A Parallel Neural Circuit For Decision Making Ahead Of The Firing Gun

Decision making is one of the basic advanced cognitive behaviors.Human and animals have to make a choice all the time in their existence and development.Growing evidence develops a more profound and comprehensive undstanding of decision making.Much data has been obtained based on the experiments of awake macaque monkeys,fruit flies and other animals with simple structures.Reareaches on dicision making are carried out from three level which contains micro level,macro level and mesoscale in between,and in the meanwhile have gone through the process from qualitative description to quantitative calculation.Such a situation has greatly advanced the science of Computational Neuroscience.Wang's group has put forward an attractor network for decision making which is seen as an important breakthrough.They set forth the mechanism of decision making from the level of neural circuits and realize the implementation of computability on the cognitive process,providing a plausible computational framwork and theoretical basis.We design a neural circuit based on the competitive relationship between SEF and LIP regions to explain monkeys' early dicisions with the theory and techniques of Computational Neuroscience.LIF model is used to construct the neural network which is composed of different neuron groups.The system consists of a large number of parameters leading to complex dynamic phenomenons.The model is considered reasonable to a certain extent for it is able to reproduce the statistical characteristics of the experimental data with suitable parameters,which is bimodal distribution namely.We propose a computational model to explain the formation of bimodal distribution of monkeys' reaction time.The paper focuses on the temporal attribute in decision making on mesoscopic scale or specifically at the level of neural circuits.Excitatory neuron populations in SEF and LIP regions encode the time information,through the competition in the rate of ramping activity.The competitive relationship above is the key to reproduce the bimodal distribution and early decisions before a starting gun.Our result supports the idea that time information is encoded in the ramping activity of neuron groups,which express the process of time estimation.Raming activity,as an electrophysiologic behaviour,plays a very significant role to understand time perception.This article is divided into five chapters.Chapter 1,as an introduction,mainly gives a summary of the background and methods of our work.This part introduces the development of Computational Neuroscience,its main problems and models for time perception.Chapter 2 describes the attractor network for decision making,a biophysical model from the view of neural computation,which provides a theoretical basis for our study.Main work is illustrated in Chapter 3 and Chapter 4,in which we give a very brief description of the experimental paradigm and results,and make a detailed explanation of the design and construction of the parallel neural circuit with two competitive regions.Kinetic equations are established to simulate the bimodal distribution observed in the real experiment.At last in the summary and outlook,we sum up the main dynamic phenomenons of the neural system and the key mechanism for bimodal distribution,and explore how to improve our work in the future.