Figure 1.

Task design and accuracies over time. (A) Task design. Subjects were shown two images of different classes (faces, objects or letters) per trial. These images were followed by a delay period. After the second delay period a cue was presented, informing the subject which target stimulus had to be kept in memory. After a final delay period, the subject was probed with a same-difference task. In this study we only used data of the period around the presentation of the first target, depicted as the period enclosed by the red square. All presentation timings were in reality 16.7 ms longer than depicted due to a fixed delay in stimulus presentation. The face image belongs to the KDEF dataset (M64; Lundqvist et al., 1998), the tool image is part of the BOSS set (Brodeur et al., 2010; CC BY-SA 3.0). (B) Averaged accuracy traces for the face-letter (green), face-object (red), and letter-object (blue) contrasts. The first black vertical line indicates stimulus onset, the second line coincides with stimulus offset. Bright colors indicate accuracies significantly above the accuracies obtained at baseline (before stimulus onset). We only tested post-stimulus onset accuracies against baseline. The dashed vertical line denotes the chance-level value of 0.5.