Vision, Self-Location, and the Phenomenology of the ’Point of View’

Presenter: John Schwenkler, Mount St. Mary’s University

Commentator 1: Kranti Saran, Harvard & Jawaharlal Nehru University

Commentator 2: James Stazicker, NYU

Commentator 3: John Campbell, UC Berkeley

  • John is running late and will join us shortly


  1. Wow, I am gratified by these tremendously thoughtful comments — as I said in an e-mail to Kranti (and as James knows), I have been writing this paper for going on five years, but now I feel like I need to start over again! I am going to try to begin responding over the weekend, but it will likely take me a while to get through all of this …

  2. Okay, so here is my initial attempt at a response, though without any italics (probably it is better off). A PDF copy is also available, here.


    First, let me say again how grateful I am to have my work engaged with in such a deep, philosophically rich way. It is going to take me a long time to incorporate adequately all the points raised in Kranti’s and James’s commentaries. Below I make an initial attempt.


    In Section 3.4 of my paper I argue that the only way to explain the difference in content between the illusion of vection I and the corresponding experience of world-motion V is by attributing to I a visual representation of one’s own motion. To prove this, I consider a few other ways we might try to explain this difference in terms of what is represented in visual experience proper is in terms of some sort of demonstrative content, say that in an illusion of vection “it is this person, or perhaps this body, that appears to be in motion in an otherwise stationary environment”. I object, however, that “the experience of a moving object in the surrounding world can equally well be described by using perceptual demonstratives to refer to that individual, and even if the demonstratives are being put to different uses in the respective cases … that difference is not exhibited, as it were, simply by citing demonstrative contents in this very generic way” (p. 17). Against this argument, Kranti objects:

    Why don’t tokenings of ‘that body’ towards spatially distinct individuals suffice to distinguish the case of my body spinning from anothers? … Perhaps Schwenkler means that demonstratives inadequately distinguish the cases by failing to capture the fact that only one of the demonstrated bodies is my body. But the Minimal View would presumably reject such an explanatory demand on the contents of visual experience: the relevant phenomenology isn’t a product of visual contents. (pp. 4-5)

    I am going to get to the question of visual vs. non-visual contents later on, but it occurs to me that the point I was after can be made more effectively than I originally did. For consider two things. First, in the kind of case we are considering the perceiver’s body is supposed to be entirely out of view. But it seems to be a condition on the meaning of a visual demonstrative that the object it refers to be experienced visually: how then can vision represent the motion of this when this is not visually represented? Second, we should note that though the sort of response I am envisioning here does manage to fall short of the Self-Location Thesis, it does require giving up on the attempt to explain the spatial content of vision just in terms of monadic spatial predicates: for if it can be part of the content of visual experience that this body is moving, then it can also be part of the content of experience that, say, a computer is in front of this body. But then what could be the motivation for refusing to say that this body is also represented as me? (Of course things are even worse for the Minimal View if the moving object is described as a person instead.) I wonder what others will think of these arguments.

    Phenomenal Properties

    Section 3.3 of my paper argues against the attempt to explain the difference between I and V in terms of a “purely phenomenal” difference between the corresponding visual experiences: I show that on the most popular account of which phenomenal properties there are (namely, ones corresponding to the positions of things in the visual field conceived of as a two-dimensional plane placed before the eyes) no such differences between these experiences at all. But James worries that I am being uncharitable:

    Couldn’t there be some other purely phenomenal respect in which V and I do differ? Even if it’s hard to say what this difference is, we might reasonably expect purely phenomenal differences to be hard to describe. And if this purely phenomenal difference is reliably correlated with a difference between self-motion and other-motion, it might explain the naturalness of Palmer’s description of V and I. (p. 3)

    The short answer is that yes, I suppose there could be such a respect (and I will turn in the next section to the possibility, which James suggests right after this, that it could be non-visual). But it strikes me as very important that this purely phenomenal difference be of a sort that we have some independent reason to think might exist. (Here compare what James says from pp. 3-4 of his commentary.) That is, imagine a defender of the Minimal View who tries to say that since vision can’t be self-locating, the visual difference between the two experiences must be non-representational – and then perhaps tries to use this to defend the claim that there are such things as purely phenomenal properties of experience! The problem with this argument is that in this case we have on hand a perfectly plausible explanation of the representational difference between the experiences, namely that an illusion of vection represents one’s own motion. Of course that would involve accepting the Self-Location Thesis, but so what? Compared to this, an “explanation” of the difference between them in terms of some ineffable phenomenal property is simply no explanation at all.

