Conférenciers invités

University of California, San Francisco School of Medicine
(Department of Physiology)


Vocal learning in adult songbirds

Birdsong is a complex motor skill that is learned and maintained in a manner that exhibits striking parallels to certain aspects of human speech acquisition. In particular, song and speech rely on hearing both to learn the species-appropriate sounds of others and to evaluate auditory feedback of self-generated vocalizations during the learning process. Moreover, because song is subserved by a well-delineated and accessible set of brain nuclei it provides a tractable model for investigating neural mechanisms that underlie production and learning of vocal behavior.

In this talk I will focus on behavioral and neurophysiologcal investigations of how auditory feedback contributes to production and plasticity of song in adult birds. Adult birdsong in many species is 'crystallized' in that song is highly stereotyped in its structure and normally changes little over time. However, we and others have used a variety of auditory feedback perturbations in adult birds to demonstrate that crystallized song is malleable. These experiments indicate that given appropriate instruction adult birds can rapidly and adaptively modify the structure of their songs. Song modification can be elicited both by a process of externally guided reinforcement and by a process of self driven error correction. These experiments demonstrate that the remarkable stability of adult song does not reflect an incapacity for adaptive vocal modification. Rather, they indicate that adult song remains fixed due to an ongoing process of feedback evaluation in which birds match their songs to a stable sensory target.

I will also describe experiments implicating an avian cortico-basal ganglia circuit, the anterior forebrain pathway (AFP), in these forms of adult vocal learning. Previous work has identified a primary song motor pathway that is required throughout life for normal song production and that is the presumed locus of much of the plasticity that reflects a bird's acquisition of his own unique song. In contrast, the AFP appears not to be an obligatory structure for the production of adult song. However, disruptions of the AFP prevent a variety of forms of adult vocal plasticity, suggesting a crucial role for this pathway in adult vocal learning. Here, I will describe evidence from the work of our laboratory and others that supports a specific hypothesis about the role of basal ganglia circuitry in learning of adult birdsong, that may apply more generally to learning of other motor skills including human speech. According to this hypothesis, the AFP contributes to three distinct components of learning: 1) it actively generates variability in motor performance on a trial-by-trial basis that constitutes a form of motor exploration in the service of discovering which behavioral variants give rise to better versus worse outcomes, 2) it retains information about which behavioral variants give rise to better outcomes and actively biases motor output towards the production of those variants during the initial stages of learning and 3) by virtue of this initial adaptive biasing of vocal output, it drives a gradual consolidation of learned changes in the primary motor pathway itself. Because the AFP is a simplified cortico-basal ganglia circuit that contributes to a single, quantifiable behavior, it may prove to be a useful system for further testing mechanisms whereby such circuits contribute to learning more generally.

University of Wisconsin-Madison
(Department of Communicative Disorders)


A Tentative Approach to a Speech Production Model and/or Theory of Motor Speech Disorders

This presentation will begin with a brief overview of speech production models and theories (if there is a difference, between the former and latter, in what we have available to us at the current time), and then discuss how we might be able to understand speech production problems in persons with motor speech disorders from this “normal model/theory” perspective. Some years ago, we (Bernstein & Weismer, J. Phonetics [2000], 28, 225-232) argued that a truly useful model/theory of “normal” speech production should be able to account for the facts of the communication deficit in motor speech disorders (or any speech production disorder). Ten years have elapsed since that paper, and the field seems to moving in this direction, especially if the imaging literature on speech production and language production and comprehension is examined carefully. However, we are still in a data-poor, technique-rich phase of speech production studies. Speech production data from normal speakers and those with motor speech disorders will be reviewed, and a list of phenomena that needs explaining by theory, or prediction by model, will be proposed. I will conclude the presentation with the claim (with what I take to be support from the literature) that disputes over the types of data that are important, as well as the phenomena that need explaining, are the primary obstacles to joining normal and disordered theories of speech production.