Journal of Comparative and Physiological Psychology 1977, Vol. 91, No. 1, 8-16
Role of Auditory Feedback in Canary Song Development Peter Marler and Mary Sue Waser Rockefeller University As in other songbirds, early deafening had drastic effects on the song of the roller canary, a cardueline finch, resulting in a song that was much simpler and more variable than the normal. The repertoire of syllable types, of which the song is made, was reduced from 30 to a mean of 5.0. Loud white noise was successfully used as a reversible method of cutting off auditory feedback from vocal behavior. Although suffering permanent elevation of hearing thresholds, birds reared in noise to 200 days, singing at first like deaf birds, subsequently increased their syllable repertoires significantly. Birds reared in noise to weaning at 40 days, again partly deaf, achieved a normal repertoire size when stimulated with a singing adult. Without such stimulation the repertoire was significantly reduced, showing that canary song is not fully innate, as had been thought. Although abnormal, the song of deaf canaries retained more species-specific features than did the song of emberizine sparrows when the songs developed without auditory feedback. The results are interpreted in terms of a sensory template theory.
Studies of the effects of deafening on vocal development in birds have implicated auditory feedback as a key factor in vocal ontogeny in many songbirds (Konishi, 1963, 1964, 1965a, 1965b; Konishi & Nottebohm, 1969; Nottebohm, 1968). In one subfamily of sparrows, the Emberizinae, deafening of males early in life, before they have begun to sing, results in a very elementary pattern of sound from which most species-specific characteristics are lacking. Thus, auditory feedback plays a prominent role in the generation of species-specific song characteristics in sparrows, a fact on which a special theory of vocal learning has been based that invokes species-specific auditory templates, modifiable through experience, as being involved in vocal development (Marler, 1976). Closely related to the emberizine sparThis research was supported by Grant MH 14651 from the National Institute of Mental Health. The authors are indebted to Ann Lutjen for help with the experiments, to Fernando Nottebohm for performing deafening operations and, with Masakazu Konishi, for valuable discussion and criticism. Requests for reprints should be sent to Peter Marler, Rockefeller University, York Avenue and 66th Street, New York, New York 10021.
rows are the cardueline finches, including such common species as housefinches and goldfinches. Their song patterns tend to be different from those of typical sparrows. Instead of the short, discrete, individually stereotyped patterns that are typical of sparrows, cardueline finch songs are often more drawn out and variable, many times longer than those of sparrows, and often delivered in flight rather than from a perch. It is conceivable that auditory feedback might have a different role to play in the ontogeny of such songs. The canary is a domestic representative of the Carduelinae. This article explores the role of auditory feedback in its song development. The results reveal that a bird's ability to hear its own voice plays a different developmental role in cardueline and in emberizine finches, indicating a phylogenetic difference in the strategy of song development in these two subfamilies. Method Subject and Apparatus These studies were conducted with an inbred strain of Belgian "Wasserschlager" canaries bred selectively by aviculturalists in Europe for particular patterns of singing behavior. Each 30 x 30 x 50
AUDITORY FEEDBACK AND SONG DEVELOPMENT cm wooden cage, within an acoustically insulated chamber (IAC Model AC-1) was fitted with three 5in. (12.5-cm) loudspeakers at one end. In some experiments white noise, with known frequency characteristics (Marler, Konishi, Lutjen, & Waser, 1973), was broadcast with a total sound pressure level of 95-100 dB re 20 /xN/m- at 20 cm from the speakers, set with a General Radio Type 1551-C sound level meter. This pressure level corresponds to 90-95 dB above the auditory threshold of canaries at their most sensitive frequencies.
Procedure Young birds were raised by their parents to independence (30-40 days after hatching) and sexed by laparotomy. "Weaning" is used synonymously with "independence from parental feeding" in this article. All experimental groups were subjected to an approximately normal photoperiod cycle bringing them into breeding condition at about the same age. On the assumption that the different treatments might have hindered gonadal development in varying degrees, all birds were implanted sc with a 10mg pellet of testosterone propionate in the neck at about 200 days of age to ensure maximal motivation for singing. Sound recordings were made of subsong and song of birds in all conditions. Selected samples were analyzed sound-spectrographically by the Kay Electric sound spectrograph (Model 6061B) and the Federal Scientific "Ubiquitous" Real Time Spectral Analyzer (Model UA7B). For estimating syllable repertoire size, we selected analyses of several goodquality recordings from the first season of singing, beginning at 200 days of age. These were inspected serially; each new syllable type was identified as encountered until the number of new ones reached an asymptote. Other samples from the same bird were then inspected in the same way; thus we reached an estimate of the size of the repertoire of different syllable types. In a few cases available samples failed to reach an asymptote and the asymptotic value was then estimated. Normally, each male has a repertoire of about 30 syllable types. Attention is focused in this study on effects of various experimental treatments upon the size of this syllable repertoire. Syllable types were identified by morphology and by timing. Thus, if there was a consistent stepwise change in the rate of a syllable type similar in morphology, a second type was nevertheless designated. Gradual changes in timing were not used, however, as criteria for a change in syllable type. If a single syllable occurred regularly as a stable transitional form between two other types, it was designated as a third type. However, if the transition between two types was variable in form through the sample, a separate type was not designated for the transitional patterns. Song samples were generally accepted for this analysis if they exceeded 1 sec in duration and contained more than one syllable type. Exceptions were made with experimental subjects
9
with very small repertoires. To distinguish song from other calls uttered by canaries, we excluded sounds separated from obvious song by intervals of more than .5 sec. As a result, utterances regularly occurring close to the beginning of singing were included in the syllabic analysis.