    Non-Visual Differences (1)

    Here is where things get difficult. In order for my argument to work, the difference between I and V has to be a difference in the representational contents of visual experience, as opposed to experience in some other modality, say the kinesthetic sense of bodily location. I argue for this in Section 3.2 of my paper, but what I say is very sketchy (see pp. 13-14): namely, that there is a kind of “cross-modal unity” in an illusion of vection (that is, what is represented in the different modalities of experience isn’t discordant), and that this is part of what makes the illusion so compelling: I write that “in the absence of any significant perceptual input suggesting a lack of change in one’s position, everything in experience suggests that the surrounding world is stable while one’s own body is in motion”, and that “if things did not appear this way to vision as well, then the illusion would be much less vivid than it is”. Kranti’s objection to this argument is appropriately pointed:

    Doesn’t this response beg the question against the Minimal View? The objector can grant Schwenkler’s counterfactual: the illusion would be even more powerful if there were a distinctly visual contribution, but the actual difference is [say] purely kinesthetic. The vividness of the vection illusion, by the objector’s lights, is purely a function of a kinesthetic contribution. How has Schwenkler shown otherwise? (p. 4)

    I suppose I haven’t. But here is a better way to make my point. If the only difference in visual content between I and V is the apparent motion of the cylinder in V, then there is less of a visual representation of motion in the vection illusion than the experience of world-motion. But if this were so, then it should be possible to undermine the vection illusion by focusing one’s attention on visual experience (as opposed, say, to the way things sound or the absence of a sensation of the air moving against one’s skin). But in fact the opposite seems to be true: the illusion is more vivid in visual experience than in other modalities, and the way to undermine the illusion is by focusing on the non-visual dimensions of experience (think e.g. of what you can do when you experience the apparent motion of your moving train). And this is just as we should expect: after all, it’s input to the visual system that gives rise to the illusion in the first place, and the dizziness one feel and the illusory experiences in other modalities seem to be a product of the fact that one’s own apparent motion is experienced visually, as opposed to something we could appeal to in order to explain this visual illusion away.

    Non-Visual Differences (2)

    James suggests something similar in response to my “cross-modal unity” argument on p. 4 of his commentary, but then goes on to develop a position in a way rather different from Kranti’s appeal to kinesthetic sensations, which ultimately cuts to the heart of my argumentative strategy in this paper. The first thing I have to say about it is that I’m really sympathetic to the idea that the boundary between perceptual and cognitive representation is “liminal”, as he puts it in the footnote on p. 6, and that in general there are no clear answers to questions concerning the precise locus of a given experiential representation. But having conceded this, we should note that this is just as much a problem for defenders of the Minimal View as it is for me: if the phenomena I am describing occur at some liminal boundary, then just as I have to admit that there is no clear sense in which visual experience is self-representational, so the defenders of the Minimal View will have to admit that there is no clear sense in which it is not. (Right?) In principle I would be happy to reach this sort of irenic position.

    Having said this, let me raise a somewhat different point. Once the defender of the Minimal View concedes that there is a difference between I and V that (i) is a difference in representational content and (ii) is at least quasi-perceptual (that is to say, it’s not just a matter of “cognitive experience”, as James argues on p. 4 of his commentary), then what reason is left for insisting that the difference can’t be visual? That is, once we’ve conceded that there is some perceptual modality in which one’s own location can be represented, why not think that vision has this characteristic? Indeed, on some standard ways of thinking the representational content of vision is richer than that of the other sensory modalities? What is there in the very idea of vision that should make one insist that it cannot be self-locating? A more general question in the vicinity is that of where the burden of proof lies: in my paper I take on the challenge of arguing for the Self-Location Thesis in a way that meets all comers, but given the naturalness of describing visual experience in self-locating terms it is not clear to me that this is actually a standard I should have to meet.

    Unarticulated Contents

    In the final section of his commentary, Kranti asks why we should think the visual representation of one’s own location in an experience like I is – as I put it in my paper – explicitly first-personal, as opposed to having some sort of unarticulated content. He considers Perry’s account of the demonstrative judgment “There is an apple there”, according to which “while lacking an explicit representation of a first-personal indexical … [the judgment] nevertheless concerns a fixed unarticulated object: oneself” (p. 5). And now Kranti asks us to consider “the perceptual judgment ‘I am in front of a door’”, writing:

    Why can’t this too lack an explicit representation at the level of visual content, while nevertheless articulating the object in the judgment? The thought is that the object – oneself – oneself is unarticulated at the level of visual contents in both cases because the same object always occupies one of the n-ary perceptual relation, so we don’t need to track its occupant … (pp. 5-6)

    I am not sure just what is being suggested here. If the content of the judgment “I am in front of a door” is supposed to be explained as only implicitly first-personal, then I think the account would suit my purposes: for it will still hold that visual experience and judgment have self-locating content in the very same respect (that is, their “face value contents” are the same), which is the claim I put at the heart of the Self-Location Thesis from pp. 1-4 of my paper. By contrast, if it is supposed that the judgment “I am in front of a door” (or “I am spinning”, say) has explicitly self-locating content while the self-locating content of the corresponding experience is only implicit and unarticulated, I would want to know what motivates this distinction, and moreover how the content of the experience is then supposed to be described. I took Campbell’s description of the spatial content of visual experience in terms of monadic spatial predicates to be a way of trying to do this, and attempted to show in my paper that it fails. Does this address the worry? I fear that I do not fully comprehend the dialectic here, and so am looking forward to more discussion of this point.

