Which Theory of Mind? – And other questions

In my final post I would like to wrap up by sketching some of the implications of my proposal – in particular concerning our theorizing about social cognition – as well as raising some questions that are being left open.

There exists quite a large controversy in philosophy and psychology with respect to the question of how we are to account for our social cognitive abilities. The most important competing theories are the theory-theory, the simulation theory, the interaction theory and the narrative theory (with the latter two generally being combined into a hybrid model).

The theory-theory assumes that social cognition is to be seen primarily as the ability to ascribe mental states to others for the purpose of predicting and explaining their behaviour, and that this ability relies on a folk psychological theory concerning the way the mind functions (i.e. a set of rules that connects beliefs, desires, and other mental states to each other and to specific types of behaviour).

In contrast to the theory-theory, the simulation theory argues that it is more parsimonious to assume that rather than having to rely on a theory of how the mind works, we simply rely on our own minds to predict and explain the behaviour of others. According to this theory, the way we come to understand the mental states of others is by generating equivalent states in ourselves. That is to say that in order to anticipate how someone else will behave in a certain situation, we ‘put ourselves in the other’s shoes’ and simulate how we would behave in the situation.

Traditionally, both simulation theory and theory-theory are in agreement with regard to the fact that subjects need to be in possession of mental state concepts in order to engage in the practice of ascribing mental states to others for the purposes of predicting and explaining their behaviour. However, there is also an interpretation of simulation theory according to which simulation need not imply a conceptualization of the other’s mental state — we may internalize another’s facial expression, for instance, without explicitly identifying the emotion it expresses.

A third, more recent alternative — interaction theory — argues that we neither need theoretical inference nor simulation to explain and predict the behaviour of others, for we can directly perceive their emotions, intentions and motives. And a fourth alternative is narrative theory, according to which the development of a theory of mind can be attributed to the narrative practices that we are engaged with. According to this theory, it is the child’s engagement in socially guided and supported story-telling activities that leads to the development of a theory of mind. Because the narrative theory must assume that the child already has a basic, ‘practical’, grasp of mental states before it can engage in story-telling activities, these last two accounts are usually defended in combination with each other.

Although the debate is standardly couched in terms that suggest that theory-theory, simulation theory and interaction/narrative theory are competing and mutually exclusive accounts of social cognition, on the view I propose, it seems plausible that they rather describe different and complementary processes.

Insofar as both theory-theory and simulation-theory presuppose the possession of mental state concepts for the ascription of beliefs and desires in order to explain and predict the behaviours of others, both theories seem to be incompatible with those primitive forms of social cognition that rely on an implicit, nonconceptual grasp of the other. However, if simulation is instead taken to consist in a kind of self-other matching (without an explicit differentiation between self and other), the forms of social cognition and interaction associated with the phenomena of primary and secondary intersubjectivity described in my previous post might involve just such a process of simulation.

This need not necessarily be incompatible with the interaction theory either. Indeed, we might interpret simulation theory as providing a (partial) account of what is going on in basic intersubjective processes, such as imitation, social referencing and shared attention, thus explaining how we can engage in these processes based on direct perception – where the perceptive information about the other is automatically and implicitly matched with our first-person experience – without having to rely on conceptual abilities or theoretical inferences.

The ability to differentially ascribe mental states to self and other requires a redescription of the information that is implicit in these forms of basic social interaction into a more explicit format. Thus, the kind of theorizing presumed by a theory-theorist would begin as the self-other matching that is implicit in primary and secondary intersubjectivity becomes redescribed into a more explicit format, allowing for a distinction between self and other and, ultimately, for the explicit ascription of mental states to self and other.

Finally, it seems plausible that the ability to theorize about mental states is directly influenced, and perhaps even enhanced, by the narrative practices that the child is embedded in. That is to say, once the child begins to speak and to access representations about the intentional relations of self and other in an explicit, conceptual and linguistic format, these representations will be shaped by the social narratives that the child is engaged with.Thus, on my view, narrative theory is not to be seen as an alternative to theory-theory, but rather as describing a complementary process that modifies and shapes how we theorize about ourselves and others.

In short, the suggestion is that simulation theory, theory-theory, interaction theory and narrative theory all have the potential to contribute to a better understanding of social cognition, and that they are therefore best seen as complementary rather than competing theories.

What I’m less sure about is how this proposal fits with the dual-systems account of mindreading that has recently been proposed by Stephen Butterfill and Ian Apperly (see the 2013 symposium hosted on this blog for a discussion).  On their view, we can distinguish between two distinct cognitive capacities: A basic heuristic system that is fast but inflexible and accounts for many of the early social cognitive abilities found in infants; and a second system that is flexible but demanding and deals in explicit attributions of propositional attitudes. While I find aspects of this view quite attractive, I am not quite sure whether and how it can be integrated with the multi-level account that I propose. I’d very much welcome thoughts on this by readers of the blog.

Another question that arises concerns the issue of self-consciousness (and social cognition) in non-human animals. I address this question in chapter 8 of my book, where I suggest that although the evidence to-date on this issue is not clear, we have some reasons to think that non-human animals possess some basic forms of self-awareness, similar to those found in human infants.

Finally, there are a number of questions that are left open by my proposal, and I would welcome thoughts on these. These concern, among other things, the role of emotions in the development of self-consciousness and social cognition, as well as the role of language. For instance, do symbolic or linguistic abilities provide the necessary prerequisites for the development of explicit (conceptual) representations, or is it rather the other way around? And do our social cognitive abilities enable the development of language, or is it rather the linguistic abilities that enable the development of complex social forms of interaction? My intuition is that while we need some basic social cognitive abilities to enter the realm of linguistic communication – and while these do indeed seem to exist before children acquire a language – it is likely that more complex forms of social cognition  depend on linguistic abilities. I’d be interested to hear what others think about this.




Back to Top