A common intuition treats memory like a video recording — accurate, stable, and unchanged by playback. Neuroscience shows the opposite: every time a memory is recalled, it becomes temporarily unstable and is re-stored (reconsolidation), during which it can be subtly altered by current context, mood, or new information encountered around the same time. This is why eyewitness testimony is notoriously unreliable after repeated retelling — psychologist Elizabeth Loftus's research, including her famous studies successfully implanting entirely false childhood memories in study participants, showed just how malleable memory can be even without any repeated recall involved.
This isn't a flaw so much as a tradeoff: reconsolidation is likely what allows memories to be updated with new relevant information (useful for keeping knowledge current), at the cost of also allowing them to drift or be distorted over repeated recall — vividness and confidence in a memory are not reliable indicators of its accuracy.
Loftus and her collaborators recruited study participants and, working with family members, presented them with three true childhood stories alongside one entirely fabricated one — typically, being lost in a shopping mall as a young child and rescued by an elderly stranger. A striking share of participants not only came to believe the fabricated event had happened, but elaborated on it over subsequent interviews with specific invented sensory details: what the mall smelled like, what the stranger was wearing. Some participants, told afterward that one story had been fabricated, couldn't identify which one it was — their confidence in the false memory matched their confidence in the real ones.