Each of us experiences our own body in a distinctive way. Part of that distinctiveness is that we each experience our body and our limbs as our own. This is often characterized by saying that we each have a sense of ownership of our own bodies. But in what does that sense of ownership consist? Where does it come from? And how is it grounded? This complex of questions is often termed the problem of bodily ownership.
Some authors have taken what I have termed an inflationary approach to the problem. Inflationary theorists argue that we experience our bodies as our own in virtue of a distinctive feeling of “myness” or “mineness” that is incorporated into all bodily phenomenology, whether tactile, proprioceptive, homeostatic, or kinesthetic. The idea here is that there is a positive quale of bodily ownership. Frédérique de Vignemont has provided a thoughtful development of the inflationary view (de Vignemont 2013—see also her 2007). She takes issue with my defense in Bermúdez 2011 of a deflationary approach to bodily ownership. In that paper (reprinted as chapter 6 in this volume) I proposed an argument deriving from Elizabeth Anscombe’s various discussions of what she terms knowledge without observation (Anscombe 1957, 1962). My basic argument there was that the putative feeling of mineness cannot be “separately described” (to use Anscombe’s term) in a way that would allow it to ground those features of thought and experience that it is being called on to explain.
De Vignemont is not convinced and in her 2013 paper appeals to the rubber hand illusion to undercut my appeal to Anscombe. Section I of this essay restates and extends the case I made in 2011 against the putative quale of ownership. Section II turns specifically to de Vignemont’s objections, and in particular her claim that the rubber hand illusion reveals the existence of precisely the type of separately describable feeling of ownership that I denied was possible. In section III, I return to Anscombe’s concept of separability, exploring some of the questions that it raises about the relation between bodily awareness and “ordinary” perceptual awareness.
It can be hard to disentangle what is at stake in discussions of the phenomenology of ownership. But here is a basic claim that I think would be accepted by almost all participants in the discussion:
(A) When we experience our bodies, we typically experience them as our own.
This way of experiencing our bodies as our own is immediate and seems to be built into the content of somatosensory experience. So, for example, when I experience my legs as crossed, my legs are given to me within experience as my own. I do not (in normal circumstances) experience crossed legs and then go on to identify those legs as my own. One index of this is that certain types of error are not possible when we experience our bodies in normal ways. In normal circumstances (with my information being proprioceptively derived, and so forth), I can’t experience someone else’s legs being crossed while misidentifying those legs as my own—and nor, alternatively, can I experience my legs being crossed while misidentifying those legs as someone else’s. The impossibility of the second type of error is standardly described by saying that bodily experience is immune to error through misidentification relative to the first-person pronoun (Shoemaker 1968).1
Claim (A) is the starting point for discussion. The debate begins when we ask why claim (A) is true. This is where the inflationary and deflationary views come apart. Taking an inflationary perspective, de Vignemont maintains that claim (A) holds because our bodily experiences all share a common phenomenological feature. Here is her central claim:
(B) We experience our bodies as our own in virtue of a “felt ‘myness’ that goes over and above the mere experience of one’s bodily properties” (2013, 643).
As she puts it in her article, when I touch the table my tactile sensations include not only sensations of resistance, texture, temperature, and location, but also a distinctive experience of the hand as my own—a “nonconceptual intuitive awareness of ownership” (2013, 650). According to (B), there is a distinctive quale of ownership that is incorporated in our experience of our own bodies, and it is through this quale that we experience our bodies as our own.
Claim (A) says, in effect, that ownership is phenomenologically salient. Claim (B) offers one account of this phenomenological salience, identifying it with a distinctive quale of ownership. To appreciate that these are two very different claims, consider an analogy from discussions of perceptual consciousness. An equivalent of Claim (A) might be, for example, the claim that when we experience blue things we typically experience them as blue. One (relatively popular) way of explaining that fact about experience might be to postulate the existence of a felt quale of blueness. But of course, there is a range of other possible explanations. Likewise, there are other possible models of explaining the phenomenological salience of ownership identified in Claim (A).
De Vignemont’s explicit target in her paper is the deflationary view of ownership that has been articulated by various authors, including Michael Martin (1992) and me (Bermúdez 2011). According to the deflationary view,
(C) There is no distinctive qualitative experience of ownership in virtue of which we experience our bodies as our own.
