Kenneth D. Harris - US grants

DESCRIPTION (provided by applicant): The brain dysfunctions underlying schizophrenia are poorly understood. Nevertheless, it is likely that a critical aspect of this disease is a breakdown of the normal information processing functions of the neuronal assemblies. This project would study the activity of neuronal populations in sensory neocortex and investigate how neuronal assembly activity is disrupted in the dissociative anesthetic (PCP) model of schizophrenia. Experimental investigation of this question will require recording large numbers of cells in functioning neural circuits. However, obtaining this data is only the beginning: the computational and statistical machinery to draw meaningful conclusions from such data must also be developed. Here we propose a collaborative research project between a mathematician (Kenneth Harris) and an electrophysiologist (Gyorgy Buzsaki), with the aim of recording, analyzing, and modeling the activity of large neuronal populations in primary sensory cortex and its disruption by psychotomimetic drugs. The project will rely on two techniques we have developed over the last years: large-scale neuronal recordings using silicon microelectrodes; and the data analysis method of peer prediction. The use of silicon probes will allow for estimation of the location of recorded cells, identification of monosynaptic connections between cell pairs, and characterization of neurons as pyramidal cells or interneurons. Experimentally identified assembly structure will be interpreted in the context of this circuit-level information. We will investigate the hypothesis that psychotomimetic effects of low doses of dissociative anesthetics are caused by a partial distortion in assembly organization, whereas larger doses cause a more complete distortion resulting anesthesia. If reliable signatures of psychotomimetic doses on assembly structure are found, this will suggest a novel method of drug screening for antipsychotics, whereby candidate drugs are evaluated by their ability to reverse these signatures.

[unreadable] DESCRIPTION (provided by applicant): The auditory cortex is a critical structure for processing of sound information in mammals. However, as with all sensory cortices, neural activity in the auditory cortex is steered, but not deterministically controlled by sensory input, and the auditory cortex is capable of generating coordinated activity patterns even in the absence of acoustic stimulation. The aim of this proposal is to measure and model the way sensory-evoked and internally generated information interact in the auditory cortical column. We will do this in the context of a specific hypothesis. Previous anatomical and in vitro studies of sensory cortex suggest that information from thalamus primarily follows a "transcolumnar" path from layer 4, via layers 2 and 3, to layer 5; and that in layer 5 a network of large pyramidal cells is capable of generating spontaneous activity through recurrent excitation, as well as receiving "feedback" input from higher cortical areas. We hypothesize that integration of sensory input with internally generated activity occurs primarily in layer 5, whereas neurons of the superficial layers are less involved in sensory-independent processing. Outputs to subcortical structures, which arise from layer 5, thus reflect the processing of sensory information in the context of the brain's internal state, whereas "feedforward" corticocortical projections arising from superficial layers provide a more faithful representation of the external world to higher cortical regions. [unreadable] To investigate this hypothesis, we will record from neural populations across cortical layers, simultaneously with individual morphologically reconstructed neurons. We will analyze this data with a novel method called predictive dynamical modeling, in which a dynamical system model fit to spontaneous activity preceding a sensory stimulus is evaluated by its ability to predict the subsequent sensory response. We will show that a simple self-exciting system model can quantitatively predict the global dynamics of cortical networks, and further investigate how the information content of individual neurons is shaped by the interaction of global dynamics and sensory input. We will show that information is coded densely in layer 5 but sparsely in the superficial layers, consistent with a higher threshold for self-excitation in superficial layers. We will study how the laminar flow of activity changes with cortical state. The state of the cortex varies from a "desynchronized" state during alert wakefulness, to a "synchronized" state during drowsiness and sleep that is characterized by coordinated stimulus-independent activity. By recording across the wake-sleep cycle, we aim to show that cortical dynamics in the desynchronized state are close to linear, but become progressively more nonlinear as cortical state becomes synchronized, accompanied by a shift in the control of cortical activity from sensory input to internally generated activity. We will show that stimulus-independent activity is primarily confined to layer 5, but that in the synchronized state spontaneous patterns also reach the superficial layers, consistent with a role of this activity in replay of sensory experience. Broader Impact. Understanding the natural function of auditory cortex is critical for learning how it misfunctions in disease states. The research proposed here will require personnel with knowledge of both mathematics and biology, and interdisciplinary training will be a central part of this project. Involvement of underrepresented groups is also a priority, and our location at Rutgers Newark allows us a unique opportunity promote diverse participation in this program. [unreadable] [unreadable] [unreadable]