Hierarchical Bayesian Modeling of Autonomic Responses during the Concealed Information Test

Abstract

The autonomic-based concealed information test continuously measures physiological indices such as skin conductance, normalized pulse volume, heart rate, and respiration. This study proposes a hierarchical Bayesian model to analyze these physiological responses’ time series data. The proposed approach uses a state-space model that comprises observation equations with truncated normal distributions and state equations with hierarchical structures. We applied the proposed method to experimental data from 167 participants to illustrate the technique’s efficacy. This method allows us to visualize the magnitude of typical physiological responses in their respective units of measurement and individual differences, which had previously been overlooked in conventional analysis that requires standardization of observed data. Furthermore, this hierarchical state-space model can be easily adapted to various experimental designs.