Case File 259 · Investigation Psychology

How Does Time Affect Eyewitness Memory?

Time weakens eyewitness memory. The longer the gap between witnessing an event and being asked to recall it, the less complete and less accurate the account tends to be. This is not controversial; it

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

Time weakens eyewitness memory. The longer the gap between witnessing an event and being asked to recall it, the less complete and less accurate the account tends to be. This is not controversial; it is one of the most consistent findings in memory research, dating back to Ebbinghaus's work on forgetting in the 1880s and confirmed repeatedly in applied eyewitness studies since. The decline is steepest in the first hours and days, then levels off into a slower, ongoing fade. Fine-grained details, such as faces, clothing and exact sequences, are lost faster than the gist of what happened. None of this means a delayed account is worthless, but it means the timing of an interview shapes the quality of the evidence it produces.

Key takeaways

Memory for an event declines most sharply in the first hours and days, then fades more gradually.
Fine details such as faces, colours and sequences are lost faster than the general outline of what happened.
A meta-analysis found a reliable association between longer retention intervals and reduced accuracy for face recognition.
An early recall opportunity can slow subsequent forgetting, acting as a partial inoculation.
Delayed accounts are not useless, but they should be weighed against the interval and any intervening influences.

What does the forgetting curve look like for eyewitnesses?

Recent research has mapped the trajectory in more detail than the classic Ebbinghaus curve. A study examining eyewitness recall over intervals from minutes to months found a non-monotonic, three-phase pattern: a rapid decline within the first hour, a period of relative stability between one hour and one day, and a more gradual decline extending out to ninety days. The early and late phases followed logarithmic functions, consistent with the general shape Ebbinghaus described but with a notable plateau in the middle.

In practical terms, this means that a witness interviewed one hour after the event will typically remember substantially more than one interviewed the next day, and both will remember more than one interviewed weeks later. Studies have found that witnesses recalled approximately 43% fewer details about a simulated crime event after a three-week delay compared to an immediate interview.

Which details are lost first?

Not all information decays at the same rate. Research consistently shows that coarse or basic-level information, such as "a man ran across the car park," is more durable than fine-grained information, such as the man's hair colour, build or the exact sequence of movements. This matters because investigations often depend on exactly the details that are most fragile: facial features for identification, clothing descriptions, vehicle colours and registration numbers.

A meta-analysis by Deffenbacher and colleagues examined fifty-three studies on facial memory and found a reliable link between longer retention intervals and reduced accuracy for face recognition. This has direct implications for eyewitness identifications: a witness asked to pick someone from a lineup weeks after the event is working with a weaker memory trace than one who viewed the lineup the same day.

Does delay only affect quantity, or accuracy too?

Both. As memory fades, the gaps are increasingly filled by reconstruction, inference and post-event information. A witness who cannot remember a specific detail may unconsciously fill it in with something plausible, drawn from general knowledge or from information encountered after the event. This reconstruction process, described in can new information change an old memory?, means that a delayed account is not just shorter but potentially different in content from what the witness would have reported immediately.

The longer the delay, the more opportunity there is for external influences to reshape the memory: news coverage, conversations with other witnesses, repeated thinking about the event, or leading questions from interviewers. Each of these can introduce details that feel like genuine memories, as explored in the misinformation effect.

Can anything slow the forgetting?

Yes. One of the most practical findings in this area is that an early recall opportunity can significantly reduce later forgetting. When witnesses are asked to give a full, free account soon after the event, even if no formal interview is possible, the act of retrieval appears to strengthen the memory trace. Researchers have described this as an "inoculation" effect: the early recall does not prevent all forgetting, but it slows the decline for the details that were successfully retrieved.

This has clear implications for practice:

Interview as soon as possible. Even a brief initial account taken at the scene is valuable.
Use open-ended recall. Letting the witness narrate freely activates more of the memory than a series of specific questions.
Record the initial account. If the witness is later interviewed again, the first account provides a baseline for assessing what has been forgotten, added or changed.

These steps do not make time irrelevant, but they capture the memory at its strongest point and provide a benchmark for later comparison.

How should delayed accounts be weighed?

A delayed account is not automatically unreliable, but it should be evaluated in context:

How long was the delay? The difference between one hour and one day may be smaller than the difference between one day and one month.
What happened during the interval? Exposure to media, other witnesses or repeated questioning all increase the risk that the memory has been altered.
How detailed is the account? A delayed account that includes vivid, specific details warrants closer examination: those details may be genuine, but they may also be reconstructed or acquired from other sources.
Is there corroboration? Physical evidence, recordings or independently gathered accounts can confirm or challenge the delayed testimony, as discussed in why does corroboration matter?.

The principle is not to reject delayed accounts but to treat the interval as a factor in assessing their weight, just as viewing conditions, stress and questioning method are factors.

How does Murder Book reflect the fragility of memory over time?

Murder Book: The Art of Interrogation gives players accounts that may have been shaped by time, retelling and outside influence. The game, developed by Gungrounds, an independent studio in Zagreb, Croatia, is inspired by real police footage and psychological interviews. Players examine unreliable stories and must judge how much weight a particular detail deserves, just as an investigator would weigh a statement taken days or weeks after the event differently from one taken at the scene.

That mirrors the research: a memory is strongest at the moment of the event and changes with every day that passes and every retelling that occurs. 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.