Why know myself? Flexible behaviour and the need for self-modelling: Poster

Abstract

In this paper I argue that some forms of the capacity for behavioural flexibility entail a specific kind of representation, a self-model. This means that systems with that capacity, among them human beings, must have self-models. In its basic form, the capacity for behavioural flexibility allows a system to respond to the same sensory stimulus differentially, depending on the values of parameters with which it represents the world. On seeing a street, I might cycle straight ahead or take a sharp turn left – depending on whether I represent it to blocked off just around the corner. More advanced forms expand on this. Self-models are a form of self-representation in which states are represented by placing a token in a model of the world. The relations this token bears to the modelled features represent the system’s states (Ismael 2007). A useful analogy are smartphone navigation apps, where a central blue dot indicates the location of the user. With my contribution I hope to, first, clarify the cognitive advantage of subject/object differentiation. Second, I want to improve on Ismael’s very promising proposal by extending it to non-map-like formats of representation and system states other than spatial and temporal properties. This, I hope, should convince authors in the burgeoning literature on self-models that paying close attention to broadly ‘Ismaelian’ accounts of self-modelling – rather than the much more widely discussed proposals by Metzinger (2007) and Hohwy and Michael (2017) – could significantly advance our understanding of the mechanisms and uses of self-representation.