Abstract
A polygraph examiner is required to make accurate judgments as to whether examinee's autonomic responses to a critical item (the item that a perpetrator should know) differ from those to non-critical items. Several statistical methods have been proposed to support this decision making. These methods are classified into the database-based approach and the within-individual approach. Recently, these approaches have been improved for dealing with individual differences in autonomic responses. The latent class discrimination method has been developed as a new database-based approach and the hidden Markov discrimination method has been formulated as a new within-individual approach. These new methods were applied to experimental polygraph data obtained from 34 participants; higher discrimination performances were seen with the new methods than with the conventional methods. In future studies, it is expected that the database-based and within-individual approaches will be merged.
