Test-Retest Reliability of EEG Spectral Parameters During Cognitive Tasks: II Coherence

Publisher Website
Google Scholar
Abstract
We analyzed test-retest reliability of EEG coherence during rest and during two cognitive tasks: one verbal, the search for a synonym, and the other consisting of mental mathematical calculations. The experiment was performed twice, with a month's interval between each session. Coherence between all pairwise combinations of the EEG recorded in 15 leads was computed for delta, theta, alpha and beta bands. Comparing the changes observed between sessions for each condition, few significant changes were observed during rest, and the condition which manifests the most changes was calculation. Differences between conditions were observed in the second session in P3-T5, P3-T6, P3-T3, P4-T6, P4-T5, P4-Cz and Cz-T6 delta coherences ordered in a decreasing order from calculation to synonyms to rest. In order to ascertain whether the pattern of coherences across the head in each subject varied in relation to the session and the condition, we computed the correlation coefficients between all coherences values for one condition versus the other condition per band. For all subjects correlation coefficients near unity were observed between conditions in each session. However, correlation coefficients between sessions, even for the same condition, were much lower. We also computed the correlation coefficients for the coherence values between subjects in the same condition separately for each session. Correlation coefficients between subjects were much higher in the second session than in the first one. We concluded that coherence is not as variable as power during rest conditions; that it is strongly dependent on the state of the subject; and that it is sensitive to previous experiences which may change the subject's strategy for the accomplishment of cognitive tasks.