Results Effect of Total Auditory Deprivation on Song Noise plus deafening. The effects of deafening on the subsequent development of song are known to be profound, especially when deafening occurs before song has begun to develop (Konishi & Nottebohm, 1969). Steps were taken in the present study to ensure that the deprivation from auditory feedback from song and other vocal behavior was complete. Not only were the subjects deafened at weaning but they were also reared by their parents in acoustical chambers in which high-level masking noise was broadcast, sufficient to prevent them from hearing sounds of the parent and their own vocalizations. Five young male canaries were treated in this fashion. As shown in Table 1, these birds developed very simple songs, with an average of five syllable types per bird, the number ranging from one to eight. For comparison, Table 2 presents data on the syllable repertoires of six young males reared normally with parents and siblings to maturity under similar caging conditions, without masking noise and with their hearing intact. Their syllable repertoires ranged from 29 to 42 with a mean of 34.3. Figure 1 shows the syllable repertoire sizes of these normally reared birds (I) and of the noise-reared, deafened subjects (V). As shown in the inset, there is a significant difference, indicating a drastic reduction in the complexity of singing behavior of birds deprived of all auditory stimulation other than that derived from the white masking noise in early life. The relative simplicity of the deaf bird's song is also manifest in its syllabic structure, as in other species (Konishi & Nottebohm, 1969). In spite of this drastic simplification of the song and the variability in
10
Individually isolated
PETER MARLER AND MARY SUE WASER
p
P
a
Ui
%
12 % 1 a n Is .5
c5 'Ca
1 o
m
OO
g
f-
S o
05
rH
V. 0
"§
e
e a •i^io
!
la .1
rt ?
01 *—t >—ifi
Role of Auditory Feedback in Canary Song Development Peter Marler and Mary Sue Waser Rockefeller University As in other songbirds, early deafening had drastic effects on the song of the roller canary, a cardueline finch, resulting in a song that was much simpler and more variable than the normal. The repertoire of syllable types, of which the song is made, was reduced from 30 to a mean of 5.0. Loud white noise was successfully used as a reversible method of cutting off auditory feedback from vocal behavior. Although suffering permanent elevation of hearing thresholds, birds reared in noise to 200 days, singing at first like deaf birds, subsequently increased their syllable repertoires significantly. Birds reared in noise to weaning at 40 days, again partly deaf, achieved a normal repertoire size when stimulated with a singing adult. Without such stimulation the repertoire was significantly reduced, showing that canary song is not fully innate, as had been thought. Although abnormal, the song of deaf canaries retained more species-specific features than did the song of emberizine sparrows when the songs developed without auditory feedback. The results are interpreted in terms of a sensory template theory.
Studies of the effects of deafening on vocal development in birds have implicated auditory feedback as a key factor in vocal ontogeny in many songbirds (Konishi, 1963, 1964, 1965a, 1965b; Konishi & Nottebohm, 1969; Nottebohm, 1968). In one subfamily of sparrows, the Emberizinae, deafening of males early in life, before they have begun to sing, results in a very elementary pattern of sound from which most species-specific characteristics are lacking. Thus, auditory feedback plays a prominent role in the generation of species-specific song characteristics in sparrows, a fact on which a special theory of vocal learning has been based that invokes species-specific auditory templates, modifiable through experience, as being involved in vocal development (Marler, 1976). Closely related to the emberizine sparThis research was supported by Grant MH 14651 from the National Institute of Mental Health. The authors are indebted to Ann Lutjen for help with the experiments, to Fernando Nottebohm for performing deafening operations and, with Masakazu Konishi, for valuable discussion and criticism. Requests for reprints should be sent to Peter Marler, Rockefeller University, York Avenue and 66th Street, New York, New York 10021.
