Functional relationships in small groups of neurons in the dorsal cochlear nucleus.

This dissertation describes research involving the investigation of functional relationships in small groups of individually recorded neurons in the dorsal cochlear nucleus (DCN) of the guinea pig. The first section is a critical review of the general problem of neural system identification from multi-unit recordings, i.e., from observations of multiple single-neuron spike discharges, with an emphasis on the problem of identifying functional properties of the DCN. The second section reports the results of an experimental study in which multi-unit recordings of two to five neurons in the DCN were analyzed with respect to complex acoustic stimuli. The general neural system is considered a nonlinear multiple-input, multiple-output system composed of many elements whose interconnections are governed by underlying anatomical structure. The system identification problem is one of describing global functional properties of the system in terms of connections between elements and representations of external stimuli by groups of elements. Multi-unit recordings allow for estimation of stimulus-to-unit and unit-to-unit relationships for the observed units. Most of the system's elements are unobserved and their effects on system function cannot be directly estimated. Several unit-to-unit analysis methods are reviewed, with particular emphasis on correlation methods and intensity models. The experimental study investigated functional relationships between neurons in the DCN in response to noise stimulation. Multi-unit recordings of two to five neurons were obtained using thin-film, solid-state, multi-channel microelectrodes having multiple recording sites spanning 100 to 600 microns in a single plane. Twenty-three neural ensembles were observed, each of which consisted of units either in the same approximate isofrequency sheet or different isofrequency sheets. The spectro-temporal receptive field (STRF) of each individual unit was calculated to estimate stimulus-to-unit functional relationships. Unit-to-unit relationships for unit pairs were estimated with correlation analysis. Shared-stimulus driving between units was found to be associated with overlaps in the spectral extents of the units' STRFs; temporal patterns of shared-stimulus driving were consistent with temporal patterns of the STRFs. Units also exhibited a high degree of effective connectivity separate from stimulus driving and not related to tuning. These results are compared to previous work that used tonal stimuli.PhDBioengineeringUniversity of Michigan, Horace H. Rackham School of Graduate Studieshttp://deepblue.lib.umich.edu/bitstream/2027.42/105553/1/9135623.pdfDescription of 9135623.pdf : Restricted to UM users only