Sujaya Neupane, Ph.D.

Dr. Sujaya Neupane is a postdoctoral fellow in the Jazayeri lab at MIT interested in understanding the neural basis of learning cognitive behavior. He obtained his Ph.D. from McGill University under the mentorship of Christopher Pack and Daniel Guitton, studying the effects of eye movement in visual cortical neurons. In his Ph.D. work, Sujaya showed eye-movement-related transient changes in neural connectivity and correctly predicted that subjects over-trained to make a particular eye movement demonstrate perceptual effects of training during fixation. In his postdoctoral work, in order to enquire how learned skills and concepts are stored and used flexibly, he and Dr. Jazayeri have developed a cognitive task which requires subjects to establish long term memory of a sequence of images and make inferences based on this memorized 1D map. Using computational approaches and neurophysiological recordings from primate medial temporal lobe and posterior parietal cortex, they are testing the computational hypotheses of mental simulation for inference and planning.

Principal Investigator: Mehrdad Jazayeri

Fellow: Xiaomei Fan

Project
Humans and animals can flexibly infer the vectors connecting any two points in their spatial environment. It is thought that neural representation of space code underlies such relational inference, but what computations enable the inference behavior is not known. We have designed a task to investigate the neural basis of mental computation when subjects mentally navigate from one point of a map to another. To test our computational hypotheses, we use normative models to characterize the behavior of humans, monkeys and recurrent neural networks trained to do the task; we also analyze electrophysiological signals recorded from multiple brain areas of monkeys engaged in the task. Potential projects for the candidate include – (i) simulating alternate models of neural representation of a map to make predictions of behavior (ii) implementing unsupervised sequence detection algorithms on primate hippocampal data to test the hypothesis of mental simulation.