Qualia! (Now Showing at a Theater near You)

Eric Lormand
University of Michigan
Philosophical Topics, 22: 127-156, 1994

Substantive notes are marked with a double asterisk**

The characters are fantastic!
–Quining Quarterly

This show brings the house down!
–Cannes Explained

Despite such widespread acclaim, there are some influential theater critics who have panned Qualia!–most recently and vehemently, Dan Dennett. In his review, "Quining Qualia,"<1> he laments several characters of Qualia! as unrealistic and even kitschy. This literary criticism begets social criticism in Dennett’s Consciousness Explained,<2> where he puts the blame for Qualia! and similar productions on any "cartoonsian" theater that promotes them, and ultimately on the rather imperceptive "audience" itself. My aim here is to respond to these charges. I first summarize the issues at stake, with brief excerpts from Qualia!’s playbill.

Playbill

Qualia! is a bold drama about qualia. On more cautious stories, to say that a mental state has "qualia" is to say that there is something it’s like to be in that mental state. If there are any states with qualia, any states it’s like something to be in, the clearest examples are conscious "experiences," namely: conscious perceptual representations, such as tastings and visual experiences; conscious bodily sensations, such as pains, tickles, and itches; conscious imaginings, such as those of one’s own actions or perceptions; and conscious streams (or trains) of thought, as in thinking "in words" or "in images." We sometimes try to describe the particular qualia of these experiences, for example, by saying that a given pain is "sharp" or "throbbing" to some degree, or that a given visual image is "blurry" or "moving." Qualia! is bold in its portrayal of qualia properties as having several second-order properties (or characters) that seem to resist familiar kinds of scientific explanation.

Cast of Characters (in order of appearance)

Intrinsicness (also appearing as: nonrelationality, nondispositionality, reaction-independence)

To say that a quale Q is "intrinsic" to a conscious experience is to say that only things that are part of the experience help constitute what the experience is like as regards Q. Other things, such as the stimuli that may cause the experience or the behavior and further mental states that it may cause, do not seem even partially to constitute its having the quale.

Directness (also appearing as: immediacy, intimacy, acquaintance, noninferentiality)

To say that one has "direct" access to a quale Q is to say that one can acquire evidence about the Q-ness of one’s experiences without the kinds of inference one needs about the mental states of other people. One does not appear always to use such inferences in one’s own case–most clearly, in detecting what it’s like to have one’s current conscious experiences.

Reliability (also appearing as: privilege, intimacy) – Understudies: infallibility, incorrigibility

To say that one has "reliable" access to a quale Q is to say that one has a source of evidence about the Q-ness of one’s experiences that is more reliable than one’s access to other empirical facts, or perhaps that is even infallible.

Unanalyzability (also appearing as: atomicity, simplicity, homogeneity, grainlessness)

To say that a quale Q is "unanalyzable" is at least to say that what an experience is like as regards Q is not wholly constituted by what it or any other experience is like in other regards. Perhaps few think that all qualia are unanalyzable; for example, what we might be tempted to call a "round, red quale" may be analyzable into "roundness qualia" and "redness qualia." But many would say that some qualia–perhaps, "points" of color qualia–are unanalyzable into simpler qualia. What it’s like (normally) to see red does not seem to reduce, perhaps not even partially, to what it’s like to see anything else.

Ineffability (also appearing as: inexpressibility, incommunicability)

To say that a quale Q is "ineffable" is at least to say that only subjects who have had Q-experiences understand what it is for an experience to have Q. It seems impossible for one to understand what it’s like without having undergone what it’s like, and impossible for a subject to specify what it’s like verbally or otherwise.<3>**

Privacy (also appearing as: subjectivity)

To say that a quale Q is "private" is to say that it is impossible for one to confirm adequately the hypothesis that someone else’s experience has quale Q, even when one is in a position to form the hypothesis (i.e., even neglecting or overcoming the alleged ineffability of Q). This notion of "adequate" confirmation must be specified carefully. For example, on views according to which qualia supervene on precise total brain state (or machine state, or even soul state), one might in fanciful cases know that two subjects are having the same qualia by knowing that they are in precisely the same total brain (machine, soul) state. On other views according to which qualia supervene on limited aspects of brain (machine, soul) state, one might know such comparative facts in less fanciful cases. But this would merely give one knowledge that the subjects have two sets of qualia that are the same or different, whatever they turn out to be, specifically. In other words, it would not give one noncomparative knowledge of another’s qualia. We might then cast the alleged privacy of qualia as the impossibility of noncomparative tests of another’s specific qualia.<4>**

Friends of the Theater

Friends of the Theater recognize with their contributions that Qualia! and offerings with similar themes (such as Consciousness!) would not exist without the theater and a faithful audience. In fact, qualia and consciousness are popular themes because they serve as reminders of the importance of the audience. Every Friend recognizes, to take an analogy from experience, that a mental state can only have qualia, or be conscious, if it is (or is disposed without much effort to be) itself the object of some sort of mental representation. This representation of the mental state is akin to an audience in a theater. The theater and audience are necessary conditions for the show to go on. If you would like to become a Friend of the Theater, we invite your support in response to these needs. We are pleased to reward exemplary contributions (at various levels listed below) with reserved seats (at a variety of distances from the stage).

Participating Friends ($10,000 and above)

Participating Friends hold reserved seats on the stage itself, and therefore become parts of performances such as Qualia! and Consciousness!. In the analogy from experience, Participants would claim that a mental state has qualia, or is conscious, only if it is somehow self-representational or "reflexive" (perhaps in addition to representing other things). The Theater gratefully acknowledges Franz Brentano for his outstanding past fund-raising efforts at the Participant level: "The presentation which accompanies a mental act and refers to it is part of the object on which it is directed."<5>

Perceiving Friends ($1,000 and above)

Perceiving Friends hold reserved seats a bit removed from the stage, but still within excellent viewing distance. In the analogy from experience, Perceivers would claim that a mental state has qualia, or is conscious, only if it is (disposed to be) the object of an inner-perceptual representation, a representation distinctively analogous to the products of ordinary "outer" perception of nonmental objects. The theater wishes to acknowledge David Armstrong for his valuable recent fund-raising efforts at the Perceiver level.<6>**

Believing Friends ($100 and above)

Believing Friends hold reserved obstructed-view seats farthest from the stage, but may receive descriptions of the action from the ushers, upon request. In the analogy from experience, Believers would claim that a mental state has qualia, or is conscious, only if it is (disposed to be) the object of a certain kind of belief. The Theater acknowledges David Rosenthal for his current contribution at the Believer level. Rosenthal claims that a mental state is conscious only if one has a "higher-order" occurrent belief that one is in the state.<7>** Similarly, he tries to explain the qualia of a mental state–what it’s like to have it–partially in terms of higher-order beliefs about it.<8>**

Response to the literary critic: Are the characters fantastic?

In "Quining Qualia" Dennett argues that "conscious experience has no properties that are special in any of the ways qualia have been supposed to be special";<9> the characters of Qualia! are utterly untrue to reality. He concludes, "tactically," "that there simply are no qualia at all."<10> As he mentions, there is some temptation to retreat to a more cautious conception of qualia, on which qualia need not have the offensive cast of characters. In this part I want to argue that the bolder conception of qualia–according to which they do have these characters–survives his arguments. This is merely a negative point; although I mean to assert that for all Dennett argues qualia have these characters, I do not mean to assert that qualia have them.

Intrinsicness

Dennett denies that an experience has qualia "independent[ly] of how [the subject is] subsequently disposed to behave or believe."<11> He argues that intrinsicness is untrue to the intuitive facts about experience. His several examples concern changes in liking or disliking reactions to tastes: professional coffee-tasters who gradually come to dislike a certain kind of coffee, experienced beer drinkers who gradually acquire a liking for a certain kind of beer, and cauliflower-haters who take a pill that instantly cures their dislike. Although we might be tempted to describe such cases as changes in reactions to the same taste, Dennett argues that this is a mistake, or is at best groundless. Changes in likes and dislikes seem to change taste experiences. Dennett concludes:

. . . [I]f it is admitted that one’s attitudes towards, or reactions to, experiences are in any way and in any degree constitutive of their experiential qualities, so that a change in reactivity amounts to or guarantees a change in the property, then those properties, those ‘qualitative or phenomenal features’, cease to be ‘intrinsic’ properties and in fact become paradigmatically extrinsic, relational properties. . . . When [someone] thinks of ‘that taste’ he thinks equivocally or vaguely. He . . . need not try–or be able–to settle whether he is including any or all of his reactions or excluding them from what he intends by ‘that taste’. . . . Of course I recognize that the taste [of the cauliflower after the pill] is (sort of) the same–the pill has not made cauliflower taste like chocolate cake, after all–but at the same time my experience is so different now that I resist saying the cauliflower tastes the way it used to taste. There is in any event no reason to be cowed into supposing that my cauliflower experiences have some intrinsic [and mental?–EL] properties behind, or in addition to, their various dispositional, reaction-provoking properties.<12>

Before directly assessing this argument, I want to consider where it leaves matters if it is successful. If qualia are not intrinsic, but are reaction-dependent in Dennett’s sense, a theory of qualia should also say which reactions determine qualia, and why.

