Developmental Psychology1999, Vol. 35, No. 5, 1338-1348Copyright 1999 by the American Psychological Association, Inc.0012-1649/99/S3.00Origins of Autobiographical MemoryKeryn Harley and Elaine ReeseUniversity of OtagoThis study tested the predictions of M. L. Howe and M. L. Courage's (1993, 1997) theory of infantileamnesia compared with a social-interactionist account of autobiographical memory development (R.Fivush & E. Reese, 1992; K. Nelson, 1993b). Fifty-eight mother-child dyads were assessed for maternalstyles of talking about the past and for children's self-recognition, language production, and nonverbalmemory when the children were 19 months old. Children's shared and independent memory reports werethen assessed from 19 to 32 months. Maternal reminiscing style and self-recognition uniquely predictedchildren's shared memory reports across time, even with children's initial language and nonverbalmemory factored out. Self-recognition skills also predicted children's later independent memory. Theseresults support a pluralistic account of the origins of autobiographical memory.Autobiographical memories are specific, long-lasting, and usu-ally of significance to one's self-concept (K. Nelson, 1993b). Oneunusual feature of the autobiographical memory system is that asadults, we are typically unable to recall events that occurred beforethe age of around 3 years, a phenomenon known as infantileamnesia (e.g., Dudycha & Dudycha, 1941; Pillemer & White,1989; although cf. Eacott & Crawley, 1998, and Usher & Neisser,1993, for earlier age estimates). What leads to the development ofthis memory system? Current theories of infantile amnesia focuson the neural (C. A. Nelson, 1995), cognitive (Perner & Ruffman,1995; Welch-Ross, 1995), and social (K. Nelson, 1996) factorsinvolved. In this study, we examine the cognitive and socialcontributors to children's autobiographical memory from V-h to2'/2 years of age, during the period of infantile amnesia.Self-knowledge theorists (Howe & Courage, 1993, 1997) haveproposed that infantile amnesia is not due to a memory deficit perse. Recent research clearly demonstrates that infants and youngchildren can retain event memories over surprisingly long delays(e.g., Bauer, Hertsgaard, & Dow, 1994; Rovee-Collier & Hayne,1987). Howe and Courage argued, rather, that it is the absence ofa "personal frame of reference" enabling memories to be stored asautobiographical that accounts for infantile amnesia. Howe andCourage believe that the emergence of the cognitive self during the2nd year of life, commonly measured by the mirror test of self-recognition (Lewis & Brooks-Gunn, 1979), is critical to this neworganizational system. Once children achieve a cognitive self, theself becomes a category for organizing personally related memo-Parts of this research were presented at the Jean Piaget Society AnnualSymposium, Chicago, Illinois, in June 1998.This research was funded by a grant from the Marsden Fund of theRoyal Society of New Zealand.We thank Rebecca Brookland, Kate Farrant, Rhiannon Newcombe,Sandra Powell, Stephanie Read, and Bridget Sly for their help on thisproject. Most of all, we thank the families who participated so willingly.Correspondence concerning this article should be addressed to KerynHarley or Elaine Reese, P.O. Box 56, Psychology Department, Universityof Otago, Dunedin, New Zealand. Electronic mail may be sent [email protected] or [email protected] in the same way that categories in general enable children toorganize and retrieve information. According to Howe and Cour-age, as children's self-understanding becomes more complex overtime, more features are available to organize events in memory.Children's ability to use these self-features, in conjunction withchildren's increasing ability to maintain memories in storage,results in the increases seen in children's autobiographical memoryabilities from the age of 3 and beyond. To date, no empirical dataexamining the relation between children's self-understanding andthe development of memory prior to the age of 3 have beenpresented.In contrast, another group of theorists has focused on the role ofsocial interaction in the emergence of the autobiographical mem-ory system (e.g., Fivush & Reese, 1992; Hudson, 1990; K. Nelson,1993b). The primary functions of autobiographical memory are todevelop a life history and to tell others what one is like throughrelating one's past experiences (Fivush, Haden, & Reese, 1996).Social interaction theorists propose that through talking about thepast with parents and other adults, children learn both the form forreporting past experiences and the social functions that talkingabout the past fulfills. A number of studies have found thatchildren 2-2 V2 years of age are able to talk about specific novelevents with their parents and with other adults and that childrentake on more responsibility for such conversations as they growolder (e.g., Eisenberg, 1985; Fivush, Gray, & Fromhoff, 1987;Hudson, 1990). Children also start initiating talk about past eventsfrom an early age (Hudson, 1990; Sachs, 1983) but by the age of3 are only just gaining the ability to use a narrative or story formfor talking about these events (e.g., Fivush, Haden, & Adams,1995).Individual differences in the way parents talk to their childrenabout the past lead to individual differences in children's reportingskill. Two parental styles of talking about the past have beenidentified: high elaborative and low elaborative (Engel, 1986;Fivush & Fromhoff, 1988; Hudson, 1990; McCabe & Peterson,1991; Reese & Fivush, 1993). Parents in the high-elaborativecategory provide a large amount of detailed information about pastevents regardless of how much information children are providing.Highly elaborative parents accept and expand on the pieces of1338ORIGINS OF MEMORY1339memory information children provide. In contrast, parents with alow-elaborative style tend to repeat their questions over and overin an attempt to get a specific answer from the child. By the timetheir children are of preschool age, mothers' styles of elaborationare consistent over time (Reese, Haden, & Fivush, 1993) andacross siblings (Haden, 1998). Importantly, the high-elaborativestyle is positively associated with children's provisions of memoryinformation, both concurrently and longitudinally (McCabe &Peterson, 1991; Reese et al., 1993). Reese et al. found that anelaborative style displayed by mothers when children were 40months