Auditory perceptual learning and the cochlear implant.

Abstract

Recent studies of the perception of complex nonspeech sounds have shown that individual parts of spectral-temporal waveforms become more salient through experience or selective training. One consequence of this is that the same sound can be perceived quite differently, depending on prior experience and also on the listener's expectations. The time course of learning to identify initially unfamiliar speech sounds by normal-hearing listeners has been shown to extend over many months, possibly even years, before reaching a stable high level of performance. If implant users or users of other types of speech aids (acoustic hearing aids, vibrotactile aids) are required to learn to interpret sounds as unfamiliar as the nonspeech sounds used in research are to normal listeners, similar lengthy experience should be required for them to achieve maximum performance. Why this has not been the case in some clinical studies is a puzzle. A possible explanation is that speech perception strongly depends on abilities other than sensory acuity or resolving power. Studies of individual differences in auditory temporal and spectral acuity have not shown strong correlations between individual differences in those abilities and individual differences in speech perception. It is argued that one way to interpret this observation is that differences in the ability to hear speech (as by two hearing-impaired listeners with the same audiogram) may be the result of central differences in the ability to infer an original stimulus from its fragments; this ability may be independent of the sensory modality in which the information is presented.