MNEMONIC ADVANTAGESOF THE SELF
degree that SR results in more organization than the comparison
Given these ideas about the joint roles of relational and item-
task, an SRE should be observed.
SR as Organized and Elaborative
Under certain circumstances, either organization or elabora- tion may produce good recall or prove superior to the other. Although some researchers have discussed the idea that either organizational (relational) or elaborative (item-specific) pro- cessing may explain the mnemonic superiority of SR, only Klein and Loftus (1988) have performed an empirical test of the hypothesis that both elaboration and organization are invoked during SR. By doing so, they raised certain issues that had not been addressed directly in the SRE literature. The basic question that has driven investigationsof SR is whether there is something inherent in the SR task that makes it likely to produce more recall than a semantic or OR task. The important focus that Klein and Loftus emphasized was the joint effects of elaboration and organization and the mnemonic advantages that occur when both processes are simultaneously invoked by a particular task.
In a review in which they discussed the joint role of item- specific and relational processing across a wide range of cogni- tive phenomena, Hunt and McDaniel (1993) summarized the basic logic that appears in different domains of the memory literature, including Klein and Loftus's (1988) experiment. In their article, Hunt and McDaniel discussed the idea that, when words are already related, similarities between words are natu- rally encoded during comprehension and attention to differences between the words (item-specific processing) facilitates mem- ory more than attention to similarities. However, when stimulus words are unrelated, differences between words are naturally encoded during comprehension and attention to similarities be- tween the words (relational processing) facilitates memory more than attention to differences (Einstein & Hunt, 1980; Hunt & Einstein, 1981 ). On the basis of this logic, Klein and Loftus (1988) hypothesized that, if organization underlies SR, then an SR task should have effects similar to that of an organi- zational task. Thus, if the stimulus words are already related, then attention during comprehension is naturally drawn to rela- tionships between words and both organizational processing and SR should be redundant. In this situation, elaborative processing should facilitate more memory than organization or SR. Klein and Loftus further hypothesized that, if elaboration underlies SR, then an SR task should have effects similar to an elaborative task. Thus, when words are unrelated, attention during compre- hension is naturally drawn to differences between words. In this situation, both elaboration and SR should be redundant and organization should serve to enhance recall more than either SR or elaboration. Klein and Loftus's results show a pattern that supports a joint elaboration-organization explanation. Specifi- cally, when list words were related, SR resulted in effects similar to elaboration: Both resulted in memory superior to organiza- tional processing. However, when the list words were not related, SR also resulted in effects similar to an organizational task. In this case, both SR and organizational tasks enhanced recall more than the elaborative task. Klein and Loftus, therefore, advocated what they called a "dual-processing approach" and concluded that both organizational and elaborative processing are invoked in an SR task.
specific processing and consistent with Klein and Loftus's (1988) experiment, several predictions that distinguish SR from other kinds of processing arise. (a) SR tasks, which presumably
promote both relational processing and item-specific processing, ought to promote better memory than any other task that pro- motes only one of these processes. (b) When list words are already related, an SR task should promote more memory than a task that promotes organization (because, when the list is related, organizational processing is redundant Hunt & Ein- stein, 1981, and elaborative processing, theoretically inherent in an SR task, promotes optimal recall Klein & Loftus, 1988 ). (c) Conversely, when list words are already unrelated, an SR task should promote more memory than a task that promotes elaboration (because, when the list is unrelated, elaborative pro- cessing is redundant Hunt & Einstein, 1981, and organiza- tional processing, theoretically inherent in an SR task, promotes optimal recall Klein & Loftus, 1988).
Boundaries for the sample of studies. We attempted to maintain in our sample only those studies that used the standard DOP incidental learning paradigm, in which an encoding (ori- enting) question is posed and a stimulus word is presented for the participant's judgment. By eliminating atypical studies from the sample, we were able to confine our recta-analysis to studies or portions of studies that were relatively consistent on all meth- odological dimensions except those that are important for meta- analytic moderator testing (Johnson, 1989). Thus, we excluded studies that used unusual methodologies relative to the rest of our sample. For example, some studies involved the reading of passages of prose, which participants were instructed to read and relate to themselves as much as possible (e.g., Reeder et al., 1987). Other studies used atypical stimuli that were not similar to those used in most of our sample (e.g., faces; Mueller, Bailis, & Goldstein, 1979) or participants were not asked to recall stimuli in a manner that conformed to the vast majority of SRE studies.
Theoretical plan. Our general plan for the meta-analysis was first to test our expectation that the SRE is significant yet inconsistent across all studies in the literature. We then planned to test our expectation that the SRE would be larger in studies where researchers compared SR with semantic encoding rather than SR with OR. Because most of our other hypotheses are specific to one or the other of these two general classes of studies, we then planned to examine our hypotheses separately within these two classes. Indeed, some moderators that we ex- amined could only be examined within one of the classes (e.g., type of semantic-encoding task).
On the basis of Klein and Loftus's (1988) conceptualization of the joint effects of relational and item-specific processing, we expect our meta-analysis to show that the SRE is smaller when the comparison task is judged to elicit both relational and item-specific processing rather than either one or the other of the two processes. Thus, for example, the SRE should decrease as intimacy of a rated target increases because the rating of a