Many experimental data suggest that a listener who focuses attention on a specific spectral region (or regions) becomes more sensitive to sounds falling there than elsewhere. Most often, the subject's task has been to detect a pure tone in noise. For example, in a two-interval, forced-choice paradigm, scores were around 90% when the signal was at an expected target frequency (0.5, 1, 2, or 4kHz) and were near chance, 50%, when the signal was at an unexpected frequency more than a half critical band away. In other tests, subjects detected unexpected temporal patterns (repeated tone bursts) just as well as expected patterns, provided spectral frequencies were the same. It has also been shown that when listening to a sequence of tone bursts that rise (or fall) in frequency, subjects miss a burst at a frequency significantly different from the value it normally has in the sequence. Other data suggest similar spectral specificity in the discriminationof complex sounds, including speech. Just what mechanism improves performance remains unclear. Nevertheless, some measurements of evoked physiological responses leave open the possibility that spectral specificity is achieved by efferent input to the auditory periphery, perhaps even to the cochlea.
View full abstract