Expectations about precision alter speech perception

Bayesian models of the mind suggest that we estimate the reliability or ‘precision’ of incoming signals to guide perceptual inference and to construct feelings of uncertainty about our percepts. However, accurately estimating precision is likely to be challenging for bounded systems like the brain. One way observers could overcome this challenge is to form expectations about perceptual precision and use these to guide awareness. Previously, we found that agents were able to acquire and use probabilistic expectations about the strength of visual signals – with expectations altering their confidence and subjective awareness without altering perceptual performance. Here we test whether the same principles hold in the domain of auditory speech perception. Participants made perceptual decisions about the clarity of different speech sounds, which they could expect to be clear or ambiguous. We found that these expectations biased subjective clarity ratings – such that listeners judged speech as clearer when clearer signals were expected. Computational modelling revealed that this effect could be well-explained by a predictive learning model that infers the precision (strength) of current signals as a weighted-combination of incoming evidence and top-down expectations. These results support an influential tenet of Bayesian models of cognition: agents do not only ‘read out’ the reliability of information arriving at their senses but also take into account prior knowledge about how reliable or ‘precise’ different sources of information are likely to be. We thus reveal that expectations about precision influence subjective impressions across sensory domains, pointing to a general principle that can explain how we come to trust or doubt our senses.