Stephan C. Meylan, Ph.D.

I am a computational cognitive scientist who focuses on reverse engineering how children learn language, both to support the development of machine intelligence and to better understand early developmental processes. More specifically, I work on understanding how children and adults communicate and coordinate to solve progressively more complex problems using language.

I make heavy use of computational models (neural networks and Bayesian probabilistic models) and my work is distinguished by an emphasis on multi-agent and multi-task learning contexts. I am deeply preoccupied with Moravec’s Paradox. I also use experimental methods like eyetracking to characterize children’s knowledge. See Publications and Research Topics for more details.

I’m currently a research scientist in the Department of Linguistics at UC Berkeley, working on animal and human models of communication with Gasper Begus. Previous to that, I was a research sscientist at the Department of Brain and Cognitive Sciences at MIT (in Dr. Roger Levy’s Computational Psycholinguistics Lab). I have previously worked with the Bergelson Lab (Dr. Elika Bergelson, at Duke and now Harvard). I completed a Ph.D. in the Computational Cognitive Science Lab at UC Berkeley (with Dr. Tom Griffiths). My work has been funded by the NIH, NSF, U.S. Air Force Office of Scientific Research, DARPA, and the Simons Foundation.

Before focusing on human cognition, I was a data scientist at a crowdsourcing startup and a geospatial data analyst at the U.S. Geological Survey, where I worked on tools for characterizing ocean acidification. I spend my non-academic time on large-format film photography, backpacking and trekking, scuba diving, and vegetable gardening.