COGS 346 Cognition and Deep Learning
This course will cover current topics at the intersection of cognitive neuroscience and deep learning. Recent successes in deep learning and AI have hinted that networks of simple units can produce surprisingly complex behaviors, and these models often even mirror the coding properties observed in the human brain. However, unlike artificial agents, humans excel at thinking creatively and reasoning about cause and effect, allowing us to behave flexibly in novel and ambiguous environments. What neural mechanisms support these behaviors? Which of these mechanisms are necessary for producing analogous capabilities in artificial agents? We will gain insight into these questions by reading primary literature and algorithm development exercises.