Electroencephalographic Measures of Depressivity: Alpha Asymmetry and Fractal Dimension

Abstract

Recent research has suggested that neurofeedback, utilising alpha asymmetry or fractal dimension as an index of depression, may be an effective treatment for depressed individuals. In this thesis the relationships between frontal alpha asymmetry (FAA), parietal alpha asymmetry (PAA) and Higuchi’s fractal dimension (HFD) with PID-5 depressivity were investigated to assess their potential as signals for neurofeedback. Resting EEG previously recorded from a general sample of 66 individuals was analysed. The data of male and female participants was analysed separately. Optimised eye condition and bandwidth were determined with one way, repeated measure ANOVAs. Optimal specific measures of FAA, PAA and HFD were then identified by the proportion of PID-5 depressivity accounted for. The optimal FAA measure was obtained from the frontopolar electrode pair (Fp2 – Fp1) in the 10-12Hz sub-band. It was the only measure that was reliable in both male and female participants. PAA between the lateral electrode pair (P8 – P7) in the 8-10hz sub-band reliably correlated with depressivity in female, but not male, participants. HFD was reliable at every electrode in female, but not male, participants and displayed intercorrelation between all electrodes in both genders. A combined model using all three optimal measures showed that the proportional variance of FAA, PAA and HFD was mainly additive, with little variance shared between measures. The results suggest that FAA at the frontopolar electrode pair (Fp2 – Fp1) in the 10 – 12Hz band may be the optimal AA measure for neurofeedback protocols targeting depression. Correlations between AA and depressivity were often site- and band-specific; reliable correlations observed in one location did not necessarily generalise to other locations. Correlations between HFD and depressivity were not site-specific with most variance shared between sites. There was little overlap between the variance accounted for by FAA, PAA and HFD; indicating that the information conveyed by each is due to distinct neural processes, which may be associated with distinct aspects of depressivity and, potentially, other trait measures. Future work should assess replicability and the extent to which the results are specific to depressivity.