probably relying upon theoretical knowledge. Theoretical knowledge
might make mind-readersconsistently commit mistakesabout the products
of quirks andXaws in human reasoning and decision-making systems. But
if mind-readers were simulating and using their own reasoning and deci-
sion-making systems, they wouldsharethe quirks andXaws of their targets
and so wouldnotconsistently mispredict their behaviour.
Taken in the abstract, this line of argument is impeccable. But it does
depend upon being able to supply the requisite empirical evidence –
namely, instances of surprising irrationality where we as observers also
tend to make incorrect predictions. And it just is not clear that there are
any examples which unequivocallyWt this speciWcation. Stich and Nichols’
prime example is the Langer eVect, which is a matter of subjects showing
an unwarranted preference for lottery tickets which they have chosen
themselves.But there is dispute over the conditions under which this
preference is exhibited (Ku ̈hbergeret al., 1995). To be sure, there are
plenty of examples of common irrationality, but the argument also re-
quires that mind-readers shouldWnd these cases surprising, and should
also get them wrong. That meansmispredicting the responses of subjects:
giving a wrong prediction, not just failing to predict by not knowing what
to expect. Until clear-cut examples of this type are provided, the cognitive-
penetrability/theoretical-fallibility argument remains an argument in
search of its premises.
3 A hybrid view
So far our approach has been resolutely anti-simulationist. Yet we do want
to acknowledge that there is a place for simulation, as an enrichment of the
operation of theory. Note that sometimes the only way of dealing with a
mind-reading question is to use your own cognitive resources. Suppose the
question is: ‘What will the President say when asked to name the state
capital of Nebraska?’. It may be reasonable to assume that the President
will know the answer and, given the context, be ready to give it. But your
ability to predictwhathe will say depends, at least in part, onyou knowing
that Lincoln is the capital of Nebraska.
Heal (1986, 1995, 1996) has argued in favour of a mix of simulation and
theory in folk psychology, with simulation handling the content-aspects of
mind-reading problems while theory is required for non-content aspects
(that is, for handling the types of intentional states and their relationships).
We also want to accept this hybrid or mixed position, while stressing the
fundamental role of theory in giving us our conceptions of the diVerent
mental state types. SpeciWcally, we think that what simulation is involved
in, is the process ofinferential enrichment– that is, that we can input the
A hybrid view 89