Case File 256 · Investigation Psychology

Can Witnesses Change Each Other's Memories?

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

When witnesses to the same event talk to each other before giving their accounts, their memories can merge. Details one person saw can transfer into another person's recollection, and the recipient often has no idea the information came from the conversation rather than from their own observation. This is co-witness contamination, and it is one of the most common ways eyewitness evidence becomes unreliable. Research shows it happens naturally, without any intent to mislead, because human memory treats socially acquired information much like directly perceived information. The effect is strong enough that a single inaccurate detail shared between witnesses can appear in multiple independent-looking statements, creating a convincing illusion of corroboration where none exists.

Key takeaways

Witnesses who discuss an event before giving statements can absorb each other's details, including wrong ones.
Co-witness contamination creates false corroboration: matching accounts that trace back to a shared conversation, not independent observation.
The effect is largely unconscious; witnesses genuinely believe the borrowed details are their own memories.
Contamination is more likely when one witness is perceived as credible or confident.
Separating witnesses before interviews is one of the simplest and most effective safeguards.

How does co-witness contamination happen?

The mechanism is straightforward. When two or more people who witnessed the same event discuss it, they exchange details. Some of those details may be accurate, but others may be wrong, incomplete or interpreted differently. Research by Fiona Gabbert, Amina Memon and colleagues demonstrated this using a method where two people watched different versions of the same filmed event, each containing details the other did not see. After discussing the event together, 71% of witnesses went on to report details in their own accounts that they could only have learned from their co-witness, not from what they actually saw.

What makes this dangerous for investigations is that witnesses are typically unaware of the transfer. They do not tag the new detail as "something my friend told me." Instead, it feels like a memory of their own experience, a failure of source monitoring that makes the borrowed detail indistinguishable from a genuine one.

What is the real-world impact?

A well-documented example comes from the 2003 murder of Swedish foreign minister Anna Lindh in a Stockholm department store. Twenty-nine people witnessed the attack. During discussions among witnesses, one person incorrectly stated the attacker wore a camouflage-patterned jacket. Several other witnesses subsequently included this detail in their police statements, despite it being wrong. The false detail spread through conversation and became part of the official record because it was reported by multiple apparently independent sources.

This pattern, where a single error multiplies through witness interaction, is precisely why co-witness contamination matters. Investigators and jurors naturally give more weight to accounts that agree, but if the agreement comes from discussion rather than independent observation, it tells you nothing about what actually happened.

Does confidence matter?

Yes. Research on what is sometimes called "blame conformity" found that witnesses were significantly more likely to adopt another person's account, including blaming the wrong person, when the source of the information seemed credible. In studies where the co-witness information appeared to come from a police officer or an academic, participants were more willing to incorporate it than when it came from a fellow bystander. Confidence works in a similar way: a witness who sounds sure can shift the memories of a less confident co-witness, as explored in does eyewitness confidence mean the memory is accurate?.

This is not the same as deliberately lying to please an authority figure. The witnesses genuinely come to believe the details they absorbed, which is what makes the contamination so difficult to detect after the fact.

How does co-witness contamination differ from other memory effects?

Co-witness contamination overlaps with several related phenomena but has its own specific features:

The misinformation effect describes how misleading information from any external source can alter a memory. Co-witness contamination is a specific delivery route for misinformation: through conversation with another witness.
Memory conformity is the broader term for witnesses' accounts becoming more similar after interaction. Co-witness contamination focuses on the transfer of specific details, particularly wrong ones.
Post-event information can come from media, investigators or other witnesses. Research suggests that information from a co-witness may carry more weight than information from news reports, because the co-witness was there and seems to have first-hand knowledge.

Studies comparing these routes have found that co-witness discussion can be more potent than media exposure or leading questions in reshaping a memory, partly because conversation feels collaborative and credible rather than imposed. The witness is not being told what to remember by an outsider; they are piecing together a shared experience with someone who was there, which makes the borrowed details feel earned rather than planted. This social dimension is what gives co-witness contamination its particular power and what makes it so difficult for the witness to detect in their own memory afterwards.

What can be done to prevent it?

The single most effective step is simple: keep witnesses apart before they are interviewed. This is why best-practice guidelines, including those from the National Research Council's 2014 report on eyewitness identification, stress separating witnesses at a scene and taking individual accounts before any discussion can occur.

Other practical safeguards include:

Interviewing witnesses as soon as possible, before natural conversation fills gaps in memory.
Asking each witness what contact they have had with other witnesses or media coverage since the event.
Using open-ended questions that let the witness report freely, rather than confirming details that might have come from discussion.
Recording all interviews, so the origin of each detail can be examined later.

These measures do not eliminate every risk, but they remove the most avoidable source of contamination. The goal is to preserve each witness's account as close to their independent experience as possible, a principle explored further in why are witnesses interviewed separately?.

How does Murder Book approach unreliable accounts?

Murder Book: The Art of Interrogation puts players in situations where accounts do not line up and the question is why. In real investigations, co-witness contamination is one reason two stories might match suspiciously well or share the same error. The game, developed by Gungrounds, an independent studio in Zagreb, Croatia, is inspired by real police footage and psychological interviews. It asks players to compare statements against each other and against the evidence, noticing where apparent corroboration might reflect influence rather than independent memory.

That design reflects the research: agreement between witnesses is only meaningful if their accounts are genuinely independent. 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.