There are many distinct layers of self-consciousness. Obvious examples are the capacity to think of one’s body as one’s own; to recognize oneself as the bearer of mental states; to master the grammar of the first-person pronoun; to view oneself an one object in the world among others, or as one person in the world among others; to have memories about one’s past self; to construct autobiographical narratives; to formulate long-term plans and ambitions. Whichever one of these one is considering, however, it is tempting to think of it as somehow parasitic on a more primitive and already-existing form of self-awareness. So, for example, it seems intuitively hard to imagine that one could formulate long-term plans without having some sort of autobiographical narrative at one’s disposal or, indeed, that one could formulate such an autobiographical narrative without having a stock of memories about one’s past self. If this is so, however, and if a regress is to be avoided, then it seems plausible to suppose that all these layers must eventually be grounded in a form of self-awareness primitive enough not to depend on a more basic self-awareness.
One of the attractions of the Gibsonian concept of ecological perception is that it seems to provide us with a basic level of self-awareness that could serve as a core for such comprehensive accounts of the phenomenon (or phenomena) of self-consciousness. On the ecological understanding of perception, a form of sensitivity to self-specifying information is built into the very structure of perception from the earliest stages of infancy in such a way that, as Gibson famously put it, all perception involves coperception of the self and the environment. In this essay I propose to explore the implications of this suggestion. I will argue that by starting with ecological perception and seeing how it needs to be built up, we can reach an understanding of the features that a basic form of self-awareness must incorporate.
For many psychologists and philosophers the five senses are directed “outward”—they are exteroceptive or exterosensitive, designed to inform us about objects and events in the world. They can, of course, be turned on oneself, as, for example, when one looks at oneself through a mirror, but doing this provides a distinct sort of information about oneself, information that objectifies the body, failing to do justice to the sense in which the subject of perception is also the object of perception. This objectifying form of perceiving oneself is often contrasted with the form of self-perception from within, gained through what has been termed a “body sense.” Receptors in the skin, muscles, tendons, and joints, operating in conjunction with the vestibular system, yield proprioceptive information about bodily position and movement that is crucial in orienting and acting within the world. This has led to a firm distinction in both operation and function, with the five exteroceptive senses deemed to provide information about the external world, while the proprioceptive system provides information about the self, in particular about bodily posture and movement.
It is instructive to view the Gibsonian theory of ecological perception as challenging precisely such a strict distinction between proprioceptive and exteroceptive senses (see also Neisser 1988, 1991). Gibson claims that the five ostensibly outward-directed senses provide both exteroceptive and proprioceptive information, rejecting the traditional division of labor between the five exteroceptive senses and proprioceptive body sense.
A deep theoretical muddle is connected with proprioception. … In my view, proprioception can be understood as egoreception, as sensitivity to the self, not as one special channel of sensations or as several of them. I maintain that all the perceptual systems are propriosensitive as well as exterosensitive, for they all provide information in their various ways about the observer’s activities. … Information that is specific to the self is picked up as such, no matter what sensory nerve is delivering impulses to the brain. (Gibson 1979, 115)
Although propriospecific and exterospecific information are distinct types of information, they are simultaneously available to each sense, rather than each being the province of a distinct sensory system.
Perhaps the most basic form of propriospecific information arises through the structure imposed upon the visual array by the perceiver’s body. As Gibson stresses, every animal has a field of view that is bounded by its body, and the particular way in which each animal’s body blocks out aspects of its environment is unique to that animal. “Ask yourself what it is you see hiding the surroundings as you look out upon the world—not darkness, surely, not air, not nothing, but the ego” (Gibson 1979, 112). In this limited but important sense the self is actually present in visual perception, as the frame of the field of view, as what surrounds and gives it structure.
A second type of propriospecific information emerges through what Gibson terms visual kinesthesis. This is how he thinks the visual system solves the fundamental challenge created by the subject’s movement through the environment. How can the massively changing visual inputs generated by movement be parsed and interpreted so that subjects perceive that it is they themselves who are moving through a broadly stationary environment? How can the visual system separate out and compensate for those changes in visual information that are due to the subject’s own movement? Gibson’s basic idea is that the visual system solves this problem through sensitivity to patterns in the optical array, and in particular the interplay between flow in the array (specifying movement) and nonflow (specifying rest). Invariant features of the optical array specify, for example, the aiming point of locomotion, which is the stationary point from which the optical flow originates. Relative to those invariants, the optical flow can indicate, for example, the speed with which a goal is being reached—think of the base of a solid angle at an object as an invariant, and the rate at which that angle increases as an indication of speed. This interplay between flow and nonflow can be manipulated, as in the moving-room experiments where subjects are placed on the solid floors of rooms with independently moveable walls (Lishman and Lee 1973; Lee and Aronson 1974). Young children sway and lose their balance as they try to compensate for their own apparent movement, when the walls are moved in the saggital plane.
