Fig. 3

Relating changes in attention task performance to changes in saccade metrics. a The change in detection rate is plotted against the saccade latency asymmetry, for each session involving SC inactivation (red), FEF inactivation (purple), or sham control (black). Both the SC and FEF data show a significant linear relationship, but the slope is steeper for the SC data than for the FEF data. Alternating white and gray-shaded regions indicate 30-ms bins in which we compared the changes in hit-rate asymmetry caused by SC and FEF inactivations; asterisks indicate bins for which the values were significantly different (Wilcoxon rank-sum, p < 0.05). b The change in detection rate is plotted against the saccade peak velocity asymmetry. Alternating white and gray-shaded regions indicate 75°/s bins in which we compared the changes in hit-rate asymmetry caused by SC and FEF inactivations. Other conventions same as in a. c The probability of observing a detection deficit is plotted against the saccade latency asymmetry for SC (red) and FEF (purple) inactivation sessions. Each circle shows the result from one experiment involving either the SC or FEF. The smooth curve shows the median, and shaded regions indicate 95% region of the posterior predictive distribution of the y variable. d The probability of observing a detection deficit is plotted against the saccade peak velocity asymmetry. Same conventions as in c. See also Supplementary Figure 3 for direct comparison of the changes in saccade latency and peak velocity caused by inactivation of SC or FEF, and Supplementary Figure 4 for control analyses that address the difference between the numbers of SC and FEF experiments