Aaron Z. Zimmerman
May 7, 2002
Sydney Shoemaker (2000) claims that a central class of our introspective judgments about our sensations and experiences are infallible. He arrives at this claim by modeling our introspective knowledge of our experiences on Tyler Burge’s (1988) account of our knowledge of our own thoughts. My aim in this brief essay is to explain Burge’s account, to describe Shoemaker’s extension of the Burgean account, and to raise and respond to several prima facie difficulties facing Shoemaker’s project.
Let content-mismatch be understood as a subject’s mistakenly judging that she is thinking that p when she is not thinking that p but instead thinking that q. Content- mismatch is extremely difficult—perhaps impossible—to imagine. One way to explain why first- and second-order mental states are linked in such a way as to rule out content-mismatch would be to say that the content of a first-order mental state is literally constitutive of the content of the appropriate higher-order belief. Burge claims that this relation holds for what he calls “basic self-knowledge” when he claims that judging that one is thinking that p is a higher-order mental act that involves thinking that p as a part. According to Burge, when one executes the relevant judgment (i.e. the judgment that one is thinking that p) one thinks the very thought (i.e. the thought that p) that makes this judgment true.
If Burge is right about the nature of basic-self-knowledge then there is a class of introspective judgments that are, as a conceptual matter, infallible. If our judgments about what propositions we are entertaining are self-verifying, then there is no possible world in which we make these judgments incorrectly.
Shoemaker has recently extended Burge’s model of our knowledge of our own thoughts to cover introspective knowledge of our sensations and experiential states. He begins by arguing that representationalism is correct and that our experiences and sensations are essentially representations of dispositional and occurrent “appearance” properties (2000: 25-9). The details of Shoemaker’s account of what is represented by experience and how it is represented will not concern us here. But I will suppose that Shoemaker is right at least in claiming that our sensations and experiences are representational states and, consequently, that feeling cold means being conscious of the fact that one is cold or being conscious as of being cold. If we make this supposition, applying Burge’s “containment” model of introspective judgment to our introspective knowledge of our feelings of cold means claiming that in typical cases when one judges that one is conscious as of being cold, one’s consciousness as of being cold is literally part of one’s introspective judgment. If typical introspective judgments about one’s feelings of cold are self-verifying in this way—if one is conscious as of being cold in the very act of judging that one feels cold—then these introspective judgments, like our judgments about what we are thinking, are infallible.
There are, however, several difficulties in extending the containment model to cover our knowledge of bodily sensations. For one thing, Burge’s claim that judging that one is thinking that p implies thinking that p derives much of its plausibility from the fact that judgment is a subspecies of thought. On Burge’s conception, judging that one is thinking that p means thinking about: (1) oneself, (2) thought, and (3) p, and endorsing a proposition with these three items (or representations of them) as constituents. It is, in other words, because judgment just is thought plus endorsement that judging that one is thinking that p means thinking that p. So it seems that we can only straightforwardly apply Burge’s account of basic self-knowledge to our knowledge of our own sensations if we can plausibly claim that judgment necessarily involves phenomenal consciousness. If judgment involves a kind of phenomenal consciousness then judging that one is conscious as of being cold will mean being phenomenally conscious as of: (1’) oneself, (2’) consciousness, and (3’) being cold, and endorsing a proposition with these three items (or representations of them) as constituents. If judgment involves a kind of phenomenal consciousness, then in the very act of judging that one is conscious as of being cold one will be phenomenally conscious as of, among other things, (3’) being cold.
To fully extend the Burgean model, then, one must defend the claim that judgment involves a kind of phenomenal consciousness; and this is not an entirely uncontroversial claim. It has been denied by those, like David Chalmers (1996), who argue for the possibility of zombies: creatures who are possessed of the full range of intentional states but who lack all phenomenal consciousness. It must be admitted, though, that Chalmers’ intuitions are not terribly widespread. Some distinction must be made between non-occurrent standing beliefs (such as one’s belief that the moon is more than 500 miles away) and the kinds of occurrent judgment that result from protracted, attention-occupying deliberation. It at least seems as though phenomenal consciousness distinguishes these two kinds of states by attaching itself to the latter but not the former. Indeed, several philosophers—Peacocke (1999) is a notable example—have recently mounted involved arguments for the claim that necessarily, judgment (in contrast with “mere” belief) involves phenomenal consciousness.
