Flashing Light Recognition On Free-moving Rats

Research on external stimulus dependence of activity in the brain is a meaningful field to understand central nervous system (CNS). Through observation of the way brain reacts to external stimulus, we could explore in depth how brain extracts biologically relevant information from sensory inputs, study the distribution of functional brain regions, neural circuit and network, and recognize the cluster of different neuron types. On the basis of well understanding of the sensory perception of the brain, it is possible to conduct perception information related behave control of experimental animals, or perform sensory information feedback through deep brain stimulation (DBS), intra-cortical micro-stimulation (ICMS), and optogenetics to implement simulation of perception.Meanwhile, the novel method brain machine interfaces (BMIs) makes it possible to read out neural signal on free-moving experimental animals, and combine with machine learning algorithm to separate different kinds of perception of CNS, further uncover inner information of brain signal. A variety of experiment methods have been conducted to explain the mechanism of flashing light evoked neural activity. However, much still remains to be learned about how the brain represents frequency dependence of visual experience.In this paper, we chose rats as experimental subjects, and set the experimental purpose to be extracting visual related neural signal from free-moving rats, and used machine learning algorithm to inference the type of flashing light with neural signal recorded, and built an integrated system. The main content of the studies are:(1) Analyzed the visual pathway of rats, chose primary visual cortex as the target region and invasive micro-electrodes array as chronic fixed BMI, chose flashing light as visual stimulation which proved be correctly distinguished by rats through behave experiment.(2) Collected and analyzed pattern the Spike signal under high and low frequency flashing light stimulation. Through correlation between external cues and neural signal, we observed positive linear correlation on above 90% channels. Statistical results showed, significant difference between high and low stimulus. The features of Spike signal were classified with support vector machines (SVMs) algorithm to build a model. The feasibility and stability of the recognition model were tested in multi circumstance.(3) Furthermore, we investigated the Spike and local field potential (LFP) signal in primary visual cortex in rats under UV flashing light stimulation. We also acquire robust increasing of Spike signals and stimulus-related firing rate under UV flashing light stimulation. Meanwhile, low frequency LFP signal under 50 Hz also shows increment of power spectral density (PSD) in certain frequency band.In the end, we delineated the online flashing light recognition which can be further connected with electric stimulus based rat robot navigation system. The integration of those two systems can combine kinetic intelligence and perceptual intelligence of rat robots, and substitute the artificial environment information collecting process with the own perception of experiment animal, which contribute to a higher integration of biological intelligence.