Closed-loop response properties of a visual interneuron involved in fly optomotor control

Due to methodological limitations neural function is mostly studied under open-loop conditions. Normally, however, nervous systems operate in closed-loop where sensory input is processed to generate behavioral outputs, which again change the sensory input. Here, we investigate the closed-loop responses of an identified visual interneuron, the blowfly H1-cell, that is part of a neural circuit involved in optomotor flight and gaze control. Those behaviors may be triggered by attitude changes during flight in turbulent air. The fly analyses the resulting retinal image shifts and performs compensatory body and head rotations to regain its default attitude. We developed a fly robot interface to study H1-cell responses in a 1 degree-of-freedom image stabilization task. Image shifts, induced by externally forced rotations, modulate the cell’s spike rate that controls counter rotations of a mobile robot to minimize relative motion between the robot and its visual surroundings. A feedback controller closed the loop between neural activity and the rotation of the robot. Under these conditions we found the following H1-cell response properties: (i) the peak spike rate decreases when the mean image velocity is increased, (ii) the relationship between spike rate and image velocity depends on the standard deviation of the image velocities suggesting adaptive scaling of the cell’s signaling range, and (iii) the cell’s gain decreases linearly with increasing image accelerations. Our results reveal a remarkable qualitative similarity between the response dynamics of the H1-cell under closed-loop conditions with those obtained in previous open-loop experiments. Finally, we show that the adaptive scaling of the H1-cell’s responses, while maximizing information on image velocity, decreases the cell’s sensitivity to image accelerations. Understanding such trade-offs in biological vision systems may advance the design of smart vision sensors for autonomous robots.