But the intuitively plausible claim that judgment involves some kind of phenomenal consciousness is not enough to license full application of the containment model to introspective knowledge of sensation; because even if judgment does involve some form of phenomenal consciousness it surely involves a different mode of consciousness than that mode involved in first-order consciousness of being cold. Consider a subject who judges not that she is conscious of being cold, but that she is thinking about being cold. (She might judge that she is thinking about being cold in response to a question about what she is thinking about.) If introspective judgment always involved the type of phenomenal consciousness present when one is conscious as of being cold, then such a subject would be conscious as of being cold in virtue of judging that she is thinking about being cold. But surely a subject who judges that she is thinking about being cold needn’t be conscious as of being cold in the way that we are when we feel cold. So though introspective judgment might always involve a form of phenomenal consciousness, it often does not involve the same kind of phenomenal consciousness involved in our first-order sensations. There is, therefore, an important disanalogy between our knowledge of our thoughts and our knowledge of our sensations—a disanalogy that impugns any straightforward application of the Burgean model to our knowledge of the latter.
Still, we can acknowledge this disanalogy and still apply the Burgean model to a certain extent; for even if it is conceptually possible to judge that one is conscious in way W as of being F without one’s being conscious as of being F in that way, we can insist that it is impossible for one to be wrong that one is, in some way or other, conscious as of being F. That is, we can insist that though mistakes about one’s mode of first-order consciousness are possible, mistakes about the content of first-order consciousness are not. This would, of course, not amount to a claim of genuine introspective infallibility, but it would achieve something of the result for which Shoemaker argues.
This weakening of the Shoemakerian position is not, I think, obviously mandatory. For what sort of error are we being asked to imagine? Is it really coherent to suppose that a subject might mistakenly judge that she feels cold when she is really just thinking about being cold? An error of this sort would indicate a form of derangement extreme enough to undercut our confidence that the subject in question had the conceptual competence necessary to make judgments about her experiences. And if this is right—if avoidance of this kind of error is a condition that must be met for possession of sensational and experiential concepts—we can at least claim that a conceptually equipped subject who judges that she is conscious in way W as of being F cannot be wrong about this.
Even if one does think that a conceptually competent subject cannot be wrong about which mode of phenomenal consciousness she occupies, and so does not for this reason balk at embracing Shoemaker’s view, there are further, more difficult obstacles in the way of extending Burge’s model into this area. For there seem to be fairly obvious examples of introspective error where the conceptual competence of the subject in question cannot reasonably be challenged. Gareth Evans describes a counter-example to infallibility where what is at issue is not a simple sensation like feeling cold, but a somewhat detailed perceptual experience:
Consider a case in which a subject sees ten points of light arranged in a circle, but reports that there are eleven points of lights arranged in a circle, because he has made a mistake in counting, forgetting where he began. Such a mistake can clearly occur again when the subject reuses the procedure in order to gain knowledge of his internal state: his report ‘I seem to see eleven points of light arranged in a circle’ is just wrong. (1982: 228-9)
Evans’ description of the case is surely right. The subject in question is not just mistaken in thinking that there are eleven points of light before him; he is mistaken in thinking that it looks to him as though there are eleven points of light. Moreover, the error in question is not merely verbal. If, for example, the subject wanted to snap his fingers once for every light he seemed to see he would snap his fingers eleven times, not ten. But if he had merely misspoken, or if he had thought that ‘eleven’ means what ‘ten’ actually means, he would have snapped his fingers ten times, not eleven.
Still, immediately after describing this case of substantive introspective error, Evans notes its limitations.
However, when the subject conceptualizes his experience in terms of some very elementary concept, such as a simple color concept like ‘red’, it is not easy to make sense of his making a mistake. (1982: 229)
We must ask, then, what is the difference between the two cases? Why can I make a substantive mistake about whether my visual experience is an experience as of eleven lights, when I cannot make a substantive mistake about whether it is an experience as of red?
