Fig. 4

Bump shifting in the P-ring. a–f Same as Figs. 3a–f. Like the C-ring circuit, the P-ring circuit was able to a develop an activity bump in response to a visual cue; b to maintain only one bump when presented with two cues (indicated by the red ticks); and c to follow the moving visual cue. However, the bump was less stable than that of the C-ring, as indicated by the occasional drift (a, b), by the larger variability in the bump deviation (d) and by the wider bump width e. In d and e, the data of the P-ring (black) are displayed together with those of the C-ring (gray, from Figs. 3d, e) for comparison. The error bars indicate the within-trial s.d. f The most distinct difference between the two rings was in their responses to the simulated body rotation in the dark. The activity bump shifted in the P-ring circuit but not in the C-ring circuit (Fig. 3f). g The trial-averaged moving speed of the bump was positively correlated with the strength of the unilateral input to PB. The error bars indicate s.d., while the straight line is the linear regression of the data points. h Left: The PB developed two activity bumps (gray bars, trial-averaged firing rate of the PEN neurons that innervate the corresponding PB regions) of the same height in response to the visual input. Right: The model predicted asymmetric bumps (gray bars) in the PB during horizontal body rotation in darkness due to unilateral input to the PB. The black curves in both panels depict the Gaussian fit to the activity bumps. The x-axis represents the EB region index relative to the peak of the bump on each side