M31. An Investigation of Feedback-Guided Decision-Making in Schizophrenia

Abstract

Background: Evidence from probabilistic reinforcement learning tasks have revealed impaired reward-driven learning in schizophrenia. This has been examined exclusively in the context of binary probabilistic choice paradigms. In real-world decision-making, however, individuals must also make choices when there are more than 2 competing options that vary in the frequency and magnitude of potential rewards and losses. To advance our understanding of decision-making in schizophrenia, it is important to examine how patients manage choices in the face of concurrent rewards and losses-especially when the immediately rewarding choice is not necessarily the advantageous option in the long run. Thus, the current study examined Win-Stay/Lose-Shift (WSLS) behavior on the Iowa Gambling Task (IGT) in order to examine the infuence of immediate rewards and losses in guiding real-world decision-making in schizophrenia. Method(s): Fifty-one patients with schizophrenia and 39 healthy controls completed the IGT, as well as a series of cognitive and clinical measures. We assessed WSLS by quantifying trial-by-trial choice behavior following wins and losses. Total Win-Stay refers to the proportion of times the same deck was chosen immediately after a reward, whereas Total Lose-Shift is the proportion of choice-shifts after receiving a loss. Additionally, Advantageous Win-Stay and Lose-Shift variables were calculated in order to index optimal decision-making on the IGT. Result(s): Group comparisons revealed that patients demonstrated signif-cantly lower Total Win-Stay rates (t =-3.3, P =.001), but higher Total Lose-Shift rates (t = 2.3, P =.026) compared to controls. This same effect was also seen for Advantageous WSLS rates. Further, patients made more disadvantageous choices, shifted their choices more often, and performed signifcantly worse on the task overall compared to controls. However, groups did not differ in total number of rewards or losses received. After partialling out the effects of working memory, correlational analyses revealed that for patients, depression and apathy severity were signifcantly related to lower Total Win-Stay rates, and higher levels of choice-shifting overall. Further, overall performance on the IGT was correlated with WSLS rates for both groups. Conclusion(s): The results of this study suggest that patients with schizophrenia experience impaired reward-driven decision-making in the context of multiple choices with concurrent gains and losses. This appears to be driven by a reduced propensity for Win-Stay behavior, accompanied by excessive Lose-Shift behavior. With the importance of reward processing and decision-making in generating goal-directed behavior, these fndings suggest a potential mechanism contributing to the motivation defcits seen in schizophrenia.