A Caltech Library Service

Neuronal Control of Bird Song Production


McCasland, James Stacy (1983) Neuronal Control of Bird Song Production. Dissertation (Ph.D.), California Institute of Technology. doi:10.7907/z6h5-1n28.


Male songbirds sing to establish species and individual identities, to maintain territories, and to stimulate reproductive behavior in conspecific females. The ability to produce a stereotyped song is therefore necessary for reproduction. In many species the patterns of song are learned by young birds from adults in a process involving two stages - an auditory phase involving storage of a song model, and a sensorimotor phase in which the bird learns to reproduce the model by using auditory feedback. Lesion results (Nottebohm et al., 1976) demonstrated that at least three discrete nuclei -- HVc (Hyperstriatum ventrale, pars caudale), RA (n. Robustus Archistriatalis), and nXIIts (n. Hypoglossus, pars tracheosyringealis) -- are indispensable to normal song production. These findings opened the way to study of a discrete vertebrate neural system which, uniquely, mediates production of an acquired yet stereotyped behavior.

However, the functional specializations of these nuclei cannot be discerned through lesion studies. For this reason I developed techniques for examining directly the neural correlates of song production -- by making neuronal recordings from the freely-behaving, singing bird. From my studies I can draw the following generalizations about the relative roles of vocal control nuclei: (1) the telencephalic nucleus NIf (Nucleus Interfacialis, of Nottebohm, 1980), which provides an input to HVc and is anatomically the "highest" nucleus in the descending motor pathway, is uniquely placed among vocal control nuclei to be a generator of timing cues for song; (2) consistent with the unidirectional and serial connections between nuclei of the descending tract, NIf, HVc, RA, and nXIIts are activated sequentially prior to sound onset; (3) single-unit recordings demonstrate that there are neurons specialized for production of specific song elements; (4) a re-examination of hemispheric dominance in song control shows that both hemispheres normally make similar contributions to all song elements; and (5) at least two types of inhibitory interactions between auditory and motor activities can be observed in the behaving bird.

The source of timing cues for song. Previous lesion and neuroanatomical studies have implicated eight brain nuclei in the control of song. I recorded neural activity from seven of the eight nuclei, in order to assess the presence and patterning of song-related activity, and to localize the site or sites where timing cues for song are generated. I confirmed the suggestion from lesion studies that the serially connected nuclei HVc, RA, and nXIIts each produce song-correlated neural activity and showed quantitatively that these nuclei are activated sequentially when the bird vocalizes. I further demonstrated that many multiple-unit recordings from these areas show a clear modulation of activity pattern which corresponds to the temporal pattern of song. This finding implicates these nuclei in the genesis or transmission of timing cues for song. I then made recordings from Nucleus Interfacialis (NIf), which provides an input to HVc, and found a similar song-related pattern of activity. The pre-sound latency of recordings from NIf was longer than that of comparable recordings from HVc, suggesting that NIf provides timing information to HVc. To test this interpretation, I sectioned the pathway from NIf to HVc; postoperatively, these subjects were unable to produce stereotyped song. A sham operation involving the same amount of tissue damage but sparing NIf had no effect on song. HRP studies show only one input to NIf (with the possible exception of the auditory nucleus Field L), from nucleus Uva of the thalamus. Recordings from this nucleus showed no changes in activity during vocalization in the adult, and bilateral lesions of this nucleus had no effect on song. To test for other inputs to NIf which might not transport HRP, I made a series of recordings from eight sites in the vicinity of NIf. None of these recordings showed song-related activity. The combination of three findings -- song-related patterning of activity in NIf, the necessity of NIf for normal song patterning, and the absence of song-related activity in inputs to NIf -- imply that NIf is a source of timing cues for song, and is the only song control nucleus to which this statement can be unequivocally applied. The fact that elimination of input to NIf has no effect on song suggests that NIf produces a learned central motor program for song.

Recordings made from two other putative song control nuclei, MAN (Magnocellular nucleus of the Anterior Neostriatum) and area X, showed no changes in activity during song or other vocalizations. Transection of the pathways linking these two nuclei to HVc had no effect on song. Thus my recording experiments demonstrate that lesion effects provide a better indication of the necessity of a nucleus for adult song production than do studies of anatomical connections. Whether Uva, MAN, or area X plays a crucial role in song development remains to be determined.

Single units with specialized roles in song production. Elucidation of the neural mechanisms of song control ultimately requires an analysis at the single-unit level: the clues to the interactions involved may be masked in multi-unit recordings. Accordingly, I have developed a new technique for recording from single neurons in song system nuclei of the freely-behaving, singing mockingbird. The method employs an X-Y microdrive which, when chronically implanted, is sufficiently stable so that units can be isolated and held for periods of up to several hours.

My recordings from mockingbird HVc revealed single units with several classes of specialized roles in song production. Many cells exhibit premotor activity for all song syllables, and do not respond to those same syllables presented as auditory stimuli. Some of these cells show long-latency (e.g., 500 msec) "anticipatory" activity at the initiation of song and between syllables of a long song bout, thus demonstrating a role for HVc beyond the purely motor aspects of sound production. A few cells show more selective premotor activity, producing highly stereotyped bursts of spikes for only a few syllables. The distinctions between sounds for which these cells do or do not show activity may be quite subtle, suggesting that HVc units may encode subtle variations in sound production. This interpretation is supported by the high degree of temporal specificity of unit firing with respect to timing of the syllable.

