Fig. 2

Joint bias map and conditional bias plots. A Joint bias map. Response errors are plotted as a function of both the previous stimulus and response direction relative to the current stimulus. For each subject, we binned the response errors within 10° bins according to the relative direction of the previous stimulus and response. Group means with less than fourteen subjects’ data were excluded before illustration. The color of each pixel represents response error on the current trial (warm color: positive error; cool color: negative error). Positive values on the x- and y-axis indicate that the previous stimulus and response direction was more clockwise than the current stimulus, respectively, and positive errors indicate that the estimated direction was more clockwise than the true stimulus direction. Estimation responses on the current trial are systematically repelled away from the stimulus direction of the previous trial, while they are strongly attracted toward the response direction of the previous trial. NA, not available. B Bias plot conditioned on relative direction of previous response. Response errors split by relative direction of the previous response were plotted as a function of relative direction of the previous stimulus. Bins with less than five trials within a subject and group means with less than fourteen subjects’ data were excluded before illustration. As can be seen across five lines with negative slopes, estimation responses are negatively biased away from the previous stimulus. C Bias plot conditioned on relative direction of previous stimulus. Response errors split by relative direction of the previous stimulus were plotted as a function of relative direction of the previous response. Estimation responses are positively biased toward the previous response, as indicated by the positive slopes. Shaded regions represent 95% confidence intervals