on white-crowned sparrows are consistent with the hypothesis that seasonal changes in volume of HVc and RA may be specifically related to the capacity for learning" (p. 88). The fact that we observed young white-crowned sparrows, as well as adults, that are singing sub- and plastic-song, producing clear full song when stimulated to do so, suggests that the memory system remains intact throughout the year.
If the development of song each year for white-crowned sparrows involved the reinstatement of a sensorimotor memory trace, then a gradual development would be expected. Because the birds are capable of singing a full song at any time, provided the proper stimulation occurs, the lack of singing during some months could well be due to a lack of hormonal support to activate song engrams. The effect of these changes in hormonal levels would be to change the threshold for song production: With low titers of testosterone the elicitation threshold is high, and with high titers the threshold is low.
The revised theory outlined here involves a set of assumptions that are based on, and do not contradict, well-accepted observations and principles. The assumptions avoid the postulation of a sensory-gating mechanism as currently conceived, or of an auditory template; only genetically biased and differentially tuned sensory systems sensitive to combinations of song components are involved. Effective song stimuli, then, are those that fire receptor units, perhaps in HVc, and which, in turn, stimulate activity in an association area. The differential ease of learning various types of song is the result of the differential frequency of neural firing in association areas. In this view, genetic tuning of receptors is a threshold-adjusting mechanism and does not involve gating--anything that evokes a sensory response often enough, and with the proper timing, can be effective. Once the song engram is established the major factor determining its ease of retrieval is hormonal state. This state could regulate both the threshold of access to the system and the likelihood of vocalization occurring. Konishi ( 1985) has demonstrated that such self-produced vocalization is a critical factor in the development of song. I believe these assumptions, if favored with more supportive evidence, can remove some of the mystery surrounding models that evoke comparators, or which involve some unspecified template to match.
The proposed theory resolves the anomalies produced by the research results. Live tutoring would be expected to be more effective than tape tutoring because in the tape-tutoring experiments birds are exposed to the repeated and stereotyped presentation of a song. This repetitious stereotype would produce a maximum amount of habituation, and the student should make fewer orienting responses. A live tutor would provide a constantly changing stimulus milieu that would make it difficult for habituation to develop, and would maintain a constant level of short-term sensitization ( Shalter, 1984; Petrinovich, 1984). In this