Journal of Experimental Psychology: Copyright 1991 by the American psychologicalAssociation, Inc. Learning, Memory, and Cognition 0278-7393/91/$3.00 1991, Vol. 17, No. 4, 779-792 Aging and Memory for New Associations: Direct Versus Indirect Measures Darlene V. Howard, Astrid F. Fry, and Caitlin M. Brune Georgetown University Three experiments examined adult age differences in memory for new associations between unrelated words. On the indirect (word-stem completion) test, age differences appeared under the less-than-optimal study conditions of Experiments 1 and 3 but not under the self-paced study conditions of Experiment 2. In contrast, all experiments revealed significant age-related deficits on the direct tests, even those (word-stem cued recall) in which the cue was identical to those provided on the indirect test. The findings show that although age-related deficits in memory for new associations are particularly pronounced on tests requiring conscious recollection, age differences are not limited to such tests. Age deficits on both direct and indirect tests are related to the precision of elaborations produced during study. Similarities and differences between normal aging and amnesia are discussed, as are functional dissociations between direct and indirect measures. Memory deficits are among the most notable characteristics of normal aging. Complaints of failing memory appear when samples of older people are interviewed (e.g., Zelinski, Gi- lewski, & Thompson, 1980) and in scholars' autobiographical accounts of aging (e.g., Cowley, 1982; Hebb, 1978; Skinner, 1983). That memory deficits do accompany aging has been documented in many laboratory studies (see the reviews by Hultsch & Dixon, 1990; Light, 1990) and even in a recent large-scale survey study that overcomes many of the sampling problems inherent in laboratory research (Herzog & Rodgers, 1989). Nonetheless, the exact nature and causes of age-related memory deficits are unclear, as is the extent to which these deficits in normal aging are similar to the memory pathology seen in amnesia (e.g., Light & Burke, 1988). This article examines what is spared and what is lost in normal aging by comparing memory for new associations on direct versus indirect tests of memory. Direct tests are those in which the subject is asked to make an introspective judg- ment about learning or remembering some past event; these include the typical tests of recall and recognition. In contrast, This research was supported by National Institute on Aging Grants R01 AG02751 and R37 AG02751. Papers based on these data were presented at the Second International Congress of Applied Psycho- linguistics, Kassel, Federal Republic of Germany, July 1987, and at the meetings of the Gerontological Society of America, Washington, 13(2, in November 1987. We are indebted to Lisa Conomy and Mary Beth Quig for their help in collecting and analyzing data, to Meghan Gibbons and Michael Whamond for coding the elaborations, to Jim Howard for his comments on an earlier version of the article, to Peter Graf for suggesting Stein's method of coding elaborations, to Fergus Craik and Leah Light for additional advice on coding elaborations, and to the research participants for their interest and cooperation. Correspondence concerning this article should be addressed to Darlene V. Howard, Department of Psychology, Georgetown Uni- versity, Washington, DC 20057. Electronic mail may be sent to D_HOWARD@GUVAX.BITNET. 779 on indirect tests, the subject is not asked to recollect a past event; instead, memory is inferred from some other aspect of performance (e.g., Johnson & Hasher, 1987; Richardson- Klavehn & Bjork, 1988). For example, when people name words that had been presented in an earlier list more quickly than words that were not in the list, such priming provides indirect evidence of memory for the words, even in the absence of conscious recollection (i.e., direct memory) of the list. Direct and indirect tests are also referred to as explicit and implicit tests, respectively (e.g., Howard, 1988b; Light & Burke, 1988; Schacter, 1987). The distinction between direct and indirect measures is prominent in contemporary theory and research (cf. Johnson & Hasher, 1987; Richardson-Klavehn & Bjork, 1988; Schac- ter, 1987) because it requires a reexamination of almost all generalizations about memory. Direct and indirect tests are often influenced differently by the same experimental vari- ables (e.g., Schacter, 1987), are sometimes stochastically in- dependent of each other (e.g., Eich, 1984; Tulving, 1985) and are affected differently by pathology, such as amnesia (e.g., Schacter, 1985). Recent studies of normal aging have revealed a pattern in aging similar to that seen in amnesia; the age deficits that are obtained so consistently on direct tests are reduced or elimi- nated when indirect tests are used (cf. the reviews by Howard, 1988b; Howard, in press; Light, 1988). This pattern raises the possibility that the memory deficits of normal aging are seen only when conscious recollection is required. It is potentially important, however, that most previous research on aging in which indirect tests have been included has focused only on memory for the occurrence of preexisting representations (e.g., for the occurrence of well-known words or for well- known associations among words such as dog and cat) and did not investigate memory for newly formed associations (e.g., between previously unrelated words). Following Graf and Schacter (1985), we refer to these as item and associative memory, respectively.