Research group: Visual Perception and Attention

Vision is the main sensory foundation of cognitive abilities in intelligent agents which actively explore the environment and exhibit task-adequate and goal-directed navigation. In the context of Cognitive Science, and specifically computational neuroscience, human vision is now seen as an intricate interplay between pre-attentive, that is primarily stimuli-driven, and attentive, that is mainly concept-guided, computational neural processes which act upon the appropriate representations. Such representations typically encode either visual qualities of stimuli or items from conceptual memory, both being related to the external world. In general, process-oriented representations, e.g. cortical maps with retinal topography, implicitly encode those aspects of the visual world that are relevant for the organization of behaviour emerging at different levels of complexity.

In vision science, it is now widely accepted that modelling the structure and function of visual systems requires a multi- and transdisciplinary approach. From the perspective of Cognitive Science, such an approach, embedded within the information processing framework, renders possible the fusion of theory and observation with modelling and experimental computer simulation. In particular, the collaboration between psychophysics, neurosciences, and computer science promises deeper insights into the mechanisms of preattentive and attentive vision. Our research draws upon both subsymbolic and symbolic modelling approaches and integrates theories and methods from the neighbouring disciplines.

The topics to be tackled by our research group are centred on the interplay between preattentive and attentive vision. One example is the experimental investigation and modelling of visual search. For an understanding of the mechanisms underlying visual search, the study of perceptual organization and pop-out phenomena as well as attentional control play a key role. Moreover, the functional connections between attentional control, gaze control, and intentional control of cognitive processes in general are of importance, as well as issues of perceptual learning, e.g. for understanding the effects of extended training. As another example, the computational implications of principles of the representation of visual qualities in maps along the visual pathway are analyzed, particularly with regard to the spontaneous organization of spatio-temporal stimuli and to binocular fusion. Modelling here aims at improved models and computational architectures for preattentive visual perception which are consistent both with neurobiological knowledge and with psychophysical findings. The topic of representation of stimuli in space-time is also closely related to the problems studied by the research group on "Representation of Space and Time in Cognitive Processes".

(Last change: December 2, 1997)
The WWW-GRK Webmasters