PSYC 3310 Computational Cognitive Neuroscience

Our eyes receive ambiguous 2D images of the world, yet we perceive objects, places, people. We instantly know where to turn our attention, sifting through the continuous stream of sensory input and combining it with years of experience to make thousands of split-second decisions every day. We continuously learn and adapt to our world, becoming ever more efficient at the tasks we do. How does the noisy, stochastic, resource-limited wetware in our brains support such remarkable cognitive abilities? The central premise of this course is that cognition can be understood as computation: transformations of information that map sensory inputs onto behavior. These transformations can be abstracted into computational models that formalize how inputs (stimuli) lead to outputs (behavior) through internal computations over latent variables. Behavior alone is compatible with many verbal theories. Computational models force those theories to become mathematically precise, producing quantitative predictions that can be tested against behavioral and neural data. Through the lens of Marrs levels of analysis, the class will examine models of vision, memory, navigation, plasticity, cognitive control, learning, and decision making. By the end of this course, you should be equipped with the background to formulate computational hypotheses about cognitive functions and evaluate them using behavioral and neural data. Recommended background: This course is designed for students from a variety of disciplines, including psychology, neuroscience, computer science, mathematics, engineering, and related fields. Students will be most successful if they have basic experience with programming (e.g., CSCI1101) and probability and statistics (e.g., MATH4426). Beyond that, it would be helpfulthough not essentialif students have taken coursework in psychology, neuroscience, and/or cognitive science (e.g., PSYC1110), as well as some calculus and linear algebra.