Deflationary theorists accept (A)—that is, they accept that when we experience our bodies we experience them as our own. So, they accept that there is a phenomenology of ownership. But they deny that this phenomenology of ownership is epistemically based on a quale of ownership. As Martin puts it, “what marks out a felt limb as one’s own is not some special quality that it has, but simply that one feels it in this way” (1992, 201).
Of course, one might reasonably ask for more detail. What exactly does it mean to say that one feels a limb to be one’s own even though it does not have some special quality? I have gone into this question in more detail elsewhere (see particularly chapter 8 in this volume). But (in brief) the deflationary theorist plainly has to identify features of bodily awareness and bodily experience that can provide an experiential ground for the sense of ownership. Whereas the inflationary theorist, in effect, takes the sense of ownership to derive from an irreducible and unanalyzable phenomenological quality, the deflationary theorist, derives the sense of ownership from other, more fundamental facts about the phenomenology of bodily experience. Here is a way of making the distinction clear in a way that will be helpful for the subsequent discussion. For the inflationary theorist, the aspect of bodily experience in which the sense of ownership consists has to be elucidated through the concept of ownership. The deflationary theorist disagrees, holding that the sense of ownership can be grounded in aspects of bodily phenomenology that can be characterized without employing the concept of ownership.
How might the disagreement between (B) and (C) be resolved? Clearly some sort of argument is required, as trading intuitions about the putative (non-) existence of a sensation is unlikely to be successful. In my 2011 paper, I adapted a powerful line of thought from Anscombe in support of (C). In effect, I used Anscombe’s discussion of why bodily awareness counts as knowledge without observation to argue that there could not be a quale of ownership underwriting the phenomenology of ownership. Anscombe is concerned to argue that how we learn about our bodily states through bodily sensations is very different from how we learn about the world through, say, visual perception. Her basic line of argument can be adapted, I claim, to argue against the possibility of any quale of ownership doing the job that the inflationary theorist thinks it can do. (This needs to be emphasized. The claim is that, even if there were such a thing as a quale of ownership, it would be an idle wheel, serving no explanatory purpose.)
Let’s start with Anscombe’s reasons for thinking that bodily awareness is knowledge without observation. Here is her presentation of the view that she rejects (and she explicitly says that we could just as well speak of knowledge without clues as of knowledge without observation):
He [David Braybrooke] thinks that feelings of resistance, weight and pressure serve as clues by which one judges that one’s legs are crossed, when they are crossed and one knows it in the ordinary way. (Anscombe 1962, 55)
The view she is attacking sees judgments about limb position and how one’s body is disposed as ultimately derived by a process of inference that starts with bodily sensations.
What is wrong with the idea that bodily sensations serve as an inferential basis (clues) for how one’s limbs are disposed? The problem, according to Anscombe, is that bodily sensations are not, as she puts it, “separable”:
When I say: “the sensation (e.g. of giving a reflex kick) is not separable” I mean that the internal description of the “sensation”—the description of the sensation-content—is the very same as the description of the fact known; when that is so, I should deny that we can speak of observing that fact by means of the alleged sensation. (Anscombe 1962, 56)
Anscombe’s point, as I understand it, is that for very few, if any, proprioceptive and kinesthetic sensations is it possible to provide internal descriptions that can serve as a guide to how one’s limbs are disposed. Bodily sensations can normally only be described either in very general internal terms (e.g., sensations of pressure, contact points, muscle stretch), or in very specific external terms (e.g., the sensation of having one’s legs crossed). Described the first way, bodily sensations are too general and vague to be the epistemic basis for one’s awareness of limb position. But described the second way, they simply recapitulate the very knowledge that they are supposed to be grounding and so are not sufficiently independent to provide an epistemic basis. Under neither horn of the dilemma can bodily sensations do the work required to justify judgments about how one’s body is disposed.
The epistemological background here is that (epistemic) grounding must satisfy the twin constraints that I term focus and independence. The focus constraint is, in effect, that nothing can serve as a ground or warrant unless its content is suitably aligned with the content of that for which it serves as a ground or warrant. The notion of “suitable alignment” can, of course, be fleshed out in many different ways, but at a minimum a content a can be suitably aligned with a content b only if the former can serve as evidence for the latter. Anscombe’s claim, then, is that when bodily sensations are described internally, they cannot satisfy the focus constraint. To the extent that they have a content at all, that content is too vague and nonspecific to serve as evidence for one’s knowledge of how one’s limbs are disposed.