The third relevant form of self-specifying information available in the environment is encapsulated in the Gibsonian notion of affordances: “At any given moment the environment affords a host of possibilities: I could grasp that object, sit on that chair, walk through that door. These are examples of affordances: relations of possibility between actors and environments. It is affordances that animals most need to see: here is prey that I might eat, a predator who might possibly eat me, a tree I might climb to escape him” (Neisser 1991, 201). The claim here is that the perception of affordances is relativized to the perceiving subject, so that, for example, in looking at a window one perceives not just an aperture but an aperture that presents the possibility of one’s looking through it. The ecological suggestion is that the perception of affordances is partly a mode of self-perception. Furthermore, it is such constitutively. The whole notion of an affordance rests on relating environmental information to one’s own possibilities for action and reaction.
Ecological self-perception falls a long way short of full-fledged self-consciousness. There is a crucial difference between having information about oneself as part of one’s ecological experience and being fully self-conscious, where full-fledged self-consciousness is taken to involve the capacity to entertain “I” thoughts or to maintain some form of detached perspective on oneself. Neither of these two sophisticated capacities is required for ecological self-perception to take place. What the ecological approach might more plausibly be argued to provide is a way of understanding the notion of a point of view on the world. The most developed account of this notion has been offered by Strawson in his discussion of Kant’s Transcendental Deduction in The Bounds of Sense, where he suggests it as a necessary but not sufficient strand in a complete account of self-consciousness. I will consider his position in some detail because it sets the framework for this essay.
Strawson argues that possession of a point of view rests on the possibility of a subject’s experiences being such “as to determine a distinction between the subjective route of his experiences and the objective world through which it is a route” (Strawson 1966, 104). That a subject’s experiences contain such a distinction over time is what yields the subject’s point of view on the world:
A series of experiences satisfying the Kantian provision has a certain double aspect. On the one hand it cumulatively builds up a picture of the world in which objects and happenings (with their particular characteristics) are presented as possessing an objective order, an order which is logically independent of any particular experiential route through the world. On the other hand it possesses its own order as a series of experiences of objects. If we thought of such a series of experiences as continuously articulated in a series of detailed judgments, then, taking their order and content together, those judgments would be such as to yield, on the one hand, a (potential) description of an objective world and on the other the chart of the course of a single subjective experience of that world. (Strawson 1966, 105–106)
Strawson’s conception of a point of view is intended to draw together two distinct (sets of) conceptual capacities: the capacity for the self-ascription of experiences and the capacity to grasp the objectivity of the world. Why do they need to be brought together?
Strawson is considering the question, What conditions have to be fulfilled for experience to be possible? He approaches it through the hypothesis that there might be an experience whose objects were sense data, “red, round patches, brown oblongs, flashes, whistles, tickling sensations, smells” (1966, 99). This would not count as anything recognizable as experience, he maintains, because it would not permit any distinction to be drawn between a subject’s experiences and the objects of which they are experiences. In such a case, the esse of the putative “objects of experience” would be their percipi. There would be no distinction between the order of experiences and the order of the objects of experience, and this, according to Strawson, effectively means that we cannot talk either about objects of experience or about a subject of experience, and hence we cannot talk about experience at all.
The general idea is that no creature can count as a subject of experience unless it is capable of drawing certain very basic distinctions. What is important about a point of view, as Strawson conceives it, is that experience that reflects a temporally extended point of view on the world will ipso facto permit those basic distinctions to be drawn, and this is so because experience reflecting a point of view has the double aspect outlined in the passage quoted earlier. Experience reflecting a point of view just is experience that permits the right sort of distinctions to be drawn between a subject’s experiences and the objects of which they are experiences.
Suppose we grant that such a purely sense-datum “experience” is impossible. And suppose we also grant that experience with the sort of double aspect that he describes qualifies as genuine experience by making available the sorts of distinctions that could not be drawn in a purely sense-datum experience. This still leaves open an important question. Are the capacity to ascribe experiences to oneself and the capacity to grasp the objectivity of the world necessary conditions of any possible experience that has the dual structure impossible in a putative sense-datum experience?
One reason for thinking that these conceptual capacities might not be necessary conditions is provided by research on object representation in infancy. Much recent work in developmental psychology strongly suggests that very young infants are capable of primitive forms of object representation which involve, for example, the capacity to perceive object unity and to employ certain basic principles of physical reasoning, such as the principle that objects move on single connected paths (Spelke 1990; Spelke and Van de Walle 1993). Such a level of object representation does not demand a conceptual grasp of causality or of the connectedness of space, and hence does not involve possessing the sort of comprehension of the objectivity of the world built into Strawson’s notion of a point of view. Nor, of course, does it involve having a theoretical grip on objects and the principles of naive physics.
It would be rash to suggest that these experiments alone provide evidence that the infants have anything like a point of view, but they at least suggest the need to formulate the basic distinction at the heart of the notion of a point of view in such a way that it does not come out as a matter of definition that the infants cannot have experience that reflects a point of view. Just as there are ways of representing objects that do not require mastery of the relevant concepts, might there not be ways in which a creature’s experience could incorporate the basic distinction at the heart of the notion of a point of view without a conceptual grasp of the distinction? In this essay I would like to suggest that there are indeed such ways. Before going any further, however, I need a way of understanding that basic distinction that at least leaves this possibility open.
