INVESTIGADORES
MINDLIN Bernardo Gabriel
artículos
Título:
Automatic reconstruction of physiological gestures used in a model of birdsong production
Autor/es:
BOARI, SANTIAGO; SANZ PERL, YONATAN; AMADOR, ANA; GABRIEL B. MINDLIN
Revista:
JOURNAL OF NEUROPHYSIOLOGY
Editorial:
AMER PHYSIOLOGICAL SOC
Referencias:
Lugar: Bethesda; Año: 2015 vol. 114 p. 2912 - 2922
ISSN:
0022-3077
Resumen:
Automatic reconstruction of physiological gestures used in a model ofbirdsong production. J Neurophysiol 114: 2912?2922, 2015. Firstpublished September 16, 2015; doi:10.1152/jn.00385.2015.?Highlycoordinated learned behaviors are key to understanding neural pro-cesses integrating the body and the environment. Birdsong productionis a widely studied example of such behavior in which numerousthoracic muscles control respiratory inspiration and expiration: themuscles of the syrinx control syringeal membrane tension, whileupper vocal tract morphology controls resonances that modulate thevocal system output. All these muscles have to be coordinated inprecise sequences to generate the elaborate vocalizations that charac-terize an individual?s song. Previously we used a low-dimensionaldescription of the biomechanics of birdsong production to investigatethe associated neural codes, an approach that complements traditionalspectrographic analysis. The prior study used algorithmic yet manualprocedures to model singing behavior. In the present work, we presentan automatic procedure to extract low-dimensional motor gesturesthat could predict vocal behavior. We recorded zebra finch songs andgenerated synthetic copies automatically, using a biomechanicalmodel for the vocal apparatus and vocal tract. This dynamical modeldescribed song as a sequence of physiological parameters the birdscontrol during singing. To validate this procedure, we recordedelectrophysiological activity of the telencephalic nucleus HVC. HVCneurons were highly selective to the auditory presentation of the bird?sown song (BOS) and gave similar selective responses to the automat-ically generated synthetic model of song (AUTO). Our results dem-onstrate meaningful dimensionality reduction in terms of physiolog-ical parameters that individual birds could actually control. Further-more, this methodology can be extended to other vocal systems tostudy fine motor control.