How the Level of Reward Awareness Changes the Computational and Electrophysiological Signatures of Reinforcement Learning

Abstract

The extent to which subjective awareness influences reward processing, and thereby affects future decisions, is currently largely unknown. In the present report, we investigated this question in a reinforcement learning framework, combining perceptual masking, computational modeling, and electroencephalographic recordings (human male and female participants). Our results indicate that degrading the visibility of the reward decreased, without completely obliterating, the ability of participants to learn from outcomes, but concurrently increased their tendency to repeat previous choices. We dissociated electrophysiological signatures evoked by the rewardbased learning processes from those elicited by the reward-independent repetition of previous choices and showed that these neural activities were significantly modulated by reward visibility. Overall, this report sheds new light on the neural computations underlying reward-based learning and decision-making and highlights that awareness is beneficial for the trial-by-trial adjustment of decisionmaking strategies.

Monetary feedback modulates performance and electrophysiological indices of belief updating in reward learning

“…In environments characterized by uncertain contingencies or unpredictable dynamics, belief updating has obvious survival implications; however, many of the cognitive and neurophysiological factors arbitrating this component of decision making remain unclear. In particular, one open question is how belief updating might be affected by the presence of reward and whether reward directly impacts on motivational state and, by doing so, improves belief updating (Achtziger, Alós‐Ferrer, Hügelschäfer, & Steinhauser, ; Correa et al, ). This question is of particular importance given the ongoing debate in educational psychology and behavioral economics regarding the efficacy of using rewards to incentivize performance (Hidi, ; Lazear, ).…”

“…remain unclear. In particular, one open question is how belief updating might be affected by the presence of reward and whether reward directly impacts on motivational state and, by doing so, improves belief updating (Achtziger, Alós-Ferrer, Hügelschäfer, & Steinhauser, 2015;Correa et al, 2018). This question is of particular importance given the ongoing debate in educational psychology and behavioral economics regarding the efficacy of using rewards to incentivize performance (Hidi, 2016;Lazear, 2000).…”

Dissociable effects of subliminal incentives on initiation and velocity control of goal-directed eye movements

“…This was exemplified in situations where only conscious compared to subconscious reward incentives resulted in an optimal speed-accuracy trade-off (Bijleveld et al, 2010), adjustment of performance based on the attainability of the expected rewards (Zedelius et al, 2012), and learning to modify choices depending on the outcomes (Correa et al, 2018). To understand our results in light of these previous findings, we note that by initiating more saccades participants could increase their total number of hits in the task we employed.…”

Pupil dilation and the slow wave ERP reflect surprise about choice outcome resulting from intrinsic variability in decision confidence

“…The slow wave amplitude on each trial was defined as the mean electrophysiological response in the window 0.5 to 0.8 seconds after feedback presentation, measured in a central region of interest (ROI): the averaged signal of electrodes F1, Fz, F2, FCz, FC1, FC2, Cz, C1, C2, CPz, CP1, CP2, Pz, P1, P2. The region of interest (electrodes) as well as time-window of interest for the singletrial slow waves (0.5 to 0.8 sec) were a priori selected and were identical to a previous study of our group on a similar topic (Correa et al, 2018). ERPs were calculated by taking the mean across all trials.…”

“…For exploratory analysis on the feedback-related negativity (FRN) we used exactly the same region of interest as for the slow wave, again identical to a previous study (Correa et al 2018). The FRN peaked around 400 ms at central electrodes, similar to Correa et al (2018). We used 350-450 ms post outcome stimulus as our time window of interest for the ANOVA's.…”