Temporal Processing in the Auditory System: A Nonlinear Dynamics Approach with Theory and Experiment

New York University, New York NY

Investigators

Abstract

Rinzel 0078420 The neural computation by the mammalian auditory system to localize low frequency sound sources relies on processing in the brain stem. Neurons and circuits there, in the medial superior olive (MSO), have specialized biophysical properties for processing and preserving precise timings. These neurons have distinctive firing properties. When a steady stimulus is presented they fire only once, at stimulus onset; many other types of neurons would behave tonically, firing throughout the stimulus presentation. This property of phasicness is believed crucial for the MSO cell's role in precise temporal processing. In contrast, tonic cells are assumed to be less capable of tracking rapidly changing signals. The MSO cells have a special potassium current, IK-LT, that underlies their phasic behavior. The investigator and his colleagues systematically study how the temporal processing ability of a neuron changes as the neuron is transformed from phasic to tonic mode, say by gradually adjusting the strength of IK-LT (using pharmacological blockers and electronic methods of re-introducing the blocked current). When a cell is in phasic mode does it track a time-varying signal better, or does it perform better coincidence detection, than when it is in tonic mode? The research combines both experimental and theoretical approaches. The experiments involve electrical recording from individual MSO neurons while stimulating them with periodic and other time-varying signals. Various quantitative criteria are applied to assess the quality of temporal processing. From the theoretical side, biophysically-based mathematical models are developed that mimic the MSO neurons, including a term for IK-LT. The model's performance for temporal processing is evaluated just like the real cells. In addition, concepts from nonlinear dynamical systems are applied in order to reveal and understand the underlying mathematical structure. This mathematical understanding will shed light on the significance of phasicness in other neural systems where the mechanism might not involve IK-LT. This project also explores the influence of randomness on the temporal processing abilities. Some randomness is intrinsic to the auditory nervous system, and it is believed to be functionally important. Without sources of variability, these nonlinear neuronal systems tend to phase-lock too well, impairing a system's ability to perform discrimination tasks. The investigator therefore also assesses the effects of randomness on the temporal processing power of phasic and tonic cells and on the theoretical models. This work seeks to test whether the commonly accepted notion, that phasicness enhances temporal processing power, passes a set of quantitative criteria and, if so, to develop a theoretical foundation that supports the notion and that extends to other neural and possibly some chemical and physical systems as well. A related subproject is to develop computational models that help explain the dynamic effects seen experimentally as interaural phase (or amplitude or frequency) is varied dynamically. A deeper understanding of some surprising effects, as seen in the auditory mid-brain, should contribute to developing a theory for how motion of sound sources is analyzed in the brain. It is believed that some neural computations involve cellular and circuit properties that enable encoding and decoding based on precise timing of action potentials. Sound localization in the auditory system offers a compelling example. It serves as the case study for this research, that seeks a more qualitative characterization of cellular properties that correlate with precise temporal processing. Many cells in the auditory brain stem contribute to the system's ability to detect coincidence of interaural signals. These neurons have distinctive firing properties. When a steady stimulus is presented they fire only once, at stimulus onset, while neurons of many other types will continue to fire until the stimulus is turned off. This property of phasicness is believed crucial for precise temporal processing. In contrast, tonic cells are assumed to be less capable of tracking rapidly changing signals. The biophysical basis, a special potassium current, IK-LT, appears to underlie phasicness in the brain stem neurons. This project systematically addresses how the temporal processing ability of a neuron changes as the neuron is transformed from phasic to tonic mode, say by gradually adjusting the strength of IK-LT. When a cell is in phasic mode does it track a time-varying signal better, or does it perform better coincidence detection, than when it is in tonic mode? The research combines both experimental and theoretical approaches. The experiments involve electrical recording from individual neurons in vitro while stimulating them with periodic and other time-varying signals, including random components. From the theoretical side, biophysically-based mathematical models are developed that mimic the neurons, including a term for IK-LT. Various measures are applied to the computer and cellular models to assess reliability and precision of processing. In addition, concepts from nonlinear dynamical systems are applied in order to reveal and understand the underlying mathematical structure. This understanding will enable us to generalize about the significance of phasicness to other neural systems where the mechanism might not involve IK-LT. A related subproject is to develop computational models that help explain the dynamic effects seen experimentally as interaural phase (or amplitude or frequency) is varied dynamically. A deeper understanding of these surprising effects, as seen in the auditory mid-brain, should contribute to developing a theory for how motion of sound sources is analyzed in the brain. This project is supported by the Applied Mathematics and Computational Mathematics programs and the Office of Multidisciplinary Activities in MPS and by the Computational Neuroscience program in BIO.