Context-Dependent Memory for Meaningful Material:Information for StudentsHARRY M. GRANT, LANE C. BREDAHL,JOHN CLAY, JENNIFER FERRIE, JANE E. GROVES,TIMOTHY A. McDORMAN and VERONICA J. DARK*Iowa State University, USASUMMARYContext-dependency eects on memory for lists of unrelated words have been shown more oftenwith recall than with recognition. Context dependency for meaningful text material wasexamined using two standard academic testing techniques: short answer (recall) and multiplechoice (recognition). Forty participants read an article in either silent or noisy conditions; theirreading comprehension was assessed with both types of test under silent or noisy conditions.Both tests showed context-dependency eects in which performance was better in the matchingconditions (silent study/silent test and noisy study/noisy test) than in the mismatching con-ditions (silent study/noisy test and noisy study/silent test). Context cues appear to be importantin the retrieval of newly learned meaningful information. An academic application is thatstudents may perform better on exams by studying in silence.#1998 John Wiley & Sons, Ltd.Appl. Cognit. Psychol. 12: 617±623 (1998)Environmental context-dependency eects occur when there is better performance on amemory test when the test occurs in the same environment in which the to-be-remembered material was originally studied (the matching condition) than when the testoccurs in a dierent environment (the mismatching condition). The literature containsmany instances of context-dependency eects for recall (e.g. Godden and Baddeley,1975; Jerabek and Standing, 1992; Martin and Aggleton, 1993; McDaniel et al., 1989;Smith, 1979; Smith, Standing and de-Man, 1992; see Smith, 1988, for a comprehensivereview; see Fernandez and Glenberg, 1985, for exceptions). Context-dependency eectsfor recall memory are typically interpreted as showing that the characteristics of theenvironment are encoded as part of the memory trace and can be used to enhanceretrieval of the other information in the trace (e.g. Eich, 1980; Smith, 1988).Context-dependency eects for recognition also have been found (e.g. Canas andNelson, 1986; Smith, 1985, 1986; Smith and Vela, 1992), but the literature containsmixed results in that many studies do not show such eects (e.g. Bell et al., 1984;Godden and Baddeley, 1980; Smith, Vela and Williamson, 1988). The fact thatcontext-dependency eects are more easily obtained with recall tests than with recog-nition tests has been used to suggest that there are dierences in the retrieval processesCCC 0888±4080/98/060617±07 $17.50#1998 John Wiley & Sons, Ltd. Accepted 11 December 1997APPLIED COGNITIVE PSYCHOLOGY, VOL. 12, 617±623 (1998)*Correspondence to: Veronica J. Dark, Department of Psychology, Iowa State University, Ames,IA 50011-3180, USA.involved in the two types of tests. Smith's (1988) `outshining' hypothesis, for example,suggested that recognition items themselves act as such strong retrieval cues (copycues) that they overshadow, or outshine, the relatively weak contribution of environ-mental context cues. Smith (1986) suggested that conditions emphasizing moremeaningful encoding might mask any eects of environmental context at test even forrecall, because such encoding would produce retrieval cues that would outshinecontext as a retrieval cue (also see McDaniel et al., 1989).Although Smith's (1988) review pointed out that most of the research assessingcontext-dependency eects have used lists of unrelated words as the to-be-rememberedmaterial, the review included ®ve studies that examined the eect of changing class-rooms on exam performance. Only one published study (Metzger et al., 1979) reportedthat switching rooms harmed perfor mance while three published studies (Abernethy,1940; Farnsworth, 1934; Sau¯ey, Otaka and Bravaresco, 1986) reported that it did not.Smith noted, however, that students actually study for exams in many dierentenvironments, so that there was a potential mismatch between study and test contexteven when students were tested in their regular classroom and that an unpublishedstudy (Mellgren, 1984), which did control both the study and test environments,showed context dependency. Thus, Smith (1988) concluded that the extent to whichenvironmental context aected retrieval of meaningful material was unresolved.We were interested in determining whether environmental context-dependencyeects would be found with the type of material and the type of tests typicallyencountered in school. For practical reasons, our focus was more on study conditionsthan on dierences in classroom testing conditions: Students have more control overtheir study environments than over their test environments. It was our collectiveobservation that many high school and college students study material in environ-ments very dierent from those in which they are tested. The study environmentsoften include background music or background noise from either family, friends, ortelevision, while the test environments are typically quieter. If context dependencyoccurs with meaningful course material, then students' study habits could be harmingtheir test performance.We chose to manipulate the presence or absence of general background noiserather than to manipulate the presence or absence of background music, becausepeople vary widely in the type of music they prefer. To distinguish between context-dependency eects and the possibility that background noise simply interferes withencoding (see Bell et al., 1984), a factorial design was used in which both the studycontext and the test context were varied. Participants read a two-page article undersilent or noisy conditions and were then tested under matching or mismatching con-ditions. The tests were designed to mimic standard classroom tests, and thus assessedcomprehension of new material (i.e. memory for meaning), rather than verbatimmemory of what was studied. Because of the possibility that context-dependencyeects might dier with dierent types of tests, participants completed both a short-answer recall test and a multiple-choice recognition test.METHODParticipantsEight members of a psychology laboratory class served as experimenters. Each experi-menter recruited ®ve acquaintances to serve as participants. There were 39 participants,618 H. M. Grant et al.#1998 John Wiley & Sons, Ltd. Appl. Cognit. Psychol. 12: 617±623 (1998)ranging in age from