contribute to the content of any given representation. This would then
entail that no two thinkers ever entertain a belief with the very same
content, since there will always in fact be some diVerences in belief between
the two. So content would becomeidiosyncratic, which would render it
unWt to serve as the basis for a scientiWc psychology. For the latter surely
seeksgeneralintentional laws to apply acrossall the thinkers in the
population. We shall return to the issue of holism in section 4.4.
2.1 Misrepresentation and the disjunction problemThe obvious problem that informational theories have to overcome, is to
make room for the possibility of (normal)misrepresentation. For plainly it
is possible – indeed, quite common – for our beliefs and thoughts to
misrepresent the world. But it appearsimpossible for one state to carry
misinformationabout another, in the objective causal sense of information
which is in question (except by violation of theceteris paribusclause
governing co-variation). Thus if heavy clouds do not just co-vary with
rain, but also with strong winds, then heavy clouds which are not followed
by rain do not misrepresent the state of the weather (althoughwe,as
observers, may draw a false inference from them); rather, what heavy
cloudsreallyrepresent (that is, causally co-vary with), is rain-or-strong-
winds. Similarly, if the spots normally caused by measles can also be
caused by, say, toxic metal poisoning, then the presence of those spots in
the latter sort of case doesnotmean (that is, carry information about)
measles (although they may lead a doctor into a misdiagnosis); rather,
what spots of that sort really mean is measles-or-toxic-metal-poisoning.
Applied to the case of mental states, then, the diYculty for informa-
tional theories is to avoid what is sometimes called ‘the disjunction prob-
lem’. Suppose that I reliably mistake certain kinds of shrew for mice. That
is, not only does the presence of a mouse in my environment reliably cause
me to thinkmouse, but so too does the presence of a certain kind of shrew
cause me to thinkmouse. What, then, doesmouserepresent, for me? If
mental symbols mean what they reliably co-vary with, then it looks as if
mousemust meanshrew-or-mouserather thanmouse. So I am not, after all,
mistaken when I thinkmousein the presence of a shrew. Indeed, if this
problem generalises, it looks as if, according to informational semantic
theories, it is going to be impossible foranyone everto be mistaken (other
than through breakdown of the mechanisms mediating perception)! And
that, of course, is absurd.
The best-developed approach to this problem has been presented by
Fodor (1987, 1990). His preferred solution is to formulate his theory in
terms ofasymmetric causal dependence. That is, he claims that a Mentalese
164 Content naturalised