So much in this rich commentary deserves response. I welcome Rocco’s suggestion that HOT theories and AIR theories might be integrated (a dissertation project for someone: The HOT AIR Theory of Consciousness). There are also many interesting and original moves in the commentary that might help the case for restrictivism; I like the point about fine grained content distinctions at the end.

For reasons of time and space, let me focus on my reasons for resisting HOT and conceptualism (the view that the content of our phenomenal states can be specified in terms of concepts possessed by the possessor of those states). One reason to resist the latter is that discrimination outstrips recognition. I think concepts are mental items that allow for re-identification (recognizing the same object or kind from occasion to occasion). The fact that we can discriminate things (say shades of color) without being able to recognize them (store exact colors in memory) suggests to me that consciousness does not reply on concepts. Another argument against conceptualism involves learning. We often seem to acquire a concept by first experiencing something, as when tasting a durian fruit for the first time. It must be granted that concepts can influence our percepts, as in the case of seeing-as or wine tasting, but it doesn’t follow that concepts make a constitutive contribution. There is evidence that seeing-as involves a shift in attention, at least in familiar cases like Necker cubes and duck-rabbits. That suggests that concepts are having an indirect impact.

With wine, there is some controversy about whether concepts change our experience or merely what we notice in experience. I am inclined, with Rocco, to think that experience itself can change, but here again attention may do the work. When you learn what tannins are, you may selectively attend to them, thereby increasing their salience in a sip of wine. I think attention just changes salience space. If I am right, then someone without the concept of tannin could have a comparable experience to the wine expert if given a sip of wine that had exceptionally strong tannins, thus drawing attention to that feature bottom-up rather than top-down. The novice might experience the salient tannins, but still fail to form a concept. If these explanations of top-down effects prove right, then concepts cause phenomenal changes indirectly.

The same arguments (from descrimination and learning) can also be used against the HOT theory, though replies to such arguments exist in the literature. My main reason for opposing HOT theory is I haven’t seen any convincing empirical evidence for it. We don’t know where HOTs are in the brain and no psychological studies directly test for HOTs behaviorally. In contrast, there is a massive literature on the neural correlates of attention and the impact of attention on consciousness. So AIR enjoys more support. Though I will stay tuned for the HOT AIR theory.

One final thing: I agree that the prefrontal cortex is not needed for consciousness. I think attention is based in posterior areas (except for chemical senses). Attention allows perceptual information to get into working memory, and working memory requires frontal cortex. But consciousness does not require working memory, only availability to working memory. I am delighted to find another point of common ground with Rocco, and I am very grateful for his commentary.

I really like the exercise of thinking about how we grasp a common element across different houses and manifestations of the word “house.” There is a debate in cognitive neuroscience about whether there are meaning-centers in the brain the contain abstract representations of meanings. I’m on the other side of that debate. I think meanings are grounded in perception. But that puts me in the embarrassing situation of saying that there is no single thing we grasp, phenomenal or otherwise, across different house encounters—no mental symbol that arises for every mansion, shack, and brownstone. I’m inclined to say that what comes to mind for each instance is (some subset of) the whole family. Which is to say, each house brings others to mind, as well as the word “house.” So the common denominator is not a single mental symbol, but a shared web of association. I’d welcome ways to test between these alternatives.

On HOT vs. AIR, fully agreed. These are not theory-neutral conclusions, and my intuitions are colored by (I would say informed by) the theory of consciousness I endorse. So much so that I see that apparent lack of cognitive phenomenology as a problem for HOT theories. Badly stated, though, this begs the question.

On cognitive Mary, I’m inclined to say that discovery of a first-time attitude is like discovery of a first time emotion. Just as Mary can learn what love is by falling in love for the first time, she can have a first bout of perplexity. Both are emotions and all emotions are embodied.

And this leads me to say something similar about delusions (a very fruitful topic in this area that Richard introduced into our discussion). I think there is a feeling of believing, and that that feeling can attach itself to things for which one has no good evidence. I think I agree with Richard on the phenomenon, but disagree with the claim that the phenomenology of believing is not sensory. I think it is. I think it is bodily. I like the suggestion that there may be different belief-like feelings and some may not lead to action. Food for thought.

On moral (and logical) phenomenology, Richard should again be applauded for enriching the discussion. I am a Humean about morals, so I think feeling that something in morally good or bad or required involves strongly felt (and embodied emotions). Feelings of obligation may involved anticipatory guilt, for example, or a desire to act. Respect for duty may involve a feeling of respect when confronted with a command and commands can be recognized by their affective force. Of course, such feelings do not give these rules their modal force–there is no feeling of necessity–but they give them strong motivational force, and that’s all, I think, we find in the phenomenology. Logical inferences are a bit different. “Primitive compulsions” in Peacocke’s sense, may infuse our modus ponens inferences, and these compulsions may involve a process of automatic belief formation. Once we believe P->Q and P, we may feel compelled to believe that Q. When this doesn’t happen (Q may be something horrible), we may still observe that we *should* believe Q, and this may be felt in the form of dissonance between our non-Q thoughts and the intruding Q thought, or else a skeptical attention delivered upon the premises.

