Fig. 1.

Changes in Prediction Error as a result of modulating the difference between the mean of the prior and evidence (Experiment 1), modulating the precision of prior expectations (Experiment 2a–c) or modulating the precision of the sensory evidence (Experiment 3). The curves represent probability distributions over a hypothetical dimension along which stimuli vary in similarity (x-axis). The blue, solid line represents the prior probability of one stimulus (e.g., word or face) given a cue (e.g., scene), while the red, dotted line represents the sensory evidence (likelihood) for that stimulus. Prediction Error is the divergence (lack of overlap) between these two distributions (i.e., greater when shaded area of overlap in prior and evidence is smaller). (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)