Research On Decoding Of Pigeon Route Selection Based On Multi-scale Neural Information Features Of Hp Area

The brain must choose among multiple behavioral goal options to support highly adaptive behaviors of the intelligent agent,a recent research trend is to explore the mechanism of goal-directed selection.The hippocampus(Hp)is a key participant in goal-directed decision-making tasks,and is believed to dominate the neural information encoding of the current location and goal.However,the current research seldom involves the characterization of the route selection from the current location to the goal by the neural information of the Hp area.Therefore,this study takes pigeons with excellent spatial navigation ability as the model animal,designed a goal-directed route selection behavior experimental paradigm,and simultaneously collected multi-channel neural signals in the Hp area during the experiments.Then,the neural information encoding characteristics of pigeons' Hp area on route selection were analyzed based on the spike firing characteristics,the time-frequency characteristics of local field potential(LFP)and the functional network,and the route selection behavior was decoded based on the neural information features at different scales.Finally,the effectiveness of fusing features of different scales to improve the decoding performance of route selection was explored.The main research work of this thesis includes:(1)The experimental design and data collection of the pigeon goal-directed route selection task were completed.Firstly,designed the behavior paradigm of route selection and completed the experimental platform.Then,spike and LFP signals in Hp area of pigeons were collected when they completed the goal-directed route selection task.Finally,the collected neural signals were preprocessed,and the behavioral data of the pigeons to complete the route selection task were statistically analyzed.(2)The multi-scale neural information coding characteristics of pigeon Hp area on route selection behavior were analyzed.First,based on the dynamic change law of the spike firing rate in the route selection task,it was found that there are significant differences in the average spike firing rate under different route selection behaviors.Then,based on the spectral analysis of the LFP signal,the response frequency band related to the route selection behavior was determined and the time-frequency analysis results showed that the time-frequency energy characteristics in the fast-gamma band encoded the different route selection behavior of the pigeon.Finally,the local brain function network of the multi-channel spike sequence and LFP signal in the Hp area was constructed respectively,and the coding characteristics of its topological characteristics on different route selection behaviors were analyzed.(3)The route selection decoding research based on the neural information features of different scales in the pigeon Hp area were completed.Firstly,the features of multi-channel spike firing rate,LFP signal time-frequency energy features and functional network connection features were extracted.Then,dimensionality reduction was performed through feature selection and fused with k-nearest neighbor classification to achieve route selection decoding based on neural information features of different scales.Finally,the multi-scale neural information features of the Hp area were fused to carry out joint decoding research,the results showed that the fusion of neural information features of different scales can significantly improve the decoding accuracy of route selection behaviors,which provides a new idea for subsequent neural information decoding research.