ArgLab • Permanent Seminar

Paul Smart

Dream Machines: Predictive Processing, Generative Models, and the Mind of Society

An emerging consensus in cognitive science views the biological brain as a hierarchically-organized predictive processing system, one that is constantly engaged in the effort to predict the flow of information originating from the sensory surfaces. Such a view seems to afford a great deal of explanatory leverage when it comes to a broad swath of seemingly disparate psychological phenomena, such as learning, attention, perception, action, emotion, reasoning, imagination, and conscious experience. It is also a view that establishes an interesting point of contact with recent research into machine learning, especially that which goes under the heading of deep learning. In this talk, I will explore the idea that hierarchically-organized predictive processing systems can be used to model the behaviour of human collectives, such as groups, teams and (at the largest scales) societies. I will suggest that socially-oriented generative models are poised to yield a ‘deep understanding’ of human collective behaviour, with potential methodological implications for the social sciences. I will also explore the idea that such models may herald the emergence of a new kind of mind—a mind whose main objective is to establish a predictive grip on the human social realm.