The Dominant Role of Visual Motion Cues in Bumblebee Flight Control Revealed Through Virtual Reality

Abstract

Flying bees make extensive use of optic flow: the apparent motion in the visual scene generated by their own movement. Much of what is known about bees' visually-guided flight comes from experiments employing real physical objects, which constrains the types of cues that can be presented. Here we implement a virtual reality system allowing us to create the visual illusion of objects in 3D space. We trained bumblebees, Bombus ignitus, to feed from a static target displayed on the floor of a flight arena, and then observed their responses to various interposing virtual objects. When a virtual floor was presented above the physical floor, bees were reluctant to descend through it, indicating that they perceived the virtual floor as a real surface. To reach a target at ground level, they flew through a hole in a virtual surface above the ground, and around an elevated virtual platform, despite receiving no reward for avoiding the virtual obstacles. These behaviors persisted even when the target was made (unrealistically) visible through the obstructing object. Finally, we challenged the bees with physically impossible ambiguous stimuli, which give conflicting motion and occlusion cues. In such cases, they behaved in accordance with the motion information, seemingly ignoring occlusion.

Keywords: bee; closed-loop; flight; free flight; motion; optic flow; virtual reality; vision.

Figure 1

Schematic diagram of experimental setup. Camera positions are not to scale. The origin of the co-ordinate system is defined as the center of the monitor surface.

Figure 2

Virtual reality conditions. Left: schematic diagram showing surfaces at virtual altitudes of 0 mm (i.e., static, pink) and 60 mm (closed-loop, purple). Right: representation of bee's ventral visual field; the center of the circle is the nadir. Pink and purple coloring is to denote z-depth only; actual chequerboard patterns are all pink/white. Chequerboard texture is not to scale. Bottom: Table summarizing the parameters that differ between conditions. BG stands for background; FG for foreground. Shading indicates which group (control, congruent VR, incongruent VR) each condition belongs to. Conditions incHtv₅ and incPtv₅ are not shown, but are simply incHtv and incPtv respectively with the spatial frequency of the background (purple) texture doubled.

Figure 3

Physical vs. virtual floor. (A) Sample trajectories (1 min duration) from the same individual in the C0 (static, blue) and C60 (virtual reality closed-loop, red). Dashed line indicates the level of the virtual floor, gray shaded band is the range of altitudes shown in (B). (B) Histogram of time spent at altitudes between 40 and 300 mm, averaged across all individuals. Shaded regions are ±1 SE. (C) Event-triggered mean altitude profiles; t = 0 is the first instance of the bee descending below z = 60 mm. Dashed red line is C60 data shifted downwards by 60 mm, i.e., expressed relative to the virtual as opposed to physical floor. (D) Mean altitude at which first leg extension is observed. Inset: stills from overhead camera showing a flying bee with legs retracted (left) and extended (right).

Figure 4

Virtual object avoidance. (A) (x,y) positions at which individuals initially descended through the plane of the virtual platform (i.e., z = 60 mm); the extent of the virtual platform is indicated by gray shading. Thus, descents occurring within the shaded area correspond to “collisions” with the virtual obstacle. Coloured “+” symbols are for the realistic occlusion condition (P), and “×” symbols for the condition where the target is always visible (Ptv). Black diamonds are the control condition, where no obstacle exists (C0). (B) As A, but for the hole obstacle. (C) Proportions of descents resulting in collision vs. avoidance. For C0 no virtual obstacle is present, but we assess whether each descent would have collided with the obstacle in question, thereby providing an estimate of the avoidance rate due to chance alone. “^***” denotes p < 0.001 (Fisher's exact test); numbers of individuals are as follows: n_P = 26, n_Ptv = 26, n_H = 25, n_Htv = 26, n_C0 = 28. (D) Altitude of initial leg extension as a function of distance from the edge of the virtual obstacle. Negative distances correspond to collisions (orange shading). Dashed line is the virtual obstacle plane.

Figure 5

Incongruent virtual objects. (A) (x,y) positions at which individuals initially descended through the plane of the virtual “background” (i.e., z = 60 mm); the extent of the platform (z = 0 mm, but occluding the background) is indicated by gray shading. Coloured “+” symbols are for the case where the background chequerboard is composed of squares of the usual size (10 mm), “×” symbols for the case where they are 5 mm, and black diamonds are the control condition where no obstacle exists (C60). (B) As A, but for the incongruent hole obstacle. (C) Comparison with the congruent conditions (Figure 4) showing how frequently animals descended through the z = 60 mm plane inside vs. outside of the platform/hole. “^***” denotes p < 0.001 (Fisher's exact test); numbers of individuals are as follows: n_Ptv = 26, n_incPtv = 24, n_Htv = 26, n_incHtv = 25.