Poster E40, Monday, March 27, 2:30 – 4:30 pm, Pacific Concourse
The feedback-related negativity indicates different use of feedback in two spontaneous strategies for handling changing values
Sucheta Chakravarty1, Isha Ober1, Christopher R. Madan1,2, Yvonne Y. Chen1, Esther Fujiwara1, Jeremy B. Caplan1; 1University of Alberta, 2Boston College
We investigated how feedback is incorporated in learning the values of a large set of items: 48 words, divided half high- and half low-value. Participants had to bet with (high value) or against (low value) each word to earn 10 or 1 point reward, respectively. Importantly, this equated utility of high- and low-value words in maximizing total reward. After 16 blocks of value learning, the values of half of the words were reversed without warning. One third of participants (N=21), when faced with changing values, responded based on their previous knowledge of value (“conservative”). Another third (N=19) appeared to guess randomly (“exploratory”). The groups were equivalent in performance during prior value learning. Analysis of the mean amplitude at FCz during the FRN window (200–300 ms post feedback onset) for this surprise reversal revealed that for the conservative group, there was no significant effect of correct (i.e. 10 points) vs. incorrect (1 point) feedback, whereas this effect was significant for the exploratory group (p<0.05). This suggests the exploratory participants used prediction error, as indexed by the FRN, to guide their learning of changed values, whereas the conservative participants used the feedback differently.
Topic Area: EXECUTIVE PROCESSES: Monitoring & inhibitory control