Oscillations and information transfer in neocortex and hippocampus

Abstract

Understanding the mechanisms of signal generation and information flow in neuronal networks is a fundamental problem in neuroscience. Although a full investigation of the brain requires experiments in-vivo , computer models can explicitly quantify what is known and allow full control and measurement of the in-silico system.

We have developed several computer models of neuronal networks based on the anatomy and physiology of hippocampus and neocortex with varying levels of complexity. We made tradeoffs between complexity at the cell and network level, allowing us to increase the scale of the networks and increase simulation duration. To assist exploration of simulation and experimental data, we developed algorithms and automated data-mining tools.

We combined our simulations with data-mining of experimental data on animal models of schizophrenia and epilepsy to investigate the dynamical generation of physiological and pathological oscillations and the use of these oscillations to carry information. We employed in-vivo electrophysiological data from two experimental laboratories in order to look at activity across scales, including single neuron activity, local field potentials, and electroencephalography (EEG), in order to relate their oscillatory components. This research has established the following results: Study 1: Data-mining of rat visual cortex driving. Cortical areas can readily entrain to driving. Analysis in the time domain demonstrates particular patterns of LFP associated with normal driving, abnormal driving and seizures. Study 2: Developed time domain, information theoretic methods for quantification of single-unit identification from ensemble recordings. Utilized the techniques to demonstrate high identification quality in different brain areas and states. Study 3: Data-mining of ECoG from animal model of schizophrenia. Patterned abnormalities in EEG recurred in the abnormal brain. Brain activity in the neonatal ventral hippocampal lesion (NVHL) rat model of schizophrenia demonstrated periods of abnormal coordination that recurred, supporting the neural coordination hypothesis and suggesting a linkage between epilepsy and schizophrenia that has been reported clinically. Study 4: Simulation study of psychotomimetic effects. In-silico replication of in-vivo ketamine-induced theta/gamma changes. Demonstration that the augmented gamma associated with ketamine would reduce information flow through the network. Study 5: Simulation study of alterations in network information transmission. Information in network was transmitted not only via effects of excitatory transmission ("what is said"), but also through the effects of inhibitory transmission ("what is left out"). Study 6: Simulation study of dynamics of oscillation. Oscillation frequencies can be partially dissected into positive effects of excitation driving emergent oscillation activation and sculpting effects of inhibition providing selective oscillation damping that carves out spectral contours. Study 7: Simulation of learning of oscillation. Learning via spike timing dependent plasticity (STDP) altered dynamical balances to produce and sculpt oscillation in particular spectral patterns of value for information transmission.