In Consciousness Explained, Dennett claims that "there is no [mental?–EL] reality of conscious experience independent of the effects . . . on subsequent action (and hence, of course, on memory)."<13> He is reasonably clear about which effects are relevant to qualia. He seems to favor a kind of holism about qualia, that is:

. . . identifying "the way it is with me" in perceptual experience with the sum total of all the idiosyncratic reactive dispositions inherent in my nervous system as a result of my being confronted by a certain pattern of stimulation.<14>

Taken strictly, this holism is very implausible. Consider, for example, unexercised dispositions of an experience. Suppose a person is very poor, and is therefore such that (unbeknownst to her) the taste of coffee disposes her, if paid $1 to say "Shazam" when she tastes coffee, to say "Shazam." Does this "idiosyncratic reactive disposition" help to determine what it’s like for her to taste coffee? Presumably not. Suppose she is slowly drinking a cup of coffee while watching the stock market reports, according to which her wealth is oscillating considerably, which in turn (unbeknownst to her) oscillates her disposition to say "Shazam" if paid. Desperately trying to console herself, she moans, "Even if I can rely on nothing else in this world, at least I can rely on the taste of this coffee." We have no reason to doubt the constancy of her taste-qualia, simply because of the change in her unexercised dispositions to say "Shazam." Similarly, qualia seem unaffected by an experience’s exercised dispositions to produce wholly and clearly unconscious reactions–suppose one retains a conscious liking for coffee over a period of time, but during this time for the usual sordid reasons one develops a deeply subconscious Freudian dislike for coffee (say, it comes unconsciously to symbolize Father). Would coffee taste-qualia be altered simply by this wholly unconscious change? Would a fickle Freudian subconscious continually change them?

We can keep a principled line against Dennett’s qualia holism as follows. Dennett’s best cases of reaction-dependence involve only consciously exercised dispositions to liking or disliking reactions. For all he has argued, then, dispositions which are unexercised or which are exercised wholly unconsciously may not be constitutive of qualia.<15>** But once we ask why they are not constitutive of qualia, we can preserve the claim that qualia are intrinsic, even given his kind of reaction-dependence! As I will now try to illustrate, the key is that the consciously exercised liking and disliking dispositions may trigger desires which have their own qualia (or which cause thoughts and images with their own qualia).

The relevant qualia in a case of disliking a taste may be produced in the following way, compatibly with Dennett’s arguments:

(i) Tasting a certain food causes a purely gustatory experience g with intrinsic qualia T (i.e., the experience and its qualia are intrinsically neutral with regards to liking or disliking reactions).

(ii) g activates disliking.

(iii) The disliking in turn causes (or perhaps partly consists of) further experiences e with their own intrinsic qualia Q distinct from T, such as imagined muscular tension, inner "yuck!" speech, visual or gustatory images of refuse, etc.

When one comes to like the food, (i) may remain the same even though (ii) and (iii) change. The food may still cause g with qualia T, but the further states may be as follows:

(ii¢ ) g activates liking.

(iii¢ ) The liking in turn causes (or perhaps partly consists of) further experiences e¢ with their own intrinsic qualia Q¢ distinct from T, such as imagined muscular relaxation, inner "yum!" speech, visual or gustatory images of other enjoyable foods, etc.

In this way, when one’s likes and dislikes change, one’s total set of qualia produced by a food may be different, even though these sets are of the same type in respect of T. These claims are not ad hoc, since they would explain how we can detect taste similarities through changes in liking reactions–e.g., why the cauliflower taste "is (sort of) the same," as Dennett argues. They would also explain why thoughts of "that taste" are equivocal and vague–in the examples such thoughts can refer to T or instead to the whole complex containing T and Q or Q¢ . But the main point of the example is to block Dennett’s argument against intrinsicness, and to avoid his path to qualia holism–T can be intrinsic to g, while Q and Q¢ are intrinsic to e and e¢ , and while the whole set of qualia in each case is intrinsic to the whole set of experiences in that case.

Directness

Dennett denies that qualia are properties "essentially directly accessible to the consciousness of their experiencer," properties "with which one is intimately or directly acquainted," or "immediate phenomenological qualities."<16> He argues against direct access on the grounds that we can be wrong about qualia:

[F]ar from being directly or immediately apprehensible properties of our experience, [qualia, if they exist,] are properties whose changes or constancies are either entirely beyond our ken, or inferrable (at best) from ‘third-person’ examinations . . .. [People aren’t] introspectors capable of a privileged view of these properties, but . . . theorists whose convictions . . . are based not only on their ‘immediate’ or current experiential convictions, but also on . . . events they remember from the recent past.<17>

This argument seems to confuse issues about direct access with issues about reliable access. These can come apart. One might have direct (noninferential) evidence about one’s qualia that is less reliable than others’ indirect (inferential) evidence, or one might have reliable evidence that is no less indirect than others’ unreliable evidence. So direct access doesn’t require infallibility, incorrigibility, or any weaker kind of reliable access, and Dennett’s argument therefore does not count against direct access, even if it counts against reliable access (but see the next section).

Direct access could be threatened if the mistakes one makes about one’s qualia are systematically explainable as the result of faulty inference. To take an analogy, we make regular mistakes in identifying some of our propositional attitudes–e.g., some beliefs, desires, and emotions–and these mistakes follow a pattern: they reflect our expectations about the attitudes we rationally should have in our circumstances.<18> Presumably we make the same kind of mistakes about other mental states, such as our moods, erring systematically in the direction of the moods that we think appropriate in our circumstances. Most of our access to our attitudes and moods, then, seems to be based on self-directed inferences, analogues of the indirect access others have to our mental states, inference via commonsense rational explanation. The illusion of direct access may be fostered by the rapidity or unconsciousness of these inferences.

Could such a "hidden-inference" model work for all access to qualia? It would have difficulty explaining how we access features of our conscious experiences that are irrelevant to commonsense rational explanation, and so involve no standards of rationality or appropriateness for us to use in thinking "too highly" of ourselves. For example, no folk-theoretic principles of rationality suggest that one should feel a stinging pain rather than a throbbing pain when a limb has restricted blood flow, yet untutored subjects offer consistent (and apparently reliable) reports of stinging pain ‘feelings’. What premises could we use, consciously or unconsciously, to draw inferences about how these experiences feel? A hidden-inference account seems less suited to (all) qualia than to what we might call "rationalia," properties appropriately related to commonsense rational explanation, such as the force and content of some attitudes and the identity of moods. (I forego here such questions as whether some qualia are rationalia, whether conscious attitudes and conscious moods–like conscious "experiences"–have qualia in addition to rationalia, etc. I merely take properties such as the stingingness of pains to suggest that some qualia are not rationalia, and properties such as the contents of unconscious beliefs to suggest that some rationalia are not qualia.)

Friends of the Theater have several options for explaining the apparent directness of access to qualia. Believing Friends, for example, may at their discretion deny that access is direct, and attempt to explain it as hidden inference. A second view (natural to Believing Friends and Participating Friends) is that there is direct access to qualia, but this access is an analogue of our psychologically primitive abilities to undergo transitions from one mental state to another. Just as the transition from believing that p and q to believing that p presumably takes place without intermediate inference, so might the transition from (say) believing that p to believing that I believe that p, or the transition from having quale Q to believing that I have Q. While there presumably isn’t such primitive access to (all) rationalia, it might be thought that there is such access to (some) qualia. But this would leave a mystery about why there isn’t primitive access to (all) rationalia. If direct access requires such simple mechanisms as primitive state transitions, why would it be generally unavailable for rationalia? A third view would be more natural for Perceiving Friends: direct access is some inner-directed analogue of our perceptual (and in some sense noninferential) access to outer objects. By postulating a distinctive, nonprimitive but also nonrational, means of access, this view promises to explain why (some) rationalia seem less reliably accessible than (some) qualia.

