Visually guided orientation behaviour in complex environments

Insects, such as flies and bees, are renowned for their aerobatic flight manoeuvres which are taken to the extremes during chasing flights in the context of reproduction. Nonetheless, flies share more widespread yet essential flight behaviours, including obstacle avoidance, landing behaviour and course control, with other insect species. Insects are often also engaged in a completely different mode of locomotion, i.e. walking. In both locomotion modes, flies rely on spatial information. However, the constraints imposed on gathering and processing visual information are likely to differ considerably.

Our behavioural analysis is designed to challenge the visual system in order to quantify its capacity and significance in mediating orientation in various behavioural contexts and to unravel the underlying mechanisms. This approach encompasses the presentation of visual stimuli containing conflicting cues to tethered animals in virtual reality, e.g. when walking on a treadmill, in order to judge their respective significance during normal behaviour, and especially in situations where decisions need to be made. Behavioural analysis of largely unrestrained animals by using cameras in a variety of visual worlds bridges the gap between systems analysis in tethered animals and behaviour in complex cluttered environments. It also forms the basis for reconstructing and characterising the specific behaviourally generated spatiotemporal retinal input sequences with their characteristic dynamics.

Behavioural experiments are combined with electrophysiological analysis of neural activity in visual and higher-order areas of the insect brain to complement the analysis of the computational mechanisms involved in visually guided orientation behaviour. For testing whether experimentally established hypotheses about computational mechanisms are sufficient to explain behavioural performance we create computational models formalising the hypotheses of the animal’s visual motion pathway and of the interface linking the sensory system with the motor output. Extensive testing of the models with a large range of stimuli allows us, in addition, to identify new critical experiments and to study questions inaccessible to experimental approaches. In this way, the functional relevance of neuronal modules can be carefully analysed.

 

Research issues that are currently being investigated comprise:

  • Head-body coordination and the consequences for the visual input as well as for object-induced walking behaviour of flies in virtual environment of variable complexity.
  • Visual attention and decision making in visually guided behaviour in virtual reality and the underlying neural mechanisms.
  • Collision avoidance based on optic flow information in simple and complex cluttered environments.