An attractive explanation for the disparity would have it that in the latter case the subject’s visual experience represents a surface as being red, whereas in the former case the subject has an experience as of some lights which are in fact ten in number, but his experience does not represent them as being ten—just as the ancients represented some object, Hesperus, which was in fact Phosphorus, without their representing it as Phosphorus. Detailed information about the number of lights is not immediately available as the content of the subject’s experience; the subject must (further) conceptualize his experience—by, say, counting the lights—if he is to arrive at a state from which he has access to this information.
If this is an accurate description of the contents of the two experiences that Evans relates, then we have, on hand, an explanation for why only the judgment concerning the seeming number of lights is subject to error. According to the containment model, the content of an introspective judgment will in part consist in the content of the experience it takes as its subject matter. If the introspective judgment concerns the experience’s being an experience as of F, and F is the kind of property represented by experiences of the type in question, then the content of the judgment will itself, in part, contain the content of the perceptual experience—here, a representation of the property of being F (or the property itself). In such cases, substantive introspective error will be impossible. But if the introspective judgment concerns the experience’s being as of G and G is not the type of property represented by experiences of the type in question, then the content of the introspective judgment will not even in part contain the content of the perceptual experience, and the introspective judgment may well be false.
There are, according to this account, two different kinds of introspective judgments concerning one’s experience; and the Shoemakerian must allow that the containment model accurately depicts only one of these kinds. When a subject’s visual experience represents a surface as being red, the content of her judgment that she is having an experience as of something’s being red will in part consist in the content of the experience that makes that judgment true. She will judge that she is having an experience as of such-and such color, and because her experience is the sort that represents things as colored, its content will function as part of the content of her introspective judgment. But when a subject’s visual experience represents a circle of lights, the content of her judgment that she is having an experience as of eleven lights will not consist (even in part) in the content of her visual experience. Though she judges that she is having an experience as of a certain number of lights, there is no number such that her experience represents the lights as having exactly that number; the relevant “part” of the content of her judgment cannot, therefore, simply consist in the content of her experience. In such a case the subject’s judgment will instead be based on the mediating descriptive state that results from her actively counting the lights before her, and it will inherit whatever errors occur when, by counting those lights, she moves from her experience of the lights to an assessment of the number of lights represented.
Shoemaker’s account therefore has no real difficulty in allowing for the kind of error Evans describes. In fact, the containment account puts us in a position to explain why some kinds of introspective judgment are infallible while others are not. Properly hedged, Shoemaker’s claim has it that when conceptually competent subjects make judgments about what they are experiencing, where these judgments do not move beyond the content of their experiences except in applying the concepts expressed by a first-person use of “I am conscious (in way W) as of . . .”, then these judgments cannot be wrong. Introspective judgments that add more to the content of experience than concepts of self and mode of consciousness relinquish their claim to infallibility.
The kind of obviously descriptive error that Evans describes is not the only sort of introspective mistake proffered by those who deny the infallibility of introspective belief. Adopting Timothy Williamson’s (2000) terminology let us say that a mental state is luminous if a subject in that state, who considers whether she is, is always in a position to know that she is in that state. In the course of arguing against the luminosity of all mental states, including sensations, Williamson claims that our introspective beliefs must be fallible. His reasons stem from an assessment of psychological continua.
Williamson describes one such continuum as follows:
Consider a morning on which one feels freezing cold at dawn, very slowly warms up, and feels hot by noon. One changes from feeling cold to not feeling cold, and from being in a position to know that one feels cold to not being in a position to know that one feels cold . . . Suppose that one's feelings of heat and cold change so slowly during this process that one is not aware of any change in them over one millisecond. Suppose also that throughout the process one thoroughly considers how cold or hot one feels. One’s confidence that one feels cold gradually decreases. One's initial answers to the question ‘Do you feel cold?’ are firmly positive; then hesitations and qualifications creep in, until one gives neutral answers such as ‘It’s hard to say’; then one begins to dissent, with gradually decreasing hesitation and qualifications; one's final answers are firmly negative. (2000: 96-97)
Imagine Williamson’s subject at the crucial mid-morning moment when she moves from judging that she feels cold to refusing to judge that she feels cold. Let t2 be the moment at which she judges that she feels cold, t3 be the moment at which she reconsiders whether she feels cold, and t4 be the moment at which she decides that she cannot judge that she feels cold. (To claim that these moments are consecutive millisecond intervals is, of course, to assert a psychological impossibility. But we can imagine that the moments are as close as is possible for the subject in question to register the kind of qualitative change that would warrant reconsideration of her introspective belief.) Williamson rightly insists that if the subject truly knows that she feels cold at t2 there cannot be a nearby possible world in which she falsely believes at t4 that she feels cold. If there were such a world the subject’s judgment at t2 would not be reliable and so would not constitute knowledge. And there must be a close possible world in which the subject’s judgment at t4 is false, reasons Williamson, because her feelings at t2 and t4 are almost identical. To claim that there is no such nearby possible world is to claim that the subject in question can reliably discriminate between the kind of feeling she is having at t2 and the kind of feeling she is having at t4. But normal human subjects are not this reliable in their discriminations of their feelings.