The independence constraint goes in the opposite direction. In effect, what it requires is that a ground or warrant must be capable of providing independent support for that for which it is a ground or warrant. And a necessary condition on a content a providing independent support for a content b is that the two contents be distinguishable (or separately describable, in Anscombe’s phrase). If a content b simply duplicates a content a, then this requirement of distinguishability and independence is not satisfied—any more than one justifies an assertion by simply repeating it. And yet (says Anscombe) we only get duplication when bodily sensations are described externally. If I say that I know that my legs are crossed because I feel that they are crossed, then the appeal to bodily sensation adds no new information.
In effect, Anscombe offers a dilemma for the view that she opposes. Bodily sensations must be described either internally or externally. When described internally, they fail the focus constraint. But when described externally, they fail the independence constraint. In my 2011 paper, I applied a version of Anscombe’s dilemma to de Vignemont’s postulation of a “non-conceptual intuitive awareness of ownership.”
It is highly implausible, I claimed, that there is a determinate quale of ownership that can be identified, described, and considered independently of the mineness that it is supposed to be communicating. Certainly no one has provided such an internal description. And this is why inflationary theorists tend to talk about a quale of mineness, or a preflective awareness of oneself as the owner of one’s limbs and body. So, in effect, inflationary theorists are committed to the feeling of mineness being externally describable. And yet a feeling of mineness that can only be described in those very terms is not sufficiently independent of the phenomenon that it is claimed to justify. So, the postulated “non-conceptual intuitive awareness of ownership” falls foul of Anscombe’s dilemma.
De Vignemont (2013) raises two objections to my argument. The first is that I overstate the import of Anscombe’s argument, even assuming it to be sound. The argument cannot show that there is no intuitive awareness of ownership. The most it can show is that such an intuitive awareness cannot serve as an epistemic basis for our judgments of ownership. This is a fair point. But one might reasonably wonder what the point is of postulating an intuitive awareness of ownership if it demonstrably cannot do any explanatory work. And de Vignemont herself does propose the quale of ownership as “a useful, simple explanatory tool, which allows for a single unified explanation of ownership illusions, for phenomenological differences between sensations in one’s limbs and in tools, and for disownership pathologies” (2013, 650). Tu quoque!
More persuasively, de Vignemont argues directly that there really is a separable awareness of ownership. Her argument is based on an understanding of perceptual awareness (and hence knowledge through observation) that she finds in an intriguing paper by Edward Harcourt (2008). As Harcourt puts it, “where there is perception, there must be belief-independent perceptual appearances” (Harcourt 2008, 309). The existence of belief-independent perceptual appearances is shown in turn by the existence of perceptual illusions, where things perceptually appear one way while one knows (and hence believes) that that is not how they really are.
As an observation about the nature of perception, this seems unassailable. De Vignemont thinks that it helps us understand the relation between separability and ownership because if we can find belief-independent appearances of ownership, then those will count (she claims) as examples of the type of separable description of the quale of ownership that I have denied exists. She writes:
For experiences of ownership to be separately describable there must be cases in which one reports feeling as if a body part belonged to one while correctly judging that this is not one’s body part. (2013, 644)
And, she argues, there are indeed such cases. Her prime example is the rubber hand illusion, first described by Botvinick and Cohen (1998) and much replicated since.2
The rubber hand illusion illustrates the dominance of vision over touch. To generate it, subjects are seated in front of a table on which there is a rubber left hand. Their own left hand is hidden by a screen. The experimenter simultaneously and synchronously use paintbrushes to stroke the subject’s concealed hand and the rubber hand. Typically, after a couple of minutes subjects report feeling sensations of touch in the rubber hand (Botvinick and Cohen 1998). When questioned, a significant proportion report feeling as if the rubber hand belongs to them. At the same time, of course, subjects are perfectly well aware that the rubber hand is not a part of their own bodies. So, we have a belief-independent appearance of ownership and (according to de Vignemont) precisely the separability that Anscombe denies.
