Gordon has constructed an ambitious and ingenious account of self-
knowledge. But we do not think it will do. The problem here is to see how
we could ever acquire the capacity which we surely do have: to ascribe,
with understanding, occurrent thoughts to ourselves immediately,noton
the basis of any sort of self-interpretation of our own behaviour. I can
know what I am thinking even where those thoughts are unrelated to my
overt behaviour or to my current circumstances. As I gaze silently out of
the window I know that I have just entertained the thought, ‘Avignon
would be a good place for a holiday’. But I cannot have self-ascribed this
thought by simulating another person simulating myself, since no one else
could have ascribed such a thought to me in the circumstances. Simulation
cannot take place under these conditions: therehas to besome salient
circumstance or behaviour to simulate.
2.3 The problem of mutual cognitionOne way of approaching this problem is by wondering how simulationism
can explain anyone as cunning as Iago, the villain of Shakespeare’s
Othello, who is hell-bent on bringing down his commander by making him
think he has reason to be jealous of his wife (Desdemona) and one of his
oYcers (Cassio). Suppose Iago can simulate what Cassio will think and
want to do when he takes it that Desdemona is attracted to him. That is a
Wrst-level problem. But can Iago simulate what Othello will think Cassio
will think and do in those circumstances? That is a second-level problem,
and altogether trickier to model in terms of oV-line processing. Suppose
subjects can run their decision-making or inferential systems oV-line. Then
how can they do ittwice at the same time, labelling one pass for Cassio and
another pass for Othello-on-Cassio?
In factthisproblem is not so very diYcult for a simulationist, perhaps.
For in one sense, in order to simulate Othello simulating Cassio,you just
have to simulate Cassio. What can make your simulation of Cassio into a
second-order one, is what initiates it, and what you do with it on com-
pletion of the simulation-process. To predict what Othello will think about
Cassio’s thoughts you begin (naturally enough) with a question about
Othello. You then ‘drop down’ a level to simulate Cassio. And then on
exiting the simulation you have to embed the resulting belief or intention
twice over –Wrst in the belief ‘Cassio will think...’, and then in the belief
‘Othello will think that Cassio will think...’. Now there may well be a
problem about how children could ever be supposed tolearnhow to
execute this process. But the mature process itself seems an intelligible one.
There is, however, a real problem for simulation-theory in accounting
for our ability to explain or make predictions concerning various forms of
Problems for simulationism 87