Philosophy, Film, Politics, Etc.

Saturday, June 27, 2009

Discussion with Stevan Harnad, continued

This is the second email I just sent to Stevan Harnad. (See here for an explanation.)

Hello again,

Here is the rest of my response. It includes a response to earlier points about Descartes and skepticism, as well as a couple of points about categorization and definitions.

SH: "
the category in question is color, not difference in color or shape."

I don't see a difference here. Categories are based upon differences. The category of "color" implies perceived differences in color, or else it is meaningless.

“People no more need a definition of feeling than they need a definition of green.”

On the one hand, you say an ostensive definition is sufficient. On the other hand, you say that no definition is necessary. Isn't that inconsistent?

Some kind of definition is necessary. And people can define green. The category “green” is complemented by other colors. We can ostend green. You cannot provide an ostensive definition for what you mean by "feeling," however, because you claim to be talking about something with no complement. So you must opt for no definition at all. That makes your argument incoherent.

SH: “there is one fact – and one fact alone – of felt experience that is not open to skeptical doubt: that it is felt.”

I do not doubt the existence of feelings. But on what grounds can you assert that this is the only indubitable fact of experience?

SH: “Let me restate ['I am not thinking'] in the form that makes the contradiction more obvious.”

I think the contradiction is obvious enough, and replacing "thinking" with "feeling" doesn't make it more obvious to me.

I never said Descartes was wrong to think “I am not thinking” implies a contradiction. But it does not surprise us with a new kind of certainty; it only reminds us of a grammatical convention we already know.

SH: “what the method of doubt further reveals is that there is, surprisingly, a second kind of certainty, over and above logical necessity.”

The certainty of the cogito is grounded in logic, in the contradiction implicit in “I am not thinking (or feeling).” One might say, following Hintikka, that there is an existential inconsistency here in addition to the inferential one. Earlier I challenged this point, on the grounds that the "feeling" or experience of a contradiction is open to interpretation. That leaves us the cogito only with the certainty of logic, which is a matter of grammatical convention. So why claim there are two kinds of certainty here?

What is logical necessity, if not grammatical convention? What constitutes the necessity of truth—in mathematics or anywhere else—if not grammar?

Descartes experienced a contradiction when he thought “I am not thinking” because he was tied to his grammar. This does not make “I think” a necessary truth. It is contingent upon the grammar itself, and this may or may not be necessary. The question we might ask is, necessary for what? Or, to put it another way, how is a grammatical rule established as such?

Grammatical rules are established on the basis of their application. In what sense is an application of a grammatical rule necessary? Again, the question is: necessary for what?

SH: “No interpretational issues at all.”

I think this is demonstrably false. We can imagine plenty of scenarios in which the sentences “I do not feel anything” and “I am not thinking” are sensical, non-contradictory utterances.

A doctor asks, “Can you feel this?”

The patient replies, “No, I don’t feel anything.”

A wife says to her husband, “You’ve put on one red sock and one green sock!”

The husband replies, “I’m not thinking. My mind is somewhere else today.”

It takes an effort, however slight, to think of “I am not thinking” (or "I am not feeling") as involving a contradiction, and this act of intellection is just as subject to error as any other. This is not to say that the contradiction you are talking about is not there; it is only to say that seeing the contradiction amounts only to remembering a grammatical convention.

You accuse me of misunderstanding skepticism by confusing “truth” with “certainty.” I have made no such mistake. The issue here is what it means to doubt, and what it means to be certain. You can say that you are not certain that you have a body, but saying it does not amount to anything. It is a meaningless claim. Saying “this isn’t real” is similarly without meaning, unless it only means, “this isn’t what it appears to be.” But not-being-what-it-appears-to-be is not not-being-anything-at-all. One can doubt one’s interpretation of what is real, but not the reality of what one is interpreting.

