A Path Not Taken
Review by John T. Bruer
Whether or not a child becomes a toxic or nontoxic member of society is largely determined by what happens to the child in terms of his experiences with his parents and primary caregivers in those first three years" (Barton, 1998). This statement, made by the actor Rob Reiner to county government officials in 1998 promoting his I Am Your Child campaign, is a strong formulation of the thesis of infant determinism. Infant determinism is the doctrine that early childhood experiences have irreversible, lifelong effects. A child's first steps on the path of life determine the child's final destination. judging from policy discussions, media coverage, and conversations with parents, infant determinism is extremely popular.
"The effects of early experiences, important at the time, will only be prolonged if similar experiences follow" (p. 22). This quote is Ann and Allen Clarkes' thesis in Early Experience and the Life Path. Their statement captures a view of child development that emphasizes long-term complexity and indeterminism. The first steps on life's path are important, but still leave many routes open to diverse outcomes. This indeterminist thesis is relatively unpopular in policy, parenting, and even some academic circles.
Indeterminism's persistent unpopularity prompted the Clarkes to publish their book. Since the early 1950s, their research has indicated that children are highly resilient and show remarkable recovery from early privation, if their life circumstances improve. In 1977, their edited volume, Early Experience: Myth and Evidence, reviewed the scientific evidence on the long-term effects of early experience and made, they thought, a compelling case for indeterminism. The new book is a sequel that the authors had hoped would now be only of historical interest. Yet, as they say, today the "quality media" and even some child development professionals hold firmly to the belief that early childhood experiences have lifelong, irreversible consequences.
Although their book is concise, logically constructed, and well written, it is unlikely to turn the popular prodeterminism tidal wave. It is an academic, not a popular work. It reads like an extremely elegant review article. Appreciating the level and subtlety of their argument requires some prior knowledge of the field. For example, they distinguish between critical periods (abrupt onset, rapid offset) and sensitive periods (abrupt onset, extended offset) in development, but do not define these terms for the lay reader. And like a good review, the book is not self-contained; rather, it encourages the reader to go beyond the text to reread and evaluate for oneself the cited studies and evidence. These factors raise significant barriers to understanding for policy diehards, on-deadline journalists, and the general reader.
Yet these same factors make the book a valuable resource for scientists or scholars who want a guidebook to assess research on the impact of early experience. Although the presentation might be too telegraphic for the general reader, it provides pointers to issues, arguments, and evidence that the interested reader cannot help but follow. It provokes the reader to explore methodological and evidentiary trails that one might otherwise ignore. The Clarkes' opening quote from A. N. Whitehouse establishes an appropriate tone for this exploration: "The doctrines which best repay critical examination are those which for the longest time have remained unquestioned" (p. 7). For these reasons, Early Experience and the Life Path would be a superb text around which to organize an upper level undergraduate or graduate seminar on early childhood experience.
A Critical Guidebook
The Clarkes' guidebook opens with a brief historical introduction to orient the reader, summarizing the philosophical and scholarly origins of both the determinist and indeterminist theses. To help the reader assess scientific evidence, the Clarkes provide methodological guidelines. Prospective studies are preferable to retrospective studies and the two study designs often lead to different conclusions. Sample loss can distort follow-up results. Sample selection and methods can limit a study's generalizablity. Finally, tests of critical or sensitive period hypotheses cannot ignore what happens to experimental subjects in the period between the end of experimental intervention and data collection. This important caveat is explicit in the Clarkes' positive thesis: Effects of early experience, although important at the time, will only have prolonged impact if similar advantageous or disadvantageous circumstances persist. This methodological caveat is worth several weeks discussion in a child development course. It is often overlooked in policy and popular discussions, as well as in some scientific contexts.
Once oriented, the Clarkes guide the reader in evaluating three "If P, then Q" hypotheses, by reviewing the evidence that not-Q to conclude that not-P. The three hypotheses (Chapters 3-5) are (a) if early experience has long-term effects, then the impact of early experience should be evident under conditions of normal development, where the quality of early experiences is most likely to persist, in the form of strong correlations between early and adult measures of characteristics; (b) if early experience has long-term effects, then severely deprived and abused children should remain permanently impaired despite improved circumstances in later life; and (c) if early experience has long-term effects, then children who