Cognitive and Translational Neuroscience of Category Learning
Is that circle or a square? Does that x-ray show a tumor? Each of these decisions involves categorization. We make hundreds of categorization decisions every day and, in general, we get better with experience. For example, a circle is rarely confused for a square. Some of these categorization decisions can be made easily and the strategy can be verbalized. These categorization strategies that are available to conscious awareness and can be verbalized are thought to be frontally mediated. Other categorization decisions cannot be verbalized. For example, the radiologist can accurately classify whether a tumor is present but will find it difficult to explain exactly how they make that decision. These categorization strategies that are not available to conscious awareness and cannot be verbalized are thought to be striatally mediated.
A major focus of our research is to examine the computational and neurobiological underpinnings of category learning. We achieve this goal through a blending of empirical data collection, cognitive neuroscience, and mathematical modeling.
Whereas the majority of the work conducted in our laboratory has focused on visually presented stimuli, over the past few years, we have begun to turn our attention to the auditory domain; specifically, auditory and speech category learning. The neural underpinnings are similar to those in vision yet multiple systems approaches have not been applied in these domains. In collaboration with Dr. Bharath Chandrasekaran from the Communication Sciences and Disorders department here at UT, we have begun to apply the dual systems framework made popular in vision to the auditory and speech domains. We find a remarkable similarity between visual and auditory category learning, but with some important caveats. We are examining changes in auditory and speech category learning across the lifespan and in special populations, such as musicians.
Cognitive and Translational Neuroscience of Decision Making
The way we make decisions often involves deciding whether to accept some short-run reward that in the long-run may be sub-optimal or whether to accept some long-run reward by forgoing some short-run reward. A major focus of our research is to examine the computational and neurobiological underpinnings of these types of decisions. In this work, we examine history-independent decision making in which the rewards available on the current choice are independent of the previous choice history, and compare that with history-dependent decision making in which the currently available rewards are dependent upon the previous reward history.