Estimation of self-motion by optic flow processing in single visual interneurons

Abstract

HUMANS, animals and some mobile robots use visual motion cues for object detection and navigation in structured surroundings^1–4. Motion is commonly sensed by large arrays of small field movement detectors, each preferring motion in a particular direction^5,6. Self-motion generates distinct 'optic flow fields' in the eyes that depend on the type and direction of the momentary locomotion (rotation, translation) ⁷. To investigate how the optic flow is processed at the neuronal level, we recorded intracellularly from identified interneurons in the third visual neuropile of the blowfly⁸. The distribution of local motion tuning over their huge receptive fields was mapped in detail. The global structure of the resulting 'motion response fields' is remarkably similar to optic flow fields. Thus, the organization of the receptive fields of the so-called VS neurons^9,10 strongly suggests that each of these neurons specifically extracts the rotatory component of the optic flow around a particular horizontal axis. Other neurons are probably adapted to extract translatory flow components. This study shows how complex visual discrimination can be achieved by task-oriented preprocessing in single neurons.