Description:

A central goal of behavioral neuroscience is to expose general principles of neural computation that give rise to cognition. Until recently, cognitive functions were thought to mainly rely on cortical areas such as the frontal cortex, while “old” subcortical areas, such as the midbrain superior colliculus (SC), have mostly been associated with sensorimotor processing. Based on these earlier work, we expected the SC to be passively inhibited while animals avoided unwanted reflexive actions. We were surprised to find that the SC played an active and pivotal role in suppressing context-inappropriate behaviors. Since then, my research has focused on understanding how the SC, in coordination with cortex, implements flexible decision making. 

My talk will cover work in two different decision-making tasks: flexible sensorimotor routing in freely-moving rats and motor planning in head-fixed mice. Using electrophysiology, optogenetics, pathway-specific two-photon calcium imaging, and computational modeling, we found that the SC is particularly important during the memory period in both tasks, when the context information or motor planning signals need to be maintained in the brain in the absence of sensory input and motor output. The participation of SC seems to increase as cognitive demand increases in both behaviors. Together, our experimental and computational work reveal the SC as a key node in the distributed cortico-subcortical network underlying flexible behaviors, and make surprising predictions for how decision making is implemented in brain circuits. If time allows, I will also talk about how we are trying to use recurrent neural networks to understand mechanisms of task switching, and other crazy new directions in my lab.