Deception detection based on repeated interrogations

This study examined how repeated interrogations affect deception detection performance. Twenty-four witnesses (12 liars and 12 truth-tellers) were interrogated three times over 11 days about a staged event, with veracity assessed by interrogators and video observers. Both groups performed poorly at detecting deception, with consistency emerging as the most commonly used but least accurate cue.

Twenty-four undergraduate witnesses viewed a staged event and were interrogated three times (after 3 hours, 4 days, and 11 days). Veracity judgments were made by face-to-face interrogators and 144 video observers, allowing comparison across presentation modes and information levels.

Background & Context

This pioneering 2001 study by Granhag and Strömwall addressed a critical gap in deception research: the fact that suspects are often subjected to multiple interrogations, which research had neglected. In real-world criminal investigations, witnesses and suspects are rarely interviewed only once, yet virtually all deception detection research at the time examined single interactions.

The study also tested the previous finding that interrogators interacting face-to-face tend to be more credulous toward suspects than observers watching the same suspects on video—what researchers call the "honesty bias." This work laid the foundation for what would become known as the "repeat versus reconstruct" hypothesis and the "consistency heuristic," concepts that profoundly shaped subsequent research on statement consistency and credibility assessment in forensic contexts.

Research Design & Methodology

After seeing a staged event, 12 lying and 12 truth-telling witnesses were interrogated three times over a period of 11 days. Half the participants were instructed to tell the truth about what they witnessed; the other half were instructed to lie, fabricating a version of events that portrayed the victim as the perpetrator.

The interrogation schedule consisted of three sessions: after 3 hours, after 4 days, and after 11 days. After the final interrogation, the veracity of each witness was assessed by his or her interrogator and by six observers who had seen the interrogations on video. In all, 144 observers were used, and all participants were undergraduate students. This design allowed the researchers to compare detection accuracy across presentation modes (live vs. video) and across exposure levels (one vs. multiple interrogations).

Results & Key Findings

Both interrogators and observers performed poorly at distinguishing liars from truth-tellers, with accuracy rates near chance levels. The main findings revealed troubling patterns about how people assess credibility over repeated interviews:

• Lie-catchers given access to consecutive statements from one suspect did not perform better than lie-catchers given access to one statement only—more information did not improve accuracy
• Access to repeated interrogations inflated the lie-catchers' confidence in the correctness of their veracity judgment, which had a detrimental effect on the accuracy-confidence relation
• Consistency over time was by far the most commonly used cue for justifying veracity judgments
• The predictive accuracy for this cue was alarmingly low

Interestingly, interrogators used verbal cues to a significantly greater extent than did the observers—a finding that contradicted previous research and suggested that direct interaction shifts attention to what suspects say rather than how they appear.

Discussion & Significance

This study introduced the critical concept of the "consistency heuristic"—the assumption that inconsistency implies deception and consistency implies truth. The data demonstrated that people rely heavily on this heuristic when evaluating repeated statements, despite its poor diagnostic value. This finding has profound implications: the very cue that legal professionals and laypersons trust most is essentially unreliable for distinguishing truth from deception.

The confidence-inflation effect is particularly concerning for criminal justice. When investigators have access to multiple interrogations, they become more confident in their judgments without becoming more accurate. This creates a dangerous illusion of validity that can lead to wrongful convictions when confident but inaccurate assessments are presented in court.

The findings support what became known as the "repeat versus reconstruct" hypothesis: liars, in an explicit attempt to maintain consistency, will repeat rehearsed answers in different interviews and across time, while truth-tellers reconstruct memories and may show more variation. This challenges the conventional wisdom that consistency equals truthfulness.

Limitations & Considerations

The study used undergraduate student participants rather than actual suspects or criminal investigators, which limits ecological validity. The mock crime scenario, while controlled, lacks the emotional stakes and consequences of real criminal interrogations. The 11-day time span, though extending beyond typical laboratory studies, is still compressed compared to many real investigations that unfold over weeks or months.

Additionally, all participants knew they were in a research study, which may have influenced both the quality of deception and the seriousness with which interrogators approached their task. The staged event was relatively simple, unlike complex criminal events that suspects must account for in actual cases.

Practical Applications

This research carries urgent warnings for criminal investigators, prosecutors, and judges. Reliance on consistency as a marker of truthfulness is demonstrably flawed—sophisticated liars can maintain consistency by rehearsing their stories, while honest witnesses may show natural variation as they reconstruct genuine memories over time. Training programs should discourage overreliance on the consistency heuristic and emphasize strategic interviewing techniques that test memory and knowledge in ways that are difficult for liars to anticipate or rehearse.

The confidence-inflation finding suggests that procedures involving multiple interviews should include safeguards against overconfidence bias. Decision-makers should be made aware that repeated exposure to a suspect does not enhance accuracy and may create false certainty. This work laid the foundation for evidence-based interviewing approaches like the Strategic Use of Evidence (SUE) technique, which leverages information asymmetry rather than relying on behavioral cues or consistency judgments.