    A Maximal View

    The last section of James’s paper raises two questions that are more constructive than critical, and as I’ve already gone on for more than 2,000 words I’ll answer only the second of them. James notes that my aim in this paper was rather modest: to show that visual experience can be self-locating just in virtue of its perspectival character, as opposed to showing that it always or often is. But is it possible to defend a stronger claim? In an earlier, even more sprawling version of this paper I attempted to do just this, but some very helpful comments by Aaron Henry persuaded me to set my sights more narrowly for present purposes. Here, though, is how the now-deleted argument went.

    Suppose you have rejected the Minimal View and are deciding between versions of the Self-Location thesis that are Modest and Maximal: the Modest version holds that visual experience is self-locating only sometimes, and the Maximal version that this is always the case, at least in ordinary human visual experience. Now, I take the best case for Modesty to be one that appeals to the so-called “experiential transparency” of the body in experience: the idea is that while one can sometimes come to experience the position or location of one’s body by focusing one’s experience on it, usually one’s body is not something one is experientially conscious of. But developing a Modest view in this way would require us to make sense of the rather unintuitive idea that having one’s attention drawn to the question of one’s own location can transform one’s visual experience from having merely perceiver-relative spatial contents at one moment to having explicitly self-locating ones at another, though without any new object ever coming into view. Prior to the shift in attention the perceiver was aware of things only as being “to the right”, “to the left”, and so on; while after the shift the perspectival spatial contents of his or her experience change significantly, and the perceiver is now aware of where those things are with respect to him or her. This is an odd proposal indeed: it is simple enough to spell out the idea of a visual experience that is perspectivally spatial but not explicitly self-locating on the basis of which self-locating judgments might be formed, but much harder to make sense of a visual gestalt shift from an experience of the non-self-locating sort to one where the perceiver’s location has suddenly shown up, though without coming into view. At the very least this consideration suggests that is the Maximal version of the Self-Location Thesis, and not the Modest one, that deserves to be the default position on the issue of visual self-location once the Minimal View has been ruled out; the burden of proof is on the proponent of the Modest version to provide positive phenomenological evidence for the kind of transformation envisioned here, rather than the other way around.

  3. Hi everybody, great paper and discussion.

    I really like your paper and I am really sympathetic to your position.

    I would like nonetheless to press a little bit more on your reply to the ‘non-visual difference objection’. I think that you are right in your claim that “once we’ve conceded that there is some perceptual modality in which one’s own location can be represented, why not think that vision has this characteristic?” so my question is whether your opponent has to concede that the quasi-perceptual modality represents one’s own location.
    Wouldn’t it be possible to account for the phenomenological changes if the quasi-perceptual modality represents something like a change in location without thereby representing one’s own location?

  4. Hi Miguel,

    Thanks for your kind words. Regarding my objection to the “non-visual difference” position, you write:

    Wouldn’t it be possible to account for the phenomenological changes if the quasi-perceptual modality represents something like a change in location without thereby representing one’s own location?

    I suppose this is possible. But what would it be a change in the location of? Here too we need to account for what I called the “phenomenological distinctiveness” of the apparent motion in the vection illusion — that is to say, the apparent motion can’t just be the motion of any old thing. And I think my arguments against the various ways of trying to do this would carry over, right? Or do you have something else in mind?

  5. I had something like the following in mind (which intends to be an analogy of the function of the vestibular system, which as James suggested might play a role in an alternative explanation):

    Imagine a glass with water, if there is some movement of the glass, then waves would be generated. Imagine a system where this waves represent a motion of the system. In this case the system can represent ‘motion’ without thereby representing it’s own location.
    A misrepresentation of motion might help to explain the change in the phenomenology in the presented case.

  6. Okay, now I see. But would the “motion” represented by such a system be appropriately fine-grained to distinguish between the motion of the system itself and that of the surrounding environment? If not, I don’t see how a system that represented motion in this way would be able to give rise to different experiences of the sort I identified. If so, then I don’t see how it could fail to represent its own motion.

  7. Hi John – thanks for your response and for presenting such a rich paper! Rather than posting a complete response, I’ll go with what I have. First up, the discussion of demonstratives. Let me make a fresh stab at what I was trying to get at.

    On behalf of his opponent, John explains the appeal of demonstratives as follows: “The idea here would be that the perceptual demonstrative “this” marks the experiential distinctiveness of the person (or body) that visually appears to move, thus explaining how such an experience differs from one where the thing in apparent motion is simply an object somewhere in the field of view” (p. 17).

    As John points out, this strategy for defending the Minimal View faces a challenge: how, using only demonstratives, can it distinguish the motion of the viewer from that of a viewed object? In response, John’s opponent could employ two kinds of demonstratives: one to
    capture the `experiential distinctiveness of the person (or body) that visually appears to move’, for example “this is moving”; and another to capture `where the thing in apparent motion is simply an object somewhere in the field of view’, for example “that (other) body is moving”. Why isn’t this sufficient to distinguish the cases?

