ABSTRACT
This paper investigates the use of thresholded random fields for the detection of significant activation in the evoked potentials in Electroencephalography (EEG) data. Three different activation states are investigated: two uni-modal signals (audio (A) and video (V) response) and one cross-modal signal (using simultaneous audio and video stimuli, A + V). The data is assessed using standard Gaussian-based hypothesis testing methods and more general methods based on non-Gaussian random fields.
