Cryptomnesia - Experimental Research

Experimental Research

In the first empirical study of cryptomnesia, people in a group took turns generating category examples (e.g., kinds of birds: parrot, canary, etc.). They were later asked to create new exemplars in the same categories that were not previously produced, and also to recall which words they had personally generated. People inadvertently plagiarized about 3–9% of the time either by regenerating another person's thought or falsely recalling someone's thought as their own. Similar effects have been replicated using other tasks such as word search puzzles and in brainstorming sessions.

Research has distinguished between two kinds of cryptomnesia, though they are often studied together. The distinction between these two types of plagiarism is in the underlying memory bias responsible—specifically, is it the thought that is forgotten, or the thinker? The first type of bias is one of familiarity. The plagiarizer regenerates an idea that was presented earlier, but believes the idea to be an original creation. The idea that is reproduced could be another's idea, or one's own from a previous time. B. F. Skinner describes his own experience of self-plagiarism:

"One of the most disheartening experiences of old age is discovering that a point you just made—so significant, so beautifully expressed—was made by you in something you published long ago."

The second type of cryptomnesia results from an error of authorship whereby the ideas of others are remembered as one's own. In this case, the plagiarizer correctly recognizes that the idea is from an earlier time, but falsely remembers having been the origin for the idea (or, having lost the specific memory of encountering it in print or conversation, assumes that it "came to" the plagiarizer as an original idea). Various terms have been coined to distinguish these two forms of plagiarism — occurrence forgetting vs. source forgetting and generation errors vs. recognition errors. The two types of cryptomnesia appear to be independent: no relationship has been found between error rates and the two types are precipitated by different causes.