    The use of that second kind of demonstrative poses no special problem and allows us to distinguish among objects given in the visual field. The use of the first kind requires care. What does such a use of “this” refer to? John’s opponent could propose either that “this”
    refers to the subject’s body or that “this” refers to the point of view.

    Does it refer to the subject’s body? By stipulation, the subject can’t see their body. John thinks that the demonstrative doesn’t refer to it, as it isn’t visually represented. In his response to my remarks, John suggests that it lacks a referent because “it seems to be a condition on the meaning of a visual demonstrative that the object it refers to be experienced visually: how then can vision represent the motion of this when this is not visually represented?” John’s opponent might object that he’s unjustly restricting the kinds of demonstratives they may rightfully appeal to: “perceptual demonstratives” (p. 17) aren’t limited to referring to objects in the field of view. For example, it’s plausible that if I say “Look over there”, pointing outside your field of view, I’m employing a perceptual demonstrative. Analogously, “this” refers to the subject’s body, despite the fact that it is not an object in the subject’s visual field. Even so, his opponent has reason to think that the demonstrative doesn’t refer to the subject’s body, but rather to the point of view. Why?

    John’s opponent may think there is a principled reason that such a use of “this” refers to the point of view rather than the body. The thought is that the point of view can come apart
    from felt bodily location: they are empirically dissociable. For example De
    Ridder et al. (2007)
    describe a case where the point of view of the subject comes apart from the subject’s felt bodily location:

    His perception of disembodiment always involved a location about 50 cm behind his body and off to the left. There was no autoscopy and no voluntary control of movements of the disembodied perception. The environment was visually perceived from his real-person perspective, not from the disembodied perspective.

    Such dissociations have also been induced in normal subjects by using virtual reality techniques and felt location and the point of view have even been independently manipulated. Such a use of “this” thus captures the point of view, not the subject’s felt bodily location.

    Such a proposal gives up on what I called the positive thesis of the Minimal View: explaining the spatial content of vision in terms of monadic spatial predicates. But it seems plausible
    that the core of the Minimal View is what I called the negative thesis: that the spatial contents of visual experience cannot suffice to represent the location of the perceiver. The proposal I’ve sketched here accords with the core of the Minimal View.

  8. Hi John,

    Consider again the analogy of the glass with water (I am not sure whether it is appropriate but I think it is useful for the moment). If the environment moves but not the glass, then there are no waves. If the the system moves there are waves. A simple system with a glass is appropriately fine-grained to detect motion. A system that only has a glass of water as its sensor will not be able to detect motion in the environment, but the system might also have a couple of sensors to detect that (like the eyes).

    Now imagine that someone blows the glass. There will be waves but no motion, so the system will be misrepresenting motion.

    Consider an over simplistic theory of representation like the following: the content of a trait is what cause the activation of the trait in normal conditions. In normal conditions waves are caused by motion. Hence, waves represent motion.

    Imagine that the sensors are connected to a bellow. In normal conditions the pump of the bellow does not cause waves. However, a certain input in the sensor may cause an huge pumping of the bellow, which in turns blows the water in the glass. I do not know whether something like that is what is going on in your case. I merely intended as a possible reply to be explored by your opponent.

  9. Miguel: Now I understand your position better. And it is a great example! But doesn’t it then seem right to say that what the waves represent is the motion of the system, as opposed to the world around it? (I.e., because this is what causes its activation under normal conditions.) Of course this representation won’t have the kind of complex semantic structure that I am attributing to visual experience, but if we extend the parallel then it would seem to support my position rather than conflicting with it.

  10. John, a very impressive paper!

    Here’s a follow-up on Kranti’s point about explicit vs. unarticulated content, and your question about it. In your comment you write

    “By contrast, if it is supposed that the judgment ‘I am in front of a door’ (or ‘I am spinning’, say) has explicitly self-locating content while the self-locating content of the corresponding experience is only implicit and unarticulated, I would want to know what motivates this distinction, and moreover how the content of the experience is then supposed to be described. I took Campbell’s description of the spatial content of visual experience in terms of monadic spatial predicates to be a way of trying to do this, and attempted to show in my paper that it fails.”

    When Campbell discussed this matter, he seems to present it in the context of something like Strawson’s or Evan’s generality constraint. For example, he writes that “it does not seem correct to say that [visual experience] represents things as ‘to my right’ using the relational notion, because of the LACK OF GENERALITY in whose right or left can be represented” (2002: 184), and “in stating the spatial content of vision, we do not need … the GENERAL CONCEPTION of something’s being to the right or left of an arbitrary subject” (ibid).

