I'm a PhD student at Stanford University, advised by Noah Goodman. I study in-context learning, representational geometry, and reasoning in neural networks and humans, drawing on normative frameworks to explain why these systems behave and represent information as they do. I've also worked on using LLMs to scale and automate scientific discovery in cognitive science.
During my PhD, I spent time as a research fellow at Goodfire AI. Previously, I completed my undergrad at the University of Toronto in psychology and computer science.
Selected papers
* equal contribution
All papers
- Language Model Activations Inhabit Privileged Error-Correcting Basins. Matthew Finlayson, Francisco Pernice, Eric Todd, …, Daniel Wurgaft, …, Ekdeep Singh Lubana. preprint 2026
- auto-psych: Automating the science of mind using agent-driven theory discovery and experimentation. Ben Prystawski, Kushin Mukherjee, Daniel Wurgaft, …, Michael C. Frank. preprint 2026
- Structuring Sparsity: Block-Sparse Featurizers Capture Visual Concept Manifolds. Thomas Fel, Matthew Kowal, Mozes Jacobs, …, Daniel Wurgaft, …, Atticus Geiger. preprint 2026
- Alexithymia and ill-being and well-being: The role of emotion regulation. David A. Preece, Ashish Mehta, Daniel Wurgaft, …, James J. Gross. Emotion 2026
- Why Larger Models Learn More: Effects of Capacity, Interference, and Rare-Task Retention. Jing Huang, Daniel Wurgaft, Rachit Bansal, …, Ekdeep Singh Lubana. preprint 2026
- Stories in Space: In-Context Learning Trajectories in Conceptual Belief Space. Eric J. Bigelow, Raphael Sarfati, Daniel Wurgaft, …, Ekdeep Singh Lubana. preprint 2026
- Manifold Steering Reveals the Shared Geometry of Neural Network Representation and Behavior. Daniel Wurgaft*, Can Rager*, Matthew Kowal*, …, Ekdeep Singh Lubana. COLM 2026
- Arithmetic in the Wild: Llama uses Standard Addition to Reason About Cyclic Concepts. Sheridan Feucht, Tal Haklay, Usha Bhalla, Daniel Wurgaft, …, Atticus Geiger. COLM 2026
- Do Sparse Autoencoders Capture Concept Manifolds? Usha Bhalla, Thomas Fel, Can Rager, …, Daniel Wurgaft, …, Ekdeep Singh Lubana. COLM 2026
- The Shape of Beliefs: Geometry, Dynamics, and Interventions along Representation Manifolds of Language Models' Posteriors. Raphael Sarfati, Eric J. Bigelow, Daniel Wurgaft, …, Ekdeep Singh Lubana. preprint 2026
- Priors in time: Missing inductive biases for language model interpretability. Ekdeep Singh Lubana, Can Rager, Sai Sumedh R. Hindupur, …, Daniel Wurgaft, …, Aaron Mueller. ICLR 2026
- Belief Dynamics Reveal the Dual Nature of In-Context Learning and Activation Steering. Eric J. Bigelow*, Daniel Wurgaft*, Ying-Qiao Wang, …, Ekdeep Singh Lubana. ICML 2026
- Diagnosing Bottlenecks in Data Visualization Understanding by Vision-Language Models. Alexa R. Tartaglini, Satchel Grant, Daniel Wurgaft, Christopher Potts, Judith E. Fan. preprint 2025
- In-Context Learning Strategies Emerge Rationally. Daniel Wurgaft*, Ekdeep Singh Lubana*, Core Francisco Park, Hidenori Tanaka, Gautam Reddy, Noah D. Goodman. NeurIPS 2025
- Scaling up the think-aloud method. Daniel Wurgaft*, Ben Prystawski*, Kanishk Gandhi, Cedegao E. Zhang, Joshua B. Tenenbaum, Noah D. Goodman. CogSci 2025
- Inverse Scaling: When Bigger Isn't Better. Ian R. McKenzie, Alexander Lyzhov, Michael Pieler, …, Daniel Wurgaft, …, Ethan Perez. TMLR 2023















