What is the strongest argument against Robert Gordons 'radical' simulation theory?

Question

in the theory-theory vs simulation theory debate it seems although a lot of philosophers dismiss radical simulation and deem it so obviously wrong that not many even go into explaining why. I am intrigued to know why this is the case. is it because simulation relies on theory to some extent?

Answer 1

It's disputed both that mirror neurons can possibly have such wide functional scope (generally they've been observed for motor function and for emotions), but it's also disputed that their firing, when it does happen, is actually involved in "reasoning" of some kind (be it merely the triggering of some emotions), as opposed to merely being a correlative byproduct. Basically the latter point is a dispute on correlation vs causation. From Wikipedia:

Neurophilosophers such as Patricia Churchland have expressed both scientific and philosophical objections to the theory that mirror neurons are responsible for understanding the intentions of others. In chapter 5 of her 2011 book, Braintrust, Churchland points out that the claim that mirror neurons are involved in understanding intentions (through simulating observed actions) is based on assumptions that are clouded by unresolved philosophical issues. She makes the argument that intentions are understood (coded) at a more complex level of neural activity than that of individual neurons. Churchland states that "A neuron, though computationally complex, is just a neuron. It is not an intelligent homunculus. If a neural network represents something complex, such as an intention [to insult], it must have the right input and be in the right place in the neural circuitry to do that."

Recently, Cecilia Heyes (Professor of Experimental Psychology, Oxford) has advanced the theory that mirror neurons are the byproduct of associative learning as opposed to evolutionary adaptation. She argues that mirror neurons in humans are the product of social interaction and not an evolutionary adaptation for action-understanding.

And to substantiate a bit my own personal view that this (ST vs TT) is somewhat of a fake dispute (between two rather implausible extremes) from a neuroscience perspective... Bastiaansen, Thioux and Keysers (who are not philosophers) wrote about a decade ago:

A variety of authors have criticized simulation theory (ST) based on the fact that it cannot explain all facets of social cognition (e.g. Jacob & Jeannerod 2005; Saxe 2005; Gallagher 2007) and that we still fail to have conclusive evidence in humans that the exact same neurons are involved in action perception and execution (Dinstein et al. 2008). As previously discussed, this criticism also applies to the case of emotions and sensations. We view the first critique as an experimental challenge that should inspire researchers in the next decade. As for the second, we believe that it is fruitless to create a competition between simulation views versus more cognitively inspired ‘mentalizing’ approaches in current social neuroscience research. There is no doubt that in many instances we rely on our knowledge of the person or the situation to make inferences about the state of mind of the other. If a salesman of second-hand cars smiled broadly while bragging about the quality of a rusty old car, our simulation circuitry could make us share his enthusiasm, but semantic knowledge about second-hand car salesmen could lead to a different conclusion. There is ample evidence that what we know about someone else can influence the simulation mechanism. For instance, the perceived fairness of the observed individual in pain influences how much pain will be shared (Singer et al. 2006). Similarly, the gender of the observed individual can influence our neural response (Simon et al. 2006). In addition, the sensory part of the pain matrix is engaged more when the perceived pain reaction of another person matches how we would respond ourselves (Lamm et al. in press). Therefore, we believe the interesting question is how these two processes are integrated in the brain (Keysers & Gazzola 2007).

I can add that the SEP article on ST isn't too far from this position in its conclusion section:

while the consensus view is now that both mental simulation and theorizing play important role in mindreading, the currently available evidence falls short of establishing what their respective roles are. In other words, it is likely that we shall end up adopting a hybrid model of mindreading that combines ST and TT, but, at the present stage, it is very difficult to predict what this hybrid model will look like. Hopefully, the joint work of philosophers and cognitive scientists will help to settle the matter.

It's also more customary in psychology and neuroscience to talk of implicit and explicit mentalizing (as opposed to the more philosophical ST vs TT terms), but there are those like Heyes who dispute that the distinction can be made.

Answer 2

Simulation theory sounds similar but also competitive to the most popular folk psychology (theory theory, child scientist theory, Bayesian learning) in the scientific debate of mindreading. Theory theory employs (naive) generalized psychological concepts or theory formed by an individual to infer the mental states of others, such as their beliefs, desires or emotions. While radical (pure) simulation theory employs an individual's own psychological skills of imagination about action, belief, desire, etc to simulate others to be understood and attribute such a imagining result to others without invoking any theory. Why we can even trust such a radical imagination result once we put in another's situation? It's mainly based on modern biological evidence such as the famous discovery of mirror neurons in neural science, which suggests we (including monkeys, birds, etc) can naturally simulate without any theorizing or conceptions.

One of major arguments against such radical simulation is it seems mirror neurons only predict emotions (such as empathy, compassion) and actions, not beliefs or desires. So radical ST seems best for explain perception-action-emotion related aspects of other's mind, not rational reasoning or belief aspects. A monkey can predict a dog's next action to find its food from its past observation, but the monkey may not believe the dog's action is the right call. Also, radical ST uses one's own skills and imagination, there may be significant differences between the simulated subject and oneself even regarding the primitive perception and emotion mental states.

Another argument against it is radical ST denies any epistemology relevance and recourse of the role of any psychology theory, which many philosophers may strongly disagree. TT is perhaps more efficient to directly explain our reasoning and JTB epistemic aspect as conception/theory seems indispensable with any rationality...

Answer 3

The theory-theory notion of learning other peoples 'mental states' is actually already in Plato when he says that a man is already born knowing. He or she must know already before they know how to know more. This is in Plato's Meno. One example of this is Chomsky's notion of a deep grammar. We do not learn grammar, we 'remember' it because it is already innate in embryonic form.

It's because of the force of this argument that the simulation theory is mostly dismissed and marginalised.