A reafferent and feed-forward model of song syntax generation in the Bengalese finch
Male Bengalese finches do not normally change their vocal patterns in adulthood; song is stereotyped and stable over time. Adult song maintenance requires auditory feedback. If adults are deafened, song will degrade within 1 week. We tested whether feedback of all sound frequencies is required for song maintenance. The avian basilar papilla is tonotopically organized; hair cells in the basal region encode high frequencies, and low frequencies are encoded in progressively apical regions. We restricted the spectral range of feedback available to a bird by killing either auditory hair cells encoding higher frequencies or those encoding both high and low frequencies and documented resultant changes in song. Birds were treated with either Amikacin alone to kill high-frequency hair cells or Amikacin and sound exposure to target hair cells across the entire papilla. During treatment, song was recorded from all birds weekly. After treatment and song recording, evoked-potential audiograms were evaluated on each bird, and papillas were evaluated by scanning electron microscopy. Results showed that hair cell damage over 46-63% of the basal papilla and the corresponding high-frequency hearing loss had no effect on song structure. In birds with hair cell damage extending further into the apical region of the papilla and corresponding low-frequency and high-frequency hearing loss, song degradation occurred within 1 week of beginning treatment and was comparable with degradation after surgical deafening. We conclude that either low-frequency spectral cues or temporal cues via feedback of the song amplitude envelope are sufficient for song maintenance in adult Bengalese finches.