Applied Physics Seminar: "Neuronal circuit interactions in search of consciousness"

Information flow in neural networks underlies perception, memory, and cognitive functions of the brain. Active neurons form dynamic ensembles through complex synaptic interactions that are ongoing and modulated by sensory input. We are interested in the state-dependent changes in neuronal interactions in neural circuits from wakefulness to sleep or pharmacologically altered states such as anesthesia. Manipulation of neuronal activity by anesthetics allows us to investigate neuronal dynamics critical to the conscious state. Anesthetics alter brain dynamics by acting on various neuronal targets including direct effects on cortical neurons and indirect modulation via thalamic and subcortical arousal sites. In our ongoing work, extracellular cortical electrical activity is recorded using chronically implanted high-density microelectrode arrays in animals to measure neuronal spiking, monosynaptic interactions, population activity, and local field potentials. Spontaneous ongoing activity of neuron populations at a mesoscopic scale exhibits a dynamic repertoire of metastable spatiotemporal patterns whose distribution follows power-law suggesting the presence of self-organized criticality. These power laws are resistant to state changes. Anesthetics dose-dependently decrease the strength of monosynaptic neuronal interactions and produce spatially and temporally fragmented activity. Extracellular spiking of individual neurons becomes fragmented to short periods of activity, interrupted by hundred millisecond gaps. Neuronal interactions quantified by mutual information are also reduced and their response to sensory stimulation is impaired. Moreover, the correlation and complexity of spike interactions drops abruptly at the critical anesthetic associated with loss of consciousness. These changes are reversed during cortical activation achieved by exogenous stimulation of the ascending arousal system. Modeling neuronal interactions as a first-order multivariate Markov process confirms these findings as they predict a distinct transition between conscious and unconscious states that cannot be fully explained by a reduction of overall neuronal firing rate alone. We tentatively conclude that anesthesia may modulate the state of consciousness by suppressing neuronal circuit interactions and the stream of information processing.