The Construction Of Learning Model And Study Of Learning Mechanisms In Cultured Neuronal Networks

For cultured neuronal networks, learning is an exploration process that involves formation and modulation of sets of associations between stimuli and responses. In recent years, the study of learning is generally focusing on behavioral, organic and molecular level. However, the evidence is still insufficient at network level. The realization of high functions of the brain such as learning and memory requires the integrality of information, and the integrality of information is performed at network level. It is necessary and important to construct one learning model at network level for learning and memory study.In this paper, we used 300-500 mV, 200μs, 0.2-1 Hz square voltage stimulation, 10-30μA, 200μs, 0.2-1 Hz square current stimulation as conditioning stimulation respectively, and performed closed-loop training on the cultured hippocampal neuronal network by stimulating a pair of associated electrodes in the multi-electrode array. As a consequent, the learning phenomenon of the neuronal network was induced, and the learning model at network level was constructed successfully. Based on the learning model, we studied respectively effects of excessive inhibition of ionotropic glutamate receptors and GABAA receptor on dynamic characteristic of response activities during learning training, and demonstrated the modulation of ionotropic glutamate receptors and GABAA receptor on temporal-spatial encoding and energy distribution of response activities in the neuronal network during learning. Moreover, we studied effects of learning training on spontaneous activities of the neuronal network, and identified that oscillation might be one of important mechanisms of learning. The overall results and review of our study are listed as following:(1) We used 300-500 mV, 200μs, 0.2-1 Hz square voltage stimulation or 10-30μA,200μs,0.2-1 Hz square current stimulation as conditioning stimulation, and performed closed-loop training on the cultured hippocampal neuronal network by stimulating a pair of associated electrodes in the multi-electrode array, consequently constructed the learning model at network level. Response/stimulus ratio and response time were used as characteristic parameters.(2) Based on the learning model, we studied the modulation of ionotropic glutamate receptors on temporal-spatial encoding and energy distribution of response activities in the neuronal network during learning. Application of 50μM APV, 50μM CNQX or 2 mM Mg2+ all decreased the rate, the amplitude and firing probability of early postsynaptic responses. All synaptic events were abolished by subsequent application of 50μM APV and 50μM CNQX. The firing pattern of response activities in the neuronal network was changed from organized pattern to disorganized pattern by 50μM APV. Power spectral density of response activities in the neuronal network was changed significantly by 50μM CNQX, the power decreased evidently in the 0-10 Hz characteristic frequencies. Correlation and synchrony of response activities in the neuronal network decreased with 50μM APV or 50μM CNQX.(3) We studied respectively modulations of GABAA receptor on temporal-spatial encoding and energy distribution of spontaneous activities and response activities of the neuronal network during learning. The rate of spontaneous activities was increased, burst durations were extended, the interval of burst was decreased, and numbers of spikes in burst were increased. And harmonic oscillation-like spontaneous activities appeared. Moreover, synchrony, correlation and rhythm of spontaneous activities were enhanced. During the training procedure, the interval of early postsynaptic responses was decreased, and the distribution of firing probability of responses was changed markedly. In addition, power spectral density of response activities was changed evidently, and the power is high in the 0-50 Hz characteristic frequencies. Correlation and synchrony of response activities in the neuronal network increased significantly, and the epileptic firing appeared.(4) We studied effects of learning training on spontaneous electrophysiological activities of the neuronal network. Successful learning induced oscillation firstly and then long term potentiation of spontaneous activities in the neuronal network. The result suggested that oscillation might be one of important mechanisms of learning. After successful learning, the rate of burst was increased, and numbers of burst were increased, but the interval of burst was decreased. Moreover, correlation and synchrony of spontaneous activities in the neuronal network were improved significantly. The distribution of power spectral density in different characteristic frequencies has been changed. However, the change was not significant for non-learning condition.