marginal cases is asymmetrically dependent upon its application to the
cases in the centre? That is, might it not be plausible that (i) Cs cause
tokenings of T, and (ii) if Cs had not caused tokenings of T, Ms would not
have, and (iii) if Ms had not caused tokenings of T, Cs still would have? In
which case Fodor’s theory would tell us that the marginal cases are not
really cases of Ts at all! In fact, it would tell us that Ms aremisclassiWed as
Ts! So orangey-reds arenotred (as opposed to marginal reds), and syn-
chronised swimming isnota sport (as opposed to a marginal one).
Let us run this objection through for the case of colour in a little more
detail, to see how it might go. Let crimson be our representative central
red, and let the marginal case be an orangey-red which speakers would still
categorise as ‘red’ in a forced choice, or when speaking carefully. Plainly,
the equivalent of (i) is true – crimson objects cause us to thinkred.Now
what would it take for crimson objectsnotto cause us to thinkred? – that
is, for the antecedent of (ii) to be true? One possibility is that we be
colour-blind. Another possibility is that the Mentalese termredbe har-
nessed to some other meaning (for example,blue,ormouse) while crimson
causes the tokening of a distinct Mentalese item – say,xyz. But in either
case, orangey-reds would not have caused us to thinkredeither – that is,
(ii) is true. Then what would it take for orangey-reds not to cause us to
thinkred? – that is, for the antecedent of (iii) to be true? Plausibly, it would
be enough if English were to introduce and make salient a distinct term
(like ‘oranred’) for just that shade of orangey-red. But in such circumstan-
ces, crimson would still have caused us to thinkred, and so (iii) is true. So
conditions (i)–(iii) are all fulWlled, and Fodor’s account entails that or-
angey-red (by hypothesis, a marginal case of red) isnotnow red. And then
we have our counterexample.
2.2 The problem of causal chainsPerhaps the biggest problem for any informational semantics, however, is
this: where in the causal chain which leads to the tokening of a mental
symbol do you stop, toWx on the meaning of the latter? Any mental symbol
will always carry information about events further out, and events further
in, from what we would intuitively take to be its referent. For example, any
term which reliably co-varies withmousewill also reliably co-vary with
mouse-mating, since the world is such that, whenever there is a mouse,
there has also been a mating between mice in the past. (We ignore the
possibilities of artiWcial insemination, cloning and test-tube reproduction
for simplicity – they would just render the background causal conditions
formousedisjunctive, while allowing essentially the same point to go
through.) So what shows thatmousemeansmouseand notmouse-mating?
166 Content naturalised