The study was aimed to reveal the subbands of correlated changes in power spectral density of brain electrical activity (EA), including the low-voltage (up to 10 mu V) high-frequency (HF) components (40-200 Hz) in dogs during instrumental conditioning by means of factor analysis. The values of the EA spectral density in interstimulus intervals calculated with a resolution of 1 Hz were used for subsequent factor analysis (the standard principle component technique with varimax rotation), Twenty factors could explain about 80% of total variance, The groups of frequencies which were presented by comparatively narrow peaks (2-3 points) of high loads (more than 0.6) of single factors ("stable" factors) were taken into consideration, In the process of conditioning the factor organization of the EA became substantially complicated, the number of the "stable" factors increased, It was originally shown that the high lends of these factors divided the HF band in comparatively narrow frequency subbands, which appear to reflect the functional mosaics in the neocortex.