Sonja Johnson-Yu

Graduate Student in Rajan Lab

Contact Information

Best way to contact me: Email

Social Media


Sonja Johnson-Yu is a third-year computer science PhD candidate advised by Kanaka Rajan. Her research aims to understand animal communication by examining the interplay between signaling and behavior, unifying agent-based models of animal behavior with autoregressive animal “language” modeling. Previously, she received a BS, MS, and BA (Music) from Stanford, where she applied CV and NLP methods to problems in sustainability and education. Outside of lab, she enjoys singing in the Harvard University Choir and Ferris Choral Fellows.

Research Focus

Johnson-Yu’s broad aim is to understand animal communication by unifying language models of communication with agent-based models of animal behavior. Currently, her work examines electrocommunication in weakly electric fish. She uses sequential modeling, and reinforcement learning with the aim to distinguish electrolocation from social signaling, identify meaningful communication units in the context of goal-oriented behavior, and understand consensus formation in group signaling.