Credibility vs Reliability: Why Believing a Witness Is Not the Same as Being Right

Every day, professionals make decisions about whether to believe what someone tells them. Solicitors assess clients. Police officers interview suspects. Employers investigate complaints. Jurors weigh testimony. In every one of these situations, a quiet, automatic process takes hold: we form an impression. We decide, often within seconds, whether the person in front of us seems honest. This instinct feels reliable. It rarely is.

A landmark paper published in Psychology, Public Policy, and Law in 2026 by Brennen, Blandhol, and Magnussen sets out a model that should give every investigator, lawyer, and examiner pause. The paper argues that legal systems routinely conflate three fundamentally different concepts: credibility, reliability, and factual truth. Getting them confused has serious consequences — not least wrongful convictions, failed investigations, and flawed workplace decisions.

At The Centre for Forensic Neuroscience, our work in investigative psychology consistently reinforces this point. Credibility is not a synonym for accuracy. A person who appears entirely believable may still be wrong. This article explains what the Brennen et al. model means for solicitors, barristers, investigators, polygraph examiners, expert witnesses, employers and members of the public involved in disputes.

Why We Judge People Before We Judge Evidence

Human beings are social creatures. We have evolved to make rapid assessments of others — are they trustworthy, are they dangerous, can we cooperate with them? These snap judgements served us well on the savannah. They serve us far less well in the interview room, the courtroom, or the workplace investigation.

When a witness gives an account, listeners begin evaluating not only the content of the account but the person delivering it. Does she maintain eye contact? Does he seem nervous? Is she consistent? Does he seem confident? These impressions are formed almost instantaneously and, once formed, they are remarkably resistant to change.

The problem is that these impressions tell us almost nothing about whether the account is accurate. They tell us how the person appears. They do not tell us whether what the person is saying actually happened.

Credibility, Reliability and Truth: Three Different Concepts

Credibility: Does This Person Seem Honest?

Credibility is an impression. It is the assessment that a person appears to be telling the truth. It is formed quickly, often on the basis of demeanour: eye contact, emotional tone, consistency, composure. Courts, tribunals, and workplace panels routinely rely on credibility assessments as a primary basis for their findings.

But credibility is an evaluation of the person, not of the account. A witness may seem entirely sincere — and may in fact be sincere — while still providing information that is factually incorrect. As Brennen, Blandhol and Magnussen note, legal systems across multiple jurisdictions use “divergent and contradictory” definitions of credibility, which compounds the problem.

Reliability: Is This Account Accurate?

Reliability is a different question. It asks whether the testimony corresponds to what actually happened. This requires a more deliberative process: comparing the account against other evidence, assessing the conditions under which the observation was made, considering whether the witness’s memory could have been affected by time, suggestion, or emotional distress.

As Brennen et al. emphasise, reliability assessments require more than intuition. They require structured analysis. This is the kind of analysis that professionals trained in statement analysis and investigative psychology are equipped to provide.

Factual Truth: What Actually Happened?

The factual truth is what occurred in the real world, regardless of what any witness believes they saw. It is not always knowable, and no assessment — whether of credibility or reliability — guarantees access to it. The authors make this point clearly: even a careful reliability assessment is itself a form of human reasoning, and therefore susceptible to error.

The central lesson is that credibility, reliability, and factual truth are different concepts. They should be treated as such. A person may appear highly credible and still be mistaken. A truthful person can provide inaccurate information. A deceptive person can appear entirely believable.

Why Looking Honest Does Not Mean Being Correct

The Demeanour Problem

Legal proceedings have long placed weight on how a witness presents. A confident witness is more likely to be believed. A calm, composed witness is more likely to be trusted. A nervous or emotional witness may be viewed with suspicion.

The psychological evidence suggests this approach is deeply flawed. Research consistently demonstrates that the cues people associate with lying — avoiding eye contact, fidgeting, hesitation — are not reliably linked to deception. A truthful witness recalling a traumatic event may appear distressed, confused, and uncertain. A deceptive witness who has rehearsed their account may appear composed and fluent.

The Confidence Trap

Witness confidence is one of the most powerful influences on juror decision-making, yet it has only a modest relationship with accuracy. A witness who says “I am absolutely certain” is not necessarily more likely to be correct than one who expresses doubt. In fact, confidence can increase over time as a memory becomes more familiar through repeated retelling, even if the underlying memory has deteriorated or shifted.

This is particularly relevant in cases involving eyewitness identification, historical allegations, and workplace disputes where testimony is given weeks or months after the events in question.

