Date of Award
Doctor of Philosophy in Psychology
W. Grant Willis
The purpose of this study was to determine whether linear combinations of absolute-power beta waves were predictive of changes in vigilance. Spectral analyzed EEG data were collected across eight bipolar sites from 102, right-handed adults during a resting condition and during an auditory continuous performance test. The most normal distribution of absolute power (AP) scores was produced by a log10 (AP) transformation and the most normal distribution of relative power (RP) scores was produced by a log10(RP/l-RP) transformation. Absolute power had greater short-term reliability than relative power.
Components were derived from Principal Components Analysis performed on AP, RP, log1o(AP), and log10(RP/l RP). A component comprising frontal, RP theta was negatively correlated with omission errors. Changes in components comprising right-sided, AP beta were negatively correlated with changes in omission errors. Changes in frontal, AP beta and posterior, RP theta components were positively correlated with changes in omission errors. None of the multiple regression equations using these components, however, accounted for more than 13% of the variance. Right-sided and frontal AP beta components appeared, however, to warrant a follow-up study for purposes of cross validation. The AP, 12-component solution provided the best model of EEG activity during a vigilance task.
Weiler, Michael David, "A Measurement Model for Spectral Analyzed EEG" (1992). Open Access Dissertations. Paper 941.