The habituation of event-related potentials to speech sounds and tones

We examined the short- and long-term habituation of auditory event-related potentials (ERPs) elicited by tones, complex tones and digitized speech sounds (vowels and consonant-vowel-consonant syllables). Twelve different stimuli equated in loudness and duration (300 msec) were studied. To examine short-term habituation stimuli were presented in trains of 6 with interstimulus intervals of 0.5 or 1.0 sec. The first 4 stimuli in a train were identical standards. On 50% of the trains the standard in the 5th position was replaced by a deviant probe stimulus, and on 20% of the trains the standard in the 6th position was replaced by a target, a truncated standard that required a speeded button press response. Short-term habituation (STH) was complete by the third stimulus in the train and resulted in amplitude decrements of 50-75% for the N1 component. STH was partially stimulus specific in that amplitudes were larger following deviant stimuli in the 5th position than following standards. STH of the N1 was more marked for speech sounds than for loudness-matched tones or complex tones at short ISI. In addition, standard and deviant stimuli that differed in phonetic structure showed more cross-habituation than did tones or complex tones that differed in frequency. This pattern of results suggests that STH is a function of the acoustic resemblance of successive stimuli. The long-term habituation (LTH) of the ERP was studied by comparing amplitudes across balanced 5.25 m stimulus blocks over the course of the experiment. Two types of LTH were observed. The N1 showed stimulus-specific LTH in that N1 amplitudes declined during the presentation of a stimulus, but returned to control levels when a different stimulus was presented in the subsequent condition. In contrast, the P3 elicited by the deviant stimuli showed non-specific LTH, being reduced across successive blocks containing different stimuli. P3s elicited by target stimuli remained stable in amplitude.