Influence of a moving textured background on direction selectivity of cat striate neurons

The influence of a moving textured background on direction selectivity for a moving bar was tested in 118 striate neurons and in 19 dorsal lateral geniculate neurons of anesthetized and paralyzed cats. In the standard conditions the background was a two-dimensional noise pattern, the bar moved at optimal speed, and its contrast was adjusted to the level producing 50% of the maximum response. These experiments revealed a new typology of cortical cells based on relative direction selectivity. Six different relative-direction-selectivity types are described. Two types of cells were found to have opposite kinds of relative direction selectivity: antiphase direction-selective cells (5% of the cortical sample) preferred the direction of the bar opposite to the direction of background motion, and absolutely direction-selective cells (20% of the cortical sample) kept their direction selectivity for bar motion independently of the background motion. Three types of cortical cells were direction selective for bar motion only in restricted background motion conditions: conditionally direction-selective cells (20% of cortical sample) only expressed their direction selectivity when the bar and the background moved in antiphase, differencing direction-selective cells (5% of the cortical sample) only expressed their direction selectivity when the bar and the background differed in speed, and limited direction-selective cells (20% of the cortical sample) only expressed their direction selectivity for near zero background speeds. The sixth type, relative nondirection-selective cells (30% of the cortical sample and all of the geniculate cells) were direction selective for none of the background motion conditions. These different relative-direction-selectivity types differed in RF organization, in ocular dominance, velocity sensitivity, in laminar distribution, and in distribution in the visual field. The relative-direction-selectivity types were invariant for changes in the contrast and bar speed. The construction of these relative-direction-selectivity types from the geniculate input requires some inhibitory, but mainly facilitatory, intracortical interactions. These experimental findings suggest that area 17 in the cat has the neuronal machinery to extract depth from motion (limited direction-selective cells) and to segregate visual scenes by motion cues (antiphase, conditionally and differencing direction-selective cells).