If we formulate the central distinction as one between subjective experience and what that subjective experience is experience of, we capture what seems to be the crucial feature missing in the hypothesized purely sense-datum experience, without immediately demanding any relevant conceptual capacities on the part of the subject. Using this as the central notion, we can reformulate the original characterization: having a temporally extended point of view on the world involves taking a particular route through space-time in such a way that one’s perception of the world is informed by an awareness that one is taking such a route, where such an awareness requires being able to distinguish over time between subjective experience and what it is experience of.
The matter can be further clarified by adverting to the concept of a nonsolipsistic consciousness, which Strawson introduces in Individuals. There he refers to “the consciousness of a being who has a use for the distinction between himself and his states on the one hand, and something not himself or a state of himself, of which he has experience, on the other” (1959, 69). We can see the notion of a point of view as fleshing out what experience must be like for any creature that is to count as nonsolipsistic in this sense. Putting it like this brings out why one might expect the notion of a point of view to play a foundational role in a comprehensive account of self-consciousness. The thought would be that being a nonsolipsistic consciousness is the most basic form of self-awareness, and since the notion of a point of view is put forward to capture what experience must be like to support such a nonsolipsistic consciousness, it would seem that here, if anywhere, we have the sort of primitive form of self-awareness that would anchor an account of self-consciousness as a whole.
Because this new formulation is trying to sidestep any demand for sophisticated concept mastery, we can term it the nonconceptual point of view. The notion of the nonconceptual at work here can be elucidated with reference to current work on representational content. It has been suggested by various writers on the philosophy of content that it is theoretically legitimate to refer to mental states that represent the world but do not require the bearer of those mental states to possess the concepts required to give a correct specification of the way in which they represent the world (Cussins 1990; Peacocke 1992). The most plausible candidates for such states are perceptual states and subpersonal computational states. For present purposes, only the former is relevant. The notion of nonconceptual content as applied to perceptual states is a reaction to the idea dominating much philosophy of perception that all perceptual experience is structured by the concepts possessed by the perceiver. Now, as formulated, the notion of nonconceptual content is neutral on the question of whether the bearer of the appropriate states possesses any concepts at all, because it is nonconceptual in virtue of the fact that the bearer is not required to possess the concepts involved in specifying it. So a point of view will be nonconceptual just in case a creature can be ascribed such a point of view without it being ipso facto necessary to ascribe to it mastery of the concepts required to specify the way in which its experience reflects a point of view. Nonetheless, it is important to distinguish between a conception of autonomous nonconceptual content, which does not require that a creature to which it is attributed possess any concepts at all, and a conception of nonconceptual content, which denies that a creature possessing no concepts at all can be in contentful states (Peacocke 1992, 90; Bermúdez 1994). I am interested in the former.
Of course, the possibility of formulating the notion of a point of view so that no relevant conceptual requirements are built into it does not mean that it is possible to have a point of view in the absence of those concepts. It is not ruled out by definition any longer, but there might be other reasons to rule it out. Nonetheless, on this weaker formulation of the notion of a nonconceptual point of view, it seems promising to elucidate it through the idea of ecological self-perception. No special argument is needed to show that it is possible to have a nonconceptual point of view, it might be suggested, because such a nonconceptual point of view is built into the very structure of perception. The propriospecific information involved in all exteroception seems to be information about the spatiotemporal route that one is taking through the world, as is particularly apparent in visual kinesthesis. One has, it might seem, a continuous awareness of oneself taking a particular route through the world that does not require the exercise of any conceptual abilities, in virtue of having a constant flow of information about oneself qua physical object moving through the world. On this view, ecological coperception of self and environment is all that is needed for experience with a nonconceptual point of view.
Before evaluating this possibility, it is worth being more explicit about why a point of view has been described as temporally extended. The reason is that a creature whose experience takes place completely within a continuous present (i.e., who lacks any sense of past or future) will not be capable of drawing the fundamental nonsolipsistic distinction between its experience and what it is experience of. We can bring this out by reflecting that being able to make this distinction rests on an awareness that what is being experienced exists independently of any particular experience of it. Such a grasp of independent existence itself involves an understanding that what is being experienced at the moment either has existed in the past, or will exist in the future, that what is being experienced at the present moment has an existence transcending the present moment. By definition, however, a creature that experiences only a continuous present cannot have any such understanding.
The important question, therefore, seems to be, What form must experience take if it is to incorporate an awareness that what is being experienced does not exist only when it is being experienced? Alternatively put, what must the temporal form of any such experience be? Clearly, such an awareness would be incorporated in the experience of any creature that had a grasp of the basic temporal concepts of past, present, and future, but we are looking for something at a more primitive level. What I would suggest is that certain basic recognitional capacities offer the right sort of escape from the continuous present without demanding conceptual mastery. Consider the act of recognizing a particular object. Because such an act involves drawing a connection between one’s current experience of an object and a previous experience of it, it brings with it an awareness that what is being experienced has an existence transcending the present moment.
