Figure 1
The experimental procedure. Each trial pairing comprises a trial of ABC retrocuing followed by a trial of 1-item delayed recall. In the ABC-retrocuing task, the presentation of two sample items (A and B) is followed by a retrocue, then by the presentation of a third sample item (C). The onset of the response wheel prompts recall of the orientation of either the retrocued item or item C. In the no-overlap condition (top) item C appears at a location different from that of the first two samples; in the overlap condition (bottom) C appears at the same location as had the IMI. In the linked 1-item delayed-recall task, both the sample and the response wheel appear centrally.
Figure 2
Results from the preliminary task, illustrating residual error of 1-item delayed recall as a function of angular difference with an item from the preceding ABC-retrocuing trial. A. Attractive bias of retrocued item in the no-overlap condition. B. Attractive bias of retrocued item in the overlap condition. C. Attractive bias of IMI in the no-overlap condition. D. Repulsive bias of IMI in the overlap condition. In each panel the thin black line represents the moving average of residual errors, with the SEMs in gray shading. The width of the moving average window is 100 datapoints and the step size is 20 datapoints. The thick black line shows the DoG fit to the data.
Figure 3
Results from the Registered Report, illustrating residual error of 1-item delayed recall as a function of angular difference with an item from the preceding ABC-retrocuing trial. A. Attractive bias of retrocued item in the no-overlap condition. B. The bias of retrocued item in the overlap condition. It is the absence of evidence for a significantly attractive bias in this condition that constitutes the failed sanity check. C. The bias of IMI in the no-overlap condition. D. The bias of IMI in the overlap condition. The width of the moving average window is 200 datapoints and the step size is 40 datapoints. Plotting conventions same as Figure 2.
Figure 4
Results from the Registered Report, illustrating the bias on 1-item delayed recall induced by items from the preceding ABC-retrocuing trial, in the subset of trials in which item A was tested. A. Attractive bias of retrocued item in the no-overlap condition. B. Attractive bias of retrocued item in the overlap condition. C. The bias of IMI in the no-overlap condition. D. The bias of IMI in the overlap condition. The width of the moving average window is 200 datapoints and the step size is 40 datapoints. Plotting conventions same as Figure 2.
Figure 5
Results from the follow-up experiment, illustrating the bias on 1-item delayed recall induced by items from the preceding ABC-retrocuing trial. A. Bias of the retrocued item in the no-overlap condition. B. Attractive bias of the retrocued item in the overlap condition. C. Bias of the IMI in the no-overlap condition. D. Repulsive bias of the IMI in the overlap condition. The width of the moving average window is 200 datapoints and the step size is 40 datapoints. Plotting conventions same as Figure 2.
Figure 6
Results from the follow-up experiment, illustrating the bias on 1-item delayed recall induced by items from the preceding ABC-retrocuing trial, in the subset of trials in which item A was tested. A. Bias of the retrocued item in the no-overlap condition. B. Attractive bias of the retrocued item in the overlap condition. C. Bias of the IMI in the no-overlap condition. D. Bias of the IMI in the overlap condition. The width of the moving average window is 200 datapoints and the step size is 40 datapoints. Plotting conventions same as Figure 2.
Figure 7
Results from the follow-up experiment, illustrating the bias on recall of item C induced by the IMI in the same trial, in the subset of trials in which item C was tested. A. Bias of the IMI in the no-overlap condition. B. Attractive bias of the IMI in the overlap condition. The width of the moving average window is 200 datapoints and the step size is 40 datapoints. Plotting conventions same as Figure 2.