    If I am on the right track, then Campbell’s idea is that that the self-locating content of visual experience is implicit in the sense that it lacks generality (in the aforementioned sense), whereas the self-locating content of the corresponding judgment is explicit in the sense that it has generality (the appeal to monadic egocentric terms is meant, it seems, to capture the lack of generaltiy in the content of visual experience). If this is so then, presumably, Cambpell (or someone with similar ‘minimalist’ inclinations) could hold that a visual experience can have the content ‘I* am spinning’ where the asterisk means that the content lacks generality in the sense that one can entertain it even if one does not grasp the conception of some arbitrary person spinning.

    I might be completely misunderstanding Campbell here. But even if I am, it seems to me that the (seemingly) minimalist position I have sketched here is reasonable and prima facie capable of answering your challenge.

  11. Assaf, thank you for your comment! (Kranti, I am going to get to yours.)

    It is true that the concept of generality plays an important role in Campbell’s position. And in fact I think he has pressed this line of argument on me before, so I am glad to have to revisit it. But I don’t think his view can be saved in the way you suggest. Here is the difficulty: I claim that in the illusion of vection, there is an experience of motion, and I take it that you concede this. But on our ordinary way of thinking motion is a property; it’s not something that can be experienced just as free-standing. So then this raises the question, what is this motion experienced as the motion of? Once we allow this much, my argument leads to the conclusion that it is the motion of the self, and so the Minimal View is finished.

    Now, perhaps you could try to head off this move by saying that in the illusion of vection the change in what’s represented doesn’t fall under (what Campbell would call) the “general conception” of motion: that’s to say, this experience doesn’t have the content “x IS MOVING”, where “__ IS MOVING” is a predicate that can be applied to some arbitrary subject. By contrast, the content is something like “It moves (here)”, where this has the same sort of semantic structure as “It is raining”, in which rain isn’t predicated of any thing. But then the experience will not really be representing motion at all, will it? It is just representing some ineffable property. And this gets the phenomenology wrong: for motion seems to be represented in the illusion of vection in the same respect (i.e., as the same property, though as possessed by a different object) as it is represented in the experience where the cylinder is seen to move; and in addition it seems like the content “It moves (here)” would be equally applicable to the veridical experience as to the illusory one: perhaps it corresponds to visual field motion or something, but as we have seen that is the same in the two experiences.

    Does that make any sense? I fear it does not, as I am writing through a fog of exhaustion.

  12. Kranti, to your question, I still don’t think the demonstrative “this” could refer to the point of view, for the same reason I give in the paper, namely that this concept seems relevant to analyzing the structure of experience rather than its content. And it seems to me that the kind of case you describe at the end is just one where one’s seen location comes apart from one’s felt one.

    Your other proposal is that the demonstrative might pick out the perceiver’s body. I objected in my reply that it seems impossible for a visual demonstrative to refer to something that’s unseen. In response, you give as an example the statement “Look over there”, where the location in question is out of view. Hmm. But now suppose instead you say “Look at this (or that)”, where the thing you refer to isn’t something I see. Hasn’t there been a failure in the act of communication? In this respect I think that “this” and “there” are as different as “here” and “there” are: the first member of each pair presupposes (as Evans would put it) a perceptual link with the referent, whereas the second member just presupposes something like the possibility of establishing one. And so I don’t think it’s possible for me to represent a proposition of the form “This is moving” without having the right sort of perceptual link with the thing I refer to as “this”: and since in the case in question the content is supposed to be represented visually, it seems to me that the body would have to be seen for the content not to be empty.

    Yes or no? Again, there is the exhaustion.

  13. John, I don’t know what Campbell would say in response to your last comment, but here is how I see things. Of course, I might be completely off here.

    On most accounts, representational content is understood in terms of accuracy conditions. Now, the accuracy conditions of ‘an apple is above’ and of ‘an apple is above ME’ are the same. Both are accurate if and only an apple is in front of me (the ‘in front of me’ corresponds to either a relational property or a centered property). In this sense, both contents are self-locating. The only difference between the contents concerns generality. That is, the first content can be entertained by a creature that does not grasp the concept of the relation ‘x is above y’, and also does not grasp the concept of ‘subject’ (on the assumption that concept-possession involves grasp of generality, which is Strawson’s generality constraint).

    And now for the crucial part: the monadic egocentric terms are no more than a way of stressing the non-generality (and so also non-conceptuality) of the content. They do not correspond to monadic PROPERTIES (they might correspond to relatonal or centered properties). Instead of using monadic ecocentric terms, Cambpell could have used hyphens, and say that the content in question is ‘an apple is above-me’. What matters is the accuracy conditions, and this provides resources for answering your challenge, as follows:
    The content of the experience you call ‘I’ is ‘I-am-spinning’. The experience is accurate if and only if the subject at the center (of a given centered world) spins (or some such). So no ineffable property (the one implicated in your ‘it moves’) is needed. The property in question is the ordinary property of spinning, except the creature that entertains the content need not be able to attribute it to creatures other than itself.

