engleski

Fingerprinting the sleep-related memory processing in EEG spectra

Aim:The aim of the study was to identify changes in the all-night EEG spectra which are specific for procedural and/or declarative sleep-related memory processing, while minimizing the effect of individual differences. Methods: Twenty one healthy male subjects were recorded in the sleep laboratory with Alice 4 polysomnographic (PSG) system. Each subject was screened for four nights: a) adaptation, b) control, and c-d) pre-sleep-task nights during which subject had performed either declarative or procedural task. The order of nights b-d was randomized. Recall performance was tested in the morning after the sleep. PSG data processing steps included visual sleep stage scoring and artefact removal. In each recording, mean log spectrum was estimated with Welch's periodogram for each (electrode derivation, sleep stage) pair. Linear regression (a0=0, a1) and correlation coefficients were used to estimate the direction and compliance of power density (PD) change in separate frequency bins. Results: Total sleep time as well as time spent in different sleep stages did not differ among nights b-d (one- way ANOVA, p>0, 05). However, mean log spectra have revealed distinctive differences. When spectra were analyzed with regression coefficients, a drop of PD relatively to the control night was evident throughout different derivations and stages, either in "procedural" or "declarative"spectra. The mildest drop was seen in the low frequency spectra (delta and theta band) where "declarative" spectra slightly decreased and there were no change in "procedural" spectra relative to the control. Within sigma band, the largest drop was detected, again for all derivations and stages. Position of this maximal drop was located at frontal positions at 12 Hz in REM and 14 Hz in non-REM stages, while for other derivations more complex patterns of local minima were observed. The highest correlation coefficients - indicating smaller interindividual differences were observed in S2 stage in: alpha and low frequency sigma band and >16 Hz beta range. Conclusion: The results indicate that although experimental procedures did not significantly alter the sleep architecture, clear cut differences were found between control night and "procedural" or "declarative"night. These changes were consistent between different derivations.

sleep ; EEG ; spectra ; memory processing

nije evidentirano