Journal of Experimental Psychology: Copyright 1989 by the American PsychologicalAssociation, Inc. Learning, Memory, and Cognition 0278-7393/89/$00.75 1989, Vol. 15, No. 6, 1192-1197 Effects of Item-Specific and Relational Information on Hypermnesic Recall Stanley B. Klein Trinity University Judith Loftus University of Texas at San Antonio John F. Kihlstrom University of Arizona Robert Aseron University of Illinois The role of encoding conditions in producing hypermnesia (increased recall over successive trials) was examined by manipulating the availability of item-specific and relational information at encoding. Our findings demonstrate that encodings providing item-specific information (e.g., elaborative encodings) produce hypermnesia by facilitating the recovery of new items over trials, whereas encodings providing relational information (e.g., organizational encodings) produce hypermnesia by protecting against the loss of previously recalled items. Thus, the effects of encodings on hypermnesia may be understood by considering the type of trace information they make available. Experiments employing multiple recall trials have repeat- edly confirmed a commonly experienced aspect of memory: the accessibility of information in memory changes over repeated testing (e.g., Ballard, 1913; Brown, 1923; Tulving, 1964). Items not recalled on one trial may be recalled on another (item gain), whereas items recalled on early trials may not be recalled on later attempts (item loss). In some multitrial studies, item gains are offset by item losses, so that from trial to trial the number of items recalled remains approximately constant (e.g., Rosner, 1970; Tulving, 1967). In other studies, however, item gains exceed item losses, resulting in a net increase in items recalled over successive trials. This phenom- enon has been labeled hypermnesia (Erdelyi & Becker, 1974). Most investigators seeking an explanation for the occur- rence of hypermnesia have focused their attention on the role of encoding conditions. From the numerous studies docu- menting hypermnesia (for reviews see Erdelyi, 1984; Payne, 1987; Roediger & Challis, 1989), three classes of encodings have been found reliably to produce hypermnesic recall. These include tasks that encourage subjects to form mental images of the to-be-remembered stimulus items (e.g., Erdelyi, Fin- kelstein, Herrell, Miller, & Thomas, 1976; Payne & Roediger, 1987; Roediger & Thorpe, 1978), tasks that encourage subjects to elaborate stimulus items by relating them to extralist ma- terial in memory (e.g., Belmore, 1981; Erdelyi, Buschke, & Finkelstein, 1977; Roediger, Payne, Gillespie, & Lean, 1982), and tasks that encourage subjects to organize the stimuli into categories (e.g., Mross, Klein, & Kihlstrom, 1988; Paris, 1978; Payne, 1986). Several investigators have speculated about a This research was supported in part by Grant MH-35856 from the National Institute of Mental Health. We would like to thank Matthew Erdelyi, Howard Klein, David Payne, and Dan Wegner for their very helpful comments on an earlier version of this article. Correspondence concerning this article should be addressed to Stanley B. Klein, Department of Psychology, Trinity University, 715 Stadium Drive, San Antonio, Texas 78284. single process that might be responsible for hypermnesia (e.g., Belmore, 1981; Mross et al., 1988), but no proposal yet offered has been able to account for the fact that these three otherwise dissimilar encoding manipulations all produce the effect. Re- cently some investigators have suggested that factors other than encoding (e.g., retrieval conditions) may be more im- portant for understanding hypermnesia (e.g., Roediger, 1982; Roediger et al., 1982). It probably is the case that encoding factors alone cannot explain hypermnesia. However, some of the difficulty en- countered in identifying the mechanisms underlying hyperm- nesia may be due to a failure to consider the specific effects of different encoding conditions on hypermnesic recall. By definition, hypermnesia requires a net increase in the number of items recalled over trials--item gains must exceed item losses. If item gains and item losses both occur over trials, a net increase in recall can result from processes that either increase the number of items gained, reduce the number lost, or both. Most attempts to account for the hypermnesia ob- tained with imaginal, elaborative, and organizational encod- ings have focused primarily on the overall capacity of these manipulations to produce a net increase in items recalled over trials. We propose that it may be more productive to focus on the way in which these encodings affect item gains and item losses. The relation between encoding conditions and item gains and losses can be understood by considering the type of information made available by a given encoding. The types of trace information provided by imaginal, elaborative, and organizational encoding tasks are well documented. Imaginal (e.g., Marschark, 1985; Ritchey & Beat, 1980) and elaborative (e.g., Einstein & Hunt, 1980; Klein & Loftus, 1988) tasks both are assumed to promote the encoding of information specific to each to-be-remembered item. This item-specific information increases the distinctiveness of each item in memory by emphasizing features that distinguish it from other items (e.g., Hunt & Einstein, 1981; Marschark, Richman, Yuille, & Hunt, 1987). Organizational tasks, by contrast, 1192