Weighing votes in human-machine collaboration for hazard recognition: Inferring hazard perceptual threshold and decision confidence from electroencephalogram wavelets

Abstract

Purpose: Human-machine collaboration is a promising strategy to improve hazard inspection. However, research on the effective integration of opinions from humans with machines for optimal group decision making is lacking. Hence, considering the benefits of a brain-computer interface (BCI) to enable intuitive commutation, this study proposes a novel method to predict human hazard response choices and decision confidence from brain activities for a superior confidence-weighted voting strategy. Methodology: First, we developed a Bayesian inference-based algorithm to ascertain the decision threshold above which a hazard is reported from human brain signals. This method was tested empirically with electroencephalogram (EEG) data collected in a laboratory setting and cross-validated using behavioral indices of the signal detection theory. Subsequently, based on numerical simulations, the decision criteria for low-, medium-, and high-confidence level differentiations characterized by parietal alpha-band EEG power were determined. Findings : The investigated hazard recognition task was described as a process of probabilistic inference involving a decision uncertainty evaluation. The results demonstrated the feasibility of EEG measurements in observing human internal representations of hazard discrimination. Moreover, the optimal criteria to differentiate between low-, medium-, and high-confidence levels were obtained by benchmarking against an optimal Bayesian observer. Originality: This research demonstrates the potential of a BCI as an effective channel for telecommunication, laying the foundation for the design of future hazard detection techniques in the collaborative human-machine systems research field.