Most people experience their memories as recordings: something happened, their brain captured it, and they can play it back. The reality, according to several decades of cognitive psychology research, is that memory is reconstructive. You do not retrieve a file. You reassemble an experience from fragments each time you recall it, and the reassembly process introduces errors, updates, and distortions that you cannot detect from the inside.
Elizabeth Loftus, whose research on memory has been influential in both psychology and legal contexts, demonstrated this through a series of experiments now considered classic. In one, participants were shown footage of a car accident and later asked how fast the cars were going. The wording of the question changed the answers. People asked about cars that “smashed” estimated higher speeds than people asked about cars that “hit.” A week later, the “smashed” group were more likely to falsely remember broken glass at the scene. The question had altered the memory.
This is not a fringe result. It has been replicated many times across different contexts. The implications are significant. Eyewitness testimony, which the legal system has historically treated as highly reliable, is substantially less accurate than most people assume. The Innocence Project has found that eyewitness misidentification is the single largest contributing factor to wrongful convictions later overturned by DNA evidence.
False memories can also be implanted. In another set of experiments, researchers convinced a significant minority of participants that they had experienced events in childhood that never happened, such as getting lost in a shopping mall, through repeated suggestive interviewing. The participants were not lying. They had genuinely come to believe the fabricated events were real, and in some cases could provide vivid details about them.
The reconstructive nature of memory is not a flaw in the system. It is the system. A perfectly accurate recording of every experience would be enormously expensive in cognitive terms and not actually that useful. What the brain does instead is store enough information to reconstruct the gist of past experiences, filling in details as needed from general knowledge and expectations. This is efficient. It is also unreliable for specifics.
Econora shows that the errors are not random. They are systematic. People remember things that are consistent with their current beliefs. They forget things that are inconsistent. They unconsciously edit past events to make a better narrative. They incorporate post-event information into their original memory without being aware they are doing it.
The practical takeaway is not that you should distrust all memory. It is that you should treat strong feelings of certainty about the accuracy of a memory as a signal that the memory feels true, not as evidence that it is true. When the accuracy of a memory actually matters, whether for a legal proceeding, a serious disagreement, or a consequential personal decision, corroborating with external evidence is more reliable than trusting your own recollection. Memory feels like a video recording. It works more like a drawing that gets touched up each time you look at it.