Of course, the plausibility of the argument here rests directly on the plausibility of de Vignemont’s interpretation of Anscombe’s separability criterion, and it seems to me that de Vignemont misses the real point of Anscombe’s discussion. In the second of the two passages quoted earlier, Anscombe certainly equates nonseparability with the identity of the sensation content and the description of the fact that is known—as she puts it, what blocks separability is the fact that “the internal description of the ‘sensation’—the description of the sensation-content—is the very same as the description of the fact known” (1962, 56). Pari passu, for separability to be achieved, there must be the right kind of gap between the description of the sensation and the description of the fact for which the sensation is claimed to be evidence. And certainly, when appearance and reality diverge (as they do in the rubber hand illusion), then there is a gap of some kind between how things seem to the subject and how things really are, just as there is in any perceptual or quasi-perceptual illusion. But is it a gap of the right kind to secure separability, in Anscombe’s sense?
I think not. When Anscombe writes about separability, she is focusing on something completely different. When, in cases of illusion, there is a gap between appearance and judgment, the content of the appearance is different from the content of the judgment. The appearance is that, say, the rubber hand is mine, while the judgment is, of course, that the rubber hand is not mine. And so, the appearance and the judgment contradict each other. This means that the appearance cannot provide any sort of grounding or support for the judgment. But Anscombe is trying to elucidate what would have to be the case for an appearance to ground or support a judgment in such a way that it would be appropriate to describe the appearance as a way of coming to know a fact about one’s body. Let’s look at this in more detail.
Extrapolating from what Anscombe says about bodily sensations and the associated judgments, we can see how it can be applied to the phenomenology of ownership. The starting point is that, when we look at the phenomenology associated with judgments of ownership it takes the form of
Sensations that ––––
Feelings that ––––
In both cases the “––––” is a placeholder for a propositional content about ownership (e.g., “this hand is mine”). The sensation content, as it were, is the same as the content of the judgment of ownership. This is why, Anscombe believes, the sensation content cannot provide an epistemic basis for the judgment. Her point is that what is doing the justifying must be sufficiently independent of what is being justified to provide genuine epistemic support.
It is true that in the rubber hand illusion, the sensation content is not the same as the content of the associated judgment, because subjects typically experience the rubber hand as theirs while knowing perfectly well that it is not. So, the sensation content is certainly not identical with the judgment content. Does this make sensation content and judgment content sufficiently independent to allay Anscombe’s concerns?
Again, I don’t think so. The sensation content and judgment content are distinct, but the former is not in any sense serving as an epistemic basis for the latter. The epistemic basis for the judgment content is completely different, as it would have to be, given that the effect of the judgment is in essence to override the sensation content. So really the rubber hand illusion is orthogonal to Anscombe’s argument. While it is true that the rubber hand illusion can be interpreted with some plausibility as a belief-independent appearance of ownership, but this does not make the phenomenology separately describable in a way that would allow us to talk about a quale of ownership in the way that de Vignemont wants.
At this point, however, a new set of problems emerges. In particular, one might think that the cure is in danger of killing the patient, because it looks as if the basic distinction we began with between knowledge without observation and knowledge through observation is in danger of disappearing. Let me explain.
The distinction between knowledge without observation and knowledge through observation is intended to highlight the differences between how we come to find out about (for example) our own bodily states and our own intentions, on the one hand, and how we find out about the world through ordinary (outwardly directed) perception, on the other. Only in the latter case, Anscombe and others have thought, is it appropriate to talk about knowledge through observation. But now, given the epistemic interpretation emphasized in the previous two sections, it looks as if ordinary perceptual knowledge may come out as knowledge without observation. The problem is that the epistemic basis for perceptual knowledge is typically taking perceptual experience at face value.3 To take perceptual appearances at face value is to judge that the world is as it appears in perception. That is to say, it amounts to forming a judgment whose content is the same as the content of the perceptual appearance on which it is supposed to be based. But now it seems as though one can raise for ordinary perceptual knowledge exactly the same questions as were raised in section 1 for bodily awareness and the putative quale of ownership. Does not ordinary perceptual knowledge fail what I termed the independence constraint, because the content of perceptual awareness is not sufficiently independent of the content of perceptual knowledge to provide evidential support for it? And if so, does not knowledge through observation collapse into knowledge without observation?
