Exploiting liars' verbal strategies by examining the verifiability of details

This 2014 study by Nahari, Vrij, and Fisher introduced the Verifiability Approach (VA) to deception detection, testing the hypothesis that liars strategically avoid mentioning details that can be checked by investigators. Using written statements from 38 participants describing truthful or fabricated accounts of a 30-minute period, the study found that liars included significantly fewer verifiable details than truth tellers, yielding classification accuracy as high as 79% — substantially outperforming methods that count all details indiscriminately.

A laboratory analog study in which 38 participants produced written statements (truthful or deceptive) about real activities from a recent 30-minute period; two coders classified each detail as verifiable or unverifiable, and classification accuracy was assessed across three analytic approaches.

The study introduced the Verifiability Approach as a breakthrough in verbal deception detection, addressing the strategic dilemma liars face between appearing credible through detailed accounts while avoiding verifiable claims that could expose their deceit. Participants provided written statements about recent activities under truth-telling or lying conditions, with researchers coding details as either verifiable (checkable through witnesses, CCTV, phone records, etc.) or unverifiable. Results showed truth tellers included significantly more verifiable details than liars, with classification accuracy assessed using verifiable detail counts, ratios of verifiable-to-unverifiable details, and combined metrics. This approach represented a paradigm shift from focusing on quantity or quality of details to examining their investigative checkability.

The Verifiability Approach offers investigators a practical framework for statement analysis that exploits liars' reluctance to provide checkable details, potentially improving accuracy in criminal investigations and providing a more systematic approach to verbal deception detection than traditional content analysis methods.

Background & Context

For decades, researchers studying deception detection have grappled with a fundamental puzzle: liars know they need to appear credible, and credibility is often associated with providing rich, detailed accounts. This creates a strategic dilemma at the heart of deceptive communication. Truth tellers are typically willing and able to provide a narrative rich in detail, whereas liars face a tension — on one hand, they are reluctant to talk, as anything they say could be used to uncover their deceit, yet on the other hand, they realize they must provide a detailed account to come across as truthful.

The 2014 landmark study by Nahari, Vrij, and Fisher introduced a novel framework — the Verifiability Approach (VA) — to exploit precisely this strategic tension. The Verifiability Approach presents a new content-based tool for distinguishing between truths and lies, built on the finding that truth tellers provide more verifiable details than liars, such that the verifiability level of an interviewee's account can be used as an indicator of veracity.

This work filled a critical gap in verbal lie detection science. Prior approaches — such as Reality Monitoring and Criteria-Based Content Analysis — focused on the sheer quantity or perceptual quality of details. Since the introduction of the verbal channel into deception detection, it has become a rapidly growing area of research, operating on the basic assumption that liars differ from truth tellers in their verbal behaviour, making it possible to classify them by inspecting their verbal accounts. The Nahari et al. study redirected focus from *how many* details are given to *what kind* — specifically, whether those details are checkable by an investigator.

Research Design & Methodology

The study was published in *Legal and Criminological Psychology*, Volume 19, Issue 2 (September 2014), pages 227–239, authored by Galit Nahari of Bar-Ilan University's Department of Criminology, Aldert Vrij of the University of Portsmouth, and Ronald P. Fisher of Florida International University. A total of 38 participants wrote a statement in which they told the truth or lied about their activities during a recent 30-minute period.

The core methodological innovation was in what participants were asked to describe and how those descriptions were coded. Two coders counted the frequency of occurrence of details that can be verified and that cannot be verified. The operationalization of "verifiable" details was carefully defined to include real-world, investigator-checkable evidence.

Key design elements of the study included:

• Participants: 38 individuals assigned to truth-telling or lying conditions about genuine recent activities
• Task: Written statement production describing activities from a real 30-minute period
• Coding scheme: Checkable details included activities someone carried out with a named person who the interviewer could consult, activities witnessed by identifiable persons, activities that may have been captured on CCTV, and activities that leave a trace such as mobile phone calls, texts, or debit/credit card purchases.
• Measures: Frequency of verifiable details, frequency of unverifiable details, and the ratio between the two
• Classification metric: Accuracy of truth/lie classification using each measure independently and in combination

The research compared three distinct classification strategies: using only verifiable detail counts, using only the ratio of verifiable-to-unverifiable details, and using the combined total of both. This multi-metric approach allowed the authors to identify which signal was most diagnostically powerful for deception detection.

Results & Key Findings

The headline finding: liars strategically suppressed verifiable details while maintaining comparable levels of unverifiable details — a pattern that produced a distinctively different verifiability ratio between the two groups. Liars, compared with truth tellers, included fewer details that can be verified and an equal number of details that cannot be verified in their statements, and the ratio between verifiable and unverifiable details was smaller in liars compared with truth tellers.

