Case File 243 · Investigation Psychology

Is Selective Memory Evidence of Lying?

By Murder Book Editorial TeamPublished by Gungrounds, the studio making Murder BookPublished 3 October 2026Updated 3 October 2026

No. Selective memory, meaning recalling some parts of an event clearly while forgetting or misremembering others, is a normal feature of how human memory works rather than reliable evidence that a person is lying. Memory is reconstructive, not a recording, so gaps, uneven detail and small changes between tellings are expected even from honest people. Forgetting is also partly an active process: people can push unwanted material out of mind without any intent to deceive. Decades of research find that behavioural and memory-based signs of lying are weak, and that observers separate truths from lies only slightly better than chance. Patchy recall can tell an interviewer where to ask more carefully. On its own, it cannot show that someone is being dishonest.

Key takeaways

Selective memory is normal: human memory is reconstructive, so recall is uneven, fades over time and shifts between accounts.
Forgetting can be an active process, and people can suppress unwanted memories without intending to lie.
Confidence in a memory does not track its accuracy, so a vividly recalled detail is not proof that it is correct.
Behavioural and memory cues to deception are weak, and people average about 54% accuracy when judging truths and lies.
Uneven recall is a reason to ask careful follow-up questions, never proof of guilt or dishonesty.

What do people mean by "selective memory"?

In everyday use, "selective memory" describes someone who seems to remember convenient facts in detail while drawing a blank on awkward ones. It is usually said with suspicion, as if the gaps were chosen. The phrase blends several very different things: ordinary forgetting, uneven attention at the time of an event, reluctance to share, and genuine inability to retrieve a detail on demand. What Is Selective Memory? looks at the term itself. The question here is narrower: when someone's recall is patchy or lopsided, does that pattern reveal a lie?

The honest answer is that it rarely reveals anything so specific. To see why, it helps to look at how memory actually behaves.

Is uneven memory normal?

Yes, and the unevenness is built into the system. Memory does not store an event like a video file that can be replayed intact. It reconstructs the event each time from fragments, and that reconstruction is shaped by expectations, later information and the way a question is asked.

The classic demonstration is Elizabeth Loftus and John Palmer's 1974 study "Reconstruction of Automobile Destruction." After watching a filmed car crash, participants estimated the cars' speed. Those asked how fast the cars were going when they "smashed" into each other gave higher estimates than those asked about cars that "hit" each other, and a week later were more likely to report broken glass that was never in the film. A single changed verb nudged the memory. This is why memory is not a recording: small differences in questioning, mood or the passage of time reshape what comes back.

It follows that two accounts of the same event, given weeks apart, will almost never match word for word. Details drop out, order shifts, and emphasis moves. Why truthful accounts can still be inconsistent covers this directly. An account that is suspiciously uniform across many retellings can be just as noteworthy as one that varies.

Can people forget on purpose?

To a degree, yes, and this matters for the "selective" part of selective memory. Research on motivated forgetting shows that people have some voluntary control over what stays accessible. In a 2001 study published in Nature, Michael Anderson and Collin Green had participants learn word pairs and then, for some cues, try to keep the associated word out of mind. On a later test, those deliberately suppressed items were recalled less well than items that had simply been left alone. A 2014 review by Anderson and Simon Hanslmayr describes the brain mechanisms involved, framing retrieval suppression as a normal form of cognitive control rather than a trick.

The practical point is that someone can genuinely find a memory harder to reach, especially if it is distressing or embarrassing, without any decision to deceive. Avoiding a painful subject and lying about it are not the same thing, even though they can look similar from the outside.

Does a confident memory mean an accurate one?

No, and this is one of the most counter-intuitive findings in memory research. People often treat a strongly, vividly held memory as more trustworthy than a hazy one. The evidence does not support that assumption.

Jennifer Talarico and David Rubin's 2003 study tracked students' memories of hearing about the 11 September attacks alongside memories of an ordinary event. Over the following weeks, the "flashbulb" memories lost accuracy at about the same rate as the everyday ones. What stayed high was people's confidence in the flashbulb memories, not their consistency. The title of the paper sums it up: confidence, not consistency, characterises flashbulb memories.

So when someone recalls one detail with great certainty and another not at all, the certainty is not a reliable guide to which parts are correct. A confidently stated detail can be wrong, and a hesitant one can be right.

Why isn't selective memory proof of lying?

Because the link between any single behaviour, including patchy recall, and actual deception is weak. The broadest evidence comes from Bella DePaulo and colleagues' 2003 meta-analysis in Psychological Bulletin, which pooled 1,338 estimates of 158 possible cues to deception. Some cues showed small associations with lying, but most were weak or absent. There is no behaviour, and no memory pattern, that marks a liar.

This weakness shows up in how well people actually detect lies. Charles Bond and DePaulo's 2006 meta-analysis of more than 24,000 judges found an average accuracy of about 54%, barely above the level of a coin toss. If trained and untrained observers cannot reliably catch lies from behaviour, uneven memory is not going to do it either.

There are also many innocent reasons for lopsided recall:

The person paid attention to some parts of the event and not others.
Time has passed, and ordinary forgetting has set in.
The topic is stressful, and stress narrows and distorts what is encoded.
They are reluctant to volunteer something embarrassing that has nothing to do with the matter at hand.
They are reconstructing, and the gaps are simply where reconstruction failed.

When someone repeatedly says they cannot recall, that too has ordinary explanations, as why people say "I don't remember" during questioning explains. None of these is a confession.

What should an interviewer do with patchy recall?

Treat it as a map of where to look, not as a verdict. Open-ended prompts that invite free recall tend to produce more and more accurate detail than pointed, leading questions, which can contaminate the memory in the way Loftus and Palmer showed. If a gap matters, the way to test it is to seek independent evidence, such as records, timings or other witnesses, that can confirm or contradict the account. A memory that can be checked against the outside world is worth far more than a guess about why someone forgot.

Above all, resist the pull of treating a single sign as an answer. There is no single reliable sign of deception, and selective memory is not the exception.

How does this connect to Murder Book?

Murder Book: The Art of Interrogation is an upcoming narrative interrogation game from Gungrounds, an independent studio in Zagreb, Croatia, inspired by real police footage and psychological interviews. Its cases are built around unreliable stories, narrative gaps and accounts that do not quite line up, exactly the territory this article covers.

That design leans into uncertainty rather than away from it. Development screenshots show an in-game casebook with entries such as Non-Response and Emotional Fading. Read through the research above, those are prompts to look closer, not labels that prove a suspect is lying. A blank in someone's story might be a lie, or it might be ordinary forgetting, stress or reluctance, and a good case has to tell the difference using evidence, not a hunch about 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.