Article Figures & Data
Figures
- Figure 1.
Walking flies fixate on single vertical bars. A, A fly walks on an air-supported ball, using its movements to turn the ball by walking forward in the x direction, side-stepping in the y direction, or by rotating the ball around the z-axis (theta). B, The fly is surrounded by a diamond-shaped LED array arena (here, viewed from above) onto which a visual stimulus (a vertical dark bar) is displayed. The fly's movements on the ball were detected by a camera and registered through FicTrac software, which was then fed back through Vision Egg software to change the location of the bar in the arena (closed-loop behavior). C, An example of a fly fixating strongly on the flickering dark bar. The flies would also adjust to 90° displacements occurring for 300 ms randomly during the experiment, by bringing the bar back to the front (cyan and magenta spikes). D, This resulted in a mean distribution of bar locations to the front of the fly, as evidenced in the polar plot. Yellow arrow is the mean vector of the location of the bars. Rayleigh test for nonuniformity of bar positions: z statistic: 20812; p < 0.0000001. E, The 2D path of the same fly from C takes into account the displacements, where the fly generally walks straight, turning only in response to the displacements to the left (magenta dots) or to the right (cyan dots). F, G, Flies will fixate on nonflickering and flickering vertical dark bars. H, This fixation behavior results in mean vector lengths that are not significantly different between flickering or nonflickering bar at 7 Hz. Kruskal–Wallis; χ2 = 2.92, p value = 0.0872. Another set of flies were tested on a round arena (black boxplots; Reiser and Dickinson, 2008). Box: flies were still able to fixate on the 7 Hz bar following electrode insertion. The letters a–c indicate statistically different groups. N for each dataset is indicated in brackets on the graph.
- Figure 2.
Multichannel recording in the fly brain reveal bar location during behavior. A, A close-up representation of a tethered fly walking on an air-supported ball with an electrode array inserted. B, A 3D model representation of the electrode array inserted into the head of the fly. C, An example trace of the bar position controlled by a fly walking on a ball with an electrode array inserted. D, The flickering bar can be detected in the brain as oscillations in the LFP across channels which can be mapped to specific neural structures (colored dots in B). E, These oscillations can be detected by measuring the MWCA over time, as represented in these spectrograms of channels 1 and 12 during the same time period represented in C, D. F, The MWCA value for the flicker frequency at 7 Hz (called the SSVEP) can be represented for each channel over time (left and right eye channels are shown, in blue and red). In this case, as the bar moves from left to right, the SSVEP amplitude in the left eye decreases and the SSVEP in the right eye increases. G–I, Mapping the SSVEP amplitude onto the bar position results in an average distribution of flicker responses around the fly. Individual channels in the same fly brain have visual flicker responses which can be mapped to the bar position. Each line represents the mean SSVEP per position and the arrows are the mean direction of the polar plot data. Peripheral brain channels, as in the lamina or retina, have more lateral and larger receptive fields (G) than central brain channels (H). I, The SSVEP mean vectors of the left and right eyes (red and blue arrows) were directed left and right, respectively, whereas the central brain channels (green arrow) pointed in front of the fly.
- Figure 3.
Closed-loop and open-loop replay reveal identical receptive fields for the 7 Hz visual flicker. A, Recordings sites from multiple flies (N = 24) were mapped onto a standard brain and combined into a common dataset. B, The regions of the brain could be subdivided into structures, which are color-coded in A and in the legend in B, mb, Mushroom body; cc, central complex; prot, protocerebrum. C–E, Flies were able to control a visual stimulus in closed loop, and the same visual sequence was replayed back to the fly during open-loop replay in a series of trials (D). The SSVEP receptive fields were summarized as mean SSVEP vectors per brain region, for closed-loop (C) and replay (E) experiments. These mean SSVEP vectors have mean directions and amplitude per brain region, color-coded as in A and B. F, G, The mean SSVEP vector amplitudes were significantly higher than the amplitudes of LFP oscillations at other frequencies outside the range of the frequency tag (7 Hz) during both closed-loop and open-loop experiments. SSVEP mean vector directions in the central brain were significantly different from the left optic lobe (P test statistic = 7.05; p = 0.0079) but not the right optic lobe (P test statistic = 1.76; p = 0.1847). H, The SSVEP mean vector amplitudes were not significantly different (n.s.) between open-loop replay and closed-loop experiments.
- Figure 4.
