EEG recordings from two groups of subjects – with high and low levels of recognition of emotions from voices were made. Comparisons were performed of the numbers of pairs of leads with different levels of coherence in baseline conditions and on recognition of emotions in six standard frequency ranges and in individual bands with 1-Hz steps. Significant differences were seen between groups 1 and 2 both in baseline conditions and during recognition of emotions: in most cases, coherence was greater in subjects with poor recognition of emotions from voices.