RECEPTIVE-FIELD EXPANSION IN ADULT VISUAL-CORTEX IS LINKED TO DYNAMIC CHANGES IN STRENGTH OF CORTICAL CONNECTIONS

1. Receptive field (RF) sizes of neurons in adult primary visual cortex are dynamic, expanding and contracting in response to alternate stimulation outside and within the RF over periods ranging from seconds to minutes. The substrate for this dynamic expansion was shown to lie in cortex, as opposed to subcortical parts of the visual pathway. The present study was designed to examine changes in cortical connection strengths that could underlie this observed plasticity by measuring the changes in cross-correlation histograms between pairs of primary visual cortex neurons that are induced to dynamically change their RF sizes. 2. Visually driven neural activity was recorded from single units in the superficial layers of primary visual cortex in adult cats, with two independent electrodes separated by 0.1-5 mm at their tips, and cross-correlated on-line. The neurons were then conditioned by stimulation with an ''artificial scotoma,'' a field of flashing random dots filling the region of visual space around a blank rectangle enclosing the RFs of the recorded neurons. The neuronal RFs were tested for expansion and their visually driven output again cross-correlated. After this, the neurons were stimulated vigorously through their RF centers to induce the field to collapse, and the visually driven output from the collapsed RFs was again cross-correlated. Cross-correlograms obtained before and after conditioning, and after RF collapse, were normalized by their flanks to control for changes in peak size due solely to fluctuations in spike rate. 3. A total of 37 pairs of neurons that showed distinct cross-correlogram peaks, and whose RF borders were clearly discernible both before and after conditioning, were used in the final analysis. Of these neuron pairs, conditioning led to a clear expansion of RF boundaries in 28 pairs, whereas in 9 pairs the RFs did not expand. RFs that did expand showed no significant shifts in their orientation preference, orientation selectivity, or ocularity. 4. When the RFs of a pair of neurons expanded with conditioning, the area of the associated flank-normalized cross-correlogram peaks also increased (by a factor ranging from 0.84 up to 3.5). Correlograms returned to their preconditioning values when RFs collapsed. Within the populaton of neurons that expanded with conditioning, the subset of neurons starting with substantial RF overlap before conditioning showed essentially no increase in cross-correlation peak area (peak areas increased by factors ranging from 0.84 to 1.75 with a mean of 1.2 +/- 0.23), whereas the complementary set of neuron pairs with low or no initial overlap showed a significant increase in cross-correlation peak area (peak areas increased by factors ranging from 1.1 to 3.5 with a mean of 1.72 +/- 0.53). RFs that did not expand showed no increases in their flank-normalized cross-correlograms (correlogram peak area changed by a factor of 0.81 +/- 0.23 with conditioning). Neuron pairs that showed no significant cross-correlation before conditioning (i.e., neurons whose orientation preference differed by >40 degrees, or were highly directional in opposite senses, or were separated by more than similar to 3 mm on the surface of the cortex, or were highly monocular with opposite ocularity) showed no significant cross-correlation after conditioning. 5. Correlation strength did not change uniformly with conditioning over the area of the RFs. We measured this inhomogeneity in the effects of conditioning by obtaining correlograms with stimuli placed in different positions within the RFs of each pair of neurons. In three neuron pairs where the RF expansion was highly asymmetric, this comparison showed that correlation strength increased significantly only in the RF subregions where the expansion was greater.