Abstract:

Artificial intelligence and automation are reshaping the landscape of scientific discovery across disciplines—from predicting new protein structures in chemistry to uncovering novel quantum states in physics.

In the natural and engineering sciences, automated discovery systems already enable researchers to explore vast spaces of hypotheses and experiments, pushing beyond the reach of human intuition and reasoning.

In this talk, I introduce automated scientific discovery as a paradigm for the study of mind and brain, highlighting its potential to overcome traditional barriers in empirical cognitive science and enable a more integrative understanding of human cognition. I then introduce AutoRA, an open-source framework designed for automating various steps of empirical research, and showcase its utility for discovering novel computational models of cognition and identifying novel behavioral phenomena. This is illustrated through case studies in psychophysics, learning, decision-making, and cognitive control. I conclude by discussing the unique challenges automated discovery faces in cognitive science, and propose directions for future research aimed at developing closed-loop discovery systems. Ultimately, I argue that these methods not only accelerate discovery but also fundamentally reshape how we study the mind and brain.