Research On Neural Representation And Decoding During Goal-Directed Route Formation And Adjustment Of Pigeons

The execution of route from the current location to the goal location is one of the most important tasks in goal-directed spatial navigation.However,it is still unclear how the brain represents key information during route execution to support route formation and adjustment.To study the neural representation and decoding problems during goal-directed route formation and adjustment,this thesis designs a multi-route selection-based goal-directed behavioral experimental paradigm taking pigeons with excellent spatial cognition and navigation ability as model animals.The behavioral data and neural signals from multiple brain regions during route formation and adjustment are collected synchronously.Multimodal neural signal analysis means,complex network theory,data modelling methods and multi-scale neural decoding model are used in this thesis for studying the roles of pigeon’s key brain regions including hippocampus(Hp)and nidopallium caudolaterale(NCL),together with the local network of them in the process of route execution,to reveal the neural representation rules of Hp-NCL circuit during goal-directed route formation and adjustment.The innovative research results are summarized as follows:1.The neural dynamic representation rules of Hp-NCL circuit during route formation is revealed: Hp participates in the neural encoding during route formation through the specifically enhanced theta-band functional connectivity and NCL through the significantly reduced multi-band functional connectivity.The communication between Hp and NCL is realized through theta-mediated interaction mechanism.The behavioral data and neural representation patterns of Hp and NCL during stable preferred route formation of pigeon are analysed,indicating that the regular changes of time-frequency and functional connectivity characteristics can be used to represent the process of route formation.It is mainly manifested in the gradual decrease of time-frequency energy in both regions,the enhancement of theta-specific connectivity in Hp and the general decrease of connectivity in NCL.The dynamic representation role of connectivity characteristics in Hp-NCL network during route formation are explored.It is found that the route formation of pigeon correlates significantly with the enhanced Hp-NCL network connectivity in theta band,and the dominant direction of information flow is from NCL to Hp.Furthermore,the linear mixed-effects model is constructed,to relize the quantitative description of the relationship between theta band functional connectivity in Hp-NCL network and stable route-formation index,and the effective prediction of the route formation degree of the pigeon.2.The neural dynamic representation rules of Hp-NCL circuit during route adjustment is interpreted: the high-frequency activation and low-frequency inhibition of Hp functional connectivity,together with the high-frequency activation of NCL functional connectivity jointly represent the spatial cognitive dissonance.Hp and NCL realize the cross-region communication through cooperatively enhanced gamma interaction under the needs of route adjustment.The behavioral indexes of pigeons in the route adjustment experiment are analysed,showing that pigeons fall into spatial cognitive dissonance when dealing with the sudden change of the spatial environment in which the preferred route is blocked.The network connectivity patterns of both regions during the experiment are analysed,indicating that spatial cognitive dissonance correlates with the selective changes of frequency band-specific network connectivity in Hp and NCL.It is mainly manifested in the high-frequency activation and low-frequency inhibition of Hp connectivity,and the high-frequency activation of NCL connectivity.The Hp-NCL network connectivity patterns in different experimental phases are analysed,and the connection strength and information flow patterns in Hp-NCL network are found to change with dissonance behavior,mainly manifested in the cooperatively enhanced gamma interaction.The multivariate discriminant analysis of spatial cognitive states based on the above network connectivity characteristics is carried out to effectively distinguish the three cognitive states including before,during and after the dissonance.The variable analysis results confirm the specific contributions of Hp theta-band,NCL gamma-band and Hp-NCL network gamma-band connectivity characteristics in representing the differences among cognitive states.3.The neural information features on the scales of neuron,cluster and network in Hp-NCL circuit are extracted to construct the multi-scale neural decoding model based on feature fusion strategy.The model significantly improves the decoding performance,verifying the joint representation roles of multi-scale information on route selection behaviors.The joint representation roles of the neural information from different scales on goal-directed route selection behaviors of pigeon are analysed including spike firing rate,field potential time-frequency characteristic and network connectivity pattern in Hp-NCL circuit.The multi-scale neural information features in Hp,NCL and the local network of them are extracted respectively.A fast dimensionality reduction method for multimodal high-dimensional neural data is proposed to effectively solve the dimensionality reduction problem involved in the high-dimensional data set,which takes into account the removal of irrelevant information and redundant information.A multi-scale neural decoding model based on neural information feature fusion strategy is proposed,effectively integrating the neural information from different scales.The decoding performance of route selection behaviors based on single-scale features and multi-scale fused features is comprehensively compared and evaluated,indicating that the accuracy based on multi-scale fused features is significantly higher than those on other scales.The research in this thesis reveals the neural representation rules during goal-directed route formation and adjustment of pigeons,and realizes the route selection decoding by fusing multi-scale neural information features.It is of great value to deepen the understanding of the neural mechanism of pigeon’s goal-directed behavior and enrich the theoretical system of brain space navigation.Meanwhile,it is also of positive significance to promote the navigational behavior-related neural decoding researches and heuristic bionic navigation modelling.