Abstract

In fear differential conditioning, stimuli that resemble the conditioned stimulus (CS+) are more likely to trigger fear responses. Excessive fear responses on stimuli not like CS + are often associated with anxiety. However, the threat judgments process and how this process manifests itself differently in subgroups with different generalisation rule applications, is unclear. This study examines whether anxiety biases the threat decision process in fear generalisation paradigm and whether subgroups characterised by different generalisation gradients was interpreted differently by drift-diffusion model. We gathered behavioural data through a binary fear generalisation judgment task and clustered participants based on their responses. Reaction time distributions and individual scale scores were analyzed using the hierarchical drift-diffusion model. The model results suggested that similarity and state anxiety facilitated evidence-gathering processes that favoured “threat” judgments, but at the same time, state anxiety weakened the effect of stimulus similarity as evidence. Further cluster analyses revealed that this effect of anxiety on threat judgments only held true for specific subgroups of participants. This pioneering computational modelling effort in fear generalisation underscores the significant role of strategy preference and its complex interaction with anxiety in shaping stimulus processing.