Classification accuracy varied meaningfully across the three analytic approaches:

• 79% correct classification based on the frequency count of verifiable details alone — the single strongest individual predictor
• 71% correct classification based on the ratio of verifiable to unverifiable details
• 63% correct classification when using the combined total of verifiable and unverifiable details

High percentages of truth tellers and liars were classified correctly based on the frequency counting of verifiable details (79%) or the ratio between verifiable and unverifiable details (71%). Those percentages were higher than the percentage that could be classified correctly (63%) based on verifiable and unverifiable detail combined. This last finding is particularly significant: lumping both types of details together — a method analogous to older detail-quantity approaches — actually *hurt* accuracy compared to examining verifiable details in isolation.

A key subsidiary finding concerns what liars did *not* do: they did not substantially inflate their unverifiable detail counts as an alternative credibility strategy. The implied liar strategy is to provide as many details as possible, but avoid the verifiable details — specifically, activities leaving physical or digital traces. This suggests that liars' primary strategy was subtraction (omitting checkable facts) rather than substitution (replacing them with fabricated unverifiable content), at least in this baseline, non-forewarned condition.

Discussion & Significance

The Nahari et al. (2014) paper is widely regarded as the foundational empirical demonstration of the Verifiability Approach as a standalone deception detection tool. The Verifiability Approach is a promising new lie detection approach that has since spawned a substantial body of follow-on research across multiple real-world settings. Its theoretical elegance lies in exploiting a dilemma that liars cannot easily escape: the more details someone provides, the more likely they are to be believed; on the other hand, liars do not wish to mention too many details because the more they provide, the more opportunity they give investigators to check those details and discover their lies — the strategic solution being to provide details that cannot be verified.

The study also made an important theoretical contribution by comparing the Verifiability Approach to competing explanations for why truth tellers and liars differ in detail quantity. The authors compared their verifiability approach with other theoretical approaches as to why differences in detail between truth tellers and liars emerge. Unlike Reality Monitoring (which attributes detail differences to memory processes) or cognitive load explanations, the VA locates the mechanism in deliberate strategic choice — liars *choose* to omit verifiable details.

Subsequent research confirmed and expanded on these foundations. Studies consistently showed liars provide fewer verifiable details than truth tellers, and a meta-analysis confirmed the VA's potential but called for more field studies and replications. The approach has since been tested in insurance claim settings, airport security screening, alibi assessment, and malingering detection — validating its breadth of applicability across investigative contexts.

Limitations & Considerations

Several important methodological and generalizability limitations should be noted when interpreting these findings:

• Small sample size: With only 38 participants, the study was underpowered for detecting small-to-moderate effects with precision. Subsequent replications found effect sizes notably smaller than the original (e.g., a replication effect size of d = 0.49 was less than half that of the original d = 1.14).
• Laboratory analog design: The use of a controlled written-statement paradigm with low-stakes consequences limits ecological validity compared to real criminal investigations where stakes are high.
• No forewarning condition: The baseline design did not test whether liars would adapt their strategy if told investigators would check their details — a variable later shown to be critical.
• Researcher overlap: There is a lack of independent VA research; a review found that all 13 published reports on the VA at that time had at least one of the three original authors (Nahari, Vrij, or Fisher) as co-author.
• Mixed replication outcomes: Results of empirical studies on VA have been mixed, with some studies finding no effect and others finding only small effects.

Additionally, the effectiveness of the VA depends on the context having sufficient opportunity for verifiable details to exist. In situations where even truth tellers have few verifiable traces of their activities — for example, when events occurred long ago or in private settings — the diagnostic power of this approach may be substantially reduced.

Practical Applications

The Verifiability Approach offers investigators a concrete, actionable framework for statement analysis that does not require physiological measurement equipment. Verifiable details include activities carried out with identifiable or named persons whom the interviewer can consult, activities witnessed by identifiable persons, activities believed to have been captured on CCTV, and activities that may have been recorded and documented, such as using debit cards, mobile phones, or computers. Training investigators to code statements along these dimensions — and to actively elicit verifiable information — gives them a structured, evidence-based tool for assessing credibility.

Importantly, while many lie detection tools are notably vulnerable to countermeasures, the VA would not just be resistant to but in fact benefit from informing the examinee what the tool is about — giving a forewarning would stimulate truth tellers, but not liars, to provide more verifiable details. This remarkable property makes the Verifiability Approach unusually practical: transparency about the method actively enhances its discriminative power, removing the ethical concerns associated with deceptive elicitation strategies and making it suitable for use across law enforcement, insurance fraud investigation, and security screening contexts.