Reliability

Perhaps it is unfair to take Dennett to be confusing directness with reliability; perhaps he simply uses "direct" and "immediate" to mean "especially reliable" rather than "noninferential." But does his argument count against reliable access? The argument as presented works only against the view that one has infallible or especially reliable memory access to past qualia, not against the view that one has infallible or especially reliable access to present qualia. Dennett argues that one can easily be wrong about "what qualia one had"<19> or wrong about their "changes and constancies" over time. But in principle one can have separate, reliable access to noncomparative (e.g., intrinsic) properties of two experiences e1 and e2 without having such access to comparative relations between e1 and e2, their "changes or constancies." This is because comparisons require comparing mechanisms in addition to accessing mechanisms, and these extra mechanisms can lack or interfere with such access. In comparing a current e1 with a past e2, one’s memory for e2 can be unreliable or fallible (or indirect, for that matter) even if current access to e1 is reliable or infallible (or direct), and even if past access to e2 was reliable or infallible (or direct). Dennett’s style of argument against reliability would face a similar problem even if he appealed to unreliable comparisons between two current experiences. For to perform such comparisons, one may have to divide attention between the two experiences, perhaps reducing the reliability (or even directness) of one’s undivided access to each.

Dennett rejects attempts like mine to find a residue of special knowledge. He criticizes the claim that "I know how it is with me right now" by asking, "if absolutely nothing follows from this presumed knowledge . . . what is the point of asserting that one has it?"<20> But substantive things do follow from this knowledge. If one’s memory (or divided attention) is reliable, or if one has evidence to that effect, one may have evidence about the relations between current and past experiences (or between multiple current experiences). And for all Dennett has argued, one can have independent evidence about the general reliability of one’s memory (or one’s divided attention). On behalf of Dennett, one might object that pure introspection could not provide independent evidence of memory reliability. But this would be irrelevant to the issue. Dennett’s opponents claim that one has reliable introspective access to one’s current qualia, not to the reliability of one’s memory (or one’s divided attention).

Once again, Friends of the Theater have many options for explaining the apparent reliability of access to qualia. Consider the views about access presented in the previous section. On the hidden-inference view, it would be natural to suppose that access is no more reliable than other kinds of empirical folk-theoretic inference. An inner-perception view of access might suggest that access is fairly reliable, at least to whatever extent outer perception is normally more reliable than folk theory. But even on this view, presumably, the alleged mechanisms of inner perception can break down or be tricked, so that one can suffer qualia illusions just as one can suffer ordinary perceptual illusions. On the primitive-transition view, access to qualia might be considered even more reliable, since there is less of an accessing mechanism to break down. Such views could provide for reliability, perhaps enough reliability for knowledge of current qualia, even if they do not provide for infallibility about current qualia. (I will discuss Dennett’s further arguments that inner theaters undermine reliability in the final section of the paper.)

Infallibility

How might a Friend of the Theater provide for infallibility, if so inclined? Dennett argues that only one move is available:

The infallibilist line on qualia treats them as properties of one’s experience one cannot in principle misdiscover, and this is a mysterious doctrine (at least as mysterious as papal infallibility) unless we . . . treat qualia as logical constructs out of subjects’ qualia judgements: a subject’s experience has the quale F if and only if the subject judges his experience to have quale F.<21>

Dennett raises a warning flag about this maneuver. Before considering it, I would like to motivate a modified version of the maneuver. Infallibilist Friends of the Theater (or, as we may call them, Friends of the Papal Audience) need not "construct" qualia out of judgments. While this would be natural for Believing Friends, Perceiving Friends would more naturally appeal to the contents of inner-perceptual representations, and Participating Friends would more naturally appeal to the (reflexive) contents of experiences. These latter options seem more plausible from the start, since beliefs (or judgments) about what current experiences are like can easily go wrong. One can even believe falsely that one is in a state it is like something or other to be in! If one can have beliefs at all while completely unconscious and lacking qualia–e.g., during dreamless sleep–there would seem to be no reason why one can’t mistakenly believe, in such a state, that one is conscious, or has qualia. Perhaps this explains certain behaviors of hypnotics, sleepwalkers, and sleeptalkers, e.g., those of us who answer the phone while apparently still in a deep sleep, and respond affirmatively when our skeptical caller asks whether we’re awake. Or, if one can have and apply a concept despite being radically confused about its proper application,<22> perhaps a blind person can radically misapply a genuine concept of conscious visual experiences, or of qualia, and mistakenly believe he is having visual experiences with particular qualia. It’s wise for infallibilist Friends to focus on reflexive or inner-perceptual access, since even if these are infallible, self-beliefs can be fallible through inattention, breakdown, or other incapacities.

Infallibility seems implausible because whatever the relevant representation r of an experience e, presumably there is in general no guarantee that (1) implies (2):

(1) r represents e as having (property) F.

(2) e has (property) F.

But, strictly speaking, the infallibilist about qualia would at most have to derive, from (1):

(3) e has quale F.

This can be done by treating (3) as an idiom, and so breaking the apparent entailment from (3) to (2). There is nothing ad hoc about this, since talk of qualia is introduced by stipulation in terms of "what it’s like," and this is probably itself an idiom.<23>** On this view, F needn’t be had by e in order for them to stand in the "is a quale of" relation, just as a property needn’t be had by an object in order for them to stand in the "exists at the same time as" or "is as interesting as" relations.

If an experience has quale F without having property F, this forces changes in the interpretations of some of the characters of qualia. First, we would need to modify the claim that qualia are properties of experience. Qualia may be real properties, with real relations to experiences, without literally being properties of experiences. Nevertheless, the property of having quale F may literally be a property of experiences, namely, the property of being represented in certain ways as F. (Similarly, having quale F may be an intrinsic property of an experience, if the experience is represented by itself as F, as Participants hold.) Such consequences are the price of nonmysterious, motivated, infallibility about current qualia. Friends of the Theater–and perhaps only Friends of the Theater–can at their discretion hold that qualia are correctly represented (or "accessed") for free in inner perception or reflexive experience, since the content of inner perceptions or reflexive experiences constitute which qualia are had.

Now, finally, to the worry Dennett raises about the infallibilist maneuvers:

Logical constructs out of judgements [or, presumably, inner perceptions or experiences–EL] must be viewed as akin to theorists’ fictions, and the friends of qualia want the existence of a particular quale in any particular case to be an empirical fact in good standing, not a theorist’s useful interpretive fiction, else it will not loom as a challenge to functionalism or materialism or third-person objective science.<24>

The dialectic is a bit tricky at this point. The friends of qualia may well consider qualia to be a challenge to theoretical explanation, without building into the notion of qualia that this challenge cannot be met. This can be so even if intrinsicness, directness, infallibility, unanalyzability, ineffability, and privacy are built into the notion of qualia, and even if these characters exist and "loom" as challenges. And, anyway, even if the quale F (as a theorists’ fiction) doesn’t itself generate insuperable difficulties for theoretical explanation, perhaps having quale F (as the nonfictional property of being fictionalized in certain ways to be F) does! Perhaps the content of the relevant fiction is something that cannot be explained by functionalism or materialism or third-person objective science. This might hold if certain claims about unanalyzability, ineffability, and privacy are true. Let’s turn to Dennett’s discussion of these characters.

Unanalyzability

Dennett denies that any qualia are "somehow atomic and unanalysable . . . ‘simple’ or ‘homogeneous.’"<25> He argues for this by considering the effects of prolonged experience:

Consider the results of ‘educating’ the palate of a wine-taster, or ‘ear training’ for musicians. What had been ‘atomic’ or ‘unanalyzable’ becomes noticeably compound and describable; pairs that had been indistinguishable become distinguishable, and when this happens we say the experience changes.<26>

Such changes can occur quickly, as Dennett illustrates:

Pluck the bass . . . string [of a guitar] open . . .. Does it have describable parts or is it one and whole and ineffably guitarish? Many will opt for the latter way of talking. Now pluck the open string again and . . . lightly [touch] the octave fret to create a high ‘harmonic’. Some people . . . will describe the experience by saying ‘the bottom fell out of the note’–leaving just the top. But then on a third open plucking one can hear, with surprising distinctness, the harmonic overtone that was isolated in the second plucking. . . . The difference in experience is striking, but the complexity apprehended in the third plucking was there all along (being responded to or discriminated). After all, it was by the complex pattern of overtones that you were able to recognize the sound as that of a guitar rather than of a lute or harpsichord.<27>

Dennett concludes that "the homogeneity and ineffability of the first experience is gone, replaced by a duality as ‘directly apprehensible’ and clearly describable as that of any chord."<28>

The trouble with this argument is that, as Dennett admits, "the difference in experience is striking" between the first and third pluckings. Given this, even if we can analyze what the third experience is like, this is no evidence that we can analyze what the first experience is like, since what they are like is different. Perhaps only the qualia of the third experience are analyzable. Similarly, although Dennett admits and even emphasizes that "the" experience changes in a case of prolonged training with wine or music, he does not seem to realize that this undermines his argument against unanalyzability. He cannot argue that the pre-training qualia are analyzable (or effable) simply on the grounds that the different, post-training qualia are analyzable (or effable). (And it is no news that some qualia are analyzable, since few would claim that no qualia are analyzable into further qualia–recall the "round, red qualia" mentioned in the Cast of Characters.)

Even if Dennett’s argument fails, we would like an account of what constitutes such changes in experience: in what way do the experiences (or the qualia) in Dennett’s cases get more complex? And how do these changes result from training? Although training may build more complex outer-perceptual discriminatory abilities, this does not seem to account for all such changes. In the guitar example, the "complexity" of the sound is discriminated "all along," so that the change in qualia does not seem to be due to a change in outer perceptions. Friends of the Theater have alternative explanations in hand. It would be natural for a Believing Friend to locate the change in the relevant higher-order beliefs, and in particular in the concepts of mental states that these beliefs involve. Training can build more complex concepts of mental states. However, such concepts do not seem sufficient for changing experience, since one can acquire complex theoretical concepts about wine-experiences and music-experiences by reading books, for example, without the associated changes in one’s experiences. On the other hand, if it is suggested that some such concepts can be acquired only by tasting wine or listening to music, why shouldn’t we think the tasting or listening changes experiences directly, rather than by detouring through concepts? It seems especially implausible to suppose that the listener in Dennett’s guitar example needs to acquire a new concept, rather than some kind of temporary perceptual modification. But what kind of modification?

A Perceiving Friend might locate the modification in one’s inner-perceptual representations of experience. Increasing attention to a nonmental object can increase the detail or intensity of one’s outer-perceptual representations of the object. Similarly, perhaps increasing attention to an experience can increase the detail or intensity of one’s inner-perceptual representations. On this view, the changes in qualia in the guitar case are due to such changes in inner perceptions. The inner perceptions on the first plucking may represent the (complex) outer perceptions simply, while the inner perceptions on the third plucking represent these same outer perceptions complexly. So the qualia on the first plucking (accessed by the simple inner perceptions) are not shown to be analyzable by the fact that the different qualia on the third plucking (accessed by the complex inner perceptions) are analyzable into simpler qualia (perhaps accessed by components of the complex inner perceptions). Perhaps training in wine-tasting or music-listening changes experiences in the same way, not merely by developing one’s theoretical concepts, and not necessarily by changing one’s outer-perceptual discriminatory abilities, but by changing one’s inner-perceptual sensitivities to one’s outer perceptions. (A similar suggestion is available to Participants.)

Ineffability

Dennett denies that "one cannot say to another . . . exactly what way one is currently seeing, tasting, smelling, and so forth."<29> He argues against genuine ineffability by attempting to undermine the following enticing example of allegedly ineffable qualitative knowledge:

Suppose . . . that I have never heard the cry of an osprey . . .. Then one day, . . . I identify an osprey visually, and then hear its cry. ‘So that’s what it sounds like’, I say to myself, ostending–it seems–a particular mental complex of intrinsic, ineffable qualia. I dub the complex ‘S’ (pace Wittgenstein) . . .. My perceptual experience has pin-pointed for me the location of the osprey cry in the logical space of possibilities in a way verbal description could not.<30>

Against this, he first downplays the knowledge claims in the example, and then upplays the possibilities for expressibility. His first point, downplaying one’s knowledge of one’s qualia, is that

. . . from a single experience of this sort I do not–cannot–know how to generalize to other osprey calls . . . [and] I do not and cannot know . . . which physical variations and constancies in stimuli would produce an indistinguishable experience in me. Nor can I know whether I would react the same (have the same experience) if I were presented with what was, by all physical measures, a re-stimulation identical to the first. I cannot know the modulating effect, if any, of variations in my body (or psyche).<31>

Here Dennett seems to confuse specifying a particular qualia complex caused by a specific osprey call with (a) specifying what is common to the sounds of all osprey calls, and (b) specifying the external property that causes the particular qualia complex. To "dub" the specific qualia complex, I need not know whether it is similar to other osprey-caused qualia complexes, or even whether it is caused by an osprey. I can dub a specific osprey egg "E" without knowing about any other osprey eggs, or even knowing whether it is an osprey egg. Why should I need similar information to dub a qualia complex? Dennett seems to trade on an ambiguity in the phrase "the osprey cry"–it can mean "this specific osprey cry (as it strikes me now)" or it can mean "the standard osprey cry." Even if the perceptual experience has not "pin-pointed for me the location of [the standard] osprey cry in the logical space of possibilities," it still may have "pin-pointed for me the location of [this specific] osprey cry [as it strikes me now] in the logical space of possibilities," and it is only the latter knowledge that a friend of ineffability should claim in Dennett’s example.

The second part of Dennett’s strategy is to upplay the expressibility of one’s knowledge of one’s qualia. The most he concedes to ineffability is that "the only readily available way" of specifying the osprey call is by a description like "the property I detected in that event."<32> He suggests that there are ways to specify qualia in principle, even if it is practically difficult. Expressing the knowledge is supposed to be difficult for the following reason:

My experience of the osprey cry has given me a new way of thinking about osprey cries . . . which is practically ineffable . . . because it is . . . such a highly informative way of thinking: a deliverance of an informationally very sensitive portion of my nervous system.<33>

But the richness of the information represented does not seem relevant as an explanation of alleged ineffability. High information sensitivity is not sufficient even for apparent ineffability. My retinal and other very early visual representations are as rich or richer in difficult-to-express information as the osprey experience, yet I can say exactly what it’s like to have them: nothing! And high sensitivity seems unnecessary for ineffability, because simplifying the information does not alleviate the appearance of ineffability. If completely blind people have trouble understanding what visual experiences are like for someone with normal vision, they would appear to have the same sort of trouble understanding what visual experiences are like for someone capable only of detecting a few points of light, without information about color, shape, motion, or depth. So Dennett succeeds neither in identifying the source of the difficulty of expressing qualitative knowledge, nor in alleviating it.

Privacy

Dennett resists the claim that "any objective, physiological, or ‘merely behavioral’ test[s] . . . of [qualia] are . . . systematically impossible."<34> To argue for this he relies on the same kinds of cases that he uses against directness and reliability, namely, cases in which, using introspection and memory alone, one may not be able reliably to compare current qualia with past qualia. In such cases, he argues, one may need to appeal to "outside" help–an "objective, physiological, or ‘merely behavioral’ test." Does the availability of outside tests threaten privacy? Not clearly, since as mentioned in the Cast of Characters a defender of privacy need not deny that all external tests are possible, but only that noncomparative external tests are possible. And, in fact, all of Dennett’s cases of external tests are comparative in just the way that makes them irrelevant. At best, the outside procedures directly test the comparative claims that the subject’s current qualia are the same or not the same as his past qualia, without "caring" how the qualia are or were, specifically. If the subject has any noncomparative knowledge of his current qualia, then coupled with this knowledge the external comparative tests can give him noncomparative knowledge of his past qualia. What the defender of privacy denies is that there are any purely external tests of noncomparative claims. And such privacy withstands Dennett’s arguments.

Dennett attempts to provide a substitute for the privacy that he denies. If his memory-comparison argument against privacy fails, then the substitute may be unnecessary. But perhaps Dennett thinks the substitute has some intuitive plausibility of its own, and so perhaps he intends it as an auxiliary argument against privacy.

. . . [W]hen we seem to ostend, with the mental finger of inner intention, a quale or qualia complex in our experience . . . [w]e refer to a property–a public property of uncharted boundaries–via reference to our personal and idiosyncratic capacity to respond to it. That idiosyncracy is the extent of our privacy. If I wonder whether your blue is my blue, your middle C is my middle C, I can coherently be wondering whether our discrimination profiles over a wide variation in conditions will be approximately the same. And they may not be; people experience the world quite differently. But that is empirically discoverable by all the usual objective testing procedures.<35>

I find Dennett’s substitute for privacy difficult to grasp. He seems to begin by denying that we can refer to our experiences by their qualia; when we try to refer to them, instead we refer to external properties. But in the next breath he says that we refer to the external properties by referring to our experiences–our personal and idiosyncratic capacities to respond. How then does Dennett think we refer to these internal capacities? He cannot say that we refer to them by referring to the public properties, since he thinks we refer to the public properties by referring to our capacities. And he denies that we can refer to our capacities by their associated qualia. So how does he think the reference to our capacities gets off the ground?

