抄録
Recognition of human vowels depends on the first and second formants. We have reported that discrimination learning of synthetic vowels with multiple formants requires the presence of the auditory association cortex in rats. In the present study, characteristics of the synthetic vowel discrimination were investigated. Vowel-like sounds with two formants were synthesized in an all-pole model. Water-deprived rats were trained to discriminate between two vowels. Licking a spout during presentation of one (S+) was rewarded with water while the other (S-) was not. Either S+ or S- was presented randomly in a trial, which was repeated every minute throughout a test session of 12 hours on consecutive 4 days. Percentage of trials in which rats licked the spout was calculated separately for S+ and S-, and test performance was estimated as the difference. Discrimination characteristics, which were analyzed using synthetic vowels with various formant frequencies, appeared to parallel those in humans. We also investigated discrimination of voiced stop consonants /b, d, g/, which are characterized by transition of formants. Bilateral lesions of the auditory association cortex impaired discrimination learning between stop consonants with multiple formant transitions, while discrimination between single formant transitions was not. These results suggest that discrimination learning between synthetic speech sounds in rats may serve as an animal model for investigating speech sound recognition in humans. [J Physiol Sci. 2008;58 Suppl:S171]
