representations of a 3-D world of tables, chairs, trees and so on. No one
believes that any more; and introspectionist/simulationist accounts of
mind-reading abilities look equally implausible. In each case the problem
is to abstract an extremely rich causal structure on the basis of something
which is a-causal.
If the argument so far holds, an important conclusion follows. This is
that simulationist models of mind-reading mechanisms whichstart out
from distinguishable psychological statesare not genuine rivals to theory-
theory. They only oVer an account of how the processing for solutions of
some mind-reading problems works within a general context of theoretical
knowledge about the mind. On such an account, knowledge of causal and
functional roles is needed to make the initial discriminations of diVerent
psychological states on which simulation might subsequently go to work.
There is, however, a more radical form of simulationism, proposed by
Gordon (1995, 1996).
Gordon thinks that we learn to mind-read by learning to pretend tobe
another person. He suggests that the default mode for simulation is ‘total
projection’, just taking the person simulated to be exactly like yourself –
without making any adjustment for situation, circumstances, or any other
personal diVerences. However, we do gradually learn to make appropriate
adjustments to match the situations of others. We also learn, more or less
successfully, to make adjustments to our belief and desire set for the
purposes of the simulation. And this need not involve any introspective
awareness of what our own beliefs and desires are, Gordon thinks. Once
we have achieved this sort of ‘partial projection’, we can then reason within
the scope of that pretence, ‘oV-line’.
Gordon attempts to give a thoroughly principled and non-Cartesian
account of self-ascription. TheWrst step on the way to knowledge of one’s
own mental states is as easy as acquiring a simple linguistic habit. We start
out self-ascribing a mental state of belief or intention by means of an
ascent-routine: learning that where we are ready to assert ‘P’ we can equally
well assert ‘I believe that P’. Note that this is not intended by Gordon to
require anything in the way of introspective awareness! The next step (on
the journey to self-knowledge through simulation) is to combine the results
of oV-line simulation with the ascent-routine for belief. So we learn to
transform pronouns in such a way as to ascribe a thought to theother
person whenever we come out of the scope of a pretence – going from ‘I
believe that P’ to ‘He/she believes that P’. Finally, according to Gordon,
we only learn to ascribe thoughts to ourselves (with understanding of the
possibility that the thoughts may be false; that is,notjust by means of an
ascent-routine) by simulating another (or ourselves at a later time) simula-
ting ourselves.
86 Mind-reading