Jacob Russin

Research Description

Jake’s research interests lie broadly in the intersection between computational cognitive neuroscience and machine learning. Within this intersection, he has focused on generalization in neural network learning systems and on higher-level aspects of human cognition such as structure-learning, compositionality and systematicity (Russin, et al., in press; Russin, et al. 2020). In particular, he has demonstrated that certain inductive biases, including a separation between structure and content pathways in the brain, can induce systematic generalization in neural networks (Russin et al 2020; O’Reilly et al, in press), and has developed a model of the complementary structure-learning abilities of the cortical and episodic memory systems in the brain (Russin et al, in press). He is particularly interested in bringing a computational perspective to the learning that emerges in the prefrontal cortex (PFC) and its interaction with other brain areas (Russin et al 2020; O’Reilly et al 2020). He hopes to shed new light on the functionality of the PFC, bringing computational tools to bear on the complexity of the disorders that can arise from its disruption (O’Reilly et al 2019).