Effects of negative affective sounds on P300

Abstract

P300 are event related potentials that are widely used; however, the performance of systems based on it decrease drastically when they are transported outside a laboratory environment where the stimuli can be isolated. Since isolation of the stimulus is not always practical it is desirable to study P300 when the subjects are receiving stimuli in other senses besides sight. Thus this paper tested the subjects in a silent environment and adding affective noise as catalogued by the Affective Digitised Sound database. The resulting EEG recordings were analysed and found that isolated P300 features increased (p<;0.05), but features that related it to the neighbourhood decreased (p<;0.05) making it more difficult to identify in half of the subjects. A preliminary test points towards models derived of events elicited in noisy environments as a solution to the decrease of performance of classifiers.