| The effect of various shapes (arrows pointing up, down, left and right) were studied to determine if EEG coherence could detected what arrows the subject(s) either saw on a screen or mentally imaged. The ultimate application of this was control of machines (e.g. robot arms) when normal limb function is impaired.; The experiment consisted of three phases. The first phase, called the visual images phase, had the subjects seated in front of a computer screen with eyes open. The second phase, called the remembered images phase, had the subjects sitting quietly with eyes closed. They visualized the shapes presented in phase 1 upon voice commands from the experimenter. The third and final phase, termed the mental images phase, had the subject assigned an image to a direction and visualize it upon command. The rest image was chosen by the experimenter and was constant across trials and subjects.; Auto and cross-spectra were computed using the Welsh window and smoothed with an exponential smoother ({dollar}alpha{dollar} =.6). Coherence was then computed and averaged over the delta, theta, alpha, alpha1, alpha2, beta, beta1 and beta2 bands. Averaged coherences for each stimulus was subtracted from the averaged coherence for the baseline period preceding it. The resulting differences were subjected to repeated-measures ANOVA, contrast analysis and discriminant analysis. Adjustment for individual differences had a significant effect, in some cases resulting in 100 per cent discrimination accuracy.; The results demonstrated that a system using EEG coherence could discriminate between shapes under the following constraints: (1) Shapes are visible; or (2) Shapes previously viewed are visualized; and (3) Individual variation is compensated for. |
