Affective and computational determinants of threat extinction biases

Pavlovian threat acquisition and extinction are fundamental processes by which individualslearn about threat and safety in their environment. Research has shown that humans learnmore rapidly and persistently to associate threatening and—somewhat counterintuitively—positive rewarding stimuli with aversive events, supporting predictions derived fromappraisal theories of emotion (Stussi et al., 2018; Stussi, Pourtois, et al., 2021). Here, thepresent study aimed to provide a confirmatory analysis of these findings and furthercharacterize their algorithmic bases. Data from the four original experiments (N = 247) usinga differential Pavlovian threat conditioning paradigm were combined and reanalyzed. In thisparadigm, threat-relevant (angry faces, snakes), positive-relevant (baby faces, happy faces,erotic images), and neutral (neutral faces, colored squares) stimuli were used as conditionedstimuli, and skin conductance response was measured as an index of learning. Computationalmodeling was applied to identify signatures of learning biases in Pavlovian threat acquisitionand extinction. An expanded model comparison indicated that a reinforcement-learningmodel differentiating between excitatory (learning from reinforcement) and inhibitory(learning from the absence of reinforcement) learning best explained the observed data.Although no evidence for differences in excitatory learning rates was found between stimuluscategories, both threat- and positive-relevant stimuli exhibited a lower inhibitory learning ratecompared to neutral stimuli, contributing to the persistence of the conditioned responseduring extinction. These results confirm the robustness of the original findings and furthervalidate the appraisal-based approach, thereby informing the affective and computationaldeterminants of Pavlovian threat extinction biases and their translational relevance.