Sreejan Kumar

This paper develops a recurrent neural network model that learns to select computations by combining rational meta-reasoning with meta-learning, demonstrating its ability to approximate optimal symbolic models and replicate neural dynamics in macaque and human planning tasks.

The paper develops a corticostriatal neural network model to show how cortical input compression enables distinct control and dynamic functions in the striatum across time-sensitive tasks.

This paper analyzes the functional specialization of transformer-based language models and finds similarities with the human brain's cortical language network.

This paper introduces a functional searchlight method to analyze fMRI data, revealing that visual, auditory, and semantic features are more distributed in the brain than previously thought.