Poster E129, Monday, March 27, 2:30 – 4:30 pm, Pacific Concourse
Goal-directed decision making incidentally recruits reinforcement learning mechanisms
Nora C Harhen1, Anne GE Collins1; 1University of California, Berkeley
In everyday life, learning is more often motivated by the desire to reach a self-chosen goal than by external reinforcement. Theoretical work suggests that treating the attainment of a (sub-)goal as a pseudo-reward is is an effective way to learn complex behaviors, which may require going through many intermediary, value-neutral sub-goals before leading to reward. There has been indirect evidence for pseudo-rewards when reaching subgoals in EEG and fMRI, but as of yet, few studies have directly compared the reinforcing effects of goals vs. rewards on value learning. We hypothesize that reaching a goal is valuable, in and of itself, and similarly to external reward, reinforces choices leading up to it. In a new behavioral protocol, participants learned to select between pairs of boxes that contained different value-neutral outcomes with differing reliability. In reward trials, participants were told that outcomes were worth either one or zero points; in contrast, goal trial outcomes brought no points, but one outcome was selected by the participant as their desired goal. We probed participants’ box preferences during a test phase in extinction. Participants’ choices revealed similar preferences within reward and goal boxes of different reliabilities, indicating that reaching a goal did reinforce the previous choice as rewards do. Furthermore, they preferred boxes leading to self-chosen goals over those leading to externally-rewarding outcomes with the same reliability. Taken together, these results support our hypothesis that reaching a goal generates a reinforcing pseudo-reward, and that goal-directed choice leverages mechanisms of habitual value learning.
Topic Area: THINKING: Decision making