Metacognition of Concepts

Meaning at subpersonal and personal levels: for the brain and for the person

A key feature of the human mind is that our mental states are meaningful. We appeal to meanings when we explain people’s behaviour in terms of what they have experienced, what they want and what they believe. Often behaviour is driven by false beliefs, so an account of meaning has to show how mental representations can misrepresent the way the world is. Philosophers have long struggled to explain this crucial phenomenon, the meaningfulness of the mental. How can there be mental representations that are about things in the outside world, whose truth or falsity turns on the way the world is?

Perhaps psychology and cognitive neuroscience should take over here, showing us how behaviour is driven by internal information processing and revealing how that works in the brain? Indeed, these burgeoning sciences do provide fantastic new resources for understanding the meaningfulness of neural representations. But neural representations are typically not available to the person performing the behaviour. They are not meaningful states for the agent as a whole.

Philosophers contrast “personal-level” mental states like belief and desire with the “subpersonal” representations typically uncovered by the sciences of the mind. This project aims to cross that divide. By understanding the variety of subpersonal representations discovered by the sciences we create a baseline for asking how meaning is constituted differently at the personal level: meaning for the brain as a way into meaning for the person.

Many scientists see a clear obstacle to this approach. There is a strong trend in psychology and cognitive neuroscience to rely on the subpersonal information-processing accounts mentioned above to claim that personal-level thought and decision play no important role in human behaviour. There are two lines of attack. First, in some cases personal level states like conscious decisions seem to do no work, since subpersonal representations are in control of behaviour. Secondly, when we do elicit personal-level states, they often appear to be deeply error-prone. Ask someone explicit questions about probabilities and they will discount the base rate, get ‘anchored’ to irrelevant numbers and show dozens of other biases. At the same time they harbour subpersonal mechanisms for forming expectations in a changing environment, for example, that make many rapid probability calculations with impressive accuracy.

We aim to address that paradox. It’s not that the personal level is doing the same thing as the subpersonal, only badly – and that it doesn’t have much influence anyway. Rather, the personal level is doing something different, and more difficult. It is widely recognised that fully domain-general probability calculations would be computationally intractable. That is why we have to use heuristics. This project points to other ways in which the personal level has a special job to do. Personal level processes often meta-represent at the same time as representing things about the world, e.g. representing their own reliability. Furthermore, where standard accounts focus on their utility within an individual, they overlook the key role of personal level representations in facilitating joint action and cooperative behaviour.

Performing these functions calls for different computations. This project will attempt to catalogue differences in the way information is processed at the personal level – difference that could affect the nature of these representations, constituting meaning differently. Contrasting that with the account of the nature of subpersonal representation developed in the first half of the project will be an important contribution towards coming to understand how personal level mental states like beliefs and desires get their meaningfulness – which remains one of the deepest unresolved problems in philosophy and the cognitive sciences.