But it is important to specify what it is that the recognitional capacities are being exercised on. A creature could recognize an experience as one that it has previously experienced without any grasp of the distinction between experience and what it is experience of. Clearly, the recognitional capacities must be exercised on something extraneous to the experiences themselves. But what? Physical objects are an obvious candidate, and any creature that could recognize physical objects would have experience that involved drawing the right sort of distinctions. However, one might wonder whether the distinctions could be drawn at a level of experience that does not involve objects. The answer one gives here depends on one’s ontological position on the issue of whether there are “things” that can be reidentified and recognized but are not physical objects. Many philosophers would deny this (including Strawson 1959, chap. 1). The debate is too tortuous to go into here, and I merely state my position without attempting to defend it. There are indeed “things” that can be reidentified and recognized and are neither physical objects nor require experience of physical objects for their recognition. These “things” are places, which can be recognized in terms of distinctive features holding at those places even by a creature that has no grasp on the notion of a physical object (Campbell 1993).
If this is right, we seem in a position to argue that any creature being ascribed a point of view must be capable of exercising the basic capacity to recognize places. If a creature can recognize a particular place, we have a nonsolipsistic consciousness, because we have an object of experience that is grasped as existing independently of a particular experience of it. By the same token, in the absence of such a recognitional capacity (and assuming that there is no grasp of basic temporal concepts nor a capacity to recognize physical objects), it does not seem appropriate to speak of a point of view in the sense under discussion. This yields a strong sense in which a point of view is temporally extended. It is temporally extended not just in that it must extend over time but also in that it must involve the use of memory and recognition to register the passing of time.
It is useful at this point to make a distinction between two types of memory. There are, on the one hand, instances of memory in which past experiences influence present experience, but without any sense on the part of the subject of having had the relevant past experiences, and, on the other, instances in which past experiences not only influence present experience but also the subject is in some sense aware of having had those past experiences. In the former case what licenses talk of memory is the fact that a subject (or an animal, or even a plant) can respond differentially to a stimulus as a function of prior exposure to that stimulus or to similar stimuli. This is not to deny that such memory can be extremely complex. Quite the contrary, such memory seems to be central to the acquisition of any skill, even the most developed. The differential response does not have to be simple or repetitive. Nonetheless, we can draw a contrast between memory at this level and the various forms of memory that do involve an awareness of having had the relevant past experiences, as, for example, when a memory image comes into one’s mind or one successfully recalls what one did the previous day. Clearly, there are many levels of such memory (which I shall term “conscious memory”), of which autobiographical memory is probably the most sophisticated. One thing that these forms of conscious memory all have in common, however, is that previous experience is consciously registered, rather than unconsciously influencing present experience. The distinction is a crude one, precisely because the term “conscious memory” is so hazy, but it will be sharpened up below.
It follows from our discussion of the notion of a point of view that the place recognition it requires must involve some conscious registration of having been there. Reflect on the case of a creature, perhaps a swallow, that is perfectly capable of performing complicated feats of navigation that involve finding its way back to its nest or back to the warmer climes where it spends the winter but nonetheless does not in any sense consciously recognize the places that it repeatedly encounters. This is not a case, I think, in which one would want to claim that the creature has the appropriate awareness of the route that it is taking through space-time, although it is sensitive to certain facts about that route, because those facts clearly determine behavior (facts, for example, about how to get from one place to another and then back to where it started from). Yet if the creature is credited with exactly the same behavior, only this time accompanied by some form of conscious recognition of the relevant places, the situation seems fundamentally different. Insofar as it recognizes a place, it is aware of having been there before, and insofar as it recognizes having been at a place before, it has the beginning of an awareness of movement through space over time. It is emerging from a continuous present and moving toward possession of a temporally extended point of view.
This enables us to evaluate the suggestion that an ecological analysis of perception shows that a temporally extended point of view is built into the structure of perception. Does ecological self-perception involve anything like conscious memory? If it does, then we will be in a stronger position to claim that the two notions are very intimately connected (although, as stressed above, conscious place recognition is a necessary rather than sufficient condition of possessing a temporally extended point of view).
On the ecological view, perception is fundamentally a process of extracting and abstracting invariants from the flowing optical array. Organisms perceive an environment that has both persisting surfaces and changing surfaces, and the interplay between them allows the organism to pick up the sort of information that specifies, for example, visual kinesthesis. The key to how that information is picked up is the idea of direct perception. The mistake made by existing theories of perception, according to Gibson, is construing the process of perception in terms of a hierarchical processing of sensory inputs, with various cognitive processes employed to organize and categorize sensations. A crucial element of this serial processing is bringing memories to bear on present experience. Gibson rejects this. Accepting that present experience is partly a function of past experience, he firmly denies that this sensitivity to past experience is generated by processing memories and sensations together. His alternative account rests on the idea that the senses, as perceptual systems, become more sensitive over time to particular forms of information as a function of prior exposure. Although Gibson was rather polemical about what he termed “the muddle of memory,” it would seem that his account involves the first notion of memory discussed above, namely a differential response to stimuli as a function of past experience. Gibson’s position seems to be that conscious recognition is not implicated in ecological perception, although it might or might not develop out of such ecological perception. It is perfectly possible for a creature to have experience at the ecological level without any conscious recognitional capacities at all. If, then, the exercise of a capacity for conscious place recognition is a necessary condition of having experience that involves a temporally extended point of view, it seems that the dual structure of experience involved in the ecological coperception of self and environment must be significantly enriched before yielding a point of view.
