Harnad, S. (1996) What to Do About Feelings? [Published as "Conscious Ecumenism" Review of PSYCHE: An Interdisciplinary Journal of Research on Consciousness] Times Higher Education Supplement. June 7 1996, P. 29.


What to do about feelings?

That Psyche should be a virtual journal, somewhat "immaterial," is quite in keeping with its subject matter. And just as there will be differences of opinion about Psyche's disembodied content, there will be differences of opinion about its disembodied form. (It is silly, however, to carp that in sanctioning a parallel paper incarnation by MIT Press Psyche has sold out: In its early years surely even its most zealous advocate would have needed a preternatural faith in the communicative power of Bell's new invention to rely exclusively on the phone for reaching the unconverted!)

The first issue is dated 1993 Volume 1, Numbers 1-2, and consists of 3 articles - one accompanied by 6 commentaries and a response, plus 7 book reviews. The majority of the contributions are from psychologists; philosophers are next, and then a few brain scientists and computer scientists. There is always some uncertainty about whether psychologists and philosophers are talking about the same things when they get together, and about whether they are really any help to one another even when they are. Psyche's ecumenism does not altogether resolve this uncertainty.

At the heart (or soul) of consciousness studies is the "feeling/matter" problem: We all know what matter is and what doing is, hence what doing-matter (say, a comet or car) is. But what kind of "stuff" is feeling-matter? how do you study it? what is it for? and how does it square with the ordinary doing-matter that is all Physics tells us there is in the world?

Philosopher Richard Dewitt's article is about a variant of the feeling/matter problem: the meaning/matter problem. An influential philosopher (Jerry Fodor) thinks our brains do what they do using a "language of thought." It is unclear, however, how the meanings of such words and sentences in our brains, as opposed to merely their motions, could play a role in what our brains (hence we) can do. Perhaps, as in mathematics, the recipes our brains follow in moving the symbols (as in doing long division) "mirror" their meanings: Long division rules, applied correctly, give all and only true results, whether applied to taxes on lottery winnings or the wind-chill factor in weather forecasting. This can be proved for mathematics. Can it be proved for a language of thought? Dewitt argues that it cannot, because words are too vague. So the connection between meaning and doing-matter remains as problematic as the connection between feeling and doing-matter.

Psychologists will be puzzled as to where to go with this, just as they would have been if the message had been the opposite: that mirroring between mental symbols and meaning can be proved. For there seems to be something more to the meaningful thoughts in my head than the meaningless symbols I use to perform long division. At the very least, the thoughts in my head somehow confer meaning on the long-division symbols; surely those thoughts themselves do not in turn derive their own meaning from yet another thinking head doing a number on them, as I am on the long-division symbols, for that would just lead to an infinite regress. "Mirroring" doesn't seem strong enough to solve the meaning/matter problem, with or without a proof.

Philosopher Selmer Bringsjord's article does rather an opposite number with symbols: turning simple meanings into less meaningful symbols. Again, the target is the view of an influential philosopher (John Searle), who thinks that feeling cannot be reduced to matter yet is a property of matter. That already sounds like a restatement of the feeling/matter problem, rather than its solution. Bringsjord might simply have noted this and stopped there, but the rest of his article tries to make this simple point into something resembling a mathematical proof, with the only result, as far as I can tell, that it becomes both less comprehensible and less convincing. What seems inescapable is that it has left the concerns of nonphilosophers far behind.

The third article is by a psychologist, Bernard Baars, who suggests that the kinds of things people do with and without consciousness (i.e., with and without feeling that they are doing it) should be systematically compared. Maybe this will help us determine what feelings are and what they are for. Six commentators bat the proposal around, but whatever insight into the nature and function of consciousness these pairwise comparisons are meant to provide, we do not quite get a glimpse of it yet here. On the other hand, contrasting our feeling and nonfeeling doings may still give us some insight into the material basis of our doings, if not our feelings; if so, that will still have been a useful exercise. Unfortunately, it will continue to leave the psyche untouched.

Apart from three further book reviews, Psyche Volume 2 is devoted entirely to commentaries (9) on the mathematician Roger Penrose's idea, expounded in two recent books, that computation cannot explain the mind because our minds can know the truth of statements whose truth is not computable; so the mind must be some sort of quantum computer. Penrose responds in the same issue.

We risk being drawn into the labyrinths of the interpretation of Goedel's celebrated undecidability proofs in joining this debate, but let as avoid them, reminding ourselves that Goedel was a mathematician talking about numbers, not a psychologist talking about thoughts. Supposing it is true that we know truths that are uncomputable: Does it follow that cognition cannot be just computation? Do we have to resort instead, as Penrose recommends, to quantum processes, with their own attendant, and even more convoluted, labyrinths of interpretative uncertainty? Isn't the feeling/matter problem problem enough? Is one field's mystery medicine for another's?

One can agree with Penrose (as I do, but for other reasons, having more to do with the problem of infinite regress mentioned earlier) that cognition cannot be only discrete computation. But then the remedy for that is just some ordinary continuous Newtonian physics, with perhaps also a dash of classical statistical uncertainty; no need to resort to quantum effects. Besides (and this is the critical point): Even though computation clearly cannot be the material basis for arriving at truths that cannot be computed, it could still be the material basis for our consciousness of those truths! Truth in the head is only a feeling, after all, at least for psychologists. There are reasons for believing that meaning can't be computational, but no one has yet shown why feelings can't be computational -- unless the feeling/matter problem and the meaning/matter problem should turn out (as I suspect they will) to be both one and the same, and insoluble. For then the only doable bit -- namely, the explanation of the material basis of our doing -- could, once done, safely leave the trick of explaining feeling to be done with mirrors after all.