Neural activity was recorded with two independent electrodes separated by 0.5-2 mm, aligned in parallel, and advanced perpendicular to the surface of the cat auditory cortex. For smaller separations a solid-state multielectrode array with interelectrode distances of 125 microns was used. The difference in recording depths for the two independently movable electrodes was never more than 100 microns; the electrode contacts of the multielectrode array were at the same depth. Thus the correlation studies dominantly explored horizontal interactions. 2. Out of 995 neuron pairs recorded, 478 represented pairs of single units, whereas the other pairs were contaminated with 5-10% misclassified spikes. Only the single-unit pairs were further analyzed. Of those pairs, 338 showed a clear correlation peak, and in 329 of these the peak heights were exceeding the Z > 4 significance level (P < 0.0001). Two hundred fifty-two of the significant correlograms (53% of total) could be attributed to common input; the remaining (16% of total) were indicative of unilateral excitation. For the 181 single-electrode pairs the percentage of unilateral excitation pairs (42%) was about the same as the percentage of common input pairs (38%). For the 297 dual-electrode pairs all but one of the 184 significant correlations were indicative of common input. No correlograms indicative of inhibition were found. 3. The correlograms with clear peaks were classified into three types: narrow (n = 40), mixed (n = 77), and broad (n = 221). Narrow and mixed types were with two exceptions found only for single-electrode pairs; broad types were found for single- and dual-electrode pairs. Narrow-type correlograms were in majority of the unilateral excitation type. Correlograms were calculated both for 1-ms binwidths (50-ms lead/lag time) and for 10-ms binwidth (500-ms lead/lag time). The correlograms were characterized by four parameters; the half width of the central peak, the peak correlation coefficient, the association index, and for unilateral excitation cases also by the effectiveness. 4. Across all three correlogram types the half width of the correlation peaks was significantly smaller for single-electrode pairs (mean, 27 ms) than for dual-electrode pairs (mean, 42 ms). For broad-type correlograms only, the mean half widths were not significantly different between single- and dual-electrode pairs. 5. The correlation coefficients (1-ms bin correlograms) were significantly larger for single-electrode pairs (mean, 0.038) than for dual-electrode pairs (mean, 0.011). The same was found for the 10-ms binwidth correlograms.(ABSTRACT TRUNCATED AT 400 WORDS)
