Extended Data Fig. 1: Analysis of bump drift during standing bouts.

a, Histograms of orientations in the ellipsoid body occupied by the compass bump at the beginning (blue) and end (red) of standing bouts for all ten flies. Note degree of overlap in the distributions, with no sign of an increase in specific orientations from beginning to restart. b, Cumulative distributions of orientations in the ellipsoid body occupied by the compass bump at the beginning (blue) and end (red) of standing bouts for all ten flies. Differences between the two distributions are not statistically significant. P-values for Watson’s U² test (flies 1-10): 0.5560, 1.0000, 1.0000, 0.9920, 0.9980, 1.0000, 0.9580, 1.0000, 0.9860, 0.1180. U² test statistic (flies 1-10): 0.0660, 0.0085, 0.0095, 0.0193, 0.0157, 0.0070, 0.0295, 0.0128, 0.0221, 0.1394. c, Drift during standing bouts for all ten flies, measured at different starting orientations of the compass bump. 8- and 16-Hz sinusoids were fit to drifts for each fly. One signature of discreteness in the performance of the compass system would be lower drift when the bump starts at stable orientations during standing bouts and higher drift when the bump starts outside of those orientations. We did not see such fluctuations in the data (see panel d). d, R² values for sinusoidal fits in panel c. In panels a-d, only those standing bouts that were greater that 0.3 s and less than 2 s were used for analyses. This resulted in the following numbers of standing bouts for flies 1-10: 980, 1005, 835, 826, 723, 714, 573, 527, 312, 949. Flies 2 and 6 correspond to flies GC7fA and GC7fB, respectively, in Fig. 1e,h–j.