The predictive power of spectrotemporal receptive fields at the single spike train level

Abstract

Event Abstract Back to Event The predictive power of spectrotemporal receptive fields at the single spike train level Auditory neurons encode the acoustic features of complex sounds according to tuning properties that make them particularly responsive to specific spectral and temporal cues. Reverse correlation models of tuning such as spectrotemporal receptive fields (STRFs) estimate the spectral and temporal tuning properties of neurons from responses to complex sounds. The accuracy of a STRF in capturing the response properties of a neuron is assessed by using the STRF to predict responses to novel sounds and comparing those predictions to peristimulus time histograms (PSTHs) of the actual responses. A PSTH averages the neural response over multiple trials and has lower temporal precision than individual spikes. Because an animal can identify a complex sound after hearing it only once, neural responses at the single spike train level, as opposed to averaged over many trials, may be highly relevant to perception. Therefore, accurate models of neural tuning should predict neural responses to a sound at the single spike train level. We asked how accurately STRFs predict individual responses to complex communication sounds. Songbirds reliably discriminate among the unique songs of individual birds. Songbird auditory midbrain neurons reliably produce spike trains that share precise temporal structure across trials. These neurons encode spectrotemporal features of songs, and their STRFs predict responses at the PSTH level well. We tested the accuracy of STRFs in predicting neural responses at the single spike train level. We recorded the responses of single auditory midbrain neurons in male zebra finches to song from 20 other males. We modeled single spike trains from those songs using a linear-nonlinear-Poisson cascade model (LNP). Using an inhomogeneous Poisson process, simulated spike trains were generated based on a STRF convolved with a song spectrogram. Simulated spike trains share temporal structure with actual spike trains, but are less reliable across trials and are therefore less temporally precise. We quantified the temporal precision of spike trains across the actual and simulated trials, and between actual and simulated spike trains from the same songs. The k-means neurometric applies the k-means clustering algorithm to Euclidean distances between spike trains as defined in the van Rossum spike timing metric. Spike trains were sorted into k groups, where k is the number of unique stimuli used to generate spike trains. Accuracy in discriminating among songs was calculated as the percentage of spike trains that cluster into the appropriate group. As a population, actual responses of MLd neurons to songs demonstrate highly accurate k-means discriminability. STRF-based LNP simulations typically fail to capture the stimulus discriminating capacity of actual neurons. This is largely due to the temporal imprecision generated by the STRF-based model. Simulated spike train clusters were more scattered than actual spike trains, and clusters were generally closer under the simulated condition. The deficiency of the STRF-based model may be due in part to an inability to resolve rapid temporal features of the stimulus. We propose methods for supplementing and improving the STRF-based LNP model, such as incorporating parameters that capture the neural response to amplitude modulations within a stimulus. Conference: Computational and systems neuroscience 2009, Salt Lake City, UT, United States, 26 Feb - 3 Mar, 2009. Presentation Type: Poster Presentation Topic: Poster Presentations Citation: (2009). The predictive power of spectrotemporal receptive fields at the single spike train level. Front. Syst. Neurosci. Conference Abstract: Computational and systems neuroscience 2009. doi: 10.3389/conf.neuro.06.2009.03.245 Copyright: The abstracts in this collection have not been subject to any Frontiers peer review or checks, and are not endorsed by Frontiers. They are made available through the Frontiers publishing platform as a service to conference organizers and presenters. The copyright in the individual abstracts is owned by the author of each abstract or his/her employer unless otherwise stated. Each abstract, as well as the collection of abstracts, are published under a Creative Commons CC-BY 4.0 (attribution) licence (https://creativecommons.org/licenses/by/4.0/) and may thus be reproduced, translated, adapted and be the subject of derivative works provided the authors and Frontiers are attributed. For Frontiers’ terms and conditions please see https://www.frontiersin.org/legal/terms-and-conditions. Received: 03 Feb 2009; Published Online: 03 Feb 2009. Login Required This action requires you to be registered with Frontiers and logged in. To register or login click here. Abstract Info Abstract The Authors in Frontiers Google Google Scholar PubMed Related Article in Frontiers Google Scholar PubMed Abstract Close Back to top Javascript is disabled. Please enable Javascript in your browser settings in order to see all the content on this page.