Of course, this does not mean that we should abandon the basic idea of ecological perception; rather, it means that the materials offered by Gibson’s own account need to be supplemented if they are to be employed in the theoretical project under discussion, and it is perfectly possible that Gibson’s concepts of information pickup and direct resonance to information in the ambient environment will have a crucial role to play in such an extension of the basic way in which ecological perception is sensitive to past experience (as they are, for example, in the account of perception and memory developed in Neisser 1976). The point is that, as it stands, the Gibsonian account cannot do all the work it was earlier suggested it might be able to do. The suggestion was that the Gibsonian account of ecological perception could show how something like a nonconceptual point of view is reflected in the very structure of perception. It now seems, however, that this will not be achieved until the appropriate capacity for conscious place recognition is added to the ecological coperception of self and environment. In the next section I will further discuss the constraints that this imposes upon the notion of a point of view.
The discussion in the previous section stressed the importance of place recognition in yielding a sense of objects of experience existing independently of their being experienced. But surely, it might be argued, place recognition involves more than this. A creature recognizing a particular place is aware not only that that place has existed in the past but also that it itself has been there before. How, it might be asked, could it have the former without the latter, since the capacity for place recognition seems to rely on both a sense of the transtemporal identity of places and a sense of the transtemporal identity of the self.
This intuition has been emphatically endorsed by Christopher Peacocke, who makes the stronger claim that any genuine attribution to a creature of a capacity for place recognition (which he terms place reidentification) entails that the creature have mastery of the first-person concept. I have taken the liberty of schematizing the rather condensed discussion within which this claim emerges (1992, 90–92):
Peacocke’s reason for maintaining (3) and (4) is that the existence of navigational abilities—however sophisticated, systematic, or structured—is not sufficient to compel the ascription of states with genuine spatial content. Simply being able to find one’s way from one place to another is not enough. What is needed is some form of grasp of having been there before, and Peacocke places a strong requirement on any such grasp. The possibility of reidentifying places, he argues, requires the capacity to entertain thoughts or protothoughts of the form “I have been to this place before,” and such thoughts could not be entertained by a creature lacking the first-person concept. The capacity for such thoughts goes hand in hand, for Peacocke, with the capacity to engage in spatial reasoning, where this “requires the subject to be able to integrate the representational contents of his successive perceptions into an integrated representation of the world around him, both near and far, past and present” (1992, 91). The thought seems to be that reidentifying places in the appropriate manner involves representing those places as existing unperceived within a spatial framework also constructed so that it is independent of any particular perception. The construction of this spatial framework implicates a rudimentary form of first-person thought, because it involves the subject’s not only representing its own location within the framework but also grasping that the location can change over time. It involves “the subject’s appreciating that the scene currently presented in his perception is something to which his own spatial relations can vary over time” (1992, 90).
This poses a serious threat to my argument. My account of what it is to possess a temporally extended point of view on the world has, as a necessary condition, the capacity for conscious place recognition. If Peacocke’s argument is sound, then any creature possessing this capacity will possess a primitive form of the first-person concept, and this creates serious difficulties for the suggestion that an account of the notion of a point of view can be given at the nonconceptual level. Evidently, if experience reflecting a point of view involves possessing a form of the first-person concept, albeit a primitive one, then it cannot count as a form of autonomous nonconceptual content in the sense discussed earlier, because any creature with such experience will have to possess at least some concepts. But a stronger conclusion also follows: that such experience cannot be a form of nonconceptual content at all. This is so because the key distinction that the notion of a point of view is intended to capture is that between subjective experience and what it is experience of, and the concept of the first person will be involved in specifying how this distinction is manifest within the experience of any particular creature. Indeed, precisely this constraint characterizes the notion of a point of view: that it can be specified only in the first person.
One way of resisting this conclusion would be to deny the strong conditions that Peacocke places on the capacity for place reidentification. In particular, one might query the suggestion that place reidentification is only available to creatures who have the capacity to construct suitably integrated representations of their environment and to engage in spatial reasoning in a way that necessarily involves grasp of the first-person pronoun. This could be done, for example, by appealing to the notion of causally indexical comprehension developed by John Campbell (1993). Grasping a causally indexical notion is just grasping its implications for one’s own actions. Examples of such causally indexical notions are that something is too heavy for me to lift, or that something else is within reach. As he points out, grasp of causally indexical notions may be linked with a reflective understanding of one’s own capacities and of the relevant properties of the object. But, on the other hand, it need not be so linked: it makes sense to ascribe to creatures a grasp of such causally indexical notions without ascribing to them any grasp of notions that are not causally indexical (even though those noncausally indexical notions might be essential to characterize the causally indexical notions) because the significance of such notions is exhausted by their implications for perception and action. And as such, causally indexical mental states qualify as states with nonconceptual content.
