Disciplines/fields: Cognitive Science, Computational Modelling, Cognitive Development

Duration: 4 sessions

Course Content

We will chart the joint history of computational modelling and cognition, with some emphasis on cognitive development. We will then pay particular attention to probabilistic generative models as a bridge between different modelling proposals, and their use in current cognitive science (including both the functional level and the algorithmic level). We will then spend a session on cognitive development and its relevance for AI, including important current concepts such as core knowledge, theory-theory and the-child-as-scientist. Finally, we will spend some hands-on tutorial time with a probabilistic programming language and see its potential relevance for different domains of cognition.

Objectives

Conceptual: To understand the current state of probabilistic computational modelling as it applies to high-level cognition, and to form links between cognitive development and proposals for artificial intelligence.

Methodological: Using current probabilistic generative programming languages for modelling cognition.

Literature

How to Grow a Mind: Statistics, Structure and Abstraction (Tenenbaum et al., Science 2011 http://web.mit.edu/cocosci/Papers/tkgg-science11-reprint.pdf).

Vita

Tomer Ullman is a post-doctorate associate researcher working in the Computational Cognitive Science group at MIT, as well as the Harvard Lab for Developmental Studies. He did his PhD in computational cognitive science at MIT, which focused on the origin of intuitive theories, particularly intuitive physics and intuitive psychology. He’s still doing that.

Link

http://www.mit.edu/~tomeru/