Feedback-related ERPs during value-learning foreshadow how participants later handle reversal learning
Sucheta Chakravarty1, Isha Ober1, Christopher R. Madan2, Yvonne Y. Chen3, Jeremy B. Caplan1; 1University of Alberta, 2University of Nottingham, 3Baylor College of Medicine
How do people handle the situation in which previously known item-values change? We used a feedback driven learning paradigm where participants learned, through trial and error, the values associated with a set of 48 words, divided equally in high-value (10 points for choosing, 1 point for not choosing) and low-value (1 point for choosing, 10 points for not choosing) words. After 16 blocks of value learning, the values of half of the words reversed without warning. Some of our participants (N=19, “exploratory”) guessed at new values, while some other (N=21, “conservative”) responded based on previous knowledge of values (presented at CNS 2017). We wondered if participants who applied these strategies had, in fact, learned item values differently. Behavioural measures failed to distinguish the two strategy groups during the 16 blocks of value learning (i.e., prior to the value reversal). However, ERPs in response to feedback stimuli, during value learning, were strikingly different. Strategy interacted significantly with latency at electrode FCz, with conservative participants showing a greater deviation from baseline at earlier latencies (negative voltage) and exploratory participants showing a greater deviation from baseline at later latencies (positive voltage). These findings suggest that the participants that were conservative incorporated feedback-information proactively, anticipating feedback and updating value knowledge earlier. In contrast, participants that were exploratory waited until feedback arrived, updating value knowledge in reaction to feedback cues. In sum, two equivalently effective approaches to learning item-values may reflect different ways of processing feedback and updating value knowledge.
Topic Area: EXECUTIVE PROCESSES: Other