Self-Relevance Effect: Why Words About You Are the Ones You Keep

The effect sits on top of an older finding about memory: what you do with information when you meet it decides how well it comes back. Craik and Tulving showed in 1975 that a word judged for its meaning is recalled far better than the same word judged for whether it was printed in capitals or whether it rhymed with another word, and they called the difference depth of processing. Rogers, Kuiper and Kirker added a fourth question, whether the word described the participant, and found that it beat the meaning question too.

Their explanation was that the self is a superordinate schema, an unusually rich and well-organized body of knowledge that is in use all day. Asking whether a word describes you makes you check it against that structure, and the check leaves the word connected to a great deal that you already know, which gives recall many routes back to it. On this account, self-relevant encoding is elaborate encoding of a special kind.

Klein and Kihlstrom challenged the special part in 1986. Their experiments showed that when semantic judgments were arranged to produce as much organization of the list as self-judgments do, free recall was virtually identical in the two conditions. Organization, grouping the words into related sets, rather than elaboration about the self, was doing the work, and it could be varied independently inside both kinds of task. Symons and Johnson's 1997 meta-analysis landed between the two positions: the self-reference advantage is real against semantic judgments and against judgments about other people, it is smaller against the second than against the first, and it shrinks when the comparison task promotes both organization and elaboration. Their conclusion was that the self works because it is a well-developed and often-used construct that encourages both, not because it is a different kind of memory.

Two refinements matter in reading the literature. Klein, Loftus and Burton showed in 1989 that "does this describe you" and "remember a time you were like this" are different tasks, drawing on trait knowledge and on autobiographical memory respectively, and that much of the inconsistency in earlier results disappears once studies are sorted by which task they used. And the effect concerns memory for the words, not accuracy about yourself: remembering a word you judged says nothing about whether the judgment was right.

Timothy Rogers, Nicholas Kuiper and William Kirker published "Self-reference and the encoding of personal information" in the Journal of Personality and Social Psychology in 1977. They borrowed the method of Craik and Tulving's depth of processing experiments, in which each word is shown with a question that forces a particular kind of encoding, and added a question that no previous study had used in this way.

Participants rated a series of adjectives, each under one of four tasks: a structural task about the word's appearance in print, a phonemic task about its sound, a semantic task about its meaning, and the self-reference task, whether the word described them. They had not been told that memory would be tested. Afterwards they were asked to recall as many of the adjectives as they could. In both experiments, the adjectives rated under the self-reference task were recalled best. The three comparison tasks were the ones Craik and Tulving had used to show that meaning is remembered better than sound or appearance, so the self-reference question sat at the top of an already ordered scale. The authors concluded that self-reference is a rich and powerful encoding process and that the self functions as a superordinate schema deeply involved in processing, interpreting and remembering personal information.

The paper opened a line of research rather than settling one. Klein and Kihlstrom's 1986 organization account, Klein, Loftus and Burton's 1989 separation of two self-reference tasks, and Symons and Johnson's 1997 meta-analysis of the field each revised the interpretation while confirming the basic result: judged against yourself, information comes back better. The demonstration on this page follows the design in miniature, with a structural task standing in for the three comparison tasks and recognition in place of free recall, so it can show the pattern without being able to measure it in you.

The rows below are illustrations of where self-relevant encoding does its work, for better and worse. They are not findings from the studies cited on this page.

The self-relevance effect is useful when you want to remember and distorting when you want to be fair about what happened, so the advice runs both ways.

• To remember something, relate it to yourself on purpose. Ask whether a fact applies to you, when you have met it, what you would do with it. The judgment is the work.
• To recall a meeting, a conversation or a review fairly, do not trust memory. The parts about you will be overrepresented in what comes back. Use notes, recordings or the written version.
• When you teach or present, give each listener a reason to check the material against themselves: a question, a case that resembles theirs, a prediction about their own result.
• When someone remembers a conversation differently from you, consider that each of you remembers best the parts that concerned yourselves. That is an ordinary property of memory, not evidence of bad faith.
• Keep the effect separate from accuracy. Remembering a word you judged about yourself says nothing about whether the judgment was right.

The self-relevance effect is about how well information is remembered after it has been judged against the self. Several neighbors concern the self in other ways.

• It is not egocentric bias, where people recall their own contributions to a shared effort as larger than they were. Self-relevant encoding may feed that bias, since your own actions are the best-remembered part of any joint event, but the two are measured differently.
• It is not the spotlight effect, the tendency to overestimate how much other people notice you. That is a judgment about others' attention, not about your memory.
• It is not the levels of processing effect, the parent finding that attending to a word's meaning improves recall over attending to its appearance or sound. Self-reference is best read as a further step on that scale, and the debate since 1986 has been about how much further.
• It is not the generation effect, where information you produce yourself is remembered better than information you read. Self-reference does not require producing anything, only judging.