Fig. 1: Task, recording locations and behavioural results.

a, The double retro-cue task: each trial began with two images (one above and one below the fixation point), followed by a spatial cue indicating the target for the detection task. After a 5 s maintenance (main.) period, a stream of eight distractor images was presented (images are presented sequentially; only three are depicted for simplicity). Subjects indicated whether the cued image appeared in the stream. After their response, a second cue indicated either the same image or the previously uncued image. This was followed by another maintenance period, a target detection task and a response screen. b, The percentage of correct responses (n = 13 sessions) for the first target, second previously cued target (‘same’) and second previously uncued target (‘switch’). The dark and light blue lines represent the mean and s.e.m., respectively. Individual sessions are connected by grey lines. A permuted F test and two-sided paired t-tests were used to test differences between conditions. c,d, Recording site locations in MNI152 space, overlaid on the CIT168 T1 brain atlas. Sagittal slice, for plotting purposes MNI x coordinates for all sites were set to −21 (c). Horizontal slice, for plotting purposes MNI z coordinates were set to −20 (d). The white lines indicate the locations of axial and sagittal slices. Amy, amygdala; hipp, hippocampus.