    The reason I think I might be deeply wrong here, is that if I am right, then the debate between (what you call) the Self-location Thesis and the Minimal View, is, in the end, an instance of the debate between conceptualists (McDowell) and nonconceptualists (Peacocke, Evans) about the content of experience. Contrary to the way you present things, it is not really a debate about whether the self is a part of the content of some experiences. It is in fact a debate concenring whether the self-locating concent of experience is conceptual (in the sense associated with the generality constraint) or not. So I must have made a mistake somewhere. Haven’t I?

  14. Assaf, I do want the debate to cut across the conceptualist/nonconceptualist divide, as well as the divide between those who accept and reject the very idea of perceptual content. As I say in the paper, I set up my argument in terms of contents simply for the sake of simplicity, and my challenge to those who reject them (a group that sometimes includes me!) would be to find some other way for accounting the phenomena I describe (in particular, those pertaining to the visual differences between vection illusions and experiences of world-motion).

    But now, you suggest that in the vection illusion the content could be described as “I-am-spinning”. And the problem as I see it is that then what is being represented is not motion at all: for motion is a property that can be possessed by a range of different objects; that’s to say, it has a kind of generality. (Though I am not sure that this is necessarily the generality of concepts: e.g. the visual system of a non-linguistic creature can represent an apple and a fire engine both as possessing the same property, even though the creature may lack the concept RED. Right?) By contrast, by your own admission what is represented when experience represents (that?) “I-am-spinning” is an entirely different thing — i.e., it doesn’t share any content — with what is represented when experience represents that “The cylinder is spinning”. (This is what I meant when I called it a representation of an “ineffable property” — though I guess it is not a property properly speaking at all; it is just some sort of semantically unstructured representation of an event (?) in the environment.) If this is right, then it seems to me that there is actually something misleading in describing the content with language like “I-am-spinning”: for this disguises the fact that, its accuracy conditions notwithstanding, the content doesn’t actually have this subject-predicate structure, and indeed doesn’t represent the property of motion (or spinning) at all. And this seems to me to run afoul of the phenomenology.

    This discussion has been really helpful; please keep pressing me on it.

  15. Hi Assaf and John,

    Assaf, your first comment is in the ball park of my suggestion that the subject is an unarticulated constituent of the contents. Campbell and Perry agree that an explicit representation of the subject is otiose. Perry draws an analogy with telling time without specifying the time-zone, since in most circumstances that isn’t required. Analogously, spatial predicates don’t need to explicitly represent the subject, since the subject is almost always one of the relata.

    On a different note, I think that John is right that how we construe the contents of experience in this case cuts across the conceptualist/non-conceptualist divide: while there is no doubt that the contents of the judgement is conceptualised, what seems to be key is that the contents of experience bear the right justificatory relation to the contents of the judgement. It is a further question what the format of the contents of experience turns out to be.

    John, I think I’m beginning to see your point against the demonstrative strategy. In effect, it faces a dilemma. Since the case stipulates that the body is not in view, either the demonstrative has no object, or it’s referring to a structural feature that isn’t part of the contents. I’m not sure I fully buy the second horn of the dilemma though. But I know what would help me out here.

    I’ll get a better grip on your view if you could flesh out what exactly you take “face value contents” to be. Now on p. 2 of the paper you say “The question whether a thought, experience, or other mental state has a given content is thus equivalent to the question whether there can be judgements with that content that are representationally dependent on such a state, i.e. whether they can be judgements with that contents that are formed just by taking such a state at face value.” If I understand this correctly what you’re saying is that on the assumption that the judgement is representationally dependent on the state, then we can read off the face value contents of the experience from the judgement. Is that right?

    If so, then the crucial bit is the assumption that the judgement is representationally dependent on the state… an assumption that both the demonstrative and Perryesque proposals deny.

  16. Kranti, what you correctly identify as the “crucial bit” is not anything I mean to assume; rather, it is what I am supposed to be arguing for in the paper. The question I am posing to one who rejects the Self-Location Thesis is: what then is the face value spatial content of visual experience, if it does not include the location of the self? I take Campbell’s appeal to monadic spatial predicates to be the most plausible way of doing this, but it comes up short in the case of vection. So we add to it: purely phenomenal properties; demonstrative properties; non-visual aspects of experience; the motion of the point of view or origin of the visual reference frame; and then there is Assaf’s suggestion of an unstructured content (as I understand it). And none of these does the trick. Absent another proposal, I conclude that the content is self-locating in this case.

    Now, what exactly do I mean when I talk about FV content? I suppose I just mean: what (visual) experience itself represents, as opposed to what we are inclined to say in our visually-based judgments. And I accept for the sake of argument that the minimizing strategy is okay in general: that is, we should not attribute more to the FV content of an experience than we have to. But the method of phenomenal contrast provides an important constraint: we have to attribute to an experience (though not necessarily its FV content; there could be non-representational aspects as well) enough to explain how it differs in the way it does from other experiences it is different from. And my claim is that there is no way to do this for the vection illusion without supposing that its FV content has a self-locating element.

