In an initial experiment, starlings (Sturnus vulgaris), cowbirds (Molothrus ater), and a mockingbird (Mimus polyglottos) learned a relative pitch discrimination between four-tone sequences that either ascended or descended in frequency within a one-octave range. They lost the discrimination completely when the frequency range was shifted either an octave above or below the initial training range. Recovery of the discrimination in the novel ranges was slow, and considerable relearning was necessary. Starlings, however (for whom the tests were conducted), easily generalized the relative pitch discrimination to new frequencies within the original training range. In a second experiment, starlings were trained on the same relative ascending-descending discrimination in two noncontiguous frequency ranges. They then lost the discrimination when transferred into a novel gap of frequencies between the original training ranges-completely counter to the outcome to be expected on the basis of stimulus generalization. The experiments show that songbirds can learn a relational discrimination between serial pitch patterns but that their ability to generalize the discrimination is constrained markedly by the frequency range in which the sound sequences are learned initially. The results have implications for human and animal perception of serial acoustic information and, more specifically, for songbirds' ability to process relative and absolute pitch.Recently, Hulse, Cynx, and Humpal (1984a) reported that European starlings (Sturnus vulgaris), a mimicking species of songbird, could discriminate between ascending and descending tone sequences on the basis of relative pitch. In this article, we describe discrimination performance when we changed the absolute pitch of the tone sequences while holding constant their

European starlings (Stumus vulgaris), a mimicking species of bird, were successfully trained to discriminate a set of temporally configured four-tone sequences that rose in pitch on a whole-tone scale from a set of otherwise identical four-tone sequences that fell in pitch. A series of transfer tests then assessed their ability to maintain the discrimination (a) when the intensity of the tones in the patterns was varied, (b) when the patterns were shortened to three, two, or one tone, and (c) when the patterns' rhythmic tempo was changed. The discrimination was maintained when intensity values changed, which indicated that apparent loudness was not a relevant cue for accurate performance. When sequences were shortened, overall discrimination performance diminished. More important, however, shortened sequences produced evidence for both absolute and relative pitch perception in sequence discrimination. That is, the discrimination depended in part on the ability to "name" particular pitches in the sequences and in part on the ability to detect that a given sequence rose or fell in pitch. Tempo changes did not affect discrimination. The results have implications for cognitive processes in serial acoustic pattern perception by animals, for a comparative study of pitch perception guided by theories encompassing absolute and relative pitch, and for research seeking the functionally significant dimensions of natural birdsong.This article describes how a species of mimicking bird, the European starling (Sturnus vulgaris), processes patterns of tones organized into arbitrary pitch structures according to rule-based principles of serial pattern learning. The research examined one of the simplest principles in pitch perception: namely, whether or not the starling can discriminate between formally organized sequences of tones that rise or fall in pitch. The research may not only add to our understanding of the basic acoustic dimensions of birdsong itself but also foster a comparative analysis of the acoustic serial pattern learning that occurs in human language and music perception.

No abstract

Zebra finch (Taeniopygia guttata) song is composed of syllables delivered in a set order. Little is known about the program that controls this temporal delivery. A decision to sing or not to sing may or may not affect the entire song. Song, once commenced, may continue or may halt. If song is halted, stops may occur only at certain points. Seven zebra finches were presented with short bursts of strobe light while engaged in song. The variables of interest were whether the birds stopped and where they stopped. The results can be summarized as follows: Ongoing zebra finch song can be interrupted, interruptions occur at discrete locations in song, and the locations almost always fall between song syllables. These results reveal a functional representation of song production and place constraints on possible neural mechanisms that underlie song production in zebra finches and probably other oscine species. The results also raise hypotheses about the elements of song perception and memory.

NeurobiologyRole of gender, season, and familiarity in discrimination of conspecific song by zebra finches (Taeniopygia uttata) ABSTRACTOperant conditioning techniques were used to assess how gender and song familiarity affect song discrimination in adult zebra finches (Taeniopy gultaft). Twentyfive males and females, divided into five cohorts, were trained to di te between conspecific songs at different times of year. Males d atig between their own and another song from their own aviary reached criterion in the fewest number oftrials, followed by males discriminating between two songs from their own aviary, then by males discriminating between songs they had not heard before. Females dirimi-

Relative pitch perception in animals has been difficult to demonstrate. This failing is due in part to stimulus sets that make an absolute pitch solution viable. In Experiment 1, starlings failed to acquire a discrimination that could be solved only on the basis of relative pitch. In Experiment 2, starlings were trained on a smaller set of pitch patterns, for which both absolute and relative pitch solutions were available, then tested with three series of unreinforced probe stimuli. Series 1 assessed stimulus control by absolute pitch. In Series 2, absolute pitch cues dictated one response, and relative pitch cues dictated a different response. Results indicate that starlings extract relative pitch from artificial pitch patterns only after acquiring a discrimination that permits both absolute and relative pitch solutions. Results are discussed in terms of the relative salience of absolute and relative pitch.

Zebra finch (Taeniopygia guttata) song syllables often include harmonically related frequency components. These harmonics may be suppressed, and this differential emphasis varies between the syllables in a song and between individual birds' songs. These patterns of harmonic suppression are timbre. Individual syllables' patterns of harmonic suppression are constant with adult males' songs. Young males that imitate the songs of older males also imitate their patterns of harmonic suppression. Syringeal denervation distorts these patterns, which suggests that they are produced through active control of the vocal organ. The selective suppression and emphasis of some harmonics creates a great number of possible timbre variants for any one syllable. These add signal diversity to the limited array of frequency modulations and range of fundamental frequencies found in zebra finch song. Analyses of bird song that disregard timbre may overlook a feature that is important in vocal communication.

Auditory input to the right or left forebrain of adult male zebra finches (Taeniopygiaguttata) was disrupted by lesloning the ipsilateral auditory relay nucleus of the tmus. These birds were then presented with two kinds of auditory discriminations: (T) between their own song and the song of a cage mate; (is) between two versions of an unfamiliar zebra finch song that differed only in the harmonic profile of one of the syllables. Right-side lesion birds did better than left-side lesion ones at discriminating between their own song and the song of a cage mate; left-side lesion birds did better on the harmonic profile task. We sugest that the two halves of the zebra finch brain process conspecific sounds differently, as seems to be the case for humans.