Following Wittgenstein, I do not view hyperbolic doubt as a viable means for establishing necessary truths. Epistemological foundations are constructed, not revealed. Grammars are necessary only with respect to particular ends and ways of living, and these may be adopted or criticized like anything else. Consider, for example, the critique of the ego made by Buddhists and some Western philosophers. Buddhists would not say “I am not thinking”; rather, they would say, “the I is an illusion.”

I would not say the “I” is an illusion, exactly. Rather, I think it is a grammatical construction, something which exists as clearly as any other social, political, or cultural institution.

Earlier, I asked you to try an experiment which I think demonstrates the fact that the “I” is a grammatical construction. (Or, to put it in the language of feeling and avoid the personal pronoun, we can say the grammatical construction is the mysterious feeler which, you say, indubitably feels feelings.) The experiment: Try to observe yourself thinking or feeling something else. For example, try to locate yourself as an observer (or as a feeler) in your own observation of a table.

It is plain to me that all one observes is the table, and that any attempt to observe something else—an observer or feeler—requires language. There is a strong temptation to observe yourself by speaking to yourself, by saying, “I am looking at the table,” or something like that. I suggest that there is no observation of an “I” (or a “feeler”) outside of such utterances; and uttering “I exist” is not observing that one exists.

Descartes was on to something. He believed that he existed as pure consciousness only so long as he was thinking. I would restate that as follows: The “I” exists only in so far as a particular grammatical convention is employed. The “I” comes into existence when it is thought, as Descartes said, but it only exists as a rule of language. It does not refer to anything beyond that language. (Similarly, democracy does not exist beyond the exercising of particular social conventions.)

For Descartes, the “I” had to be of a particular substance. For you, there is to be a mysterious, indubitable feeler. For me, these are just ways of speaking without specifiable extra-linguistic referents. We should not presume that, because grammatical conventions exist and have some use, that they must refer to anything that exists outside of that grammar, or that they reveal any non-grammatical truths.

I am not advocating global skepticism or nihilism. Rather, I am advocating practical limits on skepticism. I know I am thinking, because this is my way of life, and nothing more.

A Discussion with Stevan Harnad

Stevan Harnad is an accomplished professor, having founded and edited a respected academic journal, Behavioral And Brain Sciences. He is not a professional philosopher, but he has published articles related to cognitive science and he has a lot to say about feelings and consciousness. I've been engaging him in a discussion at the PhilPapers forum devoted to the explanatory gap. We've touched on side issues related to Descartes, skepticism, and categorization. I am continuing our discussion by email for now, at David Chalmers' suggestion. The following is the first of two emails I just sent Stevan. (The second email can be found here: "Discussion with Stevan Harnad, continued".)


David Chalmers has suggested that I take our discussion to email. I hope you do not mind. More and more of my point-by-point responses to you have been rejected by the editorial board (i.e., Chalmers) on the grounds that they are too long and too detailed to be of general interest to the forum members. (I don't have professional status, which is why my posts must pass review.) We can try to continue on PhilPapers, if you prefer, but I will not be able to address your points nearly as thoroughly as I would like. Or, if you like, we can move our discussion to another public forum of your choosing.

(This is why I have not responded to your most recent points about Descartes and skepticism. Twice I prepared responses, and both times they were rejected for being too long and detailed. I will send along the relevant content of those rejected posts in another email.)

“When I say I lifted my finger because I felt like it, I am not talking about the functionality of my brain but about the causal power of my mind. Unpacked, that amounts to telekinesis.”

I don’t think this is a fair analysis of the common language. When people say they lifted their finger because they felt like it, they aren’t necessarily postulating mind/body dualism. They could well be talking about the functionality of their brain.

It still looks like you are confusing the sense and referent of the term "feelings." You say, (1) the referent of a feeling is the feeling itself, and (2) the sense of a feeling is what allows you to pick it out, which is the feeling itself. By your definition, the sense and referent is the same.

I understand the sense of the term "feelings" differently. It is not defined by any rules we have for directly picking out feelings. Rather, it is in the rules we have for picking them out indirectly, as causes of observable behavior. (E.g., How do you know you are in pain until you observe yourself reacting to something?)