  17. Thanks John, that’s helpful!

    Right, I do realize that the “crucial bit” is what you’re arguing for, not assuming… I should have been more explicit in what I was trying to get at. It’s that only on assuming the truth of the thesis you’re arguing for can you read off the contents of experience by reading off the face value contents of the associated judgement. Hence for the purposes of your argument, the mode of access to the contents of visual experience can’t be via the contents of judgment — that would beg the question against your opponent.

    So what is the mode of access to the contents of visual experience such that we can read off its face value contents? Is it such there can be a difference — either in structure or constitutents — between the contents of visual experience and its face value contents? I think if I understand your views on these questions, I’ll be able to understand your response to the Perryesque proposal.

  18. It is certainly right that perception-based judgments can’t provide our only access to the content of experience if my argument is to work. But I think the mode of access we have is rather just introspection, or some other direct mode of awareness of what experience is like. We can tell in this way that there is an “aspect of motion” in the illusion of vection that is absent from the experience of world-motion. The challenge then is to explain this, and I grant that introspection can’t decide unproblematically between different possible answers. That’s where the mode of phenomenal contrast comes in: we consider different candidate proposals, and see which one provides the best explanation of the contrast in question (here, the aspect of motion in I that is missing from V). And my objection to the account in terms of demonstratives — is this the same as what you are calling the “Perryesque proposal”? I fear that it is not — is that the demonstrative contents that are being attributed to visual experience aren’t ones that visual experience could possibly have, since it’s a condition on the contentfulness of a perceptual demonstrative that the object it demonstrates be perceived.

  19. Hi John,
    I did not intend the example to conflate with your position but to be an example available for your opponent to explain what’s going on in your illusion.
    Let me summarize very briefly how I understand the debate because there might be important details that I am missing.

    There are two (relevant for the discussion) possible ways of characterizing my visual experience of an apple on the right side of my computer: something like ‘apple to the right of me’ and something like ‘apple to the right’. The former requires to represent certain position as my own position, the later doesn’t. The former is true true iff the actual world is such that there is an apple in certain position and that I am located in certain position. The later only requires there being an apple in certain position.

    The case you present is about motion. The question is: do we need, in this case, to represent our own position to explain the phenomenological change? My example aimed at showing that this might not be required. In the presented case, the waves are ‘correct’ if the glass is moving and not otherwise. It is correct if the glass moves and false otherwise. If I had a glass inside of me, then there is no need to represent any location as mine nor to represent the glass as part of my system.

    You write: “But doesn’t it then seem right to say that what the waves represent is the motion of the system, as opposed to the world around it? (I.e., because this is what causes its activation under normal conditions.)” I agree with you, but it also seems right to say that the content of my visual experience is ‘apple to the right of me’. I suspect that this way of presenting the content in the motion example begs the question against your opponent. The content is different when I move and when the environment moves (the glass is only able to indicate the former). In each case I represent a different state of the world without thereby represent my location or anything as mine. I thought that if this is right then your opponent can rejoin your argument. Am I missing something?

    I would like to ask you one question to get a better grip of the debate. Imagine a simple sensor that detects red light. The sensor is located in a more complex system. Imagine that its function is to indicate red object: the sensor, hence, represents red object according to some simple theories. One might (as Assef did in my paper’s thread) point out that the function would not be this one but to indicate ‘red object in certain location with respect to the sensor’s position’. So, the sensor state also represents its own position/orientation (the position/orientation is part of the correctness conditions of the sensor state).
    Does this characterization requires a sense of ‘point of view’ which is irrelevant for the debate or begs the question against your opponent who would claim that the content is something like ‘red object ahead’ (the orientation of the sensor seems to be relevant for the correctness conditions).

  20. Hi Miguel,

    You are right that the quick response about correctness conditions would be question-begging in the way I am setting the argument up: I take it to be possible in principle (and actual in reality, e.g. in simpler creatures than ourselves) to represent egocentric spatial location, and also egomotion, without having self-locating spatial representations in the sense I’m after here. So yes, it would be possible for the kind of system you describe to represent its own motion, and distinguish this from the motion of the surrounding environment, without the Self-Location Thesis holding true of any of its spatial representations. And perhaps this means that some of my earlier responses to your very helpful questions have been misdirected — if so, sorry.

    But what I hold is that the way we can tell that the spatial content of visual experience is richer than this is by analyzing what that experience is like. In the illusion of vection I, there is an experience of motion that is missing from the experience of world-motion V, yet I has in common with V that each involves an experience of motion: that is to say, the property ___ IS MOVING is represented in both I and V; it has a kind of generality, though not the generality of concepts. I don’t think this would be true, though, of the representational states of the very simple system you envision: indeed, it seems likely that its representations wouldn’t have any complex semantic structure at all. (This is why it seems wrong to describe the sensor in Assaf’s example as representing “red object in certain location with respect to the sensor’s position”, as this would require that it also be able to represent (say) “red object in certain location with respect to that other red object”, which it probably couldn’t do — indeed, it probably isn’t representing red objects at all, but rather just redness.) Once we recognize this (and I now see that this is a move I need to make explicit; it isn’t there in this version of the paper), the search is on to answer the question which object is represented as moving, and my argument is supposed to lead to the conclusion that it must be the self.