Suppose that there is a perfectly determinate point at which the subject in question moves from feeling cold to not feeling cold or, at least, a determinate point at which the subject moves from feeling cold to its being indeterminate whether she feels cold; suppose too that this point lies between t2 and t4. Williamson allows that a subject might believe (with a low level of confidence) at t2 that she feels cold, and might refrain from believing that she feels cold at t4, and that this subject might therefore accurately discriminate her feelings in a particular case. But Williamson insists that no subject could reliably make this sort of fine-grained discrimination. There must be some close possible world in which the subject mistakenly believes (with, perhaps, an extremely small degree of confidence) that she feels cold at t4. And if there is a possible world of this sort nearby, then even if the subject in fact gets things right at t2, and in fact correctly refrains from judging at t4, her judgment at t2 is not reliable enough to constitute knowledge.
Williamson’s argument against the luminosity of sensations of being cold depends then on a prior rejection of the infallibility of introspective knowledge. If we are infallible about whether we feel cold, then facts about reliability cannot be used to impugn the status of a subject’s beliefs about what she feels. In the case Williamson describes there will be no possible world in which the subject mistakenly judges that she feels cold at t4.
Of course, Williamson’s fallibilist intuitions conflict with those of Evans described above. A typical use of ‘cold’ expresses as simple a concept as does a typical use of ‘red’. If we can misapply the former concept when forming beliefs about our experiences then we can misapply the latter. Indeed, a phenomenal continuum involving color experience is just as easy to imagine as one involving sensations. The question, then, is whether the borderline cases involved in these continua can give rise to false introspective beliefs. Can one judge that one is having an experience as of red when one is really having an experience as of dark orange (or an experience that is neither determinately as of red nor determinately as of orange)? Can one judge that one is conscious as of being cold when one is really just conscious as of being cool (or conscious neither as of being determinately cold nor as of being determinately cool)?
Suppose that one can. Suppose that in some borderline case a conceptually equipped subject might falsely judge that she feels cold either because it is indeterminate whether or not she feels cold or because it is not the case that she feels cold. If she is not conceptually confused, such a subject must be like the subject who mistakenly judges that he is having an experience as of eleven points of light in that the content of her introspective judgment must not in part consist in the content of her experience. Instead, her experience and her judgment must be separated by a descriptive state like the state that results when Evans’ subject counts the points of light before him. Admittedly, our subject need not be conscious of any mediating state; she needn’t experience anything comparable to the counting procedure whereby Evan’s subject moves beyond the information that accrues to him simply in virtue of his enjoying a certain perceptual experience. But we have theoretical reasons to conclude that there nevertheless must be some such intervening state or process involved in our subject’s mental economy. For if there were no intervening state or process of conceptualization, the content of the subject’s experience would be inherited by her introspective judgment, and no error would be possible.
Perhaps, then, the concept expressed (or property referred to) by the subject’s use of ‘cold’ when she asserts “I feel cold” is not yet simple enough to be the same concept (or property) contained in the content of her experience as of being cold. This would be the case if the subject’s experience had either demonstrative or non-conceptual content.
If the subject’s experience had demonstrative content we could say that she judges that she is conscious as of this, and that this introspective judgment is true in part because it includes the very demonstrative concept (or demonstrated property) that makes up the content of her first-order experience. The subject’s mistake would then come in her judging (or assuming) that having an experience as of this is having an experience as of being cold. Her error would resemble the mistake Evans’ subject makes when he judges, “My experience as of this number of lights is an experience as of eleven points of light.”