To say that out attitudes and inferences are explained by affect, and that the affective states are embodied is, of course, hugely controversial, and I owe a story about that. I’ve written a bit on the topic, but woefully little, and would welcome further investigation from others. Psychologists talk about “feelings of knowing” and I think feeling talk something we need to push into other attitudinal arenas.

For tip of the tongue cases, I would point out that there is a rich sensory, motor, and affective phenomenology. We say related words to ourselves, we call the first letter, we try making speech sounds, we feel frustration. Of course we also know the meaning, and this may present itself to us in phenomenology by means of imagery or verbal description. If I forget Bill’s name, I might imagine his mustache or think, “the sage of Carolina” or stammer, “L-, L-, L-, argh! Lenin? Lyndon? Logan? Damnit!”

On the attitudes, I think these have an emotional phenomenology. Doubt and disappointment are emotional states, and I think emotions are felt as changes in the body. Insert long Jamesian yarn here. Likewise for changes in intensity. A firmly held belief is experienced with a kind of arousal characteristic of conviction. If you don’t believe the attitudes have distinctive bodily markers think about how easily they can be conveyed by vocal intonation.

On Siewert’s sudden onset cases, i can report only that I don’t find these phenomenally familiar. I can imagine suddenly recalling that I have an appointment without associated words or imagery, “Damn! My proctology appointment!” I would describe the imagery. In any case, this is clash of intuitions and I’d welcome suggestions on how to adjudicate. If we have any sense that the thought proceeds the imagery it may stem from the fact that there are usually little things that trigger the thought. When someone mentions that it is Tuesday, you may suddenly recall the appointment, because of an associative link between “Tuesday” and some information in your calendar. The rapid move from the name of the say to the verbalized realization may give rise to the sense that *something* preceded the verbalized realization. It was not a conscious thought, however. Just the usually benign word “Tuesday.”

I agree with Bill that I would have no case for the “must” in my restrictionism if I could show only that imagistic reductions were *possible*. As he realizes, the arguments in the end are supposed to tip the balance in an intuition stalemate. As I see the dialectic, if restricitivsm is possible, then we should embrace it on, among others, parsimony grounds. In that sense, I’d say I’m a methodological restrictivist. I treat it as the default view and wait to be budged into the bulge.

(I obviously found Prof Smith’s comment encouraging since it suggests a coherent description from a credible authority of a process similar to the one I can describe only vaguely and with no authority at all.)

My motivation for this view was the phenomenon of being able to answer an out-of-the-blue question seemingly instantaneously. Modeling the process as extracting words from the sensory input stream, parsing the resulting phrase/sentence, determining meaning, searching memory for relevant concepts, formulating a coherent reply, and sequencing the motor commands required to vocalize the reply seemed a lot to do in the available time. On the other hand, we know from google that indexed searches of databases with “canned responses” is sometimes shockingly fast.

Again following Sellars, I have a vague sense that “being conscious” might be usefully thought of in terms of “being able to justify [to offer reasons for] what one says” – ie, to “know” what one is talking about (again, whether or not the “talk” is actually vocalized). Of course, this might have the disturbing corollary that given our sound-bite culture, many might end up being “unconscious” with respect to topics about which they have much to say.

As always, many thanks to those providing this marvelous forum.

Of course, my intuitions on the case can’t prove there is no cognitive picture in the sensory frame, but I am always struck by how hard it is to find cases where nothing sensory comes for the ride. Great case to think about.

Galen counters: surely two different thoughts that are both interesting feel interesting in different ways. Consider two thoughts expressed by, “The rat entered the factory.” This could mean that a scab worker entered, violating a union strike, or that a rodent entered. Both would be interesting. But clearly they are interesting in different ways and those differences are phenomenological. I concur. But there are two features of my story that can help explain this. First, we can form different images of these sentences, and second, once our interest is captures, a sequence of thoughts and images follow, and the sequence will differ in the two cases. So there is a lot going on phenomenologically.

The hardest cases for me are very abstract thoughts. One might be interested in a math problem, for instance, even when there is no associated imagery. But the interest may work it’s way into phenomenology by the mental activity of thinking about the problem. Trying solutions. Experiencing puzzlement. Filtering out background noise. Rejoicing in each minor breakthrough. For me, this experience is nothing but a sequence of words an emotions, so, unless mathematical meaning is symbolic, we are not experiencing the actual objects of interest. But there is a phenomenology of finding the math problem interesting. And for that reason, our level of interest is phenomenally manifest. But why think it is also manifest what has captured our interest? Two cases: either we can report what’s interesting, in which case the content seems phenomenally available because of the available phenomenology of report (a kind of illusion of cognitive phenomenology); or we have no such access. In the latter case, which is quite common, we report that we are interested without knowing what in particular made us so, and that, I submit, indicates that the actual cause of often falls outside of the phenomenal spotlight.

Of course, this can happen with art as well. We may be drawn to a picture because of its composition while thinking, incorrectly, that it is the form or the content that attracts us.

My warm thanks to Galen for this.