Are fear learning, interpretation bias training and global-local processing related to social anxiety? An individual differences study

This preregistered study (https://osf.io/pktw4/) aimed to integrate multiple research domains to uncover mechanisms of psychopathology. Relationships between cognitive processing biases—fear learning, interpretation bias (training), and global-local processing—and their prediction of social anxiety (SA) levels measured by the Social Phobia Inventory were examined. Using Bayesian modeling and meta-analytic techniques, task relationships and their association with SA symptoms were evaluated across two datasets (N1 = 209, N2 = 97). Results showed non-significant between-task relationships. Negative interpretation bias was consistently associated with higher SA symptoms, and fear generalization positively predicted SA symptoms in the larger sample. No significant relationship was found between SA symptoms and global-local processing. Crucially, while initial correlations suggested transdiagnostic effects, a post-hoc bifactor analysis revealed that fear generalization was associated with general psychopathology, whereas negative interpretation bias retained a unique, specific association with social anxiety that was not explained by general psychopathology. Limitations include analogue samples and online task administration. These findings emphasize the centrality of interpretation bias as a specific mechanism in social anxiety, distinct from the broader transdiagnostic role of fear generalization.