If the subject’s experience had, instead, non-conceptual content, then, if the content of her introspective judgment were in part constituted by the content of her experience, her judgment would be ineffable. We could imagine the subject making her introspective judgment by concentrating on the properties represented in her experience and thinking “I am conscious as of . . ..” But if the ellipses in this expression were filled in—even with a demonstrative expression—it would express a frame of mind the content of which had elements beyond those involved in the content of her initial introspective judgment. As such it would not be infallible—or if it were, its infallibility could not be traced to the fact that its content literally included the content of the experience it was about.
In fact, Shoemaker tentatively endorses the second of these two options (2000: 271); for though he acknowledges that there are theoretical problems with allowing that certain judgments have partly non-conceptual contents, he accepts that the content of experience is non-conceptual and he wants to hold onto the view that the content of certain introspective judgments literally include the content of the experiences they are about.
It seems, though, that adoption of either view of experiential content would allow one to reconcile introspective infallibility with Williamson’s intuitions about borderline cases. If one allows that experience has demonstrative content and that our introspective beliefs can inherit this content, one can maintain the infallibility of a certain kind of introspective judgment. A similar option presents itself if one allows that experience has non-conceptual content that can be inherited by our introspective beliefs. And if there are infallible judgments of either one of these kinds then there will be kinds of experiential state—typed either via their demonstrative contents and/or some non-conceptual measure of qualitative similarity and difference—that are not vulnerable to Williamson’s arguments against luminosity. One might agree with Williamson that feeling cold is not luminous, but insist that states of being conscious as of this, or experiential states with non-conceptual content, survive as candidates for luminosity.
There is a great deal that must still be done if we are to adequately assess Shoemaker’s attempt to extend the Burgean model to our knowledge of our sensations. Some philosophers—following Kelly (2001)—might claim that a subject can have an experience while failing to satisfy the necessary conditions for possession of the demonstrative concepts that purportedly figure in the content of that experience. If we think that the content of experience is demonstrative, and that the content of introspective belief is partly so, we must show that this objection is based on an incorrect view of the possession conditions for demonstrative concepts. Some Wittgensteinian philosophers might claim that an ineffable introspective state of the sort Shoemaker imagines would fail to satisfy conditions that must be met by all genuine instances of judgment. If we want to reconcile the containment account with the view that the content of experience is non-conceptual we have to find some way to make sense of the claim that there are truly ineffable judgments—judgments that are in part non-conceptual in content.
But though the Shoemakerian has a great deal of explaining to do, the prospects for his view look pretty good when contrasted with those of its competitors. When in some borderline case I say that I am not certain whether I feel cold, there is surely a sense in which I know exactly how I feel. What I am unsure of is how to best describe what I am experiencing. How should we understand the pre-descriptive knowledge that I have but cannot effectively communicate in such a case? It would be absurd to say that it is inferential; and it seems too intimate to be accurately modeled on the highly fallible, highly contingent knowledge gained through observation of external objects. But if not through observation and not through inference, how can we secure introspective knowledge of our sensations? The containment model, though incomplete, is currently the most promising answer being offered.
Brewer, Bill (1999) Perception and Reason, Oxford: Oxford University Press.
Burge, Tyler (1988) “Individualism and Self-Knowledge,” Journal of Philosophy, 85, 11,
Chalmers, David (1996) The Conscious Mind: in Search of a Fundamental Theory, New York:
Oxford University Press.
Dretske, Fred (1995) Naturalizing the Mind, Cambridge, MA: MIT Press.
Evans, Gareth (1982) Varieties of Reference, Ed. John McDowell, Oxford: Clarendon Press.
Graff, Delia (2001) “Phenomenal Continua and the Sorites,” Mind, 110, pp. 905-35.
Harman, Gilbert (1990) “The Intrinsic Quality of Experience,” Philosophical Perspectives, 4,
Ed. James E. Tomberlin, Atascadero, CA: Ridgeview, pp. 31-52.
Heck, Richard (2000) “Nonconceptual Content and the ‘Space of Reasons’,” The
Philosophical Review, 109, 4, pp. 483-524.
Kelly, Sean (2001) “Demonstrative Concepts and Experience,” The Philosophical Review,
110, 3, pp. 397-420.
