The earliest recollection: A new survey

Higb scbool (N = 150) and college (N = 164) students completed a survey of tbeir earliest personal recollection (ER). Tbese memories typically were of events occurring in tbe fourtb year of life, and varied widely in terms of content and associated affect. Tbe ERs of tbe higb scbool sample were dated significantly later, contained more traumatic content, and were more likely to possess tbe qualities of a "screen memory" tban were tbose of the college sample. Upon retest tbree montbs later, 58% of tbe bigb scbool students recalled tbe same ER as on tbe first trial. For tbose recalling a different ER, tbe second was rated as more pleasant tban tbe first, and was less likely to contain traumatic content. In the college sample, tbose subjects wbose ERs were of events occurring after tbe fourtb birtbday or wbicb fit tbe definition of "screen" memories scored bigber on tbe PRF Harmavoidance scale. Implications of tbese findings for future researcb are discussed. Earliest recollections from childhood are often shared by college roommates and by guests at cocktail parties, and frequently collected from patients by psychotherapists. This activity suggests tbat botb naive and professional psycbologists sbare Hall's (1899) and Titchener's (1900) belief that they are of some importance. Intuitively, early recollections are interesting for at least two reasons. First, despite tbe wealtb of experiences which young children bave, tbeir autobiographical records are typically quite fragmentary before about age seven, and the earliest memory is rarely dated before age three. Moreover, those events that we can remember—or, depending on whom one believes, perbaps those that we cannot—often appear to be marked witb special personal significance. Most tbeories of early recollections have been concerned witb tbe apparent poyerty of adults' recollections of their first seyeral years. This research was conducted at Harvard University, and was supported in part by Grant #MH-33737 from the National Institute of Mental Health, United States Public Health Service, and in part by a grant from the William F. Milton Fund of Harvard University. We thank Beverly R. Chew, Robert H. Dworkin, Mary Harackiewicz, and Harold Rosen for their assistance in conducting the research. Address reprint requests to John F. Kihlstrom, Department of Psychology, University of Wisconsin, W. J. Brogden Psychology Building, 1202 West Johnson Street, Madison, Wisconsin 53706. Journal of Personality 50:2, June, 1982. Copyright © 1982 by Duke University Press. Earliest recollections 135 or childhood amnesia. Freud (1899, 1901) obseryed this in his clinic patients, and suggested that it was due to tbe repression of anxietyeyoking preoedipal experiences. Tbose memories that remained ayailable in consciousness were shaped by tbe defense mechanisms of condensation and isolation so that they were deyoid of affect: these were "screen memories," whose purpose was to aid in the repression of more threatening material by giving the individual something to remember. Schachtel (1947), offering an eclectic combination of Freud and Bartlett (1932), suggested that childhood memories were encoded in terms of preoedipal schemata which were incompatible with tbe scbemata employed by adults to guide tbe process of memory retrieval and reconstruction. Neisser (1962) added Piaget to Freud, and noted that tbe beginning of continuous autobiographical memory typically corresponds to tbe sbift from preoperational tbought to concrete operations. This suggests a purely cognitiye as well as an emotional reason for childhood amnesia, if the schemata employed in concrete or formal operations are incompatible with the preoperational ones which shaped the original memory encoding. Most recently. White and Pillemer (1979), working from a neopiagetian framework, haye suggested that the young child does not bave tbe attentional capacity to encode memories in easily retrievable form. Until tbis capacity develops, tbe child is unable to intentionally process episodes in narrative memory. Some isolated episodes do get encoded because of tbeir emotionally arousing nature; otbers attain this status because of tbeir emotionally arousing nature; otbers attain this status because of sheer repetition, in which case they lose their spatiotemporal particularity and are preserved as generic or semantic memories. The memories are available in storage, but because they are not elaborately encoded and lack rich connections to other knowledge structures, they are difficult to access at the time of attempted retrieval. Aside from Freud's, all of the theoretical accounts described are tied explicitly to current trends in cognitive and developmental psychology. The positions of Schachtel and Neisser anticipated more recent arguments concerning encoding specificity in memory (Tulving & Thomson, 1973), while White and Pillemer's notions are congruent with the current emphasis on depth of processing in memory (Craik & Lockhart, 1972), and are largely consistent with recent research and theory emerging from the animal laboratory concerning learning in neonates (Gampbell & Spear, 1972; Goulter, 1979; Spear, 1979). Despite this wealth of theoretical interest in childhood amnesia, empirical evidence for the existence of the phenomenon itself is sur136 Kihlstrom and Harackiewicz prisingly sparse. A number of surveys (reviewed by Dudycha & Dudycha, 1941; see also Weiland & Steisel, 1958) indicate that the average earliest recollection is of an event occurring in the fourth year of life. More extensiye samples of early childhood recollections, whether by the method of free recall (Smith, 1952; Waldfogel, 1948) or that of cued recall (Ghew & Kihlstrom, Note 1; Groyitz & QuinaHolland, 1976; Karis, Note 2) rarely elicit memories of eyents occurring before kindergarten. Howeyer, there is no conyincing eyidence from studies of humans that the alleged childhood amnesia is distinct from the ordinary forgetting that would occur in adults oyer a comparable period of time. What is required is the human equiyalent of the design employed in studies of infrahumans, where neonates, juyeniles, and adults are tested oyer equivalent retention intervals witb careful control over tbe conditions of encoding and retrieval. A substantial amount of activity bas been devoted to exploring the relations between personality and early childhood recollections. As noted earlier, Freud (1899, 1901) held that early childhood experiences were repressed, leaving only screen memories available for recall, but this hypothesis has been difficult to test empirically (Ghild, 1940): sbould neurotics bave fewer early recollections because of repression, or should they have more because repression has been less successful? Two early studies, in which subjects simply indicated by checkmarks bow many memories tbey bad at particular ages, gave contradictory results: Grook and Harden (1931) found tbat subjects witb low scores on a neuroticism scale bad more early recollections, and an earlier age of first memory, tban did tbeir more neurotic counterparts; boweyer, Gbild (1940) obtained no sucb correlations. Waldfogel (1948), employing tbe metbod of free recall, found that subjects with relatiyely few early recollections had higher neuroticism scores than did those with a relatiyely rich corpus of childhood memories. A somewbat larger number of studies bave been devoted to examining the relation between aspects of personality and otber descriptive features of the earliest recollection. Tbe two principal approaches to this topic differ in their conception of early recollections as concealing or revealing (Kramer, Ornstein, Whitman, & Baldridge, 1967). Freud (1899, 1901) held tbat tbose memories that were available to the person were only screens, whose surface features must be analyzed and interpreted to reveal tbe latent memories wbich they covered. Adler (1937; Ansbacber, 1947, 1973), on the otber hand, argued tbat the manifest content of early recollections is interesting in its own right, as it represents the lifestyle adopted Earliest recollections 137 by the indiyidual. These and otber, similar, arguments haye led to tbe use of early recollections as a projectiye tecbnique in clinical practice (e.g., Kramer et al., 1967; Mayman, 1968; Mosak, 1958; Saul, Snyder, & Shepard, 1956). Some formal studies of early recollections, especially those emanating from the psychoanalytic tradition, haye been hampered by the use of extremely cumbersome coding schemes which attempt to cover the minutiae of psycboanalytic theory (e.g., Langs, 1965a, 1965b; Levy & Grigg, 1962; Mayman, 1968). Other studies have compared the features of the earliest recollection among groups of patients differing in psychiatric diagnosis (e.g., Hafner, Gorotto, & Fakouri, 1980; Weiland & Steisel, 1958), sharing with the first type a reliance on methods of psychiatric diagnosis witb doubtful reliability and validity. Somewhat more appealing on methodological grounds are several attempts to relate surface features of tbe earliest recollection to common trait constructs as measured by objective questionnaires (for reviews see Ansbacber, 1947; Mosak, 1969; Osherson, Note 3; Taylor, 1975). While tbese studies have been relatively free of the problems besetting the other two types, evidence for systematic relations between generalized behavioral dispositions such as security and dependency on the one hand, and tbe content or affective valence of early recollections on the other, is not particularly strong. The purpose of tbe present study was to help reopen this area of inquiry by conducting a new survey of the earliest recollection, employing more adequate means of personality assessment.