In fact, the distinction between knowledge without observation and knowledge through observation seems to be under threat also from the opposite direction, with the former looking as if it might collapse into the latter. The problem here is that the discussion of the rubber hand illusion seems to have gone a long way toward assimilating the phenomenology of ownership (and of bodily awareness more generally) to the phenomenology of ordinary perceptual knowledge. In both cases, there are belief-independent perceptual appearances.
Putting these two problems together yields the following challenge. We have two ostensibly very different forms of experience (taking “experience” in the broadest of senses). We have our experience of our own bodies (including the sense of ownership), on the one hand, and perceptual experience of the external world, on the other. In each case, we have both belief-independent appearances and judgments. So, is there really the kind of epistemic difference between these two forms of experience that Anscombe envisaged? If so, in what does it consist? And finally, can that difference be spelled out in a way that will support my earlier arguments against the possibility of a quale of ownership?
John McDowell, in an interesting paper that intersects with many of the themes of this discussion, offers a radical answer. In essence, his view is that there are no (sensory) appearances of, for example, limb position. As he puts it, “the presence to one of the position of a limb, when one knows it in the relevant way, has no sensuously qualitative character” (McDowell 2011, 148). (McDowell does not explicitly consider knowledge of ownership, but I am sure that he would take the same points to apply. I cannot imagine that he would accept that there could be such a thing as a quale of ownership.)
McDowell has two connected reasons for denying that there are appearances of limb position and the disposition of our bodies. The first (which derives from a line of thought that he finds in Anscombe’s 1964 paper “Substance”) is the claim that bodily awareness does not involve the apprehension of secondary qualities, while it is constitutive of perceptual awareness that it involve such apprehension. This phenomenological claim has been disputed (e.g., by Michael Ayers—see Ayers 1991, vol. 2, 285) and McDowell backs it up with some general reflections on the nature of perceptual knowledge.
For McDowell, perceptual knowledge is essentially receptive. It is a way in which the knower is affected through the senses by what is known. Knowledge of limb position and ownership, in contrast, is self-knowledge of oneself as a bodily agent: “In receptive knowledge, what is known is other than the knower; or if that is not so it is known as other. This is exactly not so with this knowledge of limb position. This knowledge is self-knowledge; what is known is the self-conscious bodily being who is the knower” (McDowell 2011, 142).
There is much to dispute in these reflections, however. McDowell is thinking in very Aristotelian terms of individual senses and their proper sensibles (the secondary qualities, each of which can be apprehended only by one sense). On this view, each sense has its own sense organ and a corresponding proper object. The proper object of vision is the secondary quality of color. The proper object of hearing is the secondary quality of sound. And so on.4 The phenomenology of perception is much more complex and integrated than that neo-Aristotelian picture suggests, however. The boundaries between proprioception and exteroception are too blurred to carry the kind of distinction that McDowell is trying to impose with his comments about receptivity.
It is true that bodily awareness is self-knowledge. But this self-knowledge is not isolated in the way that, for example, our knowledge of our own thoughts can be. Rather, our knowledge of how our bodies are disposed is closely bound up with our awareness of the external environment. The sense of touch, for example, informs us both about limb disposition and about the configuration of outer objects, while the vestibular system tells us how our bodies are disposed relative to gravity and to supporting surfaces. At the same time, as I have stressed in other places (e.g., Bermúdez 1998 and chapter 5 in this volume), bodily awareness has a fundamental role to play in structuring exteroceptive perception. We do not gain perceptual knowledge of the external environment through the disparate and unconnected contributions of different senses, but rather through what Michael Ayers has termed “an integrated sensory field” that ineliminably incorporates bodily specific information derived from the different forms of bodily awareness.5 What integrates the various sensory modalities is that they collectively represent a single space containing three-dimensional objects. Our experience of this space is structured by our experience of our own bodies. This is exactly what one would expect, given that the body is an object among others and that bodily awareness is an important element in our understanding of how we can act on distal objects.
But now this discussion of how integrated proprioception and exteroception are seems to put us in even greater danger of losing the distinction between knowledge through observation and knowledge without observation. To see how the distinction can be preserved we need to go back to bodily experience and how it differs from (externally focused) perceptual experience.