In this context, then, one might attempt to defend the idea of a nonconceptual point of view by suggesting that the spatial abilities involved are causally indexical. Such an objection would maintain that we need not place the theoretical weight that Peacocke does on disengaged reflective and reasoning abilities. Rather, we should be looking at the way in which a creature interacts with its surroundings, because this will be how it manifests its grasp of the spatial properties of its environment (of the connectedness of places, for example). On such a view, a creature could possess an integrated representation of its environment in the absence of any capacity to reflect upon its interactions with its environment. If this line is pursued, then it seems to provide a way in which we can retain the idea of place reidentification without accepting Peacocke’s claim that it implicates possession of a primitive form of the first-person concept—precisely because it denies Peacocke’s central claim, that any creature capable of place reidentification must be capable of explicitly representing itself and its location in its surroundings.
The trouble with this suggestion, however, is that there is a strong tension between the idea of causally indexical content and my earlier insistence on the importance of conscious recognitional abilities. The idea of an awareness that is exhausted in its implications for perception and action does not mesh well with the account I have been developing of the essential features of experience reflecting a point of view. On the causally indexical conception of place reidentification, we ascribe to a creature a grasp of the spatial properties of its environment because its behavior is suitably complex and the relevant connections between perception and action seem to hold. It would seem, however, that appropriately complex behavior could exist at the primitive level of nonconscious memory and skill acquisition. On the causal indexicality account, the ability to track places and perform complicated feats of navigation of the sort regularly carried out by homing pigeons and migrating swallows would be good grounds for ascribing the relevant grasp of the spatial properties of the environment, but it does not implicate the sort of recognitional sense of the transtemporal identity of places that has been argued to be a crucial element in the notion of a point of view.
There seems, then, to be the following challenge for the idea of a nonconceptual point of view as developed in the first part of this essay. To keep the connection with conscious experience of the transtemporal identity of places that, I have argued, is required to distinguish a genuine point of view from the sort of capacity to distinguish proprioceptive and exteroceptive invariants evidenced in ecological perception, we need to reject the causally indexical account of place reidentification. In doing this we move toward understanding the representation of places in a way that is explicit and (relatively) disengaged, rather than exhausted in its implications for action. The question is whether this can be done without moving the notion of a point of view out of the realm of nonconceptual content altogether, as Peacocke does.
It would seem that the only way out here would be to resist Peacocke’s suggestion that a primitive form of the first-person concept is involved in the type of spatial reasoning involved in place reidentification and recognition. Could there not be a way of representing the self that does not count as fully conceptual but nonetheless enables the subject to engage in basic reasoning about places? Toward the end of this essay I will suggest that this is indeed a possible direction. Before doing so, however, it is important to make clear that there are theoretical reasons driving such a move that are independent of the attractiveness of the notion of a nonconceptual point of view.
An initial worry that one might have with implicating the first-person concept in spatial reasoning is that doing so seems to link spatial reasoning with unwarrantedly sophisticated conceptual abilities. Matters can be focused here by asking whether a creature’s mastery of the first-person concept involves the capacity to ascribe to itself psychological predicates. If mastery of the first-person concept does involve such a capacity, then one seems committed to the idea that spatial reasoning is available only to creatures capable of conceiving of other subjects of thought and experience, for the following reason. Suppose we take the ability to generalize as the mark of a genuine concept user, in the way suggested by Gareth Evans’s Generality Constraint, so that it is a condition on a creature’s being credited with the thought a is F, and hence of possessing the concepts of a and F, that it be capable of thinking a is G for any property G of which it has a conception, and of thinking b is F for any object b of which it has a conception (Evans 1982, 100–105). Now if a creature’s conceptual repertoire contains psychological predicates, it will have to be capable of generalizing them in the appropriate manner, and for obvious reasons, such generalization can take place only over other psychological subjects. So, on the assumption that mastery of the first-person concept requires the capacity to generalize psychological predicates, spatial reasoning is available only to subjects who have a relatively sophisticated grasp of folk psychology.
Nobody (I think) would want to maintain this. So a position like Peacocke’s clearly requires a form of the first-person concept that does not implicate the capacity for the self-ascription of psychological predicates, and indeed, he stresses that it is supposed to be primitive. We could think of it as involving only the capacity to apply certain very basic temporal, spatial, and relational predicates, and hence as requiring only a very limited conceptual repertoire. Certainly, this would completely avoid the difficulty raised in the previous paragraph. One might, however, have doubts about such a proposal. One very general worry here would concern the ontogeny of self-consciousness. The proposed primitive form of the first-person concept is clearly intended to be ontogenetically primitive. That is, it will be the key to understanding an early, if not the earliest, form of self-conscious thought. But in employing such a concept, a creature would be thinking of itself only as a subject of spatial, relational, and temporal properties. It would be thinking of itself just as a physical object. The ontogenetic story to be told here would then be one in which the higher forms of self-consciousness emerge from such a restricted but nonetheless detached mode of presentation of the self. And the worry that one might have would be that this gets the ontogenetic order the wrong way round: the capacity to think of oneself as a physical object (thought in which one features as one object among others) does not emerge until relatively late in the developmental process.
