Memory Is Not a Recording

Most of us hold, usually without noticing, a comfortable theory of memory: that it resembles a video archive, in which events are stored faithfully and retrieved on request. Remembering, on this view, is a matter of locating the right file and pressing play. The theory is intuitive, widely shared and, according to several generations of psychologists, substantially wrong. The case against it began with experiments conducted in the 1930s by Frederic Bartlett at the University of Cambridge. Bartlett asked English volunteers to read a Native American folk tale whose imagery and logic were unfamiliar to them, and then to retell it after various intervals, sometimes over a period of years. The versions that emerged were revealing. Details that seemed strange were quietly omitted or altered, the story became shorter and more conventional, and explanations that fitted the participants' own cultural expectations appeared where none had existed in the original. Memory, Bartlett concluded, is not a reproduction but a reconstruction, built each time from fragments and filled in with assumptions. Subsequent research has largely supported this view. Studies have shown that the wording of a question can influence what a person later claims to have seen, and that it is possible, under suggestive conditions, to lead people to remember events that never took place. Perhaps the most unsettling finding concerns confidence. The vividness of a memory, and the certainty with which it is held, turn out to be unreliable guides to its accuracy; a person may be entirely convinced of a recollection that is partly or wholly mistaken. These findings are not confined to laboratory curiosities; they have been replicated in many settings, from eyewitness testimony to ordinary personal reminiscence. It would be wrong, however, to conclude that memory is simply defective. The constructive process that produces occasional errors is the same one that allows us to imagine the future, to extract the gist of a conversation without recalling every word and to adapt past experience to new situations. A system that stored every detail with perfect fidelity would be, paradoxically, of less use; the few individuals who are reported to remember nearly everything often describe the ability as burdensome rather than enviable. The practical lessons are modest but valuable. Where accuracy matters, as in a legal setting, a notebook is more dependable than a conviction, and it is wise to treat our own certainty with a measure of humility. For students, there is an encouraging corollary: since each act of recall reshapes and strengthens a memory, testing oneself is generally a more effective way to learn than rereading. We are, in short, less like archivists than like storytellers, and the stories we tell ourselves are revised more often than we realise.