Show simple item record

dc.contributor.advisor Johnson, Don H.
dc.creatorGruner, Charlotte Mason
dc.date.accessioned 2009-06-04T06:41:28Z
dc.date.available 2009-06-04T06:41:28Z
dc.date.issued 1999
dc.identifier.urihttps://hdl.handle.net/1911/19385
dc.description.abstract Neurons code information about sensory stimuli temporally in the times of occurrence of action potentials as well as spatially in the joint discharge patterns of neurons within an ensemble. In this work, we present a framework for quantifying how well a particular stimulus feature is coded for a given stimulus regardless of whether the coding is temporal, spatial, or both. We develop a method for empirically quantifying response differences in terms of the Kullback-Leibler distance, an information-theoretic quantity related to classification error probabilities. A relationship between the Kullback-Leibler distance and Cramer-Rao bound allows quantification of the ultimate limits of an omniscient estimator to estimate stimulus parameters based on observation of a given ensemble's response. By studying these ultimate limits for the inputs and outputs of a neural ensemble we can quantify the processing and efficiency of the ensemble. We apply these techniques to single and multiple unit simulated data from the lateral superior olive (LSO), an auditory nucleus in cat that is responsive to the spatial location of sound. Our results support the view that the LSO is extracting azimuthal location information, but we also found that LSO outputs contain significant information pertaining to sound stimulus amplitude (loudness). The coding of these cues also has a temporal component; the early transient response tends to contain more information pertaining to stimulus amplitude than the sustained response, while both the sustained response and the transient response contain information pertaining to azimuthal location. Thus LSO responses multiplex information about both stimulus azimuth and amplitude in a time-varying manner. Our experiments with ensembles of LSO neurons suggest that extraction of stimulus angle and amplitude information by ensembles is more efficient when the ensembles are composed of different LSO unit types (fast and slow choppers) than when composed of homogeneous units.
dc.format.extent 124 p.
dc.format.mimetype application/pdf
dc.language.iso eng
dc.subjectAudiology
Neurosciences
Biomedical engineering
Electronics
Electrical engineering
dc.title Quantifying information coding limits in sensory systems
dc.type.genre Thesis
dc.type.material Text
thesis.degree.department Bioengineering
thesis.degree.discipline Engineering
thesis.degree.grantor Rice University
thesis.degree.level Doctoral
thesis.degree.name Doctor of Philosophy
dc.identifier.citation Gruner, Charlotte Mason. "Quantifying information coding limits in sensory systems." (1999) Diss., Rice University. https://hdl.handle.net/1911/19385.


Files in this item

Thumbnail

This item appears in the following Collection(s)

Show simple item record