To behave adaptively, people need to flexibly align their decisions with their immediate and long-term goals. Failures of such goal-directed decision-making can result in an array of maladaptive behaviors, such as financial risk taking, drug abuse, or gambling. While past research has helped identify the drivers of goal-directed decisions, there are still significant gaps in our understanding of how people are able to reorient those decisions as flexibly as they do, as well as when and why they fail to do so. In a series of studies that combine computational modeling with measures of behavior, attentional focus, and brain activity, we are studying the capacities and limits of human decision-makers to flexibly adjust their information search and actions to different goals and task demands. By uncovering the hidden levers and potential failure modes of goal-directed behavior, our work better disentangles sources of real-world decision-making failures, and provide a path towards targeted interventions to better align choices with long-term goals. <br/><br/>Our project addresses critical gaps in research on the neural and computational mechanisms that link decision-making and cognitive control. Past work on the computational and neural mechanisms of goal-directed decision-making has been singularly focused on a narrow subset of human goals (how people select the best option from a set). It is therefore unknown how people flexibly reconfigure to their wider array of goals – for instance, selecting under different criteria or accumulating information across options rather than comparing between them – and when and why they fail to do so. Our project develops and tests a computational framework that accommodates the range of flexibility observed in human decision-making, by representing explicit choice goals that define a) how information is translated into evidence and b) how evidence is then integrated to select responses. We are testing our framework in a series of studies that combine behavioral tasks with eye-tracking, EEG and fMRI. This multi-modal approach allows us to uncover the neural circuits and dynamics that enable people to flexibly transform and integrate information about their options to achieve their current goals, and to understand how and why people vary in these capacities. The project further supports outreach activities aimed at training researchers in computational methods for predicting and testing a wide array of decision-making behavior.<br/><br/>A companion project is being funded by the Federal Ministry of Education and Research, Germany (BMBF).<br/><br/>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.