I also find the substitute implausible in a wider range of cases than Dennett considers. What happens if I wonder whether your blue is my middle C (or, as we might be tempted to put it, whether what it’s like for you to see the sky is what it’s like for me to hear a piano note)? I am certainly not wondering whether or not you can discriminate piano sounds from purple and turquoise, given this or that light source or background noise. I assume you can make such discriminations as well as I can. Presumably, what I am wondering about is whether, when we each refer to our "personal and idiosyncratic capacity" to respond to public properties such as the color of the sky, you pick out your capacity via the same (qualitative) property that I use to pick out mine. And when I cease to wonder this, and wonder less fancifully whether your blue is my yellow, or your blue is my blue, I seem to be doing more of the same kind of wondering. I agree that Dennett describes a coherent way of wondering, and I agree that it is subject to objective test, but this does nothing to support the view that there isn’t also a coherent way of wondering about private qualia, not subject to objective (and noncomparative) test.

While Friends of the Theater can accommodate the claim that qualia are ineffable and private, in a sense that may preclude theoretical explanation, I think Friends are in a unique position to mitigate ineffability and privacy worries. If qualia are genuinely ineffable and private, they must not be identical to physical or functional properties, or else they would be specifiable, expressible, and discoverable, in any number of objective ways. There are arguments against objective (e.g., physicalist) theories on the basis of an (alleged) impossibility of knowing what it’s like for another to have certain experiences<36>–and the problems boil down to an (alleged) impossibility of belief about what it’s like (i.e., ineffability) and an (alleged) impossibility of adequate justification of this belief (i.e., privacy). If such arguments were convincing, they would weigh against any reductive theory of qualia. But they should not be convincing. A powerful but not dismissive response turns on distinguishing qualia properties from ways of representing them. Even if a creature has a special way of representing phenomenal properties that is unavailable to us, we can in principle objectively specify, express, or test for these phenomenal properties in other ways.

If a reductive theory of phenomenal consciousness is to avail itself of this alleged alternative to ineffability and privacy, it should explain the special ways of representing qualia that are allegedly available only to "owners" of the qualia. An attractive proposal compares idiosyncratic ways of representing qualia with simple demonstratives.<37> When you see a banana and think of it demonstratively–as "this"–your thought cannot literally be expressed using complex public-language expressions such as "the thing I am seeing" or "the banana in front of my face." This is because you can have the simple perceptual-demonstrative thought without using, or even without having, a concept of seeing, of bananas, or of your face. Likewise, someone who does not perceive the banana is constrained to think of it in a different way than you do–not simply as "this" (while staring at something else). By analogy, perhaps your representations of your qualia involve, at least in part, simple inner-perceptual or reflexive demonstratives of them. This would explain why these representations cannot strictly be expressed using complex public-language expressions, and cannot strictly be shared by someone who does not "perceive" the same qualia.

Even if qualia properties aren’t ineffable and private, wouldn’t someone’s ways of representing them be ineffable and private, and therefore objectively inexplicable? Here we must be careful to distinguish between specifying (or testing for, explaining, etc.) something and having it. The distinction is normally obvious, but confusion is tempting when the subject matter is representation itself. Nevertheless, in general one can specify representations without having them. Trivially, one can specify a certain thought by its contingent properties–say, as "Napoleon’s last thought"–even if one happens not to have had it, or is incapable of having it. The nontrivial task is to specify objectively the particular ways of representing involved in unavailable thoughts, by reference to their essential properties.

But this is possible, without cheating. We cannot have a thought with the same ("fine-grained") content as Napoleon’s thought that he was the emperor of France. If we try using "Napoleon," we miss his precise "I" (or "Je") thought, which he could have had even if he forgot his name; and if we try using "I," we refer to ourselves rather than to him. But so long as we can isolate the dimensions along which ways of referring to things vary, we can objectively specify Napoleon’s way of referring to himself–e.g., as using the "simple first-person singular demonstrative concept that refers to Napoleon" rather than a concept that is complex, third-person, plural, descriptive, about Josephine, etc. Even though we don’t have such a concept of Napoleon, we’ve fully "specified" it if there can be only one such concept. If so (give or take some tinkering), we can specify Napoleon’s thought objectively: it was a thought by Napoleon predicating French-emperorhood of Napoleon using Napoleon’s first-person concept.

What we need is a comparable objective way of specifying the dimensions along which the contents of inner perceptions or reflexive experiences can vary. But this is a general problem in the theory of meaning, not a specific problem about consciousness and qualia, and it is a problem we can hope to solve. Given a general theory of reference such as a causal or correlational theory, the trick would be to discover which properties of experiences (in our case, perhaps neural or functional ones) cause or correlate in the relevant way with inner perceptions or reflexive representations. And given a general theory of sense–perhaps in terms of a representation’s relations to other representations or to its representational parts–perhaps we can discover which such relations characterize inner perceptions and reflexive experiences. If so, we could objectively specify and test for someone else’s (even a bat’s) particular way of representing a quale, narrowing the possibilities to one without even being capable of sharing this experience. In this way, unshared ways of experiencing qualia would not be private.

Response to the social critic: Must we bring the house down?

The central tenet of Friends of the Theater is this: a necessary condition for a mental state to be conscious or to have qualia is that it be represented by an "audience" of inwardly-directed representation (or that it be "presented" in a "theater" that disposes it easily to be so represented). In Consciousness Explained, Dennett argues strenuously against this view. Here is a typical formulation of his claims against Theater theories of a state’s being conscious:

. . . [T]here is no reason to believe that the brain itself has . . . any inner sanctum, arrival at which is the necessary or sufficient condition for conscious experience. In short, there is no observer in the brain.<38>

As another formulation, against Theater theories of consciousness of mental states, he denies that

. . . there is a crucial finish line or boundary somewhere in the brain, marking a place where the order of arrival equals the order of "presentation" in experience because what happens there is what you are conscious of.<39>

And against Theater theories of qualia, he argues that there is no "functional place of some sort where the items of phenomenology are . . . projected."<40> In this part I want to argue that the Theater survives his main arguments. As before, this is merely a negative point; I do not intend here to provide direct positive arguments for Friendship.

Is the screen out of focus?

Dennett directs his criticisms primarily at what he calls the "Cartesian Theater" of consciousness. An inner theater is Cartesian, apparently, if it provides for a suitably precise distinction between conscious and unconscious mental states: "a crucial finish line," "a highest point, a turning point, a point such that all tamperings on one side of it are pre-experiential, and all tamperings on the other side are post-experiential."<41> He rejects such a precise distinction, in favor of a smeared (or extensionally vague) distinction. Before considering his argument to such imprecision, I want to consider his arguments from imprecision–what damage does he think imprecision would do to Theater theories?

Imprecision would certainly spell doom for Friends of the Extremely Cartesian Theater, who suppose that the conscious/unconscious distinction is absolutely precise, so that consciousness begins with this minute quantum-mechanical event and ends with that one. But such a view would have no intuitive appeal from the start, and Dennett would not need a sophisticated argument against it. No psychological (or biological, etc.) distinction is so precise, whether or not it concerns consciousness. So consider Friends of the Moderately Cartesian Theater, who hold that the conscious/unconscious distinction is psychologically precise, in the sense that it is as least as precise as any other psychological distinction. On this view, any psychological event is either conscious or unconscious (but some underlying biological or quantum-mechanical events may be determinately neither "in" nor "out" of a conscious state). This has some intuitive appeal, though the appeal can easily be undermined without much argument. There seem to be fuzzy boundaries at the edges of one’s visual field, for example, and intuitively it is hard to tell when a peripherally moving object becomes consciously represented rather than subliminally represented. Similarly, when one’s attention is drawn to various faint pressure sensations or itches around one’s body, it can be hard to answer whether they were conscious a moment earlier. It isn’t clear that reflective common sense is committed even to psychological precision. But, at any rate, Dennett argues for psychological imprecision, and by stipulation this conclusion would undermine either the extreme or the moderate Cartesian Theater.