    What do you think of this as a response to your argument?

  21. Hi John,
    I might be wrong, but I think that, if you accept that the content of experience can be non-conceptual, the fact that you need a complex expression in English to express such a content does not entail that such a content has a more complex structure.

    You think of the couple of sensors and the glass of water, following with the analogy (again only insofar as you think it is helpful and maybe it is not anymore at this point of the discussion), as ascribing (representing) the same kind of property ‘_is moving’ to the environment and to oneself respectively. I do not see why your opponent is supposed to accept this. He could maintain that they are different properties.
    Maybe rejecting your generality* principle has unacceptable consequences, but this is not obvious to me.
    Does this make sense to you?

  22. Hi Miguel,

    I think that what you say is right: it’s not necessarily the case that a content expressed using a complex English expression is a complex content. What matters is whether there are elements (e.g. properties) in that content that are capable of figuring in different contents, i.e. that have a certain sort of generality.

    But do you really think it is plausible to maintain that there are different properties (or perhaps, one property and one non-property) represented in the two cases? I understand that my opponent could say this, but it seems quite a bullet to bite, and in any case a claim that is less plausible on its face than the Self-Location Thesis. (This goes to the point I raised in my response about burdens of proof: it is not obvious to me that I should have to show that all the alternatives to SLT are simply incoherent, as opposed to just less intuitive than it, esp. given that it is not clear what motivates us to reject SLT anyway, aside from a concern to preserve some sort of explanatory economy — but the “disjunctive” view on offer here is not obviously any more economical than SLT.)

  23. I am sympathetic to representationalism, so if there is a difference in phenomenology then there is a difference in content. I do not see what is the problem of maintaining that they are different properties. But maybe your opponent does.

    “the “disjunctive” view on offer here is not obviously any more economical than SLT”
    I am not sure, it only requires to represent properties and no self-location. In any case, the view of your opponent might not be more economical in this particular case of misrepresentation but still be more economical all things considered (this is related to the dialectic of the discussion and burdens of proof)

  24. My thought is that if they are different properties (or rather, one property and one non-property, since on the disjunctive view the “motion” in the vection illusion will not be the motion of anything at all), then the similarity between the two experiences would be unaccountable. It certainly seems as if the motion you experience in the vection illusion but not in the veridical experience is the same property as the motion you experience in the veridical experience but not in the vection illusion, doesn’t it?

  25. I think I find more convincing your first reply, trying to press on the fact that they are not two different properties (I said that I saw no problem in this claim but I am not sure anymore) and elaborating on generality*.

    At first sight, I do not see problematic to account for the similarities in phenomenology through other elements in the content.

    “It certainly seems as if the motion you experience in the vection illusion but not in the veridical experience is the same property as the motion you experience in the veridical experience but not in the vection illusion, doesn’t it?”
    I am not sure but I am tempted to deny that. Alternatively , if you were right, one could try to appeal to a higher-order mistaken judgement in the comparison of the two experience.

  26. Hi John,

    The Perryesque proposal is the one on which there are unarticulated constituents in the contents of experience — it’s supposed to be distinct from the demonstrative strategy. The reason I was pressing you on how we have access to the contents is because I was envisioning your opponent pursuing the line that our powers of introspection are limited, and there may be crucial elements of the contents of experience that we miss in introspection — a particularly acute challenge when those bits of contents are unarticulated to begin with! (Your opponent could hold that the “I” is explicitly articulated in the contents of judgement, but as we agreed, that’s not the issue. The issue, I take it, is whether the subject is explicitly represented at the level of the contents of experience.) This kind of skepticism about introspection wouldn’t be a special objection against your view in particular, but perhaps against the method of phenomenal contrast in general, which while acknowledging the limitations of introspection perhaps does not go far enough in doing so.

    I’m not terribly sympathetic to the Perryesque proposal, but I don’t yet see how you’ve ruled it out — though I suspect that the limitation is mine, not yours!

    I’ve been meaning to get to some of the other interesting points raised by your paper, but have been hampered by lack of time. Regarding your response to the non-visual difference objection, you say that if the difference were non-visual, we would expect that focusing attention on visual phenomenology would reduce the illusion of vection when in fact the opposite occurs. That sounds right, but only assuming that what’s outside the focus of attention can’t make a difference to your phenomenology — which is controversial. Does that sound right to you?

  27. Folks, I have been very grateful for this discussion — things are wrapping up, but I am not going to attempt a final response. I am about to submit what will hopefully be the final version of this paper for an R&R, and in it I am going to deal as much as possible with these questions. Check my website or PhilPapers page in the next month or so …

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Google photo

You are commenting using your Google account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s