A satisfactory development of this line of argument awaits a satisfactory account of the ontogenesis of self-conscious thought. But someone tempted by it might find additional support in reflecting on Gareth Evans’s account of how first-person thought meets the Generality Constraint. According to Evans, the Generality Constraint can only be properly met (and hence one can only speak of genuine mastery of the first-person concept) when the subject is capable of conceiving of himself in an impersonal manner as an element of the objective order (1982, chap. 7, particularly 208–210). Clearly, this involves an even more detached perspective on the self than that discussed in the previous paragraph, and one may well baulk at making this a condition on the capacity for genuine place reidentification. Of course, a defender of Peacocke’s position does not have to accept Evans’s account of how the Generality Constraint is met for first-person thought, but if he does not accept it, it is incumbent on him either to give an alternative account or to explain why it is not applicable at the putatively primitive level under discussion.
What this brings out, I think, is a significant tension in Peacocke’s suggestion that a primitive form of the first-person concept is involved in basic spatial reasoning. It seems inappropriate to claim that the constraints and conditions operative in the case of full-fledged first-person thought are operative in this primitive case. But on the other hand, there is a danger of stripping away so many of the trappings of full-fledged first-person thought that it is no longer clear what the force is of claiming that we are dealing with a form of the first-person concept at all. The bottom line of Peacocke’s position seems to be that the self has to be explicitly represented for genuine spatial reasoning to take place. It is, to my mind at least, a moot point whether this forces us to conclude that a conceptual grasp of the self has to be present. And certainly the matter will remain undecided until we have a clearer account of spatial reasoning. In the remainder of this essay I would like to make some preliminary moves toward the suggestion that the self can be explicitly grasped in a manner that is not conceptual.
As a first step in this direction, we need suitable criteria for determining, first, whether something is being explicitly represented and, second, whether such an explicit representation is conceptual. A preliminary suggestion on the first issue would be that a creature can only be described as explicitly representing something when, in explaining that creature’s behavior, we need to go beyond stimulus–response (S–R) psychology. The sort of explicit representation relevant here comes into play when S–R explanations cease to be applicable. Of course, appeal to mental representations is crucial to S–R psychology (see, e.g., the discussion of the associative-cybernetic model in Dickinson and Balleine 1993), but the type of mental representations I am interested in are those that feature in (proto)intentional accounts of behavior. Clearly, this is not a sufficient condition, but it at least gives us something to work with (cf. Peacocke 1983, chap. 3). On the second issue, that of what qualifies a mental representation as conceptual, I propose that we stick to some form of the Generality Constraint, as discussed earlier. On this criterion, a representation counts as conceptual if it can be combined with any other representations the subject possesses.
Prima facie, it seems that the first of these criteria could be satisfied without the second being satisfied, that is, that we might need to appeal to an explicit representation in a protointentional account of behavior, even though that representation does not support the appropriate form of generalization. If this were the case for the first person, then we would have an explicit but nonconceptual grasp of the self. In the remainder of this essay I would like to discuss a set of experiments carried out by Watson and Ramey (1987) on 3-month-old infants that seem to offer an example of just such a way of grasping the self.
The experiments examine the responses of young infants only 3-months-old to the movements of a mobile suspended above their cribs. The mobile could move in two different ways. In the first, a control situation, it could be rotated electrically by the experimenter (in this case the infant’s mother). In the second, the experimental situation, it was set up with a pillow sensitive to pressure in such a way that it would move when the infant moved its head. For a further control group, the mobile was set up not to move at all. The experimental apparatus was set up in the homes of the 48 infants involved, and the infants were exposed to it for 10 minutes a day over 14 consecutive days. The number of pillow activations in each 10-minute period were counted to see if it increased over the 14 days of exposure. Significant increase was found in the experimental group but not in either of the control groups.
The most common way of explaining these experiments is as a form of instrumental conditioning. All that is shown is the gradual development of S–R links between head movements and the movements of the display, so that whenever the infant sees the display, it moves its head. The S–R links develop because the infants enjoy watching the display move. Such an explanation of their behavior need not appeal to any explicit mental representations at all. In this respect the explanation suggests that the experimental behavior is to be explained at the same level as the ecological behavior discussed in the previous section. I do not want to claim that this interpretation is incorrect. What I would like to do, though, is give one reason for thinking that it might not tell the whole story and, on the basis of this, sketch out an alternative interpretation. This alternative interpretation involves attributing to the experimental infants an explicit representation of the self that is nonetheless not conceptual.
