PUB - Publikationen an der Universität Bielefeld

In a rich and complex world, it is a crucial task for animals, especially for fast moving ones, to detect objects in front of their background. Fast moving animals strongly rely on optic flow, i.e., the visual motion induced on their eyes during locomotion, to guide their behavior, such as to avoid obstacles, to estimate depth or distance to environmental objects, or to prepare for landing during flight. This thesis investigates with electrophysiological recording techniques the performance of different motion-sensitive neurons in representing objects and the spatial layout of the environment as well as how this representation is affected by adaptive processes. The analysis is done in the visual motion pathway of the blowfly, Calliphora vicina. Only the translational component of the optic flow induced by an animal's self-motion contains spatial information, since the retinal images of close objects move faster than distant ones only during translatory movements, whereas during rotation, the retinal velocities are independent from the distance between objects and observers. Like several other groups of animals, blowflies pursue an active saccadic flight and gaze strategy to separate by their behavior the rotational and translational component of optic flow and, thus, to facilitate the processing of spatial information. During largely translational motion between saccadic turns, the gaze is stabilized and the spatial layout of the environments can potentially be encoded by the visual system flies. How this may be accomplished is investigated for three types of motion-sensitive neurons, the horizontal system (HS), centrifugal horizontal (CH) and figure-detection (FD) cells in the third neuropil of the fly's visual system. Among the different types of neurons, HSE/HSS (HS equatorial, southern), VCH (ventral CH) and FD1 (one subtype of FD) cells constitute major elements of a neural circuit which is assumed to be involved in object detection and distance estimation. CH cells receive retinotopic visual input from large parts of the ipsilateral visual field indirectly via dendro-dendritic electrical synapses from the large-field HS cells and transfer a GABAergic inhibitory signal to the FD1 cell and, thus, mediate its selectivity to small moving objects. In this thesis, neurons are confronted with semi-naturalistic optic flow as is seen by free-flying animals as well as targeted modifications of it. The results show that FD1 and HSE cells both respond strongly to nearby objects and are also affected by the distance to the background. The general performance of the FD1 cell not only to detect nearby objects, but also to represent spatial information is better than that of HSE. The detectability of objects under given environmental conditions by motion sensitive neurons is not fixed but may improve as a consequence of adaptive processes. Therefore, this thesis investigates the functional significance of motion adaptation for providing spatial information under the complex stimulus conditions encountered in a three-dimensional world. This is done in electrophysiological experiments on HS cells of the blowfly visual system. With manipulations of semi-naturalistic optic flow, motion adaptation is shown to facilitate the detection of objects in a three-dimensional environment although the overall neuronal response amplitude decreases during prolonged motion stimulation. Furthermore, it was tested how motion adaptation is affected by different dynamic properties of the optic flow. In particular, this thesis assessed to what extent neuronal responses to an object located close to the flight trajectory depend on the dynamical characteristics of the optic flow before the object appears in the receptive field of the HS-cell. Object-induced responses were stronger in the adapted compared than the non-adapted state. This effect holds for all types of adapting optic flow that have been used in the experiments. Adaptation with optic flow that lacked typical dynamical features resulting from natural flight dynamics, and even pure rotation at a constant angular velocity, was effective to enhance object-induced responses. The enhancement was slightly direction-selective, since preferred direction rotation was a more efficient adaptor than null direction rotation. These results provide evidence that the adaptive mechanisms are most likely distributed over different processing stages along the visual motion pathway and that the natural dynamics of optic flow is not a basic requirement to adapt neurons in a specific, presumably functionally beneficial way.