Lies divorced from context: evidence for Context Embedded Perception (CEP) as a feasible measure for deception detection

ABSTRACT

This study examined the efficacy of a novel measure for lie detection—Context Embedded Perception (CEP). CEP exploits a qualitative difference between the functional roles of provided perceptual and contextual details: By framing the perceptual details in time and space, or by annotating them semantically, the contextual details are those that confer upon the perceptual details the status of verifiability. Consequently, expanding on the tenets of the Verifiability Approach (VA), liars, who are generally motivated to adopt an ‘unverifiable-information’ reporting strategy, should also tend to avoid such contextualization. The proportion of contextual details can then be used to differentiate deceptive from truthful accounts. Participants provided deceptive or truthful accounts during a mock-crime interview; Half were warned beforehand about an upcoming verification procedure and ratification interview. For each account, the CEP measure was calculated as the proportion of contextual details out of contextual and perceptual details combined. As predicted, CEP scores were significantly lower for liars than for truthtellers; warning condition had no effect. A CEP-based discriminant function correctly identified 92% of the truthtellers and 66% of the liars, resulting in 79% total classification accuracy. CEP’s theoretical and practical strengths and weaknesses, specifically in relation to sources of verifiability, are discussed.

KEYWORDS:

• Verifiability approach
• Context Embedded Perception
• detection of deception
• verbal lie detection
• output-bound measures

Disclosure statement

No potential conflict of interest was reported by the author(s).

Notes

1 Although the potential effect of the warning manipulation on the CEP measure did not require that participants actually be called back for a second interview, we were concerned that not doing so would make the ‘deception’ regarding the expected additional ‘ratification’ interview immediately apparent, which might then be ‘leaked’ to later participants. We acknowledge, however, that this delay could potentially affect the validity of the warned participants’ responses to the post-interview questionnaire as well as the comparison of the responses made by warned and unwarned participants.

2 One female participant in the warned liars condition did not complete the post-interview questionnaire, and is therefore not included in the analyses of the questionnaire responses.

Additional information

Funding

This research was supported by the Israel Science Foundation grant to Galit Nahari (grant number 372/14).

This paper is based on the first author’s thesis, submitted to the University of Haifa in partial fulfillment of the requirements toward the M.A. degree. The study was conducted at Bar-Ilan University.