The mothers of the experimental infants almost universally reported that their children took great interest and pleasure in the movement of the display. They smiled, cooed, and laughed a lot, fixating intensely on the mobile, with this behavior developing after only a few days’ exposure to the apparatus. In contrast, those infants in the first control group, for whom the mobile moved according to a regular pattern, rather than in response to the infants’ own movements, showed considerably less interest in the mobile. If, as the S–R account suggests, what drives the reinforcement process is the pleasure taken in watching the mobile move, then it fails to account for the discrepancy between the degrees of pleasure shown by the experimental infants and the first set of control infants. Its prediction would presumably be that both sets of infants should display substantially similar behavior, because they both watch the mobile move and watching the mobile move is an interesting and pleasurable experience—just as for rats, eating food pellets is an interesting and pleasurable experience even when the food pellets arrive without the rat having had to press any levers. But this, of course, is not what happens. The experimental infants are far more interested and amused than the control infants. Why? It is very tempting to suggest that what they take pleasure in is the fact that they have made the mobile move. The source of their pleasure is a power to affect the world that they are discovering in themselves, a capacity to bring about changes in the world. They repeat the action both to confirm the discovery and for the sheer pleasure of it.
By claiming that the infants are taking pleasure in their own agency I am, ipso facto, claiming that they are aware of what they are doing, and therefore aware that they are acting on things that are distinct from them. It seems to follow that they appreciate in a conscious manner the distinction between their own intentions or acts of will and the movements of objects in the world. And this seems to involve an explicit representation of the self. If this line of interpretation is accepted, the first criterion is satisfied. But what about the second? Does this implicate a grasp of self that satisfies the Generality Constraint?
One reason for thinking that it might do would be the thought that the infants’ grasp of their causal agency could be generalized via a grasp of the causal properties of other objects, the idea being that what they are discerning in their own case is a special case of ordinary causal interactions in the world. But even if the infants were capable of grasping causal relations between physical objects, this would still not count as an appropriate generalization of their own agency, for two reasons. The first is that the infant’s power to affect the world is importantly different from the causal impact of one physical object on another. We talk about agency in the one case and not in the other precisely because the first case has an intentional dimension lacking in the other. So we do not have a generalization of the predicate. The second reason is that we do not have a generalization of the subject either, because the appropriate generalization would have to be one across psychological subjects rather than across physical objects. The Generality Constraint would be satisfied here only if the infants were capable both of generalizing the special sense of agency involved and (relatedly) of grasping the existence of other agents. And it is surely implausible to attribute such sophisticated cognitive abilities to infants at this stage of development.
Now the interpretation I have offered of the infants’ behavior may well be resisted, but it will not be resisted because it is in principle impossible. The S–R interpretation could perhaps deal with the apparently recalcitrant features of the behavior, but this would not affect my main point, which is that the coherence (and indeed plausibility) of the interpretation I have put forward suggests that there might be a way of representing the self that is explicit but not conceptual. And if this point is accepted, we have taken the first step to meeting the challenge discussed earlier in this section: the challenge of explaining how the self might be represented in spatial reasoning without it being necessary to ascribe to the subject a grasp of the first-person concept.
Of course, it is one thing to show that the self can be represented in a manner that is explicit but nonconceptual and quite another to show how such a representation can feature in spatial reasoning so as to support the sort of protothoughts that, if Peacocke’s argument is accepted, are implicated in conscious place recognition. The second of these tasks has clearly not been done. Considerable work is required here but, by way of conclusion, I offer some brief remarks on the sort of role that this explicit but nonconceptual representation of the self could play in spatial reasoning.
Remember that on Peacocke’s account a crucial element of spatial reasoning is “the subject’s appreciating that the scene currently presented in his perception is something to which his own spatial relations can vary over time” (1992, 90). This is one of the vital respects in which first-person thought enters the picture. There are various ways in which one can flesh this idea out, depending on how one construes the subject’s grasp that he can enter into varying spatial relations to a particular scene. One plausible idea, though, would be that fully grasping this is conditional on realizing that one can intentionally act in a perceptually presented scene—by realizing, for example, that if one decides to return to it from here, one can do so by passing through these intermediate places, or that if there is something there that one wants, one should take this route to obtain it. In this sense, appreciating that the scene currently presented in perception is something to which one’s spatial relations can vary over time depends on appreciating how these varying spatial relations afford different possibilities for action. Of course, there is a crucial difference between being able to react in varying ways to one’s environment and grasping that there are different possibilities for action open to one. The latter is what I am stressing here, and what is interesting about it is that it seems to implicate the subject’s representing himself as an agent. What the subject grasps, on this account, is the close connection between his own intentions and the spatial configuration of the environment. Here, I would tentatively suggest, is where we will find the connection between the primitive nonconceptual mode of representing the self as an agent and the first-person component in spatial reasoning.
If this is right, we are considerably closer to an understanding of the notion of a nonconceptual point of view, originally put forward as a way of capturing the minimal requirement on self-conscious thought that it support the right sort of distinction between experience and what it is experience of. We started with the basic coperception of self and environment described in Gibson’s ecological account of the structure of perception and built it up by arguing that creatures to which it is legitimate to ascribe a point of view should be capable of conscious place recognition and of representing themselves in an explicit manner. In the final section, following Peacocke, I discussed the connection between these two demands and suggested, pace Peacocke, that they could both be accepted without it being necessary to move beyond the nonconceptual level.
I have been greatly helped by comments from Bill Brewer, Naomi Eilan, Anthony Marcel, and an anonymous referee.