Auditory selective attention in humans: Analysis of mechanisms using event-related brain potentials.

摘要

This research investigated the neurophysiological and psychological mechanisms of auditory selective attention in humans using the technique of event-related potentials (ERPs) to gauge neural activity at different levels of the auditory pathways. Two dichotic listening experiments were performed in which conditions were optimized for highly focussed attention and the early selection of stimuli. These conditions included: (1) easily discriminable "channels" of stimuli, with attended and unattended tones distinguished by both pitch and ear of entry, (2) rapid stimulation rate, and (3) a difficult discrimination task (of infrequent deviant tones) within the attended channel.;The earliest ERP components (latency 1-10 msec), the brainstem evoked responses, were not affected by attention. However, an enhanced positivity between 20-50 msec was observed in the ERPs to attended-channel tones. The early onset of this effect strongly supports the early selection hypothesis that sensory processing can be selectively biased before full stimulus analysis has occurred. In addition, the effect of attention in the 80-230 msec range appeared to include an enhancement of the "exogenous" N1 and P2 components. This lent additional support to the view that early gating of irrelevant sensory input is one of the mechanisms involved in selective auditory attention. The "mismatch negativity" (MMN) wave elicited by the infrequent deviant tones in the unattended ear was markedly reduced, providing additional evidence for such early suppression and calling into question previous assertions that acoustic feature analysis and mismatch detection cannot be influenced by attention.;When stimuli are presented at rapid rates as in these attention experiments, the ERPs to adjacent stimuli in the sequence can overlap and confound one another. This problem was analyzed in detail using signal processing principles, and a new technique (Adjar) was developed to correct for it. Application of the Adjar technique demonstrated that the early attention effects were not the result of differential overlap from previous ERPs and enabled an analysis of the effects of the preceding stimulus type and ISI on the attention-sensitive ERPs. This latter application further supported the view that highly focussed selective attention can modulate sensory-evoked activity in the auditory pathways.