Figure 3
From: Self-beneficial belief updating as a coping mechanism for stress-induced negative affect
(a) Structure of the model space. αUni = one learning rate for the whole time course; αHigh ability/ αLow ability = two separate learning rates for the two ability conditions; αPE+/αPE- = two separate learning rates for positive and negative prediction errors; adapted from Müller-Pinzler et al.21. (b) Protected exceedance probabilities resulting from the Bayesian Model Selection procedure including the prediction error learning models depicted in (a) and a mean model (M0) assuming stable means for each ability condition instead of continuous learning (c) Performance expectation ratings (EXP, solid line) and performance expectations predicted by the winning model (EXP − pred., dashed line) over the time course of 30 trials. Ratings and predicted values were averaged across participants separately for the two ability conditions and the three experimental groups. Shaded areas represent the standard errors of the expectation ratings for each trial. (d) Learning rates derived from the Valence Model (winning model). A significant interaction effect (*) of PE-Valence x Stress group (SOC = social-evaluative stress, PHY = physical stress, CON = no stress control) indicates that a bias towards increased updating in response to negative prediction errors (αPE-) in contrast to positive prediction errors (αPE+) is absent in the social-evaluative stress group. (d) Rank-based regression plot of valence bias score predicting the recovery from negative affect (REC, ratings T2AFF − T3AFF) in the subsample of the social-evaluative stress group (n = 29) controlled for the stress-induced change in negative affect (ΔAFF, ratings T2AFF − T1AFF), i.e., residuals of the valence bias score and the recovery predicted by ΔAFF are plotted. More self-beneficial belief updating (higher valence bias score) is associated with a better recovery from stress-induced negative affect.
