Vilma Pullinen

Context has been theoretically proposed to exert greater impact on face emotion processing when expressions are ambiguous. However, evidence for such “context-weighting” as a function of expression ambiguity is very limited. We investigated the presence of context-weighting using emotive sentence cues that provided a predictive context for a neutral face that changed expression in response. Emotional expressions were either congruent or incongruent with the sentence cue. To modulate expression ambiguity we manipulated expression intensity: low (80%-intensity) and high (20%-intensity) ambiguity in Experiments 1a/b using angry and happy faces, with medium ambiguity (50%-intensity) added in Experiment 2 using disgust and sad faces. Participants categorised the face emotion. Error rates were lower when face expressions were congruent vs. incongruent with the predictive context. Crucially, congruency effects were larger when expressions were more ambiguous (Expts 1b and 2 especially), thus indicating greater context-weighting. Drift Diffusion Modelling revealed that this effect was underpinned by use of predictive context to improve the efficiency of face expression evidence accumulation. Our findings provide the first empirical evidence that emotion perception is based on flexible integration of both face and prior context within a predictive processing framework, with the degree of context-weighting determined by the level of expression ambiguity.