Analysis of Correlation in Neural Responses across Multiple Subjects or Trials during Decision-making for Newsvendor Problem

Abstract

To understand how correlated the human brain is across multiple subjects or multiple trials when people make risky decisions, economical decision-making based on the Newsvendor Problems (NP) were investigated with 64-channel electro-encephalography (EEG) taken from 23 human subjects. Specifically, in this work, we analyzed the EEG signals recorded while subjects were playing NP decision-making tasks. Two groups of subjects were assigned to take NP with distinctive difficulty levels for examining the difference in neural responses to the task challenge. The Correlated Components Analysis (CorrCA) method was then utilized to identify the EEG channels that maximized the correlations among multiple subjects or trials. Both Inter-Subject Correlation (ISC) and Inter-Trial Correlation (ITC) results revealed an apparent modulation of the attentional state of subjects across three test phases (decision-making, feedback, and recovery) and two distinct task challenges. The neural correlation across multiple subjects or multiple trials increased significantly during the decision-making phase for more challenging tasks, in which subjects needed to be more attentive. Such observations confirm the modulation of the engagement state on the correlations of the neural activity across multiple subjects or repeated trials. Moreover, the alpha-band power results for different decision-making phases affirmed that the difficulty of different test phases modulates subjects’ engagement.