On completion of training, the candidate will pursue a first tier university professorship or a position at a government research institution, and investigate human learning and decision making. His goal is to ultimately shed light on how the brain supports these behaviors to inform both clinical and neuroscientific applications. To this end, the candidate brings to bear powerful eye tracking techniques for measuring information processing of categories. The proposed research examines the interplay between the structure of the learning task and how people attend to features in the task environment. By examining such variables, the work will narrow the gap between tasks of the real world and those that are studied in the area of categories and concepts-an area that concerns the core cognitive process by which people conceive of distinct things as belonging to the same class. The proposal explores whether the effect of learning task on category representation is mediated by interplay between task dynamics and learners'sampling of the environment. To understand the interplay between task, eye movements, and concept representation, the proposal will incorporate modern reinforcement learning techniques from the machine learning literature into a category learning model. Reinforcement learning has been used to model eye movements in other areas including reading and problem solving. Neuroscientific work has found circuits linking the basal ganglia through dopamine with the frontal cortex that appear to operate according to reinforcement learning processes (Frank and Glaus, 2006). Moreover, Yechiam, et al. (2005) showed that the parameters in reinforcement learning models appear to map to specific clinical populations as they performed in a simple gambling task. Instead of gambling tasks, or artificial category learning tasks, the goal of the proposed work is to move towards tasks that might help to understand real world behaviors and their causes. The new category-reinforcement learning model proposed here will be applied first to the candidate's existing eye movement data, a rich data set on learners'attention to category information as they learn in standard tasks. Then, the model will be applied to the proposed behavioral experiments investigating the role of task dynamics on learners'sampling behavior.
The proposed research will improve our understanding of the cognitive processes having to do with learning and attention. By exploring learning tasks that are more like those experienced in the real world, the research can improve the applicability of cognitive models for diagnosing and devising treatments for deficits from ADHD, Autism, Asperger's, Huntington's disease, Parkinson's disease.
