FLY VR

Morgane, Olga & Nick

Our aim was to develop a learning paradigm for tethered flies. Thus, we created a virtual environment using Unity consisting in a room with two green checkerboard walls and two red ones. The fly behaviour was video-recorded and analysed in real-time using Bonsai, allowing us to extract the fly heading. This information was passed on to Unity, thus creating a closed-loop system in which the fly could navigate within the virtual environment. “Landing” on the red walls was rewarded by the presentation of a platform soaked with a sugar solution, while landing on the green walls triggered the presentation of a non-rewarded platform.

While the flies exhibited a large variety of trajectories in the virtual environment, when looking at their landing preferences we found that they liked the green (non-rewarded) walls more than the red (rewarded) ones. This might be because of phototactic behaviour, since the green walls appeared brighter than the red ones. This issue arose because we were technically limited to pure RGB colours by the projector. Changing the projector or switching to different patterns would have been viable solutions.