Fig. 5. The rate and time of pre-saccadic rise vary in a manner consistent with ramp-to-threshold models.

a Saccade direction choice probabilities. Data presented as mean (blue) ± SEM (gray) averaged over fixation duration (n = 19 samples examined over 16 fish). Each sample computed using 100 randomly selected fixations of constant duration. Dotted black line shows choice probability from random guessing. b Visualization of the slope statistic (gold, arbitrary units) used in c and d to quantify rate of pre-saccadic rise in deconvolved fluorescence before each saccade. Deconvolved fluorescence (black, scaled to fluorescence shown in blue) is plotted with simultaneously recorded eye movements. The arrows labeled L/R indicate the direction of leftward/rightward movements. c Histogram of slopes. Bin size = 50 (arbitrary units). d Slope versus time of pre-saccadic rise with respect to upcoming saccade (bin size = 500 ms; n = 2,715 fixations from 380 cells examined over 16 fish). e Example traces of population average, pre-saccadic deconvolved fluorescence as a function of time until upcoming saccade. Fixation duration was fixed (within 0.5 s) at either 5 s (n = 235 fixations from 107 cells examined over 10 fish) or 12 s (n = 168 fixations from 141 cells examined over 13 fish). f Deconvolved fluorescence (blue) at the time of saccade versus fixation duration (n = 2,375 fixations from 388 cells examined over 16 fish). Gray trace is from the same cells after selecting random saccade times. g Hypothetical mechanism for how a pre-saccadic signal could initiate saccades. h Predicted time until saccade versus actual time (gray dots). Blue line shows median ± SEM (500 ms bins; n = 405 samples from 388 cells examined over 16 fish). i Prediction error versus actual time until saccade. Data presented as median ± SEM (500 ms bins; n = 405 samples from 388 cells examined over 16 fish). Data in d–f are presented as mean ± SEM. Source data are provided in a Source data file.