What matters is whether the demise of Cartesian Theaters amounts to the demise of any Theaters at all. This depends entirely on whether there can be a vague Theater. Friends of the Theater do hold that inner-directed representation is a necessary condition for consciousness. But even if Friends also hold that it is a sufficient condition for consciousness, none of this would commit them to holding that it induces a psychologically precise conscious/unconscious distinction. We can give necessary and sufficient conditions for vague categories: bald grandfathers are bald fathers of parents (and probably also fathers, with virtually no hair on their head, of parents). If consciousness is vague in some way, then a Friend of the Theater should hold that the relevant kind of inner representation is vague in the same way. And, in fact, this is independently plausible. Consider first that the distinction between mere reflexive reactions to outer stimuli and (even unconscious) representations of outer stimuli may be vague; there may be stimuli that are neither (clearly) represented nor (clearly) reacted to merely reflexively. Similarly, perhaps the distinction between (Theatrical) representations of mental states and mere reflexive reactions to mental states is vague. Another potential source of Theatrical vagueness is that a Friend of the Theater may only require a suitable kind of disposition to be innerly represented. If the specification of "suitable disposition" admits of vagueness, then the resulting conscious/unconscious distinction will be vague. Given the apparent vagueness of consciousness and qualia in the visual field periphery and in bodily sensation, for intuitive plausibility a Friend of the Theater is likely to favor, from the start, psychologically vague inner-directed representation. At any rate, Dennett cannot simply move (if that is his intention) from the claim that there is no Cartesian Theater to the claim that there is no Theater and audience at all, of reflexive experience or inner perception or higher-order belief.

Does his argument work against the (moderate) Cartesian Theater, however? This depends on whether his considerations support the claim that the conscious/unconscious distinction is psychologically vague, or, in other words, that there are psychological events that are neither conscious nor unconscious. He considers a representation (call it "b") that exists very briefly, and is promptly revised and unremembered. How could two opposed theorists test whether or not b was (briefly) conscious? Dennett argues first that after b vanishes, the subject has no possible retrospective evidence that favors either claim. Then he argues that there is no possible evidence available to others either:

Your retrospective verbal reports must be neutral with regard to two presumed possibilities, but might not the scientists find other data they could use? They could if there was a good reason to claim that some nonverbal behavior (overt or internal) was a good sign of consciousness. But this is just where the reasons run out. . . . Both models can deftly account for all the data–not just the data we already have, but the data we can imagine getting in the future. . . . Moreover, we can suppose, both theorists have exactly the same theory of what happens in your brain . . .. So in spite of first appearances, there is really only a verbal difference between the two theories . . ..<42>

Must the reasons run out? Suppose we seek to test the hypothesis that consciousness is some psychologically precise, objectively specifiable property C. Even if we can’t use subjectively unclear cases like b as evidence that consciousness is C, we can in principle use subjectively clear cases, and then apply the conclusions to initially unclear cases. We might, for example, discover that in clear cases of consciousness, there is a certain precise kind of inner-directed representation, and in clear cases of unconsciousness, there is not. Then we could conclude that b was (however briefly) conscious if and only if it was (however briefly) represented in the required way.

Dennett anticipates such an objection to his argument:

Whatever could it mean to say I was, however briefly and ineffectually, conscious of [b]? If there were a privileged Cartesian Theater somewhere, at least it could mean that the film was jolly well shown there even if no one remembers seeing it. (So there!)<43>

He responds that

. . . this is metaphysically dubious, because it creates the bizarre category of the objectively subjective–the way things actually, objectively seem to you even if they don’t seem to seem that way to you! . . . Some thinkers have their faces set so hard against "verificationism" and "operationalism" that they want to deny it even in the one arena where it makes manifest good sense: the realm of subjectivity. . . . We might classify [my] model, then, as first-person operationalism, for it brusquely denies the possibility in principle of consciousness of a stimulus in the absence of the subject’s belief in that consciousness.<44>

I agree that we can’t reject Dennett’s argument simply by rejecting verificationism in general. But nothing in the defense of the Cartesian Theater depends on commitment to the "objectively subjective" in Dennett’s sense. Friends of the Cartesian Theater need only say, quite plausibly, that things can objectively have seemed some way even if they don’t now seem to have seemed that way. They needn’t say that things can now seem some way they don’t now seem to seem, or that things can have seemed some way they didn’t then seem to seem. So Dennett’s objection that the Cartesian Theater commits us to such a metaphysically dubious category of facts fails.

Even if the Cartesian Theater were committed to the objectively subjective, it is hard to see what is wrong with this. Plausibly, for something to seem F one needs to have and deploy a concept of F (or a percept, or some similar mental representation). If so, for things to seem to seem some way, one needs a concept of seeming. If, as Dennett suggests, things can only seem F if they seem to seem F, then things can only seem some way if one has and deploys a concept of seeming. But this is a sophisticated mental concept, probably lacked by some creatures (babies, lower mammals, higher nonmammals) for whom things may nevertheless seem some way. In other words, arguably, things seem some way to babies and animals, even though things don’t seem (to them) to seem some way to them. Why should this not also happen for normal human adults? The point is clearer against Dennett’s brusque requirement, for consciousness, of a "belief in that consciousness." Is it supposed to be obvious that creatures cannot have (say) conscious pains unless they have concepts of consciousness? Is it even supposed to be obvious that we who have such concepts must use them continuously at every conscious moment?

I conclude that Dennett’s imprecision argument against the Cartesian Theater is dubious, and that even if repaired, it weighs not at all against imprecise inner theaters and audiences.

Are the tickets too expensive?

When we seek to address questions about the distribution and nature of some alleged psychological phenomenon P, in the absence of more direct evidence we can sometimes make headway by appealing to "engineering" considerations. Would there be a psychological point to P’s existing in this or that creature, or having this or that feature? It is easy to give plausible examples of mental phenomena with theoretically illuminating functions: intentions are good for the stable and reliable guidance of action, beliefs and desires for the rational formation of intentions, perceptions for the stable and reliable formation of beliefs, attention and moods for the activation of coherent sets of attitudes, etc. But these states need not be conscious to fulfill these roles. When they are conscious, does this endow them with any greater capacity to fulfill these roles, or with any new roles? Even if we could settle the question of what consciousness is good for, we might be left with the question of what having qualia is good for. What is the point of having states it’s like something to be in? Just as creatures that do not overcome threats to the stability or reliability of their actions are at best unlikely candidates for intentions and the rest, it may be that Dennett can argue against the existence of qualia and inner theaters by arguing that they are pointless or worse. Consider a few challenges to the value of qualia and inner theaters.

One extreme possibility is that qualia endow a state with no extra capacities at all, whether good, bad, or neutral. A theory that accepted this would have to explain how we could have reason to believe we have qualia, since our beliefs by hypothesis could not be influenced by the facts about qualia. Perhaps, then, qualia endow a state with the capacity to cause beliefs about qualia, but endow it with no other (independent) capacities. Dennett says that "qualia are supposed to affect our action or behaviour only via the intermediary of our judgements about them,"<45> though he doesn’t say who started this rumor, and he doesn’t provide any argument for the view. At any rate, the view would also undercut beliefs in qualia: if we don’t suppose that qualia do anything except via beliefs about qualia, why should we suppose there are qualia in addition to (false) beliefs about them?

A third possibility is that qualia endow a state with capacities besides beliefs about qualia, but these all have practically neutral or even harmful effects. Dennett argues that having qualia would waste precious cognitive resources, if it involved anything distinct from mere judgments about experienced events, such as inner "seemings" or presentations in an inner theater. For example, he considers the "phi phenomenon" of vision, in which a flash of light is followed very rapidly by a second, nearby, flash, and one seems to see, not two separate points of light, but a single light moving from the first spot to the second. On any view, one’s visual or cognitive system must somehow (consciously or unconsciously) register the second point of light, and only then jump to the mistaken conclusion that there was light at the intervening points (without the registration of the second flash, we would have no good explanation of the direction of the concluded motion). A natural idea (at least for Participating or Perceiving Friends) is that one doesn’t merely conclude that there was motion, because in addition representations of the intervening spots are "filled in" so that the entire path appears lit. Dennett objects as follows:

[T]he brain doesn’t actually have to go to the trouble of "filling in" anything with "construction"–for no one is looking. . . . [O]nce a discrimination has been made once, it does not have to be made again . . .. [R]etrospectively the brain creates the content (the judgement) that there was intervening motion, and this content is then available to govern activity and leave its mark on memory. But . . . the brain does not bother "constructing" any representations that go to the trouble of "filling in" the blanks. That would be a waste of time and (shall we say?) paint. The judgement is already in, so the brain can get on with other tasks!<46>

There are three resources that Dennett says would be wasted by presentations in an inner theater: discriminatory effort, time, and representational media (a Friend of the Theater need not posit mental "paint"–neurons will do nicely, so long as they realize states that are innerly represented). Friends of the Theater have ready responses to all of these charges.

