Uncovering Gamma in Visual Tasks

Abstract

Gamma is an important frequency band of the electroencephalogram (EEG), but its study has been impaired by problems with artefacts. This paper focuses on the artefacts caused by contraction of the extra-ocular muscles at the start of a saccade, which produces spurious gamma oscillations in the EEG. An algorithm was written and tested which detects and reduces the effect of this artefact. It involves novel adaptations of standard regression techniques which have traditionally been used to reduce blink artefacts, so as to render them applicable to the gamma band ocular artefact. Before the algorithm can be applied any power-line noise must be removed by noise cancellation and not notch filtering. The sharp, broadband gamma peak at around 200 ms was substantially reduced by the algorithm in all three subjects tested. However, there may be lower amplitude, task related, modulations in gamma which are uncovered when the artefact is reduced. The amplitude of the artefact had its largest positive value at the most anterior electrodes and its largest negative value at midline central and parietal electrodes, and these two sets of locations also showed the greatest reductions in gamma band magnitude after application of the algorithm. This study demonstrates the feasibility of reducing the saccade linked gamma band artefact.