Figure 3.

Integration of postsaccadic visual information in the subsequent saccade S2 is statistically optimal. Predicted weight of postsaccadic vision in the Target+Background condition (from weights in Background and Target conditions; see Eq. 10) versus actual weight of postsaccadic vision in Target+Background condition. White squares, Group averages for the four target configurations, with confidence intervals (averages were computed by weighting the data of each subject according to their variance using an expectation-maximization algorithm; confidence intervals were obtained by bootstrap). The dotted lines mark the confidence zone for the bidimensional regression over the target configurations. Gray points, Data for each subject and target configuration; dashed line, main diagonal. “Express” responses, which did not show use of postsaccadic information (S1–S2 delay <250 ms; Fig. 5), were not considered.