You observe pain behavior, and interpret its cause as "pain." Thus we say "pain is felt," which only means "pain exists." There is no invariant feeling here, but rather a family resemblance of neural patterns which, due to similarities in observable behavior and a fair amount of fluidity in the common language, are indirectly referred to as “pain.” (Compare the pain of heartbreak with the pain of stubbing your toe. I wouldn’t assume there is an invariant feeling—or an invariant neural pattern—common to both.)

In everyday language, we talk about feelings as being felt. Yet, when we talk about those feelings in the language of neuroscience, as patterns of neural activity, we do not say they are felt. This does not mean the "feeler" is mysterious, or that the feeling is inexplicable. Rather, it means that the notion of "feeling" is dependent upon a certain ignorance of how human behavior actually works. It's not that feelings don't exist. It's just that our everyday language is not very good at picking them out. Your felt-functing/unfelt-functing distinction is thus a category error, a mistake made by taking categories from everyday language and assuming they apply to the language of neuroscience.

You can ask, "why are some neural patterns feelings, and others not?" That is like asking, "why are some chemical reactions distillation, and others not?" How you answer will depend on what you are trying to understand.

SH: “If you can categorize X's at all, you have to be able to detect whether an instance is or is not a member of the category X.”

Your point rests upon the supposition that, for all valid acts of categorization, all questions of membership presuppose a correct answer. Why make this supposition?

To categorize, you have to be able to decide whether an instance is or is not a member of the category. But that decision need not be based on a pre-existing criterion. You can just make one up. The only problem is when people assume that some criterion has to be there head of time, somewhere, hidden from view.

SH: “It matters not a bit whether you can categorize X's correctly by detecting that all X's are P and all non-X's are not-P or you can categorize X's correctly by detecting that all X's are "L or M or not-N or (more Q than R) or (if S then not-T)", otherwise they are not-X's. Just let P = "L or M or not-N or (more Q than R) or (if S then not-T)".

From this it follows that, for every P, there is an invariant Q which is shared by both P and NOT-P, where Q = (P or NOT-P). Doesn't this undermine the notion of an invariant?

SH: “What is uncomplemented is feeling itself, because there is nothing that non-feeling feels like; it is a contradiction in terms.”

If this claim makes sense, then the notion of “non-feeling” must make sense, too. Can we agree on that?

Consider the adjective, "special." It is possible that, whenever you categorize something as “special,” you are using a new criterion. This is not problematic or strange. It would not undermine the way we use the word “special.” In fact, we might think that our use of the term is in some ways dependent upon this possibility. (After all, when I ask you what is special to you, I am not asking about what has always been, or what will always be, special to you; but only about what is special to you now.) So there is no need to postulate an invariant here; and there is no need to think that there must be a correct way to categorize members of this set.

Thus it is with many adjectives in our common language. Must we draw a strong line between adjectives and nouns here? Isn’t it plausible that a number of nouns and noun phrases enjoy currency in our language, not because there is a set criterion of correctness, but precisely because there is none? Mightn’t “feeling” be such a noun, and "what it is like to be an X" such a noun phrase?

SH: “It could be (as Wittgenstein would point out), that this feeling I am (undeniably) feeling right now -- which feels to me like agony, and exactly like the agony I have felt before -- is not in fact the agony I felt before; indeed, perhaps I've never felt agony before; it only feels as if I've felt it before.”

That would be Kripke, not Wittgenstein. Wittgenstein’s aim was not to foster skepticism about how we categorize. On the contrary, it was (at least partially) to help us understand that we categorize on the basis of observable behavior, and that we should not assume that talk about “private sensations” is anything other than a grammatical convention.

It is not that doubt is impossible; rather, it is that doubt always has some motive, and we should never forget what it is that is calling some fact into question. Wittgenstein warned against doubt for the sake of doubt, or doubt for the sake of epistemological foundations.