Cryptomnesia: When Someone Else's Idea Comes Back as Your Own

Memory stores content more reliably than it stores origin. A word, a tune or an idea leaves a trace that can be strong and detailed, while the record of where it came from, who said it, when and in what room, is a separate and weaker thing that fades first. When the content comes back without its origin, the mind has to decide where it came from, and the most available answer is that it has just been thought of. That decision is what Marcia Johnson, Shahin Hashtroudi and Stephen Lindsay described in 1993 as source monitoring: a judgment made at retrieval from the qualities of the memory, not a stored label.

Some conditions make the judgment fail more often. Ideas that were elaborated or held in mind for a purpose acquire the richness of self-generated thoughts, so they pass as one's own. In the brainstorming task above, every partner's word had to be held in mind to avoid repeating it, which is exactly the kind of processing that makes a word feel owned. Brown and Murphy found that the most likely source of a plagiarized word was the partner who had spoken immediately before, whose word was freshest when the person's own turn came. Richard Marsh and Gordon Bower reproduced the effect in 1993 with a word-search puzzle played against a computer partner, and Marsh, Joshua Landau and Jason Hicks showed in 1997 that the failures track the ordinary conditions of source confusion, such as time passing and divided attention.

Delay and intention also matter. The borrowed idea can return within minutes, as in the brainstorming task, or weeks later when the original conversation has faded, with the feeling of having just been generated. Trying to be original does not prevent it, because the search for a new idea pulls up whatever is accessible, and the ideas most accessible are the ones recently heard. People in Brown and Murphy's generate-new task, explicitly instructed to produce items nobody had said, borrowed from their partners at a higher rate than in the group round itself.

None of this requires dishonesty. The signature of cryptomnesia is sincerity: the person is as surprised as anyone when the source is pointed out. The law has recognized the distinction, most famously in the 1976 judgment that George Harrison had copied "He's So Fine" in "My Sweet Lord" subconsciously rather than deliberately, and was liable all the same.

The term is older than the laboratory work. It was used around 1900 by the Swiss psychologist Théodore Flournoy for the hidden memories that surfaced in a medium's trance speech, and defined by F. Kraupl Taylor in a 1965 paper as the presence in consciousness of what are objectively memories but are not subjectively recognized as such. The anecdotes are older still: Jung in 1905 traced a passage in Nietzsche to a book he had read as a boy, and Helen Keller was accused of plagiarism in 1892 for a story she believed she had made up.

Alan Brown and Dana Murphy brought the phenomenon into the laboratory with three experiments published in 1989. In the first, 24 undergraduates at Southern Methodist University were placed in six groups of four and took turns, down a row of seats, naming members of four categories: sports, musical instruments, clothing and four-legged animals. Each person gave one new item per turn, no repeats allowed, for four rounds, so that each group produced 16 items per category. Afterward each person wrote down the four items they themselves had given in each category, then four new items nobody had said.

Borrowing happened at every stage. During the group round itself, 10 of the 24 participants repeated an item someone else had already given, 3.4 percent of all items produced, despite the instruction not to. In the recall-own test, 18 of 24 wrote down at least one partner's word as their own, and such words made up 7.3 percent of what was recalled as one's own. In the generate-new test, 17 of 24 offered at least one word that had already been said in the group, 8.6 percent of the new items. The partner who had spoken just before was the most common source, and higher-frequency category members were borrowed more often.

The second experiment made the turn-taking sequence more complex and found more borrowing; the third replaced the live partners with a standardized written list and still found it. Items from orthographic categories, such as words starting with a given letter, were borrowed more than meaning-based ones. The demo on this page uses the four-legged animals category, four rounds of four, and a partner who always speaks immediately before you, in a recognition test rather than free recall.

Cryptomnesia is most visible in creative work, where originality is the point, but it runs through ordinary collaboration. The rows below are illustrations, not findings from the studies cited on this page.

The memory of a source cannot be willed back, so the counters are external records and habits of suspicion toward ideas that arrive too easily.

• Write down who said what. A brainstorm record with names settles what the feeling of ownership cannot, since that feeling is not evidence of origin.
• Separate quotation from paraphrase in notes. A copied sentence and a summary can look the same a month later, and borrowing can start there.
• Be suspicious of fluency. An idea that arrives complete, with no memory of the work of making it, may have been made before, by you or by someone else.
• Search before you claim. For a phrase, a tune or a design, a few minutes of looking is cheap compared with discovering the source afterward.
• Give credit generously and accept it lightly. When a dispute is sincere on both sides, the record matters more than the memory.

Cryptomnesia is a failure of source memory with a specific outcome: the borrowed content is experienced as original. Its neighbors share the mechanism and differ in the result.

• It is not a false memory, which is a recollection of something that did not happen. In cryptomnesia the content is real and was encountered; what is wrong is the belief that it is new.
• It is not source confusion in general, which covers any misattribution of where a memory came from, such as thinking a friend told you something you read. Cryptomnesia is the case in which the misattributed source is oneself, now.
• It is not misattribution of memory in the sense of placing a real memory in the wrong time or place. The cryptomnesic idea is not placed anywhere; it is experienced as having no past at all.