Learning is a common ability, accompanied by gamma oscillation, across species to acquire new knowledge stored in the hippocampus and neocortex into short-term and long-term memory, respectively.
Author: Kwan Tung Li
Publisher:
ISBN: OCLC:1286934157
Category:
Page: 210
View: 889
Learning is a common ability, accompanied by gamma oscillation, across species to acquire new knowledge stored in the hippocampus and neocortex into short-term and long-term memory, respectively. Thus, memory is first stored as short-term memory quickly and then consolidated into long-term memory in a longer timescale. Excitatory to excitatory (E → E ) spike-timing-dependent plasticity (STDP), an experimentally observable synaptic plasticity, is a widely used mechanism to form synaptic clusters in neural network models, where memory is proposed to be stored in strengthened synapses within the cluster. However, the interaction between gamma oscillation and STDP is unclear. On the other hand, the role of inhibitory plasticity in memory cluster formation attracts the attention of scientists in recent years, but it is not well understood yet because of the numerous species of inhibitory neurons and their plasticity. Besides, connectivity lesion, such as induced by Alzheimer's disease, causes memory deficits and abnormal gamma oscillation, but its relation to memory cluster is still an open question. My doctoral research thus aimed to study the interaction among different types of synaptic plasticity, gamma oscillation and circuit connectivity in memory learning and recall through computer simulation of the integrate-and-fire neuronal network of excitatory and inhibitory (E-I) neurons. i In the first part of my study, we explored the interaction between gamma oscillation and E → E STDP in an E-I integrate-and-fire neuronal network with triplet STDP, heterosynaptic plasticity, and transmitter-induced plasticity. We show that the plasticity performance depends on the synchronization levels accompanied by the emergence of gamma oscillations. Moreover, gamma oscillation is beneficial to form a unique network structure through synaptic potentiation. Secondly, we were inspired by an experimental result to study the functional role of excitatory to inhibitory ( E → I ) plasticity in memory consolidation through a feedforward two-layer E-I circuit model. We found that E → I plasticity can prevent overexcitation and assist memory cluster formation. We also predict that suitable pulse input to inhibitory neurons can rescue the memory performance deficits in the absence of E → I plasticity. Thirdly, we used E-I neuronal network model to investigate the effect of connectivity reduction as a result of Alzheimer's diseases on the interaction between circuit dynamics and STDP and the rescue of memory performance by optogenetic stimulation found in the experiments. It is found that the firing rate of the persistent activity is increased if connectivity is reduced mildly because of a transition from synchronous state to asynchronous state, while the persistent activity cannot be maintained and the firing rate is reduced with severe connectivity reduction. iv Furthermore, we found that stimulation with gamma frequency in circuits with connectivity lesion is the best for memory rescue because it can suppress the activation of the memory clusters that were initially activated in the lesion circuit. Moreover, we found that connectivity reduction causes the merging of memory clusters and the deterioration of existing memories during learning new memory with STDP. The whole study gives more insight into the co-evolution between microscopic synaptic dynamics, such as synaptic weight change, firing rate and synchronization of neuron spikes, and macroscopic phenomena, like gamma oscillation, memory performance, and connectivity. Our results may have implications in clinical applications to develop suitable brain stimulation schemes for memory rescue in neurodegenerative diseases. Furthermore, the understanding of the interaction among neural connectivity, dynamics, and plasticity may also offer insight into braininspired neural networks in artificial intelligence.