What the Research Tells Us About Lie Detection

The meta-analysis by Bond and DePaulo (2006), which synthesised over 200 studies involving more than 24,000 observers, found that people correctly distinguish truth from deception only around 54% of the time — barely above chance. This figure applies to untrained members of the public, police officers, and judges alike. The research also found a consistent “truth bias”: people tend to believe others are being honest, correctly identifying 61% of truths but only 47% of lies.

The implications for anyone who relies on “reading people” — whether in a courtroom, an interview room, or a boardroom — are stark. Human beings are not effective lie detectors when relying on demeanour alone. This is one reason why the Centre advocates for structured assessment methods rather than reliance on subjective impressions.

The Problem of Memory

Memory Is Not a Recording

A common assumption in investigations is that if a witness is honest, their account should be accurate. This assumption is mistaken. Memory does not function like a video camera. It is a reconstructive process, and it is subject to decay, distortion, and contamination at every stage — from initial perception, through storage, to retrieval.

This has direct implications for how investigators, lawyers and tribunals should approach witness testimony. An honest witness with a distorted memory is not lying. They are simply wrong — and they may not know it.

How Memory Degrades

Memory decay begins almost immediately after an event. Within hours, peripheral details begin to fade. Within days or weeks, significant elements of an event can be lost or altered. Over months and years, the original memory may become substantially different from the event as it occurred.

This is not a sign of dishonesty. It is a normal feature of human cognition. A witness who provides an account six months after an event will almost certainly recall less — and recall it less accurately — than one interviewed within hours.

How Memory Distorts

Beyond simple decay, memories are vulnerable to distortion. The misinformation effect, extensively documented by Loftus and colleagues, demonstrates that exposure to misleading information after an event — through conversations, media coverage, or even the phrasing of interview questions — can alter what a witness genuinely believes they saw.

Consider a practical example: a witness to a workplace altercation initially recalls that one person “raised their voice.” After speaking with a colleague who describes the incident as “violent,” the witness may subsequently recall the event as physically threatening — not because they are lying, but because their memory has been genuinely altered by post-event information. The original memory has been overwritten, and the witness may be entirely unaware that this has happened.

Implications for Historical Allegations

These findings have particular significance for cases involving historical allegations, where the events in question may have occurred years or even decades ago. In such cases, the passage of time, repeated retelling, exposure to other accounts, and the influence of therapeutic or investigative questioning can all introduce substantial changes to what a witness recalls.

This does not mean that historical allegations are inherently unreliable. It means that they require especially careful assessment, with particular attention to corroboration and to the conditions under which the memory was formed and maintained. Where possible, accounts should be tested against contemporaneous records, physical evidence and independent witness testimony.

Lessons for Investigators

For any professional who gathers evidence — whether a police officer, a solicitor conducting pre-trial preparation, or an employer investigating a grievance — the research summarised in Brennen et al.’s model points to a number of practical principles.

• Focus on corroboration. An account that is supported by independent evidence — documents, contemporaneous records, physical evidence, other witness accounts obtained without cross-contamination — is more likely to be reliable than one that stands alone, however convincing the witness may appear.
• Test accounts against evidence. Rather than asking “Do I believe this person?”, ask “Does this account fit with what we know from other sources?” Where there are discrepancies, investigate them — but remember that discrepancies may reflect memory error rather than deception.
• Avoid overreliance on impressions. A witness’s demeanour, confidence, and emotional presentation should not be treated as evidence of accuracy. They are, at best, evidence of the witness’s subjective belief.
• Consider alternative explanations. When an account contains inaccuracies, consider whether those inaccuracies are better explained by normal memory processes — decay, distortion, or suggestion — than by deliberate dishonesty.
• Separate observation from interpretation. A witness who says “I saw him strike her” is reporting an observation. A witness who says “He was clearly aggressive” is offering an interpretation. The distinction matters, and it should be explored during questioning.
• Interview early and carefully. The sooner a witness is interviewed after an event, the more likely their account is to be accurate. The manner in which questions are asked also matters — leading or suggestive questions can contaminate the account being recorded.

These principles are relevant not only to police and legal professionals, but also to employers conducting workplace investigations, regulatory bodies, and anyone tasked with establishing what happened in a disputed situation. The Centre’s investigative psychology services are designed to support exactly this kind of structured, evidence-led assessment.

Lessons for Polygraph Examiners

Polygraph examination occupies a particular position in the investigative landscape. At The Centre for Forensic Neuroscience, we emphasise that competent polygraph practice must be informed by the very research this paper addresses.