rows are the cardueline finches, including such common species as housefinches and goldfinches. Their song patterns tend to be different from those of typical sparrows. Instead of the short, discrete, individually stereotyped patterns that are typical of sparrows, cardueline finch songs are often more drawn out and variable, many times longer than those of sparrows, and often delivered in flight rather than from a perch. It is conceivable that auditory feedback might have a different role to play in the ontogeny of such songs. The canary is a domestic representative of the Carduelinae. This article explores the role of auditory feedback in its song development. The results reveal that a bird's ability to hear its own voice plays a different developmental role in cardueline and in emberizine finches, indicating a phylogenetic difference in the strategy of song development in these two subfamilies. Method Subject and Apparatus These studies were conducted with an inbred strain of Belgian "Wasserschlager" canaries bred selectively by aviculturalists in Europe for particular patterns of singing behavior. Each 30 x 30 x 50
AUDITORY FEEDBACK AND SONG DEVELOPMENT cm wooden cage, within an acoustically insulated chamber (IAC Model AC-1) was fitted with three 5in. (12.5-cm) loudspeakers at one end. In some experiments white noise, with known frequency characteristics (Marler, Konishi, Lutjen, & Waser, 1973), was broadcast with a total sound pressure level of 95-100 dB re 20 /xN/m- at 20 cm from the speakers, set with a General Radio Type 1551-C sound level meter. This pressure level corresponds to 90-95 dB above the auditory threshold of canaries at their most sensitive frequencies.
Procedure Young birds were raised by their parents to independence (30-40 days after hatching) and sexed by laparotomy. "Weaning" is used synonymously with "independence from parental feeding" in this article. All experimental groups were subjected to an approximately normal photoperiod cycle bringing them into breeding condition at about the same age. On the assumption that the different treatments might have hindered gonadal development in varying degrees, all birds were implanted sc with a 10mg pellet of testosterone propionate in the neck at about 200 days of age to ensure maximal motivation for singing. Sound recordings were made of subsong and song of birds in all conditions. Selected samples were analyzed sound-spectrographically by the Kay Electric sound spectrograph (Model 6061B) and the Federal Scientific "Ubiquitous" Real Time Spectral Analyzer (Model UA7B). For estimating syllable repertoire size, we selected analyses of several goodquality recordings from the first season of singing, beginning at 200 days of age. These were inspected serially; each new syllable type was identified as encountered until the number of new ones reached an asymptote. Other samples from the same bird were then inspected in the same way; thus we reached an estimate of the size of the repertoire of different syllable types. In a few cases available samples failed to reach an asymptote and the asymptotic value was then estimated. Normally, each male has a repertoire of about 30 syllable types. Attention is focused in this study on effects of various experimental treatments upon the size of this syllable repertoire. Syllable types were identified by morphology and by timing. Thus, if there was a consistent stepwise change in the rate of a syllable type similar in morphology, a second type was nevertheless designated. Gradual changes in timing were not used, however, as criteria for a change in syllable type. If a single syllable occurred regularly as a stable transitional form between two other types, it was designated as a third type. However, if the transition between two types was variable in form through the sample, a separate type was not designated for the transitional patterns. Song samples were generally accepted for this analysis if they exceeded 1 sec in duration and contained more than one syllable type. Exceptions were made with experimental subjects
9
with very small repertoires. To distinguish song from other calls uttered by canaries, we excluded sounds separated from obvious song by intervals of more than .5 sec. As a result, utterances regularly occurring close to the beginning of singing were included in the syllabic analysis.
Results Effect of Total Auditory Deprivation on Song Noise plus deafening. The effects of deafening on the subsequent development of song are known to be profound, especially when deafening occurs before song has begun to develop (Konishi & Nottebohm, 1969). Steps were taken in the present study to ensure that the deprivation from auditory feedback from song and other vocal behavior was complete. Not only were the subjects deafened at weaning but they were also reared by their parents in acoustical chambers in which high-level masking noise was broadcast, sufficient to prevent them from hearing sounds of the parent and their own vocalizations. Five young male canaries were treated in this fashion. As shown in Table 1, these birds developed very simple songs, with an average of five syllable types per bird, the number ranging from one to eight. For comparison, Table 2 presents data on the syllable repertoires of six young males reared normally with parents and siblings to maturity under similar caging conditions, without masking noise and with their hearing intact. Their syllable repertoires ranged from 29 to 42 with a mean of 34.3. Figure 1 shows the syllable repertoire sizes of these normally reared birds (I) and of the noise-reared, deafened subjects (V). As shown in the inset, there is a significant difference, indicating a drastic reduction in the complexity of singing behavior of birds deprived of all auditory stimulation other than that derived from the white masking noise in early life. The relative simplicity of the deaf bird's song is also manifest in its syllabic structure, as in other species (Konishi & Nottebohm, 1969). In spite of this drastic simplification of the song and the variability in
10
Individually isolated
PETER MARLER AND MARY SUE WASER
p
P
a
Ui
%
12 % 1 a n Is .5
c5 'Ca
1 o
m
OO
g
f-
S o
05
rH
V. 0
"§
e
e a •i^io
!
la .1
rt ?
01 *—t >—ifi
