Supplementary Figure 2: Computing the just noticeable difference (JND) from two alternative forced choice data.

To compute the just noticeable difference (minimum orientation difference that can be reliably discriminated) for the Gabors presented in our experiments, we used the 2AFC discrimination data from Experiment 3, this time using only the information from the 2AFC judgments. For each individual subject, we plotted the percent of the time that the subject responded “right” as a function of the relative orientations of the Gabors presented in the left and right visual fields (positive values on the abscissa indicate that the right-hand Gabor was more clockwise). Data shown here are from one example subject. We fit a logistic function to the resulting plot as in the analysis of Experiment 3, and found the x values for which the logistic curve had a y value of 25% and 75%. One half of the difference between these two x values was taken as the JND. The mean JND for the three subjects who participated in Experiment 3 was 5.39. The same three subjects participated in Experiment 1; the mean amplitude of serial dependence for these subjects was 8.21; that is, two orientations that differed by 16.42°. could appear identical based on the stimulus that preceded them, an effect that is 3.05 times the JND as measured here in the same subjects.