the content of visual perception. Some of these arise from the features that lead Gibson to describe perceived body parts as subjective objects in the field of view—the fact, for example, that body parts can vary in perceived size only within a limited range. The way in which the body appears as the frame and boundary of the visual array is also extremely important. All of these features serve to create, for each of us, a distinctive and unique visual perspective on the world. It is the fact that self-locating judgments are made on the basis of information derived from such a unique visual perspective that explains their immunity to error through misidentification. The uniqueness of the visual perspective on the world does not just make it the case that one’s perceptually derived information about where one is relative to objects in the perceived environment could not possibly be about anyone but oneself. More importantly, the uniqueness of the visual perspective is itself directly perceived in virtue of those features of the environment we have been discussing and so, by extension, it is visibly manifest to the perceiver that the information about the self gained through vision could not be about any creature except the self. And it is this that warrants the self-specifying aspect of perceptually based judgments that are immune to error through misidentification.
These features that make up the uniqueness of the visual perspective all play a part in underwriting both types of judgment identified in the previous section—judgments about one’s own location relative to perceived objects around one and judgments about one’s own movement relative to the perceived environment. For the second class of judgments, further aspects of the self-specifying information available in visual perception also come into play. Once again, Gibson’s analysis of vision proves helpful. One of Gibson’s major contributions to the study of vision is the proposal to reconstrue the visual field as a constantly moving and constantly reconfiguring set of illuminated surfaces and concomitant solid visual angles, rather than in terms of empty space containing bounded objects (figures on a ground). We do not, he thinks, ever see empty space surrounding discrete objects. What we see is a complex and gapless structure of surfaces. Some of these surfaces are surfaces of objects, while others are not (the various surfaces in the sky, for example). To each surface there corresponds a solid visual angle with its base at the face of the visible surface and its apex at the point of observation. We can, for simplicity’s sake, think of these solid angles as cones (recalling Euclid’s visual cones), although, of course, their shape will vary with the visible outline of the surface in question. As the observer moves through the environment the solid angles change, as one surface moves in front of another (relative to the perceiver) or as the observer approaches or moves away from the surface. This is what Gibson terms optic flow.
The particular pattern of changes in the optic flow specifies the perceiver’s trajectory through the environment. What allows the perceiver to extract a meaningful indication of his trajectory from the changing structure of the optic flow is that there is an underlying invariant structure to the optic flow. For present purposes, the crucial point is that the optic flow contains a stationary point determined by the movement of the observer. The optical flow in any field of vision starts from a center that is itself stationary, and this stationary center specifies the point that is being approached, when the perceiver is moving. The aiming point of locomotion is at the vanishing point of optical flow. This means that the entire structure of the field of view is determined by the direction of the perceiver’s movement. And this creates an effect that is an almost exact counterpart of the effect created by the uniqueness of the visual perspective. The field of view is centered on the observer in the ways discussed earlier (and also in the sense that the point of observation marks the apex of all the solid angles in the field of view). But it is also centered on the aiming point of locomotion, for that is where the movement of the optic flow begins. This is what underwrites the self-specifying aspect of judgments about one’s own movement. Once again, it is manifest to the perceiver that visually derived information about movement relative to the perceived environment could not be information about anyone else’s movement relative to the perceived environment—for the very structure of the optic array is given in terms of the perceiver’s own direction of movement.
This essay began by proposing an attractively simple picture of the relation between the elusiveness thesis, immunity to error through misidentification and privileged access. The phenomenon of immunity to error through misidentification relative to the first-person pronoun is characteristic of the most fundamental sources of information providing privileged access to our own psychological and physical properties. Moreover, it seems to be more basic than the other proposed forms of privileged access, in the sense that it is presupposed by them but does not presuppose them. When one concentrates on introspective access to one’s own psychological properties, it is natural to think that immunity to error through misidentification is a function of the fact that the self cannot be an object of introspective awareness; that is, the elusiveness thesis seems to explain why introspection does not involve possibilities of error attendant upon ordinary perceptual awareness.
The burden of this essay, however, has been to show that this cannot be the whole story. The phenomenon of immunity to error through misidentification relative to the first-person pronoun extends far more widely than introspective awareness of our own psychological properties. Judgments that are immune to error through misidentification relative to the first-person pronoun can be made on the basis of somatic proprioception, as well as on the basis of the self-awareness provided in and through ordinary perceptual awareness. Here the elusiveness thesis does not apply. A different explanation is required, therefore, for the warrant attaching to judgments based on these sources of information—and, in particular, for the self-specifying aspect of those judgments corresponding to the immunity to error through misidentification of the information source on which they are based. Proposals for developing such accounts were offered in section IV. The upshot of the discussion is that, although immunity to error through misidentification is indeed an important type of privileged access, it does itself require explanation, and, moreover, there will be different explanations for the different sources of information at stake. The domain of privileged access is both wider and more complex than it is often taken to be.