Is Inconsistent Behaviour Evidence of Deception?

No, inconsistent behaviour is not reliable evidence of deception. When a person's account changes, their manner shifts, or their story does not quite line up from one telling to the next, it is tempting to read this as a sign they are lying. The research does not support that leap. Honest people are routinely inconsistent, because memory is reconstructive and fades over time, and careful studies find that truth-tellers are often no more consistent than liars. Deception cues in general are weak, and people judge lies and truths at only about 54% accuracy. Inconsistency can be a reason to ask more questions and to check the facts, but on its own it tells you that something changed, not why, and certainly not that a person is guilty of anything.

Key takeaways

• Inconsistency is a weak and ambiguous cue. On its own it does not show that someone is lying.
• Honest accounts are often inconsistent, because memory is reconstructive and decays over time.
• Classic work found truth-tellers were no more consistent than liars across repeated interviews.
• Overall, behavioural cues to deception are faint, and people detect lies at around 54% accuracy.
• Inconsistency is best used as a prompt to investigate further, not as proof of guilt.

Why do honest people behave inconsistently?

The simplest reason is memory. Human memory is not a recording that can be replayed unchanged; it is reconstructed each time it is recalled, and it is shaped by the questions asked and the time that has passed. Memory Is Not a Recording sets out this idea in more depth. An honest witness who is asked about the same event twice will often give accounts that differ in wording, emphasis and detail, add something one time and forget it the next, or correct an earlier mistake. These are the normal signs of a real memory being retrieved, not fabricated.

There are also plenty of non-memory reasons an honest person's behaviour shifts during questioning. Stress, fatigue, fear of not being believed, discomfort with a particular topic, or simply relaxing as an interview goes on can all change how someone speaks and acts. Reading those shifts as deception confuses the strain of being questioned with guilt. The broader point that truthful accounts can legitimately wobble is developed in Why Can Truthful Accounts Be Inconsistent?.

What does the research say about consistency and lying?

The intuitive theory is that liars trip themselves up by being inconsistent. Careful studies have found this is far less reliable than people assume. Granhag and Strömwall's research on repeated interrogations reasoned that liars actively try not to contradict themselves between interviews, while truth-tellers simply try to recall the original event. Because real memory is reconstructive and error-prone, truth-tellers' accounts picked up discrepancies, so the overall amount of inconsistency ended up similar regardless of whether the person was lying. In short, liars can be as consistent as truth-tellers, and consistency on its own carries little diagnostic value.

Time complicates matters further. Research on delayed interviews has described a "stability bias": liars tend to assume their memory should stay stable and so repeat details, whereas truth-tellers genuinely forget and report fewer details later. This means inconsistency over time can actually be a feature of honest recall, the opposite of the folk assumption.

This does not mean inconsistency is meaningless in every setting. A 2018 study by Masip and colleagues found that when interviewers use strategic methods, such as unexpected questions that make it hard for a liar to fall back on a rehearsed, repeated story, differences in consistency between liars and truth-tellers can emerge and support more accurate judgements. The key qualification is that this depends on a carefully designed interview, not on simply noticing that an account changed. Spontaneous inconsistency, observed without that structure, remains a weak signal.

How weak are behavioural cues to deception in general?

Very weak, which is the backdrop to all of this. The most comprehensive review, DePaulo and colleagues' 2003 meta-analysis of 158 possible cues, found that most behaviours had no clear link to deception, and those that did were faint. Bond and DePaulo's 2006 meta-analysis of nearly 25,000 judges found average accuracy of about 54%, barely above the level of a coin toss. A 2019 Annual Review of Psychology article by Vrij, Hartwig and Granhag concluded that the nonverbal cues to deceit discovered so far are faint and unreliable, and that people are mediocre lie-catchers when they rely on behaviour.

Inconsistency sits inside this picture. It is one more behaviour that is shared by the honest and the dishonest, which is why Why There Is No Single Reliable Sign of Deception treats the search for a single giveaway as a dead end. The same caution applies to the related habit of reading a changed story as a confession, examined in What Does It Mean When Someone Changes Their Story?, and to body-language "tells" more broadly in Can Body Language Tell You If Someone Is Lying?.

So how should inconsistency be used?

As a question, not an answer. A gap or a change in an account is a reasonable reason to ask more, to clarify, and above all to check the content against independent evidence. What it cannot do is settle whether a person is lying or guilty. The disciplined move is to treat inconsistency as a lead to follow rather than a verdict to deliver, and to remember that the most innocent explanation, ordinary memory, is also the most common.

Is there a game that treats inconsistency this way?

Murder Book: The Art of Interrogation is an upcoming narrative interrogation game from Gungrounds, an independent studio in Zagreb, Croatia. It is built around unreliable stories, narrative gaps and accounts that change under questioning, and its development materials show an evidence board that links statements together, for example a flat denial such as "I NEVER met her!" placed against a detailed, fabricated description. The design invites you to notice where accounts do not line up and then press further.

What keeps this honest to the research is that the game is about weighing and testing rather than delivering a guilt meter. An inconsistency is a thread to pull, and following it may reveal a lie or may reveal nothing more than a fallible memory. Cases are designed for focused 5–10-minute sessions. Murder Book is in development for PC (Steam) and mobile, with release currently planned for the end of Q2 2027. For the underlying question, see Can You Actually Tell When Someone Is Lying?.

Sources

1. DePaulo et al. (2003), Cues to deception. Psychological Bulletin (PubMed)
2. Bond and DePaulo (2006), Accuracy of deception judgments. Personality and Social Psychology Review (PubMed)
3. Vrij, Hartwig and Granhag (2019), Reading lies: nonverbal communication and deception. Annual Review of Psychology (PubMed)
4. Masip et al. (2018), Learning to detect deception from evasive answers and inconsistencies across repeated interviews. Frontiers in Psychology (PubMed Central)
5. Granhag and Strömwall, (In)consistencies as cues to deception. University of Gothenburg (GUPEA)
6. Gungrounds, Murder Book fact sheet