Demeanour Is Not a Diagnostic Tool

Polygraph examiners are not immune to the same cognitive biases that affect other professionals. If an examiner forms an impression of a subject’s honesty based on pre-test behaviour — their apparent nervousness, their willingness to cooperate, their fluency in answering background questions — that impression can contaminate the entire examination process. The research is clear: demeanour-based judgements are unreliable.

Credibility Assessment Is Not Structured Testing

There is a fundamental difference between an informal credibility assessment — “I think this person is telling the truth” — and a structured physiological test conducted under controlled conditions. Professional polygraph practice depends on the latter, not the former. Examiners must resist the temptation to allow subjective impressions to influence test administration or chart interpretation.

Question Formulation and Corroboration

The quality of a polygraph examination depends significantly on the quality of the questions asked. Questions must be clear, specific, and grounded in the facts of the case. The Centre’s guidance on polygraph question formulation addresses this in detail. The results of any examination should be considered alongside — not in place of — other available evidence.

Professional Caution and Limitations

No single investigative tool, including polygraph, provides certainty. The Brennen et al. model reinforces what responsible practitioners already know: that all forms of assessment — including physiological measurement — must be understood within their limitations. Results should be reported with appropriate caveats and should contribute to a broader evidential picture rather than serving as a standalone determination. The Centre’s approach to polygraph accuracy, base rates and predictive value reflects this commitment to honest, scientifically grounded interpretation.

Practical Investigative Checklist

The following checklist, derived from the principles discussed in this article, may assist investigators, lawyers, and examiners in conducting more rigorous assessments of witness testimony.

• What evidence supports this account? Identify all independent sources that corroborate the witness’s version of events.
• What evidence contradicts it? Identify any evidence that is inconsistent with the account, and assess the significance of the inconsistencies.
• Is the witness describing observation or interpretation? Distinguish between what the witness actually saw, heard, or experienced and what they have inferred or concluded.
• Could memory explain discrepancies? Consider whether any inaccuracies in the account could be attributed to normal memory processes such as decay, distortion, or the misinformation effect.
• Is there independent corroboration? Assess whether the account is supported by evidence obtained independently, without cross-contamination between sources.
• Has the witness been exposed to other accounts? Determine whether the witness has discussed the event with others, read media coverage, or been exposed to information that could have influenced their recall.
• What were the conditions of observation? Consider factors such as lighting, distance, duration, stress, and the witness’s physical and emotional state at the time of the event.
• How much time has elapsed? Assess the interval between the event and the witness’s account, recognising that longer intervals increase the risk of memory degradation.
• Am I relying on impressions or evidence? Reflect on whether your assessment of the witness is based on their demeanour and presentation, or on the evidential support for their account.
• Have I considered alternative explanations? Before concluding that a witness is lying, consider whether the evidence is equally consistent with honest error, misperception, suggestion, or memory failure.

Conclusion

The model proposed by Brennen, Blandhol, and Magnussen offers a clear and necessary framework for anyone involved in assessing the accounts of others. Its central message is deceptively simple but profoundly important:

A witness who appears honest may be mistaken. A witness who appears uncertain may be accurate. A witness who is entirely sincere may provide testimony that is demonstrably wrong — not because they are lying, but because human memory is imperfect.

Good investigations do not rely on impressions. They seek corroboration. They test accounts against evidence. They consider alternative explanations. They recognise the limits of human judgement — including their own.

At The Centre for Forensic Neuroscience, we advocate for an evidence-led approach to investigative decision-making. Whether through investigative psychology, statement analysis, or carefully structured polygraph examinations, the aim is the same: to move beyond impressions toward structured, evidence-based assessment. The research is clear. The question is whether we are willing to act on it.

References

Brennen, T., Blandhol, S., & Magnussen, S. (2026). Witnesses and their testimony: A model of credibility, reliability, and the factual truth. Psychology, Public Policy, and Law.

Bond, C. F., & DePaulo, B. M. (2006). Accuracy of deception judgments. Personality and Social Psychology Review, 10(3), 214–234.

Loftus, E. F. (2005). Planting misinformation in the human mind: A 30-year investigation of the malleability of memory. Learning & Memory, 12(4), 361–366.

This article is provided for general information and educational purposes. It is not legal advice, clinical advice or a substitute for professional consultation. The assessment of witness credibility and reliability should always be conducted by appropriately qualified professionals with reference to the specific facts and context of each case.

Frequently Asked Questions