Data from neurosciences can be exploited by computer sciences to explore new models of information processing. We propose to study here hemispheric specialization and its possible underlying neuronal mechanisms. More particularly, these investigations elaborate a simple dichotomy of information processing, from parallel and spatial to sequential and temporal processing, of which we propose a neuronal implementation, based on neurobiological considerations. This is illustrated by an application, also corresponding to a task performed by human subjects in functional imaging. The usefulness of such artificial perception mechanisms for biologically-inspired robotics is also discussed.