Decoding Visual Luminance From Different Color Backgrounds In The Optic Tectum Neurons Of Pigeon

Color and luminance are two basic properties of visual information.It is of great significance to study the decoding of color and luminance for understanding the processing mechanism of the nervous system,analizing the color modulation law toluminance information,and evaluating the stimulating effectofthe visual prosthesis implanted brain-computer interface.Birds have developed visual system and excellent color perception,becoming the ideal materials for visual research.The Optic Tectum is the key to the birds' visual information processing nerve nuclei,which receives the retinal ganglion cells axons message before modulating,coding,integrating and other processing.This thesis mainly reserched the luminance information decoding of neurons from OT area underdifferent color backgrounds.The luminancestimulation experiments under different color backgrounds were designed.Microelectrode arrays were used to test the cluster response signalsof OT area neurons.The decoding algorithms weredesigned toexplore the modulation function of color background on luminance decoding.In this thesis,the main work and research results are as follows:(1)The neuron receptive field position of OT aera was determined.According to the types of color-sensitive cells in the retina of the pigeon,checkerboard stimulation patterns differed in red,green,blue,black,and white were designed.Using reverse correlation method analysis for Spike sequence fired from the OT area neurons,the results found that the positions and extents of the neuronal receptive fields determined from different backgrounds were closed,coinciding in a wide range.The center coordinates of the receptive fields determined under black and white stimuli almost overlaped together.The receptive field was wider under green background,whereas narrower under red background.(2)The luminance informationunder different color backgroundswasdecoded.First,the six equal colors of the Hue ring were selected.Thosewere H0,H16,H33,H49,H66 and H82 corresponding colors which were used as the backgrounds of stimulus modes.After extracting theSpike issuance rate characteristics,a multivariate linear inverse filter andArtificial Neural Network wereconstructed respectively to decodethe luminance information from different colorbackgrounds.The correlation coefficient ofluminance between the actualand estimated value was calculatedas an index to measure the effect of decoding.The results showed that:(a)thecorrelation curve of the H66 corresponding background color was always at the bottom of all these curves,showing that the H66 corresponding background color weakened the decoding of luminance;(b)the correlation curves of some colorsadjacented in the hue ring showed very close to each other,such as yellow and green,red and magenta.(3)The influence of the four parameters on the experimental results was explored.The neurons cluster numbers,the size of time window,the responseduration time after stimulus,the start time after after stimulus were changed to compared the influence on luminance stimulus decoding.The results showed that neurons cluster numbers participated in the decodingwithin 3-10,the division of time window number less than 10,thestimulation from starting to 30 ms-150 ms,after thestimulation start time of 20 ms,resulting in thestimulation havinga higher decoding accuracy.The results of the orthogonal experimentsshowed that:the decreasing sequence among the four factor levelswas not found;there existed a strong interaction among the four factors.Combiningeach factor level that had a highcorrelation in the single factor experiment may result in adecreasing correlation.