After an excellent SSPP run by Gualtiero (who deserves great thanks!), I thought a modestly humorous post from here in Savannah was in order:
Here is a question, which I think suggests a funny criticism: if simulation does play a prominent role in mindreading (the attribution of mental states to others) shouldn’t it be a lot like playing dungeons and dragons? But I don’t think it is. Hence, while it is of course plausible that mental simulation plays some role in human mindreading I suspect that it is a very minimal. Here is a quote from the SEP entry on simulation theory:
“It is assumed that in role-taking, one’s own behavior control
system is employed as a manipulable model of other such systems. (This
is not to say that the “person” who is simulating is the
model; rather, only that one’s brain can be manipulated to model other
persons.) The system is first taken off-line, so that the output is not
actual behavior but only predictions or anticipations of behavior, and
inputs and system parameters are accordingly not limited to those that
would regulate one’s own behavior. Although this sometimes results in
vicarious decision-making, more typically it stops at the more modest
goal of establishing which options would be attractive (so that one
would not be surprised to find the other pursuing one of them) and
which unattractive (so that one would be surprised to find the other
pursuing any of them).
Compare this to playing D&D, which seems to also involve simulation. One designs a character, including various details about history, class, abilities and so forth. In deciding what actions one’s character takes, whether it be communicating with other characters or combat, one is constrained by these details; one in effect tries to imagine what one’s character “would do”, and this imagining seems to fit the profile typically given of simulation processes that are thought to be involved in mindreading. For example,iIn playing D&D, our system is “off-line” in that we do not ourselves do the acts we suppose our character would do (some, I even lack the ability to perform: e.g. cast spells). And, typically one is stopping at the “modest goal” of a plausible action that one’s character might perform in the circumstances at hand.
There are of course some disanalogies. D&D is fictional, and we are not merely prediction actions but making decisions about what our characters will actually do. But these are rather superficial difference (notice many experiments that test mindreading involve fictional scenarios, for example). In fact, the “psychology” of playing D&D seems to be very much what mindreading via simulation “should” be like, or the psychology of D&D should be very different than it is, depending on whether one thinks the simulation occurs consciously or unconsciously:
1. If one thinks the simulation occurs consciously, then one might think mindreading should be a lot like playing D&D since the simulation involved in the game seems to pretty clearly be explicit and the product of slow conscious deliberation and inference (often, it is even verbalized!). But, while in some cases of mindreading it does seem to be similar to playing D&D, often our attributions are certainly not: they are quick, effortless, and involve little to no conscious inference. So, if simulation plays a prominent role in mindreading (as, for example Goldman and Gordon think it does) one might suspect it should be more like playing D&D.
2. Alternatively, if the simulation is unconscious, then one might wonder why playing D&D relies on conscious simulationat all! If we have a well-tailored simulation process that is regularily relied on in mindreading, then why do we seem to always reply on the more deliberative conscious kind when playing the game? Plausibly, if unconscious simulation exists, that is what we would rely on in playing D&D instead of the conscious kind. So, if unconscious simulation plays a prominent role in mindreading, then one might think the psychologica states we seem to experience when playing D&D should be different than they actually are.
If 1 or 2 is correct, then I suspect simulation has a minimal role in mindreading. None of this has been argued in detail, of course, but I think this is a modest and rather interesting criticism if correct.