Zoom link: https://mit.zoom.us/j/91619236639

A prominent approach to pragmatics is the Rational Speech Act (RSA) framework, which formulates pragmatic reasoning as probabilistic speakers and listeners recursively reasoning about each other. While RSA enjoys broad empirical support, it is not yet clear whether the dynamics of such recursive reasoning may be governed by a general optimization principle. In the first part of the talk, we present a novel analysis of the RSA framework that addresses this question and show that RSA can be grounded in Rate–Distortion theory, the branch of information theory characterizing optimal data compression. These findings suggest a novel computational-level view of pragmatic reasoning. In the second part of the talk, we begin to explore the practical implications of this view, both for understanding human behavior and for informing artificial agents with pragmatic skills. This work furthers the mathematical and conceptual understanding of RSA models, and suggests that general information-theoretic principles may give rise to human pragmatic reasoning.