Neural Mechanisms of Visual Feature Binding Investigated with Microelectrodes and Models

Abstract

A single object generally activates neurones in many visual cortical areas corresponding to a distributed representation of its features. Itis still under debate how the distributed representation of an objectis bound intoa coherent whole and how unrelated features are separated. Synchronization of neural signals has been proposed to code spatial feature binding, supported by the discovery of synchronized assemblies in visual cortex. Synchronizations are either fast oscillatory (30–90 Hz) cortically generated events, or non-rhythmical stimulus-locked responses, depending on the visual stimulation. The cortical range over which synchronizations occur, transformed to visual space, is generally several times larger than the classical receptive fields (CRF) of neurones in lower visual cortex areas. However, the cortical regions synchronized by fastoscillations do notspan the representational range of larger objects but only parts of it. To relate such restricted segments to perceptual processes the concept of the association field (AF) of local neural assemblies was introduced in accordance with CRFs of single neurones. Here an AF is composed of the aggregate of CRFs of an assembly engaged in a common synchronized state. Itis argued thatspatial continuity of an object is coded by a continuum of overlapping AFs, that is, by overlapping regions of phase coupled neurones. Hence, object continuity would be represented by phase continuity. Besides feature binding, feature separation is necessary for scene segmentation. Separation may be coded by functional decoupling causing uncorrelated activities or by mutual inhibition leading to alternating activations in assemblies of separate representations. A third temporal coding aspect is temporal segmentation by the shortactivation-inhibition cycles of fast oscillations orshorttransientstimulus-locked responses, which may preventperceptual “smearing” by interrupting the flow of visual information into precisely defined frames. The presentpaperalso aims atrelating signals of stimulus dependentsynchronization and desynchronization with basic neural mechanisms and circuits. Finally, the synchronization hypothesis is critically discussed with respect to contradictory psychophysical work and supportive new recording results, including evidence for perception-related synchronizations.