We have seen (and endorsed) Anscombe’s reasons for thinking that bodily sensations cannot be “clues” on the basis of which we infer or otherwise derive judgments about the body. Any such proposal faces a dilemma, depending on whether it proposes to characterize the content of bodily sensations internally or externally. When the content of bodily sensations is characterized internally, it falls foul of the focus constraint, while when characterized externally it runs into the independence constraint.
That insight alone is not going to yield the distinction that we are looking for, since it is fairly clear that perceptual sensations cannot be clues either. The only way in which perceptual sensations could be clues in the sense that Braybrooke proposed, and Anscombe criticizes, would be if some form of sense datum theory were true, and that surely is not a viable position. But thinking further about why both types of sensation cannot serve as clues will point us in the right direction, because it turns out that they each fail to function as clues in importantly different ways.
One key characteristic of bodily experience is that the bodily location of sensations is part of their phenomenological content. We experience a tingle in the knee, for example, or a pain in the neck. There are different models of the spatial content of bodily awareness, but all agree on the basic point that sensations are experienced on a body-relative frame of reference.6 Moreover, it is not just that bodily sensations are experienced at particular body parts. Our experience of the spatial location of a sensation is given in terms of the spatial location of the body part in which it is located. The tingle in my knee is experienced at a particular point in space (in front of the chair and below the desktop) because that is where I experience the knee to be. This means that knowledge of limb position is an integral part of the content of bodily sensations. This is why sensations cannot serve as clues, or provide any other inferential or quasi-inferential basis for knowledge of limb position. McDowell himself appreciates this point, writing:
It is not that one knows where a felt tingle, say, is, independently of knowing how one’s body is disposed in space, so that an aggregation of such knowledge of the location of objects of sensation—or, better, of the sensations themselves—might enable one to get to know how one’s body is disposed in space. That gets things backwards. One locates these sensations in space only by locating them in one’s own body. Spatially organized awareness of one’s bodily self is a presupposition for the capacity to locate bodily sensations, not something enabled by that capacity. (McDowell 2011, 145)
In effect, McDowell is eloquently affirming the independence constraint.
A similar point can be made about the phenomenology of ownership. The starting point for this essay was the following observation:
(A) When we experience our bodies, we typically experience them as our own.
Another way of formulating (A) would be to say that ownership is an integral part of the content of somatosensory awareness. And, just as with bodily sensations and limb position, the independence constraint entails that our bodily experiences cannot serve as an inferential or quasi-inferential basis for our sense of ownership.
So, knowledge of limb position is an integral part of the content of bodily sensations and ownership is an integral part of the content of somatosensory awareness. This in itself is not enough to distinguish the two forms of awareness from ordinary perceptual awareness because we can make the parallel observation that knowledge of the layout of objects in the external environment is an integral part of the content of perceptual awareness? After all, we typically learn about the layout of objects in the external environment through taking our perceptual experience of the world at face value. However, the independence constraint does not apply here in the same way, because there is a crucial difference, which I will now explain.
The content of the perceptual awareness of the world that (assuming it is veridical) yields knowledge when taken at face value is propositional content. When I come to know that φ through perception, what I perceive is the same as what I come to know—namely, that φ holds. Here “φ” is a placeholder for some characterization of a propositional content, with the details to be filled out according to your preferred theory of propositions. When I come to know through perception that the cat is on the mat, for example, this is because I see that the cat is on the mat. More generally, when we specify the propositional content of ordinary exteroceptive perceptual awareness, we do so using the very same concepts and language that we use to spell out the content of the knowledge that is acquired through perception—and, of course, on the assumption that the conceptual/linguistic machinery that we use is itself reflected and applied in the perceiver’s experience of the world. If this were all there were to the content of perceptual awareness, then the independence constraint would apply. But it is not, and so it does not.
As many commentators on perception have noted, there are other ways to characterize the content of perception, at least in the case of visual perception. Dretske, for example, famously distinguished epistemic seeing (the “seeing that φ” discussed in the previous paragraph) from simple seeing (Dretske 1969). Other theorists use the terminology of conceptual content as opposed to nonconceptual content (Evans 1982; Crane 1992; Peacocke 1992; Bermúdez 1998—see Bermúdez and Cahen 2015 for further references). For the authors mentioned, and many others, the nonconceptual content of perception provides a vital epistemic underpinning for the epistemic/conceptual/propositional content of perception. So, for example, on Peacocke’s account, first put forward in his 1992, the possession conditions for observation concepts are given in part in terms of the appropriateness of applying them when enjoying perceptual experiences with relevant nonconceptual contents.
