Fig. 3.

The trans-saccadic attention shift is synchronized to the saccade, not to fixation point offset. The time-course of attention shift, plotted separately for trials with saccade latencies shorter than the median (across all attentional conditions for that recording session) and saccade latencies longer than the median, is similar when aligned to saccade offset (a, c) but delayed (b, d) when plotted aligned to fixation point offset. Time-course for shorter-latency trials in red and longer-latency trials in blue. The attention shift time-course is defined as in Fig. 2a, b as the difference between the firing-rates in the attend-in and attend-out conditions (see Supplementary Fig. 2 for the corresponding PSTHs). Plots show mean difference and s.e.m. Results for Experiment 1 are shown in the top row (a, b) and for Experiment 2 in the bottom row (c, d). Red and blue vertical lines in each panel represent the mean time of fixation-point offset (a, c) or saccade latency (b, d) for trials with short-latency and long-latency saccades, respectively. Data from both monkeys were pooled for this analysis, and trials were dropped as necessary from each condition so as to match the timing distributions of saccade offset for the attend-in and attend-out conditions for each neuron (also see Text and Supplementary Figs. 5 and 6)