Lycan, William G. (1996) Consciousness and Experience, Cambridge, MA: MIT Press.
McDowell, John (1994) Mind and World, Cambridge: Harvard University Press.
Peacocke, Christopher (1992) A Study of Concepts, Cambridge, MA: MIT Press.
-----, (1998) “Nonconceptual Content Defended (Comment on John McDowell’s Mind and
World),” Philosophy and Phenomenological Research, 58, pp. 381-88.
----, (1999) Being Known, Oxford: Clarendon Press.
----, (2001) “Phenomenology and Nonconceptual Content,” Philosophy and
Phenomenological Research, 62, pp. 609-15.
Shoemaker, Sydney (2000) “Introspection and Phenomenal Character,” Philosophical
Topics, 28, 2, pp. 247-73.
Tye, Michael (2000) Consciousness, Color and Content, Cambridge, MA: MIT Press.
Williamson, Timothy (2000) Knowledge and its Limits, Oxford: Oxford University Press.
 Other defenders of representationalism include Fred Dretske (1995), Gilbert Harman (1990), William Lycan (1996) and Michael Tye (2000). Whether feeling cold is being conscious of the fact that one is cold or being conscious as of being cold depends on what kind of property being cold is. If being cold is entirely a matter of body temperature then feeling cold cannot be type-identical with consciousness of the fact that one is cold, because one can feel cold when one’s body temperature is quite high (as when one has a fever). If, as Shoemaker claims, being cold is instead a dispositional or occurrent appearance property—e.g. the disposition to cause conscious states of a certain qualitative type—then it may be that feeling cold just is being conscious of the fact that one is cold. I am sympathetic toward Shoemaker’s account, but for the purposes of this paper I will remain agnostic on this matter. To avoid cumbersome paraphrase I will write of feeling cold as being conscious as of being cold, but this choice will not affect my arguments.
 I have tried to explain, in an informal way, how Williamson’s reasoning relies on a rejection of infallibility. But it is also clear that Williamson depends on a rejection of infallibility when he tries to support the crucial premise of his semi-formal argument against luminosity. This premise—(1i)— says that if the subject in question knows that she feels cold at some time during the day she must actually feel cold at the moment that follows. Williamson thinks we should accept this premise, because knowledge requires reliability. In Williamson’s words, “ . . . if one believes outright to some degree that a condition C obtains, when in fact it does, and at a very slightly later time one believes outright on a very similar basis to a very slightly lower degree that C obtains, when in fact it does not, then one’s earlier belief is not reliable enough to constitute knowledge” (2000: 101). Obviously, if one is truly infallible about one’s sensations, then one cannot be wrong about one’s sensation at some later time, and the purported fact that one easily could be mistaken at this later time cannot be used to impugn the epistemic status of one’s initial introspective judgment. Williamson acknowledges this when he admits that one must assume “limited discrimination in the belief-forming process . . .[to] deduce failures of luminosity” (2000: 126-7). But Williamson fails to appreciate that his general rejection of luminosity is put in jeopardy by this admission; for Williamson does not argue against the Burgean view that judging that one is entertaining the proposition that p conceptually entails that one is entertaining p, and if this entailment holds, then the assumption of “limited discrimination” necessary to argue against the luminosity of states of entertaining propositions cannot be made. Without an argument against Burge, Williamson has no convincing argument against the intuitively plausible claim that our thoughts are luminous.
 Which is not to say that it is easy to imagine in its full complexity without misdescription. See Delia Graff (2001) for the claim that prominent accounts of phenomenal sorites arguments misdescribe the experiential facts.
 Bill Brewer (1999) and John McDowell (1994) argue that the content of experience is demonstrative. Richard Heck (2001), Sean Kelly (2001), and Christopher Peacocke (1998) and (2001) argue against this view. The dissenters join Evans (1982) in claiming that the content of experience is, in fact, entirely nonconceptual. There is though a point of agreement between the parties to this debate: they all seem to have been convinced by Evans’ reflection on the “richness” of experience (1982: 229) that the content of experience does not contain non-demonstrative concepts.
 Shoemaker (2000: 269) says that he is persuaded that the content of experience is non-conceptual by arguments advanced in Evans (1982) and Peacocke (1992).