MNEMONIC ADVANTAGESOF THE SELF
obtained findings in which SR failed to facilitate memory better than other types of processing. Researchers soon observed that comparisons of SR tasks with semantic tasks were confounded: That SR denotes a social entity, whereas the semantic task does not, suggests that enhanced memory may be a mere artifact of this task feature (Bower & Gilligan, 1979). In an effort to solve this problem, researchers compared memory following SR to that following OR (e.g., "Does this word describe your mother?"). Studies usually showed that the SRE was reduced, if not eliminated entirely, when the target referenced in the comparison task was a highly familiar other (e.g., Bower & Gilligan, 1979; Kuiper, 1982; Kuiper & Rogers, 1979)• Further studies suggest other boundary conditions on the SRE. For ex- ample, the SRE is shown to be reduced or reversed when imag- ery tasks are used (Lord, 1980), the semantic comparison task is a desirability rating (Ferguson, Rule, & Carlson, 1983), or the semantic comparison task promotes organization (but the SR task does not; Klein & Kihlstrom, 1986)• Despite these null findings and reversals, our narrative inspection of the literature suggests that the SRE appears more often than not.
Self as a Cognitive Construct That Promotes Elaboration
The most popular explanation of the SRE is that SR promotes elaborative processing of to-be-remembered information (Kee- nan, 1993; Rogers et al•, 1977)• Based on DOP theory, depth is equated with the extent or amount of processing that a stimulus receives, whereas elaboration involves item-specific processing (see Eysenck & Eysenck, 1979)• When a participant processes a word using elaboration, he or she attends to the specific mean- ing of the word and the semantic associations between the word and extra list material in semantic memory (Anderson & Reder, 1979; Einstein & Hunt, 1980; Klein & Loftus, 1988). According to Klein and Loftus, the effect of this kind of processing is to provide multiple routes for retrieval and create an environment in which "inference-based reconstruction . . . is supported
. . in the event of retrieval failure" (p. 6). Exemplifying this
elaboration perspective, Anderson and Reder (1979) theorized that
it is not depth of processingper se that is important,but one's prior practice at making elaborations about various types of information and practice at interpretingthe previously stored elaborations.The "better" processing is that which generates more elaborations of the input that can be interpreted at retrieval . . . . The instructions that can produce rich elaboration and the materials that can be richly elaboratedmustbe definedwith respect to the processor.The most critical determinantof retentionis the numberof elaborations. (p. 390)
Anderson and Reder further argued that certain kinds of elabora- tions may be easier for some people because they are practiced habitually. This ease of processing that develops as a conse- quence of repeated elaborations suggests a connection to the memorial advantage of SR processing• Many researchers have concluded that processing information in a self-relevant way may be a "normal" processing mode (e.g., Catrambone, Beike, & Niedenthal, 1996; Catrambone & Markus, 1987; Fong & Markus, 1982; Wells et al., 1984)• They argued that
the self is exceptionally well learned and often used (Kihlstrom, 1993; Maki & Carlson, 1993) and that, indeed, people generally possess more expertise about themselves than about any other structure in memory (Markus, 1977). Thus, the evidence sug- gests that SR constitutes a processing task that receives a great deal of practice. An important consequence of the facility with which one elaborates on information using SR is that such processing can become exceptionally efficient.
Meta-analytic predictions for SR-semantic and SR- OR com- parisons. Given the foregoing logic, we expected the results of our investigation to show that SR should produce superior memory when compared with tasks that promote less elabora- tion. Proponents of the elaboration hypothesis argued that, under most circumstances, SR results in greater elaboration of the stimulus word than that achieved by a semantic comparison task (e.g., Rogers et al., 1977). Thus, based on the elaboration hypothesis, the SRE should be smaller (or disappear altogether) when studies use semantic-encoding tasks that engender greater elaboration and larger when the semantic-encoding tasks engen- der less elaboration than the SR.
On the one hand, it is difficult to think of tasks that would promote more elaboration than SR when the stimulus words are trait adjectives because trait dimensions are the most common attributes along which people judge themselves (Maki & McCaul, 1985; see Markus & Kitayama, 1991 ). Thus, we might predict that when traits are used as stimulus items, SR should always result in memory superior to that of a semantic task. On the other hand, such mnemonic superiority should increase to the extent that a person is practiced at making such elaborations• Certainly self-relevant judgments about traits are often prac- ticed, however, certain semantic judgments are also often used. In particular, people commonly identify certain trait adjectives as more socially desirable than other adjectives, a process that is part of socialization (Ferguson et al., 1983). We thus expect our meta-analysis to show that the act of judging a word for its desirability produces memory equivalentto that of SR (Ferguson et al., 1983; cf. McCaul & Maki, 1984). Similarly, SR should theoretically facilitate processing of traits better than nouns be- cause it is more common for, people to judge themselves along trait dimensions than to judge, for example, which careers they have considered in the past (Maki & McCaul, 1985; cf. Klein & Kihlstrom, 1986)• Thus, on the basis of the elaboration hypothe- sis and findings that are consistent with it, we expect our meta- analysis to show that the SRE is larger for studies in the literature that used traits rather than nouns•
Similar arguments hold with regard to SR versus OR tasks: To reference highly intimate others (e.g., one's mother) should result in more elaborations than to reference a less intimate other, presumably because elaboration of information relevant to intimate targets is a highly practiced task undertaken many times before. This high degree of elaboration theoretically pro- motes superior memory because it increases the likelihood of additional retrieval routes at the point of recall. Thus, to process information about a highly intimate other ought to promote superior memory relative to that produced by reference to some- one less intimate• Elaboration of stimulus words during refer- ence to a highly intimate other (e.g., one's mother) is a fre- quently occurring task. It is conceivable that information about intimate others may be nearly as well known and well elaborated