The Rise of Simulation Theory

As far as I know, the modern debate ab out folk psychology begins with Wilfrid Sellars’s view that folk psychology is a proto-scientific theory.  For a while, the whole debate was over whether such a theory was mostly correct or incorrect, reducible or irreducible to lower-level theories.  Then in the 1970s, psychologists (and primatologists) picked up on the idea and developed a whole empirical field devoted to “theories of mind”. 

Shortly after that, Robert Gordon came up with an alternative.  According to his simulation theory, we predict others’ mental states and behavior based not on a theory, but on simulating them using our own mental states and processes.  Since then, there has been considerable debate between simulation theorists and theory theorists, with some people (such as Shaun Nichols and Steve Stich) offering “mixed” simulation-theory theories.

More recently, the theory vs. simulation dichotomy has come under attack from various sides. 


On the one hand, some authors (Heidi Maibom, Peter Godfrey-Smith) have drawn from philosophy of science to argue that folk psychology should be taken to be a model rather than a theory or simulation.  I recently read Godfrey-Smith’s paper along these lines.  My impression is that he did not succeed in defining a stable position independent of the theory-simulation dichotomy.  When he explains how folk psychology is not a theory, his models look like they could be simulations.  When he explains how folk psychology is not a simulation, his models seem close enough to theories to make his view a variant of the theory theory.


On the other hand, some authors (Shaun Gallagher, Daniel Hutto) have argued that folk psychology is a narrative rather then a theory or simulation.  While there is obviously some truth to the narrative view, the capacity to construct narratives is very high-level and in need of an explanation.  So I don’t see that the narrative view is a serious competitor of the existing alternatives.

There is also Matthew Ratcliffe, whose new book argues that there is no such thing as folk psychology.  Folk psychology is just a “misguided reification of abstractions” (23).  “Social understanding . . . is a form of what is often called ‘situated’, ’embodied, embedded’ or ‘extended’ cognition.  Social understanding is inextricable from interaction with the social world” (86, see also 107; emphasis added).  (I took these quotes from Hutto’s review; I haven’t seen the book.)  Ratcliffe’s view sounds extremely implausible to me; yet another instance of situated/embodied/embedded cognition run amok.


Interestingly, in recent years there has been movement among some cognitive neuroscientists and roboticists in the direction of simulation theory (Gordon, personal communication).  For cognitive neuroscientists, the big reason is the discovery of mirror neurons, which are relatively easy to interpret within a simulationist framework and not so easy within a theory theory framework.  For roboticists, the main reason seem to be that implementing simulation routines is more straightforward and computationally less costly than attempting to give a robot an explicit theory of mind.  So for now, simulation theory seems to be on the rise.

Comments?  Anything wrong with my story?


  1. Ben

    I think the story is right, simulation is definitively en vogue theses days, especially in neuroscience and robotics. Barsalou’s research on concepts and simulation is similar (but wider in scope, since he study simulation in all domains). There is also a lot of new stuff in “Folk Psychology Re-Assessed”, Hutto & Ratcliffe eds (Springer). Besides narrativity, researchers in the fied also suggest that phenomenology is important. Kristin Andrews also argued that personality traits are fundamental features of folk-psychology. Experimental philosophy (Knobe, Prinz, etc.) also showed that philosophers are sometimes wrong about what the folk allegedly think, and that FP is more than a predictive/explanatory tools, but something we use for attributing responsibility. Knobe and Prinz recently showed that the folks are not always functionalist. So yes, simulation is en vogue, but the nature of folk-psychology is becoming more complex. Maybe folk-psychology is an invention.

  2. Geoff

    Just a note on Maibom — I seem to recall from reading her paper in Mind & Language (which I did well over a year ago, and so I might well be mis-remembering) that she was _not_ proposing that folk psychology is “a model, rather than a theory or a simulation”. Instead, she’s arguing that the right way to characterize our folk psychological theory is as a “theoretical model”, rather than a set of universal generalizations, which is the way you often see it talked about by philosophers. Part of her argument — again, as I recall — was that this is a better way to characterize scientific theories, too. So I saw her as staking out a position within theory-theory, not trying to introduce a third position within the simulation-theory debate.

  3. It’s not clear that mirror neurons really pose a problem for theory-theories, because it’s not clear what mirror neurons do, or how they would facilitate action-understanding in general, and mind-reading in particular.

    And of course, the neuro data isn’t always consistent with mirror neurons/simulation theory being involved in mind-reading (see, e.g., Wheatley et al. in Psych Science this year).

    It’s not surprising to me that mirror neurons are seen as important for simulation theories, though. No one really knows what mirror neurons do, and no one really knows what simulation is, in theory of mind or more generally (in Barsalou, for example). So it’s easy to throw them together. It doesn’t strike me as a good explanation for anything, as a result.

  4. In my ordinary experience, and i think the ordinary experience of anyone irrespective wether he would be a philosopher, scholar or scientist involve in disentangle the intricacies of social cognition, i do not procede in order to “understand others” to formulate hypothesis, contrasting hypothesis, forming models of how they (and i) shall behave etc. Too much time consuming in survival terms. Its all more fast-pace, something that arise as a virtue of emotional attunement, visual description and motor emulations.

    I don´t say some “social sripts” inhibited in memory systems, or models with cultural background with experience and practice are not in use in “folk psychology” (perhaps folk psychology must be revise or even “eliminate” as virtually not directed to reflect anything present in our mind/brains) but primordially neural mechanisms for action representation and unerstanding and empathy engagement are at work in social cognition. Maybe, we have to translate lab evidence on mirror neurons to more ecologically valid contexts, socially speaking to know rally what mirror neurons are for. The work in robotics could be doing this task by reverse engineering.

    But, that is just a way to sharp the argument that simulation theory is on the rise.

  5. Ben

    See also Stueber, K. R. (2006). Rediscovering Empathy : Agency, Folk Psychology, and the Human Sciences. Cambridge, Mass.: MIT Press. He retraces all the theory vs. simulation debates, plus proposes that mirror neurons account for “basic empathy”, but not re-enactive empathy.

  6. Ralph Ellis

    Another simulation theory that I find very helpful is by Natika Newton (see her FOUNDATIONS OF UNDERSTANDING, 1996). Beginning with the premise that we understand the world by imagining its action affordances, she defends the view that a subject S understands an object O if S is able to imagine using O in an imagined action, with imagery rich enough to guide in action. Then other kinds of understanding follow from this. For example, S understands a person P if S is able to imagine P performing an action with imagery rich enough to guide in action; S uses R to represent O if R plays the role of O in an imagined action analogous to the way S could imagine using O, and so forth. (Notice that this was long before Grush’s famous “simulation theory” of representation said essentially the same thing).

  7. shannon

    The Embodied Cognition take on social cognition actually combines the (higher order) narrative construction with what they call “primary embodied practices”. This is a non-mentalistic (or non-robustly-mentalistic) perceptual understanding. It’s sort of a combination of two of your paragraphs above. But most EC folks aren’t so unreasonable as to say that mindreading (whether theorizing or simulating) never occurs. The main claim is that mindreading isn’t the primary way we understand others. (Daniel Hutto, Shaun Gallagher and Jose Bermudez make such claims.) The EC arguments put a lot of weight on phenomenology.

Comments are closed.

Back to Top
%d bloggers like this: