| The brain is the most complex network in nature,and the different connection patterns of hundreds of millions of neurons in the brain not only encode the behavior of the animal,but also control the function of the body.Therefore,the analysis of neural network information processing mechanism corresponding to specific behavior and function has become a new research hotspot in the field of neuroscience and control science,which can help to deepen the understanding of brain specific function and mechanism.This paper sloves the problem of neural information decoding for animal goal-directed decision behavior.the neural signal of nidopallium caudolaterale was recorded by the implantable microelectrode array,and the local field of nidopallium caudolaterale was recorded by pigeon as the model animal.The coding frequency band of decision behavior is determined by the frequency band characteristic of different frequency bands of local field potential.Then,the gamma subband function network is constructed by the synchronization likelihood algorithm.The topological characteristics of the gamma subband function network is analyzed to determine the decoding time window for pigeon's turning behavior.Lastly,the network feature of the gamma subband functional network is extracted,and the decoding algorithm combining leave one out with k-nearest neightor is used to decode the turning behavior of the pigeons.The work done in this thesis and the results obtained are summarized as follows:1)Using the multi-window overlapping power spectrum estimation and wavelet transform technology,local field potential signal in the process of pigeon turning is analyzed from three aspects of time domain,frequency domain and time-frequency domain to determine characteristic band of local field potential signal related to the pigeon's turning behavior.It was found that the energy of the gamma subband(40 ~ 60 Hz)in the turning zone was significantly increased compared with the waiting zone,while the other subbands did not change significantly.The result indicates that the local field potential signal gamma subband encodes the pigeon's turning behavior.2)Using the synchronization likelihood analysis algorithm,the local field potential signal gamma subband functional network is constructed and the topological characteristics of the network are measured and analyzed.Contrast analysis of function network topological characteristics of gamma subband of the waiting area and turning area,different directions and before and after turning,found the network clustering coefficient and global efficiency of the turning zone are significantly higher than those of the waiting area,and the difference of the network characteristics of different directions in the turning area is also obvious.Further studies have found that clustering coefficient and the global efficiency after the turning are significantly higher than those before the turning,which indicates that the gamma subband function network after the turning contains the information of the pigeon's turning behavior.3)Using the connection strength value of the network and principal component dimension reduction technique extracts the feature of the gamma subband functional network.the turning behavior of the pigeon is decoded by LOO-kNN decoding algorithm.The decoding accuracy of network feature and energy feature is compared and analyzed.The results show that the decoding accuracy of network feature(0.74 ± 0.08)is higher than that of energy feature(0.61 ± 0.12).At the same time,the dynamic characteristic analysis of the decoding accuracy of gamma subband network feature in the process of pigeon turning finds the decoding accuracy of different pigeons is mostly after the turning,which indicates that the goal may play a key role in the process of pigeon turning. |
PDF Full Text Request