Abstract:
Electroencephalography (EEG) captures cerebral activity by recording electrical signals from within the brain non-invasively, and contains information on both the physical and mental state of the subject. This thesis applies signal-processing techniques to EEG analysis in two directions. Firstly, this thesis investigates signal-processing approaches for the automated classification of EEG signals based on whether the eyes of the subject are close or open at the time the EEG was recorded. Conventionally, the two eye conditions of opened or closed have been distinguished by visual inspection of the EEG activity that occurs at around 10 Hz, with increased activity being associated with a state of closed eyes and vice versa. This property can be empirically measured in the form of power spectral density in appropriate frequency range. However, a significant difference between the two eye states cannot be consistently observed in many of the subjects surveyed in this study. Therefore, an alternative automated technique based on histograms is designed in this study. The proposed algorithm achieved an overall successful recognition rate of 85%. Secondly, our next study examines the relationship between EEG signals and personality traits based on scores from the Revised NEO Personality Inventory (NEO PI-R). The NEO PI-R is a personality metric empirically designed to provide a measure of personalities in terms of the five-factor model. The five major domains described by the five-factor model are general dimensions of personality, including neuroticism, agreeableness, extroversion, openness and conscientiousness. This study found statistically significant differences in EEG signals between the extremes and the norms of both the domain scores as well as the more specific facet scores, in particular for the domain neuroticism and associated sub-traits. Additionally, pairs of the domain scores can be combined to provide a broad measure of personality known as personality styles. This study also found significant differences in the EEG signals of different personality styles. On the other hand, there are no readily apparent linear correlations between the personality scores and EEG features to be seen in the data collected for this survey. These results strongly suggest that EEG signals have the potential to serve as indicators of general trends and tendencies in a subject's personality, though not as a direct measure of the personality factor or facet scores.