Compartmental Model for Binaural Coincidence Detector NeuronsMotivationInteraural time difference is a cue to sound source azimuthBinaural Coincidence Detector NeuronsThe ModelBuilding a compartmental modelCompartmental Model of Coincidence Detector NeuronDendritic filtering and attenuationPoint vs. compartmental neuron modelsBetter coincidence detection for 3-compartment modelExtra slidesBinaural coincidence mechanism for coding interaural time differences (ITD)User InterfaceResult: ITD tuning improves as synaptic inputs get farther from soma along dendritesResult: There is an optimal frequency for every dendritic lengthStudent FeedbackWhat next?Slide 18Electrical circuit for small segment of nerve fiberSynapse positionFarungeBasic circuit elementsCompartmental Model for Binaural Coincidence Detector NeuronsBertrand DelgutteZachary Smith and Leonardo Cedolin, SHBTJonathan Simon, University of MarylandMotivation•Provide understanding of how neurons work, and how their structure defines their information-processing capabilities. •Traditional teaching formats such as lectures and discussion of literature papers do not give sufficient intuition.Specific Goals•Provide hands-on experience with modern compartmental model of a neuron.•Experiment with model parameters and learn their role in neural signal processing.Model System•Binaural coincidence detector neurons in the auditory brainstem.Interaural time difference is a cue to sound source azimuthBinaural Coincidence Detector NeuronsHigh FrequenciesLow FrequenciesSmith & Rubel, 1979Axons from left earAxons from right earThe Model•Developed by Jonathan Simon at University of Maryland•Based on coincidence detector neurons in the chick•Compartmental model: Neuron geometry is explicitly represented•Includes known membrane channels (Hodgkin-Huxley, synaptic, low-threshold K+, etc…)•All model parameters easily manipulated with GUI•Implemented in NEURON, a general, high-level language for neural modelingBuilding a compartmental modelC. Circuit model for small length of passive cable-> Also need active membrane channelsCompartmental Model of Coincidence Detector NeuronSomaLeft Dendrite Right DendriteHillockAxonSynaptic Inputs from Left EarSynaptic Inputs from Right EarDendritic filtering and attenuation•Transient response of linear cable to impulse of current at different distances from the current source.•Both latency and temporal spread increase with distance (lowpass filtering). Peak amplitude decreases (attenuation).2ci maGSpace ConstantPoint vs. compartmental neuron modelsGdGdGmCmGrEsEsGlGmCmGrEsEsGlPoint neuron 3-compartment modell rsm l rG GV EG G G 2( )(1 ) (2 )l r l r dsm l r m d l r m d dG G G G GV EG G G G G G G G G G     •Synaptic potential depends only on sum Gl+Gr for point-neuron model, but also depends on product GlGr for 3-compartment model. •Point neuron does not distinguish between monaural and binaural coincidences.Better coincidence detection for 3-compartment model•Binaural: Gl=Gr=Gb•Monaural: Gl=0, Gr=2Gb•Fixed Parameters: Es=100mV, Gm=100, Gd=20Extra slidesBinaural coincidence mechanism for coding interaural time differences (ITD)SOUNDCOCHLEAR FILTERINTERNAL DELAYXNEURAL RESPONSECOCHLEAR FILTERCOINCIDENCE DETECTORCONTRA EARIPSI EAR ITDUser InterfaceResult: ITD tuning improves as synaptic inputs get farther from soma along dendrites-180 -135 -90 -45 0 45 90 135 18000.10.20.30.40.50.60.70.80.91Interaural Phase Difference (degree)Normalized Discharge RateDistance from Soma10%30%90%Result: There is an optimal frequency for every dendritic length-180 -90 0 90 1800100200300400Interaural Phase Difference (degree)Discharge Rate (spikes/sec)500Frequency (Hz)3008001250Student FeedbackPros•The lab provides the basic understanding of a compartmental model•I am happy to work with a full-blown model and not a baby version•We had the opportunity to be creative and try different parameters•It was very user friendly•The simulations really drove home the reasons for using a compartmental model in the first placeCons•This lab was a little too complicated… I prefer something more straightforward.•All we did was load the configuration file and press `Init & Run’.•I must admit that the lab was pretty "dry".General•The labs provide the best available introduction to the field.What next?•Improve existing laboratory exercise:–Make the lab less “cookbook”–Make user interface less daunting•Connect neuron model to model of signal processing by normal and pathological ears•Develop more challenging simulations for advanced classes (e.g. requiring programming in NEURON)Interaural time difference is a cue to sound source azimuthElectrical circuit for small segment of nerve fiberSynapse positionFarungeBasic circuit