Sashank Varma

Professor

Research Interests

computational models of cognition, mathematical thinking, language understanding, cognitive science, artificial intelligence, machine learning, education

About

Sashank Varma is a cognitive scientist who uses human experimentation and computational modeling to investigate those complex forms of cognition that are uniquely human and indeed make us human.

His primary line of empirical research is in mathematical and computational thinking. He and his students are investigating, for example, the cognitive mechanisms by which people understand abstract mathematical concepts (e.g., in geometry and topology) and reason about computationally ‘hard’ problems (e.g., the traveling salesperson problem).

His research program in artificial intelligence focuses on the cognitive alignment between humans and machine learning models. His group is investigating whether large language models, vision models, and multimodal models ‘understand’ mathematics, language, and concepts as humans do, and if they ‘reason’ similarly as well. His most recent work is taking on the question of developmental alignment – whether the improving performance of machine learning models over training parallels the progression of children’s thinking over development.

Selected Publications

Bye, J. K., Marupudi, V., Koenen, R., Park, J., & Varma, S. (2025). Estimation of factorial expressions and its improvement through calibration: A replication and extension of Tversky and Kahneman (1973). Memory & Cognition. https://doi.org/10.3758/s13421-024-01662-0

Marupudi, V., & Varma, S. (2024). Human visual clustering of point arrays. Psychological Review. https://doi.org/10.1037/rev0000525

Varma, S., Sanford, E. M., Shaffer, O., Marupudi, V., & Lea, R. B. (2024). Recruitment of magnitude representations to understand graded words. Cognitive Psychology, 153, 101673. https://doi.org/10.1016/j.cogpsych.2024.101673

Varma, S. (2024). Meeting John Bransford. Educational Psychology Review, 36, 74. https://doi.org/10.1007/s10648-024-09912-y

Rao, V. N. V., Bye, J. K., & Varma, S. (2024). The psychological reality of the learned “p < .05” boundary. Cognitive Research: Principles and Implications, 9, 27. https://doi.org/10.1186/s41235-024-00553-x

Upadhyay, N., Marupudi, V., Varma, K., & Varma, S. (2025, February). Alignment of CNN and human judgments of geometric and topological concepts. In Proceedings of The 39th Annual AAAI Conference on Artificial Intelligence (AAAI’25) (pp. XXX-XXX), Philadelphia, PA.

Shah, R. S., Bhardwaj, K., & Varma, S. (2024, November). Development of cognitive intelligence in pre-trained language models. In 2024 Empirical Methods in Natural Language Processing Conference (pp. 9632–9657). Miami, Florida. https://doi.org/10.18653/v1/2024.emnlp-main.539

Gandhi, A., Shah, R. S., Marupudi, V., & Varma, S. (2024, October). Natural mitigation of catastrophic interference: Continual learning in power-law learning environments. In Proceedings of the 27th European Conference on Artificial Intelligence (pp. 2886-2893). Santiago de Compostela, Spain. https://doi.org/10.3233/FAIA240826