質問の言い方ひとつで、記憶は書き換わる。割れていないガラスを「見た」と答えた人たち

Our memories are not like videos. In a famous study, students watched a short video of a car accident.

Then they answered a question about the cars' speed. Some heard the word "smashed," and others heard "hit." One week later, more students who heard "smashed" said they saw broken glass. But there was no broken glass in the video. One word changed their memory.

Psychologist Elizabeth Loftus has shown that memory is surprisingly easy to change. In a 1974 study, students watched a film of a car accident and were asked how fast the cars were going when they "smashed" or "hit" each other. Those who heard "smashed" gave higher speeds. A week later, 32% of them said they had seen broken glass, compared with 14% of the "hit" group, although no glass was broken.

Later, Loftus asked people to read a false story that they had been lost in a shopping mall as children. About one in four began to remember it. In 2023, a team in Ireland repeated the study with 123 people, and about 35% reported a false memory.

We tend to treat vivid memories as faithful recordings, but decades of research by psychologist Elizabeth Loftus suggest otherwise. In a classic 1974 experiment, students watched footage of a traffic accident and were asked how fast the cars were going when they "smashed" into or "hit" each other. The verb alone shifted speed estimates upward, and a week later 32% of the "smashed" group reported seeing broken glass, compared with 14% of the "hit" group, even though no glass had broken. This misinformation effect supports the view that remembering is reconstruction: each time we recall an event, we reassemble fragments, and later information can slip in.

Loftus later showed that entire false events could be implanted. After reading a fabricated story about being lost in a shopping mall, about a quarter of 24 participants fully or partly remembered it, and a preregistered 2023 replication with 123 people found about 35%. Estimates vary with how a false memory is defined, however; when eight studies were rescored using common criteria, about 30% showed clear false memories, while another 23% accepted the event to some degree. Such findings have prompted courts and police to reconsider eyewitness testimony and leading questions.