Fig. 3.

Design of Experiment 2a showing a single run. At the start of each run, all scenes were presented twice to familiarise participants with the images. Participants kept count of indoor/outdoor scenes. The Training trials then established either strong priors, by repeatedly presenting the same scene with the same face (High PE condition), weaker priors, by presenting a scene-face pair only once (Baseline condition) or least strong priors, by presenting the same scene with different faces (Low PE condition). Critical Study trials (intermixed with Training trials) used novel faces, evoking the different degrees of prediction error. Participants made speeded male/female decision on the faces. In the Test phase, associative memory for faces paired in the critical Study trials was assessed, using 3AFC in which the foils came from other critical Study trials.