Discovering Symbolic Cognitive Models from Human and Animal Behavior

Pablo Samuel Castro, Nenad Tomasev, Ankit Anand, Navodita Sharma, Rishika Mohanta, Aparna Dev, Kuba Perlin, Siddhant Jain, Kyle Levin, Noemi Elteto, Will Dabney, Alexander Novikov, Glenn C Turner, Maria K Eckstein, Nathaniel D. Daw, Kevin J Miller, Kim Stachenfeld
Proceedings of the 42nd International Conference on Machine Learning, PMLR 267:6849-6890, 2025.

Abstract

Symbolic models play a key role in cognitive science, expressing computationally precise hypotheses about how the brain implements a cognitive process. Identifying an appropriate model typically requires a great deal of effort and ingenuity on the part of a human scientist. Here, we adapt FunSearch (Romera-Paredes et al. 2024), a recently developed tool that uses Large Language Models (LLMs) in an evolutionary algorithm, to automatically discover symbolic cognitive models that accurately capture human and animal behavior. We consider datasets from three species performing a classic reward-learning task that has been the focus of substantial modeling effort, and find that the discovered programs outperform state-of-the-art cognitive models for each. The discovered programs can readily be interpreted as hypotheses about human and animal cognition, instantiating interpretable symbolic learning and decision-making algorithms. Broadly, these results demonstrate the viability of using LLM-powered program synthesis to propose novel scientific hypotheses regarding mechanisms of human and animal cognition.

Cite this Paper

BibTeX
@InProceedings{pmlr-v267-castro25a, title = {Discovering Symbolic Cognitive Models from Human and Animal Behavior}, author = {Castro, Pablo Samuel and Tomasev, Nenad and Anand, Ankit and Sharma, Navodita and Mohanta, Rishika and Dev, Aparna and Perlin, Kuba and Jain, Siddhant and Levin, Kyle and Elteto, Noemi and Dabney, Will and Novikov, Alexander and Turner, Glenn C and Eckstein, Maria K and Daw, Nathaniel D. and Miller, Kevin J and Stachenfeld, Kim}, booktitle = {Proceedings of the 42nd International Conference on Machine Learning}, pages = {6849--6890}, year = {2025}, editor = {Singh, Aarti and Fazel, Maryam and Hsu, Daniel and Lacoste-Julien, Simon and Berkenkamp, Felix and Maharaj, Tegan and Wagstaff, Kiri and Zhu, Jerry}, volume = {267}, series = {Proceedings of Machine Learning Research}, month = {13--19 Jul}, publisher = {PMLR}, pdf = {https://raw.githubusercontent.com/mlresearch/v267/main/assets/castro25a/castro25a.pdf}, url = {https://proceedings.mlr.press/v267/castro25a.html}, abstract = {Symbolic models play a key role in cognitive science, expressing computationally precise hypotheses about how the brain implements a cognitive process. Identifying an appropriate model typically requires a great deal of effort and ingenuity on the part of a human scientist. Here, we adapt FunSearch (Romera-Paredes et al. 2024), a recently developed tool that uses Large Language Models (LLMs) in an evolutionary algorithm, to automatically discover symbolic cognitive models that accurately capture human and animal behavior. We consider datasets from three species performing a classic reward-learning task that has been the focus of substantial modeling effort, and find that the discovered programs outperform state-of-the-art cognitive models for each. The discovered programs can readily be interpreted as hypotheses about human and animal cognition, instantiating interpretable symbolic learning and decision-making algorithms. Broadly, these results demonstrate the viability of using LLM-powered program synthesis to propose novel scientific hypotheses regarding mechanisms of human and animal cognition.} }
Endnote
%0 Conference Paper %T Discovering Symbolic Cognitive Models from Human and Animal Behavior %A Pablo Samuel Castro %A Nenad Tomasev %A Ankit Anand %A Navodita Sharma %A Rishika Mohanta %A Aparna Dev %A Kuba Perlin %A Siddhant Jain %A Kyle Levin %A Noemi Elteto %A Will Dabney %A Alexander Novikov %A Glenn C Turner %A Maria K Eckstein %A Nathaniel D. Daw %A Kevin J Miller %A Kim Stachenfeld %B Proceedings of the 42nd International Conference on Machine Learning %C Proceedings of Machine Learning Research %D 2025 %E Aarti Singh %E Maryam Fazel %E Daniel Hsu %E Simon Lacoste-Julien %E Felix Berkenkamp %E Tegan Maharaj %E Kiri Wagstaff %E Jerry Zhu %F pmlr-v267-castro25a %I PMLR %P 6849--6890 %U https://proceedings.mlr.press/v267/castro25a.html %V 267 %X Symbolic models play a key role in cognitive science, expressing computationally precise hypotheses about how the brain implements a cognitive process. Identifying an appropriate model typically requires a great deal of effort and ingenuity on the part of a human scientist. Here, we adapt FunSearch (Romera-Paredes et al. 2024), a recently developed tool that uses Large Language Models (LLMs) in an evolutionary algorithm, to automatically discover symbolic cognitive models that accurately capture human and animal behavior. We consider datasets from three species performing a classic reward-learning task that has been the focus of substantial modeling effort, and find that the discovered programs outperform state-of-the-art cognitive models for each. The discovered programs can readily be interpreted as hypotheses about human and animal cognition, instantiating interpretable symbolic learning and decision-making algorithms. Broadly, these results demonstrate the viability of using LLM-powered program synthesis to propose novel scientific hypotheses regarding mechanisms of human and animal cognition.
APA
Castro, P.S., Tomasev, N., Anand, A., Sharma, N., Mohanta, R., Dev, A., Perlin, K., Jain, S., Levin, K., Elteto, N., Dabney, W., Novikov, A., Turner, G.C., Eckstein, M.K., Daw, N.D., Miller, K.J. & Stachenfeld, K.. (2025). Discovering Symbolic Cognitive Models from Human and Animal Behavior. Proceedings of the 42nd International Conference on Machine Learning, in Proceedings of Machine Learning Research 267:6849-6890 Available from https://proceedings.mlr.press/v267/castro25a.html.

Related Material