Once one discriminates (or checks) that p, it might indeed be wasteful to rediscriminate (or recheck) that p. But Friends don’t require rechecking or even re-representing that p; instead, they require newly representing features of a representation that p. Inner perception is not repeated, redundant, outer perception. As to the other resources, it might indeed be a waste to reach a detailed judgment (e.g., of motion) and then use "paint" to put on a seems-like-this show. But Friends can hold that the judgment itself is realized in the innerly represented medium, or caused by activity in it, and so Friends can agree that there is no need to use more paint or more time after the judgment is "already in." Is it plausible that the judgment that the light moved is especially related to an innerly represented medium, when many other judgments may not be? I think it is somewhat plausible. For starters, merely judging that the light moved is compatible with merely being told this, but what it’s like consciously to judge-by-seeing that the light moved is different from what it’s like consciously to judge-by-hearing that the light moved. This difference could be explained by the presentation of a seeming in addition to the judging. The difference is not plausibly due to a simple further judgment that I see it move. First, arguably a creature may judge-by-seeing, and undergo what it’s like to do so, without having or deploying a concept of seeing. Second, we can distinguish how we saw–the precise trajectory, the speeds, the blurriness, presence or absence of double-vision, etc.–and this information might be most efficiently encoded in a seems-like-this medium, such as an innerly represented "bitmap".

These considerations indicate that the light’s seeming to move need not happen, wastefully, after one’s judging that it moved. Dennett objects to a similar claim as follows:

Some people presume that this intuition is supported by phenomenology. They are under the impression that they actually observe themselves judging things to be such as a result of those things seeming to them to be such. No one has ever observed any such thing "in their phenomenology" because such a fact about causation would be unobservable (as Hume noted long ago).<47>

This Humean point is not quite relevant to the overall argument. Even if the seeming (or the inner representation) doesn’t cause the judging, it may coincide with it, and so not waste time or paint after the judgment is already in. And even if Dennett is right that one cannot innerly observe the causal relations that seemings bear (to judgings), one may innerly perceive the seemings (which Friends may accept to be the judgings–it doesn’t matter so long as they are innerly represented). Dennett seems to care "which came first"–the innerly-represented seeming or the judging–but what is at issue is whether there are innerly-represented seemings at all. It is no solace to Dennett if such seemings exist but are caused by or identical to judgings, since he means to argue against the existence of seems-like-this presentations. It would be no good in an "engineering" argument against a trait to allow that the trait exists, and lament its pointlessness!

Furthermore, once one accepts that there are seemings, one can argue (even if one cannot observe) that they are not caused by judgments. For example, someone familiar with the phi phenomenon might resist judging that the light moved, even though it seems to move. Of course, this may be due to having initial perceptual "judgments" without considered judgments. It would still be open to Friends to identify innerly represented seemings with innerly represented initial perceptual judgments. At any rate, if we are prepared to distinguish two kinds of judging-attitudes, we should be prepared to distinguish them even when they agree, that is, even when they are attitudes to the same content. But this would also undercut Dennett’s engineering argument against presentations. If there is some point to reaching initial perceptual judgments, and then reaching further considered judgments to the same effect, why wouldn’t this show that Dennett is wrong to think redundancy of content is wasteful? And if there is no point to reaching initial perceptual judgments, why wouldn’t their existence give us reason to resist the inference from alleged wastefulness to nonexistence, in the closely related case of seemings?

Are there too many obstructed-view seats?

Dennett argues that "the Cartesian Theater is incoherent in its own terms"<48> because it asserts that we know what our experiences are like, but at the same time it undermines this knowledge by inserting implausibly too many places for error to creep into our self-reports and self-beliefs. Since nothing in this argument turns on precision, it may be directed at any Theater, Cartesian or not. While a Friend could avoid incoherence by denying reliable (or infallible) access to qualia, Dennett’s argument is interesting for its claim that Theaters jeopardize reliability.

Dennett suggests that inner-perception and higher-order-belief theories yield implausible models of how one reports one’s conscious mental states:

In order to report a mental state or event, you have to have a higher-order thought which you express. This gives us a picture of first observing (with some inner sense organ) the mental state or event, thereby producing a state of belief, whose onset is marked by a thought, which is then expressed.<49>

This is supposed to be implausible because it yields too many potential sources of reporting error:

Suppose I have my subjective experience (that’s one thing) and it provides the grounds in me for my belief that I’m having it (that’s a second thing) which in turn spawns the associated thought (a third thing) which next incites in me a communicative intention to express it (a fourth thing), which yields, finally, an actual expression (a fifth thing). Isn’t there room for error to creep into the transition between each thing? Might it not be that I believe one proposition but, due to a faulty transition between states, come to think a different proposition? (If you can "misspeak," can’t you also "misthink"?) Wouldn’t it be possible to frame the intention to express a rather different proposition from the one you are thinking? And mightn’t a defective memory in the communicative intention subsystem lead you to set out with one preverbal message to be expressed and end up with a different preverbal message serving as the standard against which errors were to be corrected?<50>

To assess this objection we have to consider each transition, ask whether it is an implausible source of error, and, if so, ask whether Friends must posit it. I look at the steps in reverse order.

The last step goes from a communicative intention to an utterance (perhaps via communicative subintentions). Errors in this step are commonplace on plausible views of speech, since there are slips of the tongue, mispronunciations, distractions, etc. "Friends don’t let friends drive drunk," the public-service announcement says, but Friends do let Friends speak drunk. The next-to-last step is from a thought about the target state to a communicative intention. Here it seems possible for a Friend to reduce the potential error. This could be done in at least two ways. First, the thought about the state may cause the utterance without first causing a communicative intention, as in a case of "blurting out." And second, even if there is an intention, it can be framed parasitically on the thought–e.g., the intention subsystem can have a fixed pointer to whatever is in the current thought-about-myself subsystem. Now consider the step from a belief about the target state to a thought. Why does Dennett think there is any step here at all? He focuses on Rosenthal’s theory of higher-order "thoughts," but he doesn’t show that Rosenthal does or should treat thoughts as anything but beliefs (see note <7>**).

This brings us to the first step Dennett considers, from the target mental state to one’s belief about it. A Friend of the Theater need not hold that self-beliefs are necessary for self-reports, since (the Perceiving Friend’s) self-perceptions or (the Participating Friend’s) reflexive experiences may suffice. Furthermore, even though Friends treat self-representations as necessary for consciousness, they may deny that self-reports need to be generated via the self-representations. But set these possibilities aside for the moment. As I suggested in the discussion of reliability, errors in the step from a state to a self-belief about it are commonplace due to delusions of rationality, at least if the self-belief is about (nonqualitative) rationalia rather than qualia. And in the case of qualia, a Friend may either find some way to insure that this step is fairly reliable (though possibly erroneous), or else a Believing Friend may employ the infallibilist maneuver, and hold that the appropriate self-beliefs constitute qualia. (Similarly, a Perceiving or Participating Friend can hold either that their favored representations constitute qualia, or else merely find some way to insure that they are fairly reliable.) Dennett considers the infallibilist line (in the form a Believing Friend might put it), and objects that you (the subject) still "will not thereby be able to rule out the possibility that [you have] a misjudgement about how it seemed to you a moment ago."<51> But this objection is beside the point. The Believing Friend can claim that a state is conscious if it coexists with an appropriate higher-order belief, and can if so inclined claim that there is no room for error between a conscious state and this simultaneous belief. Of course, reports based on such a belief may take longer to generate than the conscious state lasts (or longer than its associated higher-order belief lasts), and so reports may always be a bit retrospective. But the infallibilist about certain self-beliefs may happily admit that self-reports admit of error, since some of the steps from the belief to the report admit of error.

