Fig. 6
From: Coupled symmetric and asymmetric circuits underlying spatial orientation in fruit flies
The maintenance of spatial orientation memory in the full EB–PB circuit in a random walk paradigm. a A schematic of the input protocol for a random walk. A random walk consists of a sequence of interleaved periods of forward walking and rotation. During forward walking the RPEN ring neurons are activated to inhibit the P-ring, while during the rotation the RPEI ring neurons are activated to inhibit the C-ring. The PB receives a unilateral input when the simulated fly performs a rotational movement. REIP neurons (not shown) are activated throughout the entire trial. To allow smooth transitions between forward walking and rotation, each input has to cease 150 ms prior to the end of the corresponding movement type (see Methods). b As shown in the region-based firing rate plot, the full EB–PB circuit produced a stable activity bump that continuously updated its position when the body rotated in darkness. c EB activity of an example trial of a full period (120 s) of random walk in darkness (cue off at t = 1 s). d The perception of spatial orientation, as indicated by the peak position of the activity bump in EB (red) closely reflected actual fly body orientation (red), though the deviation between the two increased with time. e Trial-averaged (n = 20) deviation between the perception and actual body orientation increased with time (black circles: the full model, gray squares: the P-ring-only model). To visualize the trends of the data, we fit a square-root function to the data as represented by the curves. The result indicates that although the P-ring itself can track the changes of orientation, the accuracy is much worse than that of the full EB–PB model
