joshua abbott  

Computational Cognitive Science Lab
Department of Psychology
5429 Tolman Hall
University of California, Berkeley

me  |  research  |  papers

I am currently a sixth-year PhD candidate at UC Berkeley under the advising of Tom Griffiths in the Computational Cognitive Science lab. My research lies at the intersection of machine learning and computational models of cognition where I'm interested in using mathematical models to formalize what it means to learn: an ability central to both artificial and biological intelligence. In particular, I explore ways in which people reason with semantic knowledge like colors and animals, and how we learn new words and concepts.

As a computational cognitive scientist, I follow a top-down approach to reverse-engineering human learning. This view holds that we can understand key aspects of cognition by identifying the problem being solved and comparing how people's behavior differs from an ideal observer. By generating a formal description of the problem being solved, we can develop experiments to test the types of mental representations and inductive biases people use in their everyday lives. As a result, this top-down approach provides us with a better understanding of human cognition and lets us build machine learning models that perform more like people.

Curriculum Vitae | Google Scholar