Thanks to Fixed Point, I discovered both the rubber gloves story and a paper written by Bryan Caplan on mental illness and unusual preferences. The author describes the paper as the most unpublishable one that ever got past a referree, but there is much insight to be gained from it about what goes wrong with psychiatry (Caplan's point) and with economics (my point).

Caplan uses rational choice theory to promote a Szaszian view of mental disorders; first he argues, in good humean fashion, that all preferences were made equal, and there is no reason why you should discriminate against people with highly unusual preferences. Then he tries to show that (so-called) mental disorders are nothing but unusual preferences. Of course, Caplan has a hard time with delusional beliefs and hallucination, when he tries to reframe them as very peculiar cases of very unusual preferences. But, if you forgive these philosophical blunders, Caplan has a point at least on some disorders.

One of the most savory parts of Caplan's paper is when he reminds us that classic economic theory and rational choice theory does not apply to mad individuals (a truism), then summons the 20% americans who are diagnosed with a mental condition of some sorts, then concludes that

A seemingly small loophole in the applicability of economics has grown beyond recognition.

Economics and psychiatry have bit so many bullets (one wonders whether economics hasn't swallowed common sense itself) that you can hardly imagine a fact that would remain stuck down their throat. Yet it must be disturbing to learn that one fifth of the population you study escapes your analysis. faced with that, Caplan chooses the easy way and gets rid of the dismaller science of the two: psychiatry.

One can easily see how some unusual tastes have been turned into medical diseases; homosexuality (voted off the DSM in 1973) is a case in point.

Robert Spitzer, then head of the Nomenclature Committee of the American Psychiatric Association, was especially open about the priority of social acceptance over empirical science. When publicly asked "whether he would consider removing fetishism and voyeurism from the psychiatric nomenclature, he responded, 'I haven't given much thought to these problems and perhaps that is because the voyeurs and the fetishists have not yet organized themselves and forced us to do that."

(that was true in 1973, but fetichists are not closeted any more)

Caplan plays easy on these cases, on addiction or on such nonentitites as anti-social disorder ("Homo economicus arguably suffers from this disorder by definition" how true!) and ADHD; but some other sections of the DSM are tougher, particularly those concerned with delusions and hallucinations. Caplan tries to solve the problem by stating that agents can have preferences over their beliefs. This can be argued for (as indeed such rational choice theorists as Jon Elster or Akerlof have done). But this "Will to Believe" hypothesis raises the difficult question of choice without alternatives: if I can have a preference over my beliefs, that supposes I can conceive of 1) my current belief, 2) another, more rational one, see that 2) is more rational than 1), and still prefer 1) (otherwise I would simply be making a mistake). But one can argue that seeing that 2) is more rational than 1) amounts to believing 2). Which makes it difficult for rational agent to choose what he believes in.

Anyway, I doubt an agent can have preferences over his perceptions (anticipations and habituations are not preferences). That follows clearly from the fact that our minds (at least as far as basic perceptual organs are concerned) are modular to a certain extent (at least to the extent that frontal lobes have little impact on our retina). As Szasz did, Caplan flatly denies that genuine hallucinations are ever experienced.

The belief/desire framework is too poor to account for those disorders that can shatter a whole mind; for example, I do not think that schizophrenia can be described as an extremely unlikely preference plus a change in beliefs. One does not see what could make a schizophreniac's halluciantions desirable to himself. Caplan's story of John Nash choosing to become a schizophreniac, then renouncing madness to receive the Nobel prize, is moving, but anecdotal.

Caplan nevertheless managed to throw me into perplexity as to where the line between mental illness and unusual preferences should be drawn, and I think his views are perfectly relevant to many DSM diagnosis (but not all). His paper should bring us to criticize the economic dogma as well as the psychiatric one, beginning with the humean assumption that substantive rationality is a chimaera, that there is no such thing as a reasonable preference. I think that's where evolutionary psychiatry might help.