"Social cognition is concerned with how cognitive agents understand and navigate the social world. Most social cognition research today – whether in cognitive science, psychology or neuroscience – focuses only on the individual mechanisms needed to figure out others’ mental states. But this view over-emphasises the individual and gives social dynamics no more than a contextual role. However, there is an alternative: the process of interaction itself may be crucial to social cognition. To articulate this alternative, we need 1. to know whether the interaction process can enable or even constitute social cognition, 2. to know how individuals change through interaction, 3. to re-assess the function of individual mechanisms in the context of interaction. This project proposes an original theoretical framework with direct relevance for empirical work. It uses an operational notion of autonomy developed in the philosophy of biology. The main claim is that social interaction can act as an extended mechanism that modulates and transforms individual cognition, autonomy and agency. This leads to the following proposals: 1. Explanations of social cognition should range from individual-dominant to interaction-dominant ones. 2. We can distinguish between contextual, enabling, and constitutive factors in the role of interaction dynamics for cognition. 3. Intentions are not locked and ready-made inside individuals, but can be modulated and transformed in interactions. 4. Interaction processes can acquire a form of systemic autonomy. 5. There is an interplay between individual (organismic, sensorimotor) and interactional autonomy. 6. Socio-cultural norms need to be negotiated in interaction. 7. Individual neural mechanisms are shaped - literally - by social interactions. In this new model of social cognition, the interaction does proper explanatory work. This has consequences for theory, modelling and experiments, and extends the range of explanations."