Neural mechanisms of visual feature binding / 生理学报

ABSTRACT

Integrating different visual features into a coherent object is a central challenge for the visual system, which is referred as the binding problem. Firstly, this review introduces the conception of the binding problem and the theoretical and empirical controversies regarding whether and how the binding processes are implemented in visual system. Although many neurons throughout the visual hierarchy are known to code multiple features, feature binding is recruited by visual system. Feature misbinding (or illusory conjunction) is probably the most striking evidence for the existence of the binding mechanism. Next, this review summarizes some critical issues in feature binding literature, including early binding theories, late binding theories, neural synchrony theory, the feature integration theory and re-entry processing theory. Feature binding is not a fully automatic or bottom-up processing. Reentrant connection from higher visual areas to early visual cortex (top-down processes) plays a critical role in feature binding, especially in active feature binding (i.e. feature misbinding). In addition, with electrophysiology, electroencephalography (EEG), magnetoencephalography (MEG) and transcranial electric stimulation (tEs) approaches, recent studies explored both correlational and causal relations between brain oscillations and feature binding, suggesting that brain oscillations are of great importance for feature binding. Finally, this review discusses some potential problems and open questions associated with visual feature binding mechanisms which need to be addressed in future studies.