This article examines the role of attention and automaticity in auditory processing as revealed by event-related potential (ERP) research. An ERP component called the mismatch negativity, generated by the brain's automatic response to changes in repetitive auditory input, reveals that physical features of auditory stimuli are fully processed whether or not they are attended. It also suggests that there exist precise neuronal representations of the physical features of recent auditory stimuli, perhaps the traces underlying acoustic sensory (“echoic”) memory. A mechanism of passive attention switching in response to changes in repetitive input is also implicated.

Conscious perception of discrete acoustic stimuli might be mediated by some of the mechanisms underlying another ERP component (NI), one sensitive to stimulus onset and offset. Frequent passive attentional shifts might accountforthe effect cognitive psychologists describe as “the breakthrough of the unattended” (Broadbent 1982), that is, that even unattended stimuli may be semantically processed, without assuming automatic semantic processing or late selection in selective attention.

The processing negativity supports the early-selection theory and may arise from a mechanism for selectively attending to stimuli defined by certain features. This stimulus selection occurs in the form ofa matching process in which each input is compared with the “attentional trace,” a voluntarily maintained representation of the task-relevant features of the stimulus to be attended. The attentional mechanism described might underlie the stimulus-set mode of attention proposed by Broadbent. Finally, a model of automatic and attentional processing in audition is proposed that is based mainly on the aforementioned ERP components and some other physiological measures.

In a repetitive auditory stimulus sequence, deviant infrequent tones typically elicit a component of auditory event-related potentials termed mismatch negativity (MMN). The elicitation of MMN is assumed to reflect the existence of a memory trace of the standard stimulus that has a decay time of about 10 s and is strengthened by repetition of the standards. The main aim of the present study was to test the decay time of the sensory memory trace during rapid-eye movement (REM) sleep vs. wakefulness, as indexed by the MMN. Subjects were presented 10 tone trains, separated by 3, 6, or 9 s of silence, during waking and REM sleep. Each train consisted of 9 standards of 1000 Hz and 1 deviant of 2000 Hz that occurred at position 1, 2, 4, or 6. The waking deviants elicited a frontocentral negativity with a scalp topography equivalent to the MMN component. During REM sleep, the negative component showed the same scalp distribution only for the 3-s intertrain interval (ITI). In this brain state, the MMN amplitude was smaller and decreased with prolongation of the ITI. These results suggest a weaker sensory memory trace formation and a premature decay time of such a memory trace during REM sleep as compared with wakefulness.