Neurocognitive Mechanisms Of Probability Information In Bayesian Reasoning

In everyday life, people often encounter many uncertain informations(The informations have probabilistic nature), According new informations or evidences to adjustment their existing views to draw the appropriate conclusions and make a decision. For example,"The meteorological department make weather forecasts based on the climate data","Doctors make a diagnosis based on the inspection results"," Homebuyers to make a purchase decision based on the market informations", and so on, These process called Bayesian reasoning.Upon until now, the behavioral results of Bayesian reasoning have much controversy. The debate mainly focuses on two aspects:(1) In Bayesian reasoning, whether or not people follow Bayes’theorem;(2) What causes deviation in Bayesian reasoning? The core of the problem is that, how do the two key probabilities, base-rate and hit-rate, work in Bayesian reasoning. Prior studies mainly make conclusion based on comparison between posteriori probability estimation value and the value calculated by Bayes’ rule. Even though some studies discuss the reasoning process and response, the conclusion backward reasons the reasoning process based on reasoning results, and thus lack support of the real-time data. Based on the Previous researehes,the present study aimed at provide an electrophysiological evidence of Bayesian reasoning’s time process and neural mechanism.In experiment1, using three factors mixed design, examined that the effect of the base-rate, hit-rate,and time Pressure on Bayesian reasoning.Results show that both base-rate and hit-rate affect the posteriori probability estimation value significantly.In experiment2, using two factors within-subjects design, examined that the effect of the base-rate and hit-rate on Bayesian reasoning.This is the first time we provide an electrophysiological evidence of Bayesian reasoning’s time process and neural mechanism. ERP results show that, comparing to high hit-rate reasoning tasks, low hit-rate tasks elicit a more obvious Nl (100-200ms) on the frontal and temporal lobe and a more N300(250-350ms) on the frontal-parietal, parietal and parietal-occipital lobe; comparing to low hit-rate reasoning tasks, high hit-rate tasks elicit a more obvious advanced stage positive component LPC (350-700ms).In conclusion, our experiments show that in the process of Bayesian inference, people do not follow the Bayes’ theorem, but use the anchor-adjustment heuristic. The cause of the reasoning deviation is not ignoring the base-rate information, but the cognitive deviation and the different cognitive function of left and right brain in the anchor-adjustment heuristic. At the same time, It also shows that the reasoning process mainly activates the frontal lobe, temporal lobe and parietal lobe, which is consistent with the deductive reasoning.