This is not the place to go into the complex debates surrounding the notion of nonconceptual content, or to engage with critics of the very idea such as John McDowell (1994). My point is a simple one. The fact that the content of ordinary, exteroceptive perceptual awareness is not exhausted by its propositional content opens up the possibility of finding a further epistemic basis for the propositional content of perception. This gives precisely the “separability” that Anscombe emphasized and that was discussed in sections I and II above.
The nonconceptual content of (visual) perception meets both of the key constraints identified earlier. It meets the focus constraint because, on all ways of developing the notion, nonconceptual content is representational. It represents the world as being a certain way, and so it is exactly the right sort of thing to provide evidential support for perceptual judgments. At the same time, there is no danger of an identity of content between the nonconceptual content of perception and the content of perceptual judgments and perceptual knowledge. The nonconceptual content of perception can provide evidential support for perceptual beliefs and perceptual knowledge because the two types of content represent the environment in fundamentally different ways. This is how the independence constraint is satisfied.
In contrast, as I have emphasized, there is no such possibility of finding within the content of bodily awareness an independent epistemic basis either for judgments about how one’s body is disposed, or for judgments of ownership.7 This is why I have described these forms of awareness as nonseparable, in Anscombe’s sense. In both of those cases we are confronted with what is in effect a phenomenological given. There is no way of characterizing the content of bodily sensations that can exclude the fact that we experience bodily sensations at particular locations in particular body parts. Likewise, there is no way of characterizing the content of our experience of our own bodies that can exclude the fact that we experience our bodies as our own.
I should emphasize that the key issue here is not the distinction between conceptual content and nonconceptual content. It is most definitely not the case that the nonconceptual content of perception satisfies the independence constraint simply in virtue of being nonconceptual. To say that a content is nonconceptual is (as standardly construed) to say that a subject can be in a state with that content even though she does not possess the concepts required to specify it.8 So, for example, Christopher Peacocke’s model of scenario content characterizes the nonconceptual content of visual perception in terms of ways of filling out space around the perceiver. This model can accurately describe the (nonconceptual) content of how a subject sees the world even when that subject lacks some or all of the concepts required to set up or employ the model. But the fact that the subject lacks these concepts is not what secures the satisfaction of the independence constraint. Rather, the independence constraint is satisfied because there is sufficient distance between a description of the world given in terms of scenario content, and the (conceptual) description of the world for which the scenario content provides evidence and grounding.
For this reason, the question of whether or not bodily awareness is nonconceptual is orthogonal to my overall argument. As it happens, I think that it is.9 But that cannot close the gap between bodily awareness and ordinary perceptual awareness. My point is that the content of bodily awareness cannot be accurately characterized at any level, or with any conceptual machinery, in ways that do not reflect the very facts about limb position and ownership that the content of bodily awareness might be called up to justify. That is what it means to say that knowledge of limb position and ownership of one’s limbs is a phenomenological given. And it is for that reason that the (nonconceptual) content of bodily awareness does not stand to the (conceptual) content of judgments about limb position and ownership in the same way as the (nonconceptual) content of visual perception stands to the (conceptual) content of perceptual knowledge.
And so, to summarize the discussion in this section, there is a meaningful and important distinction between bodily awareness and knowledge of limb position, on the one hand, and ordinary, exteroceptive perceptual awareness, on the other. The terminology of “knowledge through observation” and “knowledge without observation” may not be entirely happy, but it certainly marks a real distinction. Moreover, circling back to the principal argument of this essay, understanding the distinction properly shows why there is no utility to postulating a quale of ownership in the way that de Vignemont proposes. Everybody agrees that we experience our bodies as our own. Such a “sense of ownership” is part of the phenomenology of experience. But it does not follow from this that, as de Vignemont proposes, we have such a sense of ownership in virtue of a “felt ‘myness’ that goes over and above the mere experience of one’s bodily properties” (2013, 643). In fact, for the reasons I have brought out, such a quale of ownership would be an idle wheel with no explanatory power.