Fig. 3. Artificial visual rotation during saccades does not affect saccade dynamics (amplitude, duration or peak angular velocity). Flies performed spontaneous saccades in a visual panorama displaying a 45° dark stripe in the foreground (represented here in light gray for clarity) over a background of thin vertical stripes with a spatial frequency of 22.5°. When a fly began a saccade, either the foreground or background was rotated by a predetermined amount. Top to bottom: no visual rotation (control, n=134 saccades); foreground rotated with fly's turn (n=151); foreground rotated against fly's turn (n=148); background with (n=163); background against (n=142). Most of the distributions (here and in other figures) are neither normal nor lognormal (Shapiro-Wilk test, W<0.05); therefore, pink bars show the median value. None of the distributions here differed significantly from the control (Kruskal-Wallis nonparametric one-way ANOVA with Bonferroni correction for multiple comparisons, P>0.05). N=14 flies.