Dennett suggests an instructive alternative model of self-reports and self-beliefs:

It is not that first one goes into a higher-order state of self-observation, creating a higher-order thought, so that one can then report the lower-order thought by expressing the higher-order thought. It is rather that the second-order state . . . comes to be created by the very process of framing the report. We don’t first apprehend our experience in the Cartesian Theater and then, on the basis of that acquired knowledge, have the ability to frame reports to express; our being able to say what it is like is the basis for our "higher-order beliefs." . . . The higher-order state literally depends on–causally depends on–the expression of the speech act. But not necessarily on the public expression of an overt speech act. . . . We must break the habit of positing ever-more-central observers. As a transitional crutch, we can reconceive of the process as not knowledge-by-observation but on the model of hearsay. I believe that p because I have been told by a reliable source that p. By whom? By myself . . ..<52>

I think this model is very plausible, but compatible with a Theater theory of consciousness. In fact, it fits rather naturally with one. Dennett distinguishes outer speech acts from inner speech acts. Presumably an outer self-report does not cause a self-belief unless one perceives what one says out loud–presumably hypnotics or sleeptalkers who do not hear themselves do not learn from what they say. So we should ask: how does an inner self-report cause a self-belief, if not by inner perception of the inner self-report? Why wouldn’t inner saying without inner hearing be like inner hypnotic talking or sleeptalking? If Dennett doesn’t want to posit an inner hearing or self-belief about the inner saying, does he mean to substitute for it a second inner self-report about the inner saying, and a third about the second, and so on? Even if it is implausible to suppose that we innerly perceive the rationalia of attitudes and moods, it may be plausible to suppose that we innerly perceive the qualia of our conscious inner speech and other conscious experiences.

In fact, for all Dennett’s opposition to inner audiences, his criticism is not quite relentless. At some crucial points, particularly where he comes closest to convincing the reader of the relevance of his positive theory to consciousness, he flirts with a particularly strong idea of inner perception. He seeks to explain a large part of human consciousness as the result of a "Joycean machine" that innerly talks and listens to itself, and so has workings that are "just as ‘visible’ and ‘audible’ to it as any of the things in the external world" since "they have much of the same perceptual machinery focused on them."<53> Such a machine, Dennett suggests, can also draw pictures to itself, which "do indeed amount to re-presentations . . . not to an inner eye, but to an inner pattern-recognition mechanism that can also accept input from a regular (‘outer’?) eye."<54> These assertions–for they are not merely possibilities Dennett considers only to reject–are at odds with a sustained argument against metaphorical inner audiences. Maybe we can all be Friends?<55>**

 

Notes

<1>In Consciousness in Contemporary Science, ed. A. J. Marcel and E. Bisiach (Oxford: Oxford University Press, 1992, first published 1988).

<2>Boston: Little, Brown and Company, 1991.

<3>**While ineffability is often treated as an inability of a speaker to "express" or "communicate" something, the heart of the matter seems to be an inability of a hearer to grasp it. Speakers may be able to express (or even, in a sense, communicate) things that hearers can’t understand.

<4>**If one knows one’s own relevant brain (machine, soul) states and those of another, and if one has some sort of noncomparative knowledge of one’s own qualia, then one can derive noncomparative conclusions about another’s qualia. But this still counts as using a comparative test.

<5>Psychology from an Empirical Standpoint, trans. A. C. Rancurello et al. (London: Routledge and Kegan Paul, 1973, originally published 1874), 128.

<6>**See A Materialist Theory of the Mind (London: Routledge and Kegan Paul, 1968). In "Consciousness" (in The Routledge Encyclopedia of Philosophy, London: Routledge, forthcoming) I argue that a certain combination of the Participant and Perceiver theories of qualia has substantial explanatory virtues, beyond those I will describe in the present paper, and I also respond to common arguments against inner perception based on the ‘diaphanousness’ of perception and the need to avoid phenomenal sense data.

<7>**See "A Theory of Consciousness," ZIF Report 40 (Bielefeld: Center for Interdisciplinary Research, 1990). I substitute "occurrent beliefs" for Rosenthal’s "thoughts" to emphasize (with Rosenthal) that the higher-order states need not be conscious. (By an "occurrent" belief I mean a belief that is datable, active, and, for emphasis, more than a mere disposition to believe.) Colloquially at least, thoughts are occurrent states that consciously "occur" to one, as in the "stream of thought" when one makes assertions "to oneself." I argue in "Consciousness" (op. cit.) that we get a better higher-order-thought theory of consciousness by construing "thought" more along the lines of (innerly perceived) inner speech than along the lines of occurrent beliefs.

<8>**Since in my terminology a state can only have qualia if it’s like something to have the state, qualia should not be confused with what Rosenthal calls "sensory qualities", which he does not take to presuppose consciousness or higher-order beliefs or something-it’s-like (see "The Independence of Consciousness and Sensory Quality," Philosophical Issues 1, 1990).

<9>Op. cit., 43.

<10>Ibid., 44.

<11>Ibid., 45.

<12>Ibid., 61-63.

<13>Op. cit., 132.

<14>Ibid., 387.

<15>**Dennett’s qualia holism may stem from his denial that there is a suitable distinction between conscious and unconscious states. Friends of the Theater draw this distinction in terms of the presence or absence of suitable (dispositions to) innerly-directed representations. I will take up Dennett’s objections to such views in the final part of the paper.

<16>"Quining Qualia," op. cit., 46-47.

<17>Ibid., 59.

<18>See Richard Nisbett and Timothy Wilson, "Telling More Than We Can Know," Psychological Review 84 (1977).

<19>"Quining Qualia," op. cit., 60, emphasis added.

<20>Ibid., 54-55.

<21>Ibid., 55.

<22>See Tyler Burge, "Individualism and the Mental," Midwest Studies in Philosophy 4 (1979).

<23>**Here is a suggestion about how the idiom works. By stipulation, (3) is equivalent to:

(4) It is like F to have e.

The occupant of the "it" role is plausibly specified by "to have e," yielding, simply:

(5) e is like F.

We might try reading (5) literally, as:

(6) e is similar to something that is F.

However, since e may be similar to something in nonqualitative respects, if F is, say, the property of being 500 miles from Medford, (6) can be true while (3)-(5) are not true. We cannot avoid this problem by treating (5) in either of the following ways:

(7) e is similar in qualitative respects to something that is F.

(8) e is in qualitative respects as if it were F.

Suppose Q is a legitimate quale, and that where F is Q, (3)-(5), (7), and (8) are all true. Even in that case, where F is the property of being 500 miles from Medford and Q, (7) and (8) can be true while (3)-(5) are false. Such proposals as (7) and (8) would also lead to circularity, since "qualitative" is to be specified in terms of "what it’s like."

Compare (5) with an idiomatic but nonphenomenal use of "is like": "the car is like new." This doesn’t literally mean "the car is similar to something new," since everything is similar to something new in some or other respect. It seems to be idiomatic for something like "the car appears new"–which is why we don’t accept "the car is like 500 miles from Medford and new." A telltale sign of this idiomatic use of "x is like F" is (understood) accompaniment by reference to beings for which x is like F–beings to which x appears F. We can easily add "for those who see it" to the idiomatic "the car is like new" but not to the literal "the car is like everything else in some or other respects." Since "what it’s like to have e" is clearly elliptical for "what it’s like for e’s bearer to have e," it may just mean "how e appears to e’s bearer." So (5) becomes:

(9) e appears F to e’s bearer.

On a perceptual or experiential construal of "appears" (as in the "new car" case) this brings us full circle: it makes (3)-(5) equivalent to (1), where r is an inner perception of e, or e itself, rather than a belief or judgment. And (3) does not entail (2), since (9) clearly does not entail (2).

<24>"Quining Qualia," op. cit., 55.

<25>Ibid., 46.

<26>Ibid., 73.

<27>Ibid., 73-74.

<28>Ibid., 73.

<29>Ibid., 46.

<30>Ibid., 69.

<31>Ibid., 69.

<32>Ibid., 71 and 70.

<33>Ibid., 70.

<34>Ibid., 46.

<35>Ibid., 71.

<36>See Thomas Nagel, "What is it Like to Be a Bat?," reprinted in Mortal Questions (Cambridge: Cambridge University Press, 1979), and Frank Jackson, "Epiphenomenal Qualia," Philosophical Quarterly 32 (1982).

<37>For a similar suggestion, from a Perceiving Friend, see William Lycan, "What is the Subjectivity of the Mental?," Philosophical Perspectives 4 (1990).

<38>Op. cit., 106.

<39>Ibid., 107.

<40>Ibid., 127, ellipsis and emphasis in the original.

<41>Ibid., 108, emphasis on "point" and "line" added.

<42>Ibid., 124-125.

<43>Ibid., 131.

<44>Ibid., 131.

<45>"Quining Qualia," op. cit., 56.

<46>Consciousness Explained, op. cit., 127-128.

<47>Ibid., 133.

<48>Ibid., 303.

<49>Ibid., 314.

<50>Ibid., 317.

<51>Ibid., 319.

<52>Ibid., 315-316.

<53>Ibid., 225-226.

<54>Ibid., 293.

<55>**Which reminds me: thanks to Monique Roelofs for helpful suggestions on a draft of this paper, and thanks to colleagues and students at the Universities of Michigan and Maryland for discussions of my related work.