Using predictions to resolve emotional ambiguity: Facial expression intensity influences the reliance on prior expectations

Facial expressions vary in ambiguity, making emotion perception challenging. ‘Precision-weighting’ theory posits that interpretation of ambiguous facial expressions would rely more strongly on prior expectations to help resolve uncertainty, yet evidence for this is lacking. To investigate this we presented emotional sentences that set expectations for how a neutral face would then emote in response. Critically, the face expression was either reliable (high intensity) or ambiguous (low intensity). Across two experiments using angry/happy and sad/disgust emotions, we found better face emotion labelling performance when expressions matched (congruent) versus mismatched (incongruent) expectations. Importantly, congruency effects were larger when face expressions were low- versus high-intensity, indexing greater influence of prior expectations when expression information was more ambiguous. This provides the first empirical evidence of precision-weighting mechanisms in face emotion perception and prediction. Drift Diffusion Modelling confirmed that this precision-weighting effect was underpinned by use of prior sentence evidence and faster evidence accumulation.