Temporal Pattern Perception Mechanisms for Acoustic Communication

University Of California, San Diego, La Jolla CA

Investigators

Linked publications & trials

Abstract

Project Summary/Abstract: Processing temporally patterned acoustic communication signals is an important function of the auditory system and crucial to human speech and language. Understanding the neurobiological mechanisms that support these basic communication abilities holds promise for improving treatments for learning disabilities and communication disorders, including auditory processing disorder, dyslexia, and specific language impairment. While non- invasive studies in humans reveal brain loci of some language-related processes, the neurobiological mechanisms that support temporal pattern processing of communication signals remain poorly understood. Predictive coding, a general computational framework in which incoming sensory signals are compared to internal models, provides a potential mechanism for temporal pattern perception in the context of communication. This proposal investigates neurobiological mechanisms of predictive coding, by leveraging the songbird model, which permits invasive neuroscience methods, access to acoustically complex learned vocal communication signals, and natural communication behaviors. One sensory driven communicative behavior guided by internal models is recognition, i.e. matching an internal model to the sensory input corresponding to, for example, a word or face. In Aim 1, we test the cellular- and neural population-level predictions of predictive coding in the context of vocal recognition, examining how predictive errors generated in reference to learned internal models from the songs of individual birds, can guide recognition behavior. We combine state-of-the-art machine learning methods with large-scale multi-electrode electrophysiology, to test explicit models for natural stimulus representation, prediction, and error coding in single neurons and neural populations in secondary auditory cortical regions in awake birds during individual vocal recognition. A second way that internal models shape communicative behavior is in sensory feedback during vocal production. As we speak (or as birds sing) the auditory system continuously monitors the temporal structure of the sounds we produce, ensuring that the syllables, words, and phrases that emerge match our intentions. In Aims 2 and 3, we use the predictive coding framework to investigate sensory feedback and vocal-motor intention. We examine the coordination of neural population activity between sensory and motor regions, and identify neural dynamics tied uniquely to vocal-motor intentions in a secondary auditory region. We then causally manipulate this internal sensory-motor model, using both a behavioral manipulation to distort the sensory-feedback signal, and using optogenetics to directly modulate the hypothesized internal model that arises in motor control regions. Results of the proposed studies will yield understanding of neurobiological substrates foundational to communication behaviors, and a general framework within which more complex, uniquely human processes, can be proposed and tested.