Orientation in Visual Cortex: A Simple Mechanism Emerges

A. Das contribution is estimated to lie between 5% and 28% intracortical excitatory and inhibitory connections provide the rest. Other models were therefore proposed for orientation selectivity, with intracortical connections playing a more significant role. These models started A recent report by David Ferster and colleagues (1996) with the assumption that the RFs of LGN neurons provid-provides striking new evidence to help settle an issue ing input to any cortical cell form a cluster with no partic-that has been the focus of continuing debate for over ular elongation or alignment. Thus, the cortical cell 30 years, namely, how are neurons in primary visual would get roughly equal excitation from visual stimuli cortex (V1) so sharply selective for the orientation of at all orientations with at best, a slight bias for one visual stimuli when their afferent inputs lack such selec-orientation. This bias would then be sharpened by intra-tivity (Figure 1)? The major question is whether the pat-cortical interactions. These models can be divided into tern of afferent connections alone can account for the two groups, depending on the nature of the intracortical response of V1 neurons to orientation, or if it is neces-interactions proposed: one group of models involve sary to invoke some intracortical feedback process (Fig-mainly inhibitory mechanisms and the other mainly ex-ure 2). In their experiments, Ferster and colleagues show citatory. that V1 neurons maintain their sharp orientation tuning Inhibitory Models even when intracortical circuits are essentially silenced. In this family of models, the primary agent determining These results validate a major tenet of the original and the orientation axis and tuning of a cortical cell is strong most simple model for orientation selectivity and sug-intracortical inhibition. Each layer 4 simple cell gets gest that similarly simple processes may underlie other strong LGN excitation at all stimulus orientations. Inhibi-cortical response features. tion from other cortical cells at the orthogonal orienta-The original model for orientation selectivity relied tion then selectively damps the responses of the target purely on the geometry of afferent inputs (Figure 2A). neuron at inappropriate orientations (Figure 2B). Histori-This model was proposed by Hubel and Wiesel (1962), cally, such models involving intracortical inhibition were when they first described orientation tuning in V1. Over the first to be offered as alternatives to the Hubel–Wiesel the years, a substantial body of evidence has been gath-model for orientation tuning (Creutzfeldt et al., 1974; ered in favor of this model …