Control of vocal repertoire by reward in budgerigars (Melopsittacus undulatus)

The calls of some bird species may be modified by reward and punishment. However, the operant control of vocal topographies (i.e., the effect of reward or punishment on the physical dimensions of a vocal response) in such species has not been extensively explored. Using a computer-based, real-time system for rewarding vocalizations with food, the authors placed 3 budgerigars under a frequency-dependent reward schedule. During a session, the budgerigars received food for each vocalization that differed from the last N rewarded vocalizations. It was found that each of the budgerigars adapted their vocalizations to this procedure. When the value of N was 1 or 2, the birds ''solved'' the frequency-dependent schedule by developing N + 1 call types and used a simple ''win stay, lose switch'' sequencing strategy. At N = 3, 1 of the birds again produced N + 1 (i.e., 4) call types, and another solved the criterion by markedly increasing call variability. New calls developed from the elements of old call types and using multidimensional scaling techniques, the authors traced the evolution of each new call type from the previous experimental call repertoire.