Cognitive science · Dartmouth College
Academics
Computational methods for studying cognition, in biological and artificial systems alike.
Research interests
I am interested in work that apply computational methods to the study of cognition in biological and artificial systems. I currently work with Steven Frankland of Dartmouth’s Mind, Brain, & Computation Lab, on projects examining working memory in large language models.
More broadly, I’m interested in continuous learning, artificial general intelligence, and consciousness.
Publications & papers
-
2026
Backward Digit Span Benchmarks Working Memory in LLMs
Diak, C. J., Nguyen, N., Tibbetts, J. R., Webb, T. W., & Frankland, S. M.
· CogSci 2026 ·
PDF
We tested 15 frontier LLMs on digit span tasks and found a sharp asymmetry: models recall sequences forward almost perfectly but collapse when asked to recall them in reverse. That backward-span performance also tracks measures of fluid intelligence (Raven’s Matrices and ARC-AGI-1), suggesting it’s a useful benchmark for the “working” part of working memory in LLMs — and a point of contact with how humans handle capacity limits.
Notable classes
- Winter 2025 Information Theory in Cognitive Science
- Winter 2025 What is Consciousness?First-year writing seminar
- Spring 2025 Computational Methods in Cognitive Science
- Fall 2025 Theories of Consciousness