Supplementary Figure 5: Serial dependence does not arise from negative aftereffects alone.

We conducted a simulation to test whether apparent serial dependence in perceived orientation could arise from negative aftereffects alone, for example due to an interaction between negative aftereffects and random fluctuations in the stimulus sequence. Using the same trial sequences presented to subjects in Experiment 1, we simulated subjects' responses based on negative aftereffects of varying strength and persistence over time. We found that regardless of the aftereffect strength or duration, negative aftereffects alone predict a negative amplitude in the measured serial dependence (repulsion rather than attraction; panel a shows the simulation results for the stimulus sequence presented to one subject; all subjects showed comparable results), contrary to what we observed in subjects' responses. Panel b shows a sample position discrimination plot from the simulation. Predicted responses based solely on negative aftereffects are qualitatively dissimilar to the pattern of responses that subjects made in Experiments 1-6. This simulation shows that it is not possible for negative aftereffects alone to produce an apparent attraction of perceived orientation toward previously seen orientations within our experimental design and analysis.