Learning from the consequences and outcomes of the choices we make is crucial for our well-being and survival. However, how we learn in natural settings where multiple options or actions, each with many features and attributes, lead to a single reward feedback is poorly understood (e.g., feeling good or bad after consuming a few foods). The proposed research will use a combination of novel experimental and computational approaches to investigate the role of attention in reward learning in settings that better resemble the real world. It will also contribute to the development of a powerful and coherent computational framework for understanding this unexplored role of attention. The outcomes of this research will be of interest to multiple fields in psychological, cognitive, and social sciences as well as to the artificial intelligence community. This project will include direct mentoring of female undergraduate and graduate students working on research and teaching. The outreach program will introduce the tasks used in the experiments as games to high school students in order to teach them about attention, learning, and the brain through an engaging method. Overall, research activities focus on advancing basic science, on inspiration, recruitment, and retention of the next generation of STEM students, and on bringing brain-inspired ideas to engineering. This project is jointly funded by Perception Action Cognition (PAC), the Established Program to Stimulate Competitive Research (EPSCoR), and Science of Learning and Augmented Intelligence (SL).
The researchers hypothesize that in a complex, multi-dimensional environment, learning from reward feedback is strongly influenced by features or attributes of choice options that are attended to and used for making decisions, and subsequently shapes future deployments of attention. The major goal of the proposed research is to use a combination of novel experimental and computational approaches to study how attention influences reward learning in settings that better resemble the real world in two ways: 1) having more than one cue preceding an outcome, and 2) using choice options with multiple features. Experimentally, the team will investigate how exogenous and endogenous attention influence decision making and learning at different points in time (relative to reward feedback), and moreover, how endogenous attention is shaped by the properties of the decision makerâ€™s environment and influences learning strategies. The team will also develop computational models to reveal plausible mechanisms for how exogenous and endogenous attention exert their influence on reward learning in particular, and for interactions between attention, learning, and decision making more generally. The results will provide insights into neural mechanisms of adaptive learning in more naturalistic settings.
This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.