COINCIDENCE DETECTION MODELS IN AUDITORY BRAINSTEM

点击翻译按钮获取中文摘要

DESCRIPTION (from applicant's abstract): The long-term objective of this research is to characterize coincidence-detector neurons used for binaural sound localization in the avian and mammalian brainstem, using computational, biophysically accurate, models of individual neurons. The specific research goals are to: 1) build and characterize models of avian brainstem Nucleus Laminaris (NL) neurons using physiological data from chicks (Gallus domesticus), capable of reproducing the behavior and capabilities of real chick NL neurons; 2) build and characterize models of NL neurons in barn owls (Tyto alba) capable of reproducing the behavior and superior capabilities of real barn owl NL neurons, using physiological data from barn owls (supplemented by data from chicks); 3) modify and generalize the models of (1) and (2) to examine coincidence detection in mammalian brainstem Medial Superior Olive (MSO) cells, using mammalian data. The models will be designed to determine and analyze the anatomical and physiological features crucial to coincidence detection. The fundamental features are similar in chicks and barn owls, but crucial differences will give the barn owl its significantly greater abilities. The differences between NL and MSO are even larger, but their functions as coincidence detectors are sufficiently similar to permit MSO models to be built analogously to the NL models, modified according to anatomical and physiological differences. The research design and methods are based on constructing the models in the neural modeling environment NEURON (freely available from Yale University), with voltage-dependent conductance mechanisms and synaptic mechanisms written using the programming language C, for speed. The anatomical and physiological parameters describing the model cells are taken from the literature. Each model neuron has multiple dendrites, a soma, an axon hillock, a myelinated segment, and a node of Ranvier. The number, length, and branching of dendrites are free parameters and may vary tonotopically. The dendritic and somatic sections have low voltage activated (LVA) and high voltage activated (HVA) K+ conductances. The axon hillock and node of Ranvier have spike generating/propagating voltage-dependent ion channels. Non-phase-locked, diffuse, depolarizing inhibition is incorporated. Input vector strength is linked to the best frequency (BF) of the neuron, and NL versions of the model also link BF with dendritic length. The model automatically generates physiologically useful statistics (e.g. spike rate and vector strength) in numerical and graphical forms. The health relatedness of this research comes from the gained understanding of the neural mechanisms of sound localization. Understanding those mechanisms will substantially aid the quest for hearing aids and cochlear implants that would allow their users to localize sounds far better than current technology.