Adaptive mechanisms underlying visual motion processing
The spatiotemporal retinal motion patterns generated on the eyes during behaviour provide the animal, especially during flight, with information about self-motion, on the one hand, but also with information about the environment, on the other hand. In particular, during translatory flight spatial information about the environment can be gained from the optic flow without sophisticated computations. In the insect brain, visual motion is rapidly processed in an amazingly efficient way. How are the mechanisms in the visual system tailored to achieve the performance observed? Answering this question requires to understand, on the one hand, the constraints imposed by the complexities of the environments in which the animals normally operate. Environments may be complex with regard to their spatial layout and textural properties and may change on a variety of timescales. On the other hand, one needs to know the specific dynamics that is imposed on the retinal image flow by the animals’ behavioural dynamics as a consequence of the closed action-perception cycle.
Extracting behaviourally relevant information by the visual motion pathway from complex optic flow patterns relies on adaptive processes at various levels which, depending on the respective behavioural task, need to be matched to the respective spatiotemporal characteristics of retinal image flow. Motion computation and the involved adaptive mechanisms are analysed by electrophysiological analysis of fly motion sensitive neurons combined with pharmacological manipulations of the overall activity state of the nervous system as well as by computational modelling and theoretical analysis. One relevant feature of neural representation of visual motion information in the insect motion pathway is that information may be encoded by graded voltage changes and/or action potentials as output signals of neurons. Our analysis of motion information processing and the involved adaptive processes is distinguished by the non-exclusive employment of experimenter-designed visual stimuli to characterise neurons or the model performance, but also the highly dynamic retinal image sequences experienced by behaving animals during unrestrained locomotion.
Research issues that are currently being investigated comprise:
- The visual information that is encoded in the two components of neural signals, i.e. graded potential changes and action potentials, and how they complement each other.
- Adaptive processes along the visual motion pathway and how these are tailored to facilitate processing of natural spatiotemporal image sequences and to represent behaviourally relevant information about the environment.
- Representation of self-motion of the animal on the basis of the optic flow generated during locomotion in three-dimensional environments.