Effects Of Aging And Anisometropic Amblyopia On Neural Network Information Encoding Of Visual Cortex

As the most well-studied sensory system,visual system is an important window to explore the coding mode of neural networks.Here,we use aging and amblyopia as disease models,respectively,to explore the processing mechanism of the visual cortex neural network to noise in aging and bistable motion perception.Durring the senescent of visual cortex,the neuron response characteristics changed significantly,with the decreased signal-to-noise ratio(SNR)and increased noise level which caused by the increasing response fluctuations,indicated a decline of information processing of aing neurons.In order to resist the loss of information caused by the increased response variability of a single neuron,neural networks need some mechanisms to reduce the negative effects of response variability.One is the information redundancy effect,the other is weighted averaged processing of information.The redundancy effect can reduce the negative impact of the variability of single neuron response on the information carried by the neuron population by increasing the backup of information in the individual.So is the redundancy effect in the aging cortical neural network as effective as young animals?Through mutual information calculation,we found that the population of macaque secondary visual cortex(V2)neurons showed a significant reduction in redundancy during the aging process,making the neural network unable to resist the influence of noise,and then showing a decline in the group signal-to-noise ratio.We further studied the reasons for decreased encoding.By calculating the synaptic connections between neurons,we found that the decrease in the similarity of neuronal responses and their contribution to information redundancy which caused by the decrease in synaptic transmission efficiency are important reasons for the decline of the redundancy effect of the neural network of the aging visual cortex.This result provides a new perspective for the decline in information processing capacity of the aging brain.The neural network integrates the information carried by a single neuron through a weighted average method,so that the response variability of a single neuron can be eliminated by an average method,but the existence of noise correlation(Rsc)weakens the effect of noise elimination.Our previous work showed that the Rsc between neurons in the visual cortex of the old macaques was significantly higher than that of the young,and the increase in Rsc showed a clear relationship with the decline of inhibitory neuron function.But how do inhibitory neurons regulate Rsc?We further used the PV-/SST-/VIP-Cre mouse to suppresse the specific neurons in visual cortex respectively,and then detected the performance of Rsc of neuron population.The results show that suppressing both PV+ and SST+neurons can cause the increase of Rsc of the mouse visual cortex.By isolating PV+inhibitory neurons based on waveform characteristics,the further founded decreasing of decorrelation suggested that this subtype interneuron regulated the network noise by directly acting on the pyramidal neurons,i.e.by decorrelation effect,while VIP+neuron showed a reversed influence,which may be related to the de-inhibition of VIP+neurons in the cortical network.Through data modeling,we respectively explored the effects of inhibition and de-inhibition on Rsc,and the simulation results are consistant with the electrophysiological data we recorded.These results clarified the mechanism of Rsc modulation by inhibitory system,and further clarified the specific role of different subtypes of interneurons,also showed us that the decrease of the GABAergic system during the aging would lead to an increase in cortical noise,so as a decrease in noise modulation decoding,provided a new entry point to deeply understand of the neural mechanism of the aging brain.Besides natural aging,disease also affects the neural coding of the cortical network.Anisometropic amblyopia is a typical visual development disease which can be used as a good model to investigate the visual processing mechanism.We compared the performance of amblyopes and nerurotype normal subjects when observing bistable motion stimulation.By adjusting the contrast to change the noise level of the cortical network,we found that the bistability perception of amblyopes did not show a systematic change.Further using the bistable neural network model to verify it,we found that the noise level did significantly affect the transition frequency and duration of the percept in bistable perception.These results were consistent with what we observed in the control group,suggested that the amblyopes may may have an abnormal coding pattern for changes in internal noise when processing the bistable motion percption.We explore the information coding of neural network on two different models of aging and anisometropia amblyopia,and the results show that both of them can affect the information processing of visual cortex,and the processing of noise may play a certain role in it.