Mean SSVEP varies with bar position and walking speed. A–C, The distribution of walking speeds could be divided broadly into no walking and walking (A, B) classifications (movement along the x/y axes; Fig. 1A) and turning and no turning (D, E) classifications (changes in theta; Fig. 1A). Median walking speed was not significantly different (1.372 cm/s vs 1.304 cm/s for closed loop and replay, respectively; p = 0.468, Wilcoxon rank sum test). Median turning speed was also not significantly different (p = 0.26, Wilcoxon rank sum test). C, F, SSVEP was significantly higher when flies were walking compared with not walking (C), whereas SSVEP was not dependent on turning behavior (F). G, H, SSVEP was significantly higher in optic lobes versus central brain (G) and when the bar was in the front versus the back halves of the arena (H). I, No significant differences were seen in the SSVEP across the brain for closed-loop versus replay experiments. N = 24 flies.
- Figure 5.
Mean SSVEP shows different interaction effects with behavior in closed-loop and replay experiments. Significant interaction effects were detected between brain region and turning behavior (see Table 1 for associated statistics). Different interaction effects were detected depending on if the experiment was closed loop (A, C) or replay (B, D), and whether the fly was walking (forwards, sidestep movements) or not. These interaction effects were not evident when the flies were not walking (C, D). SSVEP interaction effects were examined for periods when the bar was in front of the fly, to control for the effects of bar position. N = 24 flies.
- Figure 6.
Coherence increases during closed-loop experiments. A, Coherence measures the coupling of brain activity between different brain regions, (Bokil et al., 2010; Paulk et al., 2013), here for the 7 Hz frequency tag in the fly brain. Representative examples of three LFP recordings are shown here, two are close together in one optic lobe (green, blue) and the other recording is more distal, in the opposite optic lobe (magenta). B, Coherence is higher between local brain regions and lower between distal brain regions. Coherence fluctuates through time, with high coherence indicating strong coupling between signals, whereas low coherence indicating weak coupling. C–H, Coherence was measured between electrode sites of different brain regions (indicated by red dots in fly brain schemas) when flies were active (walking and turning) during closed-loop and replay behavior. Coherence was not significantly different between opposite hemispheres of the optic lobes or central brain (C, D), or between neighboring channels in the optic lobes (E). However, coherence was significantly higher between neighboring channels within hemispheres in the central brain during closed-loop behavior (F). Coherence was significantly higher between central brain and optic lobes in the same hemisphere during closed-loop behavior (G), but not between opposite hemispheres (H). Within-hemisphere calculations were for both left and right sides of the brain. Statistical comparisons in C–F were performed using the Wilcoxon rank sum test, N = 24 flies.
Tables
- Table 1.
Multiway ANOVA testing how SSVEP varies with brain region, where the electrodes were located, whether the experiment was closed-loop or open-loop replay, the bar position (front vs behind the fly), whether or not the fly was walking, and whether or not the fly was turning
Multiway ANOVA Source p Walking/not walking Walking 0.0044 Turning vs turning left or right Turning 0.4672 Bar position (front vs back) Bar location 0.000000 Closed loop vs replay Loop 0.9538 Brain region Region 0.000000 Interaction effects Walking × turning 0.2954 Walking × bar location 0.000000 Walking × loop 0.035 Walking × region 0.4614 Turning × bar location 0.3855 Turning × loop 0.0189 Turning × region 0.048 Bar location × loop 0.2839 Bar location × region 0.000000 Loop × region 0.6365 Bold text indicates significant effects.
- Table 2.
Multiway ANOVA testing how coherence varies with brain regions, whether the experiment was closed-loop or open-loop replay, the bar position (front vs behind the fly), whether or not the fly was walking, and whether or not the fly was turning
Multiway ANOVA Source F statistic p Walking/not walking Walking 0.229706 0.631742 Turning vs not turning Turning 7.100217 0.007708 Bar position (front vs back) Bar location 3.564825 0.059017 Closed loop vs replay loop 2.423198 0.119552 Coherence estimation across brain regions Brain region comparison 437.1532 <0.00000001 Interaction effects Walking × turning 1.052304 0.304978 Walking × bar location 17.05647 0.0000363 Walking × loop 22.53801 0.000002 Walking × region 54.86743 <0.00000001 Turning × bar location 0.690296 0.406064 Turning × loop 5.459179 0.019466 Turning × region 0.384883 0.859501 Bar location × loop 1.135769 0.286549 Bar location × region 96.13367 <0.00000001 Loop × region 16.58375 <0.00000001 Bold text indicates significant effects.