A re-examination of hemispheric dominance in song control. Series of studies by Nottebohm et al. demonstrated that left-side lesions in the song system of several species inflict much more severe damage to song than comparable right-side lesions. These studies led to the hypothesis that the left hemisphere plays a dominant role in song production. My neural recordings, however, led to a completely different interpretation of the lesion results. Recordings from the left and right hypoglossal nerves innervating the syrinx invariably showed very similar activity patterns for any given vocalization. These patterns consist of neural bursts and silent periods of different durations, with each song syllable represented by a unique neural homologue. Because the hypoglossal nerve represents the "final common pathway" for song control, the neural activity it transmits must convey coded commands for song production. To the extent that my recordings reveal these commands, it appears that both right and left nerves are transmitting the same messages to their respective syringeal halves. Section of right or left hypoglossal nerve in a subject from which nerve recordings were also made led to the classical behavioral deficit, including complete disappearance of certain syllables, but these behavioral deficits were uncorrelated with any consistent pattern differences in the two nerves. Recordings from right and left HVc of the same bird also showed similar activity patterns for a given song element. These results are incompatible with any all-or-nothing mechanism of lateralization in the song control system. However, they are consistent with the absence of hemispheric anatomical asymmetries in song control nuclei.

The critical feature of the qualitative dominance theory is the independent contribution of different sets of sounds by the two sound sources in the syrinx -- the right and left internal tympaniform membranes. By blocking airflow through the right or left bronchus, I was able to eliminate the function of one membrane and observe the set of song syllables produced by the bird with only the other membrane. With this bronchus-plugging technique, I found that the right syringeal half alone is sufficient to produce easily recognizable counterparts to most syllables normally produced by both syringeal halves. This result contradicts the conclusion from nerve-section experiments that most sounds are contributed by the left side only, but is completely consistent with the nerve recordings which indicate bilateral participation in sound production. Taken together, these results indicate that both hemispheres and syringeal halves make similar contributions to the production of all song elements.

Interaction between auditory and motor activities in an avian song control nucleus. Intracellular recordings from anesthetized birds have shown that many neurons in HVc respond to auditory stimuli. I confirmed this result in multi-unit recordings from awake behaving birds, and further demonstrated responses of HVc neurons to playback of the bird's own song. The functional significance of these responses is not yet clear, but behavioral studies show that auditory feedback plays a crucial role in the development of normal song. I showed that the song-correlated temporal pattern of neural activity persists even in the deaf bird. Furthermore, in the normal bird the activity pattern correlated with production of certain song elements can be clearly distinguished from the pattern of auditory responses to the same song elements. This result implies that an interaction occurs in HVc of the singing bird between motor and auditory activity. Through experiments involving playback of sound while the bird is singing, I showed that the interaction consists of motor inhibition of auditory activity in HVc and that this inhibition decays slowly over a period of seconds after the song terminates.

The time-locking of pre-motor activity in HVc to song elements survives the loss of auditory feedback by deafening even though auditory inputs to HVc produce clear responses to sounds heard by the normal quiescent bird. Normal songs are produced by deafened adults in some species. These findings suggest the possibility of a learned central motor program for song, functioning independently of sensory input in the adult. Because the bird must make use of auditory feedback to develop normal song, the autonomy of the motor program would have to be acquired during ontogeny. Recordings from Field L neurons in the singing bird show auditory responses to song elements which persist during singing, suggesting that the sensorimotor interaction I observed occurs within HVc. If so, it is likely that the motor inhibition of auditory inputs to HVc plays a role in song development or song maintenance. If present throughout development of song, this inhibition would appear to serve the paradoxical function of rendering these inputs inaccessible for guidance of motor learning, unless such guidance does not occur within HVc or involves nonspiking interaction between or within cells. These considerations thus impose constraints on possible neural mechanisms of song control.

In single-unit recordings I have confirmed the expectation from multi-unit data that at least some auditorily responsive neurons are inhibited during singing, thus demonstrating a type of sensorimotor interaction at the single cell level. One cell in mockingbird HVc was inhibited by playback of all syllables of the bird's own song, and showed premotor activity for some syllables. Another cell which was inhibited by playback showed specific motor activity, firing for only one of two very similar song syllables. While the functional significance of these interactions is unknown, it will clearly be of great theoretical importance to know whether other types of sensorimotor interaction are exhibited by single cells in the vocal control system. Such cells would be likely candidates for specialized roles in song learning or song maintenance.

Item Type:Thesis (Dissertation (Ph.D.))
Subject Keywords:Neurobiology
Degree Grantor:California Institute of Technology
Major Option:Neurobiology
Thesis Availability:Public (worldwide access)
Research Advisor(s):
  • Konishi, Masakazu
Thesis Committee:
  • Allman, John Morgan (chair)
  • Arnold, Arthur P.
  • Hudspeth, A. James
  • Van Essen, David C.
  • Konishi, Masakazu
Defense Date:11 February 1983
Funding AgencyGrant Number
NIH5 T01 GM00086
NIH5 T32 GM07737
NIH5 R01 NS 14617
Jean Weigle Memorial FundUNSPECIFIED
Pew Memorial TrustUNSPECIFIED
Record Number:CaltechTHESIS:10212019-145504230
Persistent URL:
Default Usage Policy:No commercial reproduction, distribution, display or performance rights in this work are provided.
ID Code:11835
Deposited By: Mel Ray
Deposited On:21 Oct 2019 22:21
Last Modified:03 Nov 2021 23:43

Thesis Files

PDF - Final Version
See Usage Policy.


Repository Staff Only: item control page