Dhara Yu

Your Photo

I am a PhD student in Psychology at UC Berkeley, advised by Bill Thompson and supported in part by the Cooperative AI PhD Fellowship.

I am broadly interested in understanding the computational principles that underlie social intelligence. Some of the questions I am particularly interested in are:

  • How do people and AI systems make inferences about others' goals, and how can we study these capabilities in ecologically valid settings?
  • How do intelligent agents develop new concepts and normative arrangements for coordinating complex forms of joint action?
  • How do internal goals and conceptual knowledge transform as a result of interaction with other agents?

I view AI as both an object of study and as a tool to help us overcome methodological challenges in the study of intelligent behavior; I'm particularly excited about its use for principled measurement of more complex, open-ended forms of behavior and for the development of new kinds of theoretical artifacts that promote (human) scientific understanding.

I write about topics at the intersection of cognitive science, cultural evolution, AI and metascience on my research blog.

Before my PhD, I spent a year in startup land, and before that, I earned a BS in Symbolic Systems and an MS in Computer Science from Stanford.

Selected Publications

GoalBench: Benchmarking goal inference in multi-turn interactions

Dhara Yu, Zahra Sheikhbahaee, Dylan Hadfield-Menell, and Rachel Ma (2026).

Under review

Experimental diffusion reveals structural simplification of emerging social norms

Dhara Yu, and Bill D. Thompson (2026).

Under review