Item

Abstract

In flying insects the stabilization of gaze aids visual processing by reducing motion blur, for instance, and facilitating self-motion estimation. The gaze control system in blowflies makes use of information from a number of well-characterised sensory modalities, including the compound eyes, ocelli, halteres and campaniform sensilla (Hengstenberg 1991). Individually, these sensory pathways are sensitive to different dynamic ranges, incur different processing delays, and suffer from different levels of sensor and processing noise. Understanding their dynamics and the interplay between them would allow us to study the performance limits of the fly gaze stabilization system, and may advance our understanding of multisensory control architectures in biological systems in general. We applied a linear systems analysis to roll gaze stabilization in Calliphora based on the frequency responses of the two visual pathways, the compound eyes and the ocelli. In previous studies, stimulation of the ocelli showed little effect on compensatory head movements in the blowfly (Schuppe and Hengstenberg 1993). Recent work at the neuronal level suggests that descending neurons with coupling to neck motor neurons receive input from the ocelli (Haag et al. 2007), and that the ocelli also modulate the activity in optic flow processing interneurons receiving input from the compound eyes (Parsons et al. 2010). Using high speed videography, we measured head roll in response to visual stimulation of both the compound eyes and ocelli, and of the compound eyes alone. The behavioural data were obtained by oscillating a false horizon around the longitudinal body axis at a range of four different input frequencies (1-10Hz) and three amplitudes (10, 20 and 30 degrees). Assuming a linear system, we derived the individual visual pathway frequency responses and obtained the gain and phase plot for the open-loop behaviours corresponding to the two different sensory conditions. We found that the combination of compound eye and ocellar input reduces the phase in the processing of visual motion information compared to that of the compound eye-mediated response alone (Figure 1B), but does not significantly affect the response gain or bandwidth (Figure 1A). Over the input frequency range investigated, the ocellar contributions lead to an average reduction in the pathway delay of 5.4 ms (SE = 2.5 ms), which is likely to benefit stable and robust feedback control. Our result is consistent with previous studies in locusts which suggest that the ocelli provide a fast response preceding the slower, more precise information from the compound eyes (Taylor 1981). With increasing stimulus amplitudes we observed a reduction in overall response gain, suggesting non-linear saturation effects as the system deviates from its equilibrium point. The influence of the ocelli on the phase of the output, but not on the gain, could also be attributed to non-linearities in the system. Preliminary experiments suggest that other dipteran flies are amenable to a similar behavioural analysis. Future comparative work on the integration of ocellar information and other sensory inputs to gaze control systems across species will help us to derive general principles of multimodal biological control design.