1. Single binocularly driven complex neurons in cat striate cortex were recorded extracellularly under nitrous oxide-oxygen-halothane anesthesia and muscle relaxant. Orientational/directional tuning was initially derived for each eye in turn, with sine wave gratings of optimal spatial frequency and velocity, while the other eye viewed a uniform field. 2. For the dominant eye, previously concealed suppression was revealed against elevated levels of firing induced with a conditioning grating, drifting continuously in the preferred direction, simultaneously presented to the nondominant eye. During steady-state binocular conditioning, orientational/directional tuning was reestablished for the dominant eye. In a subset of cells, tuning curves during conditioning were also derived for the reverse configuration, i.e., nondominant eye tuning, dominant eye conditioning: results were qualitatively identical to those for conditioning through the nondominant eye. 3. Neurons were initially segregated into five groups, according to the observed suppression profiles induced at nonoptimal orientations/directions during conditioning: Type 1, suppression centered on orthogonal directions; Type 2, suppression around null directions; Type 3, null suppression combined with orthogonal suppression; Type 4, lateral suppression, maximal for directions immediately flanking those inducing excitation; and Type 5, the residue of cells, totally lacking suppression or showing complex or variable suppression. 4. Sharpness of (excitatory) tuning was correlated with directionality and with class of suppression revealed during binocular conditioning. Direction-biased neurons were more sharply orientation tuned than direction-selective neurons; similarly, neurons exhibiting lateral or orthogonal suppression during conditioning were more sharply tuned than neurons with null suppression. 5. Application of suboptimal directions of conditioning weakened the induced suppression but altered none of its main characteristics. 6. The relationship between excitation, suppression, and spatial frequency was investigated by comparing tuning curves for the dominant eye at several spatial frequencies, without and during conditioning. End-stopped neurons preferred lower spatial frequencies and higher velocities of motion than non-end-stopped neurons. Confirming previous reports, suppression in some neurons was still present for spatial frequencies above the cutoff frequency for excitation, demonstrating the tendency for suppression to be more broadly spatial frequency tuned than excitation. 7. Scatterplots of strength of suppression, in directions orthogonal and opposite maximal excitation, partially segregated neurons of Types 1-3. Clearer segregation of Types 1-4 was obtained by curve-fitting to profiles of suppression, and correlating half-width of tuning for suppression with the angle between the directions of optimal suppression and optimal excitation in each neuron. 8. Two interpretations are advanced-the first, based on three discrete classes of inhibition, orthogonal, null and lateral; the second, based on only two classes, orthogonal and null/lateral--in which null and lateral suppression are manifestations of the same inhibitory mechanism operating, respectively, on broadly tuned direction-selective or on sharply tuned direction-biased neurons. Orthogonal suppression may be untuned for direction, whereas lateral and null suppression are broadly direction tuned. Within each class, suppression is more broadly spatial frequency tuned than excitation. 9. It is concluded that orientational/directional selectivity of complex cells at different spatial frequencies is determined by the balance between tuned excitation and varying combinations of relatively broadly distributed or untuned inhibition.
Neuronal basis of perceptual learning in striate cortex
