a great many psychological generalisations relate toparticularcontents or
types of content.
Consider the claim that there exist various kinds of incest taboo, for
example. If I say that people have an aversion to the thought of sexual
intercourse between mother and son, then I have expressed a generalisa-
tion in relation to a thought which no one else but myself can share, if
strong holism is true. Since no one else can have an aversion tothat
thought (a thought individuated by the totality of the inferential connec-
tions which it has, for me), then my attempted generalisation fails.
One way of trying to respond to this objection would be to take up
Stich’s (1983) suggestion, that scientiWc psychology should operate with a
gradedsimilarity measurefor intentional contents. Psychological laws
might then be formulated for contents which are more or less similar to one
another, even if no two people ever share thought-contents which are
strictly speaking identical. However, as Fodor (1998) points out,similarity
of contents has to be cashed in terms of the extent of overlap in the sets of
thoughts with which a given pair of contents is inferentially connected.
And this in turn presupposes that we canidentify(absolutely, not merely as
similar) the thoughts in question. So this approach looks unpromising.
To see the possibility of a rather diVerent way of responding to the
objection, notice that not all of the inferential connections of the concept
incest between mother and sonare likely to be equally relevant to the
existence of the incest taboo. In order to feel aversion at the thought,
‘Jocasta is committing incest with her son’ there are probably various
beliefs which you have to have, and various inferences which you must be
prepared to make – such as the inference that Jocasta is considerably older,
for example. But remote inferential connections with beliefs about average
family size, say (let alone with beliefs about famous female novelists) are
unlikely to have any impact. What this then suggests is that there may be a
subset of the total inferential connections of a given thought-content
which bear on the truth of the law-like generalisations in which itWgures.
In which case psychological science could identify the content in question
by means of the relevant subset.
(This might lead to a situation in which one and the same mental
representation – such as, Jocasta is committing incest with her son–
would be assigned contents withdiVerentidentity-conditions in diVerent
contexts, depending on which psychological laws were in question. But
there need not be anything particularly surprising in this, let alone subver-
sive of the very idea of content-based psychology.)
A real example might help here. As we saw in chapter 4, psychologists
have discovered that there is a watershed in normal development which
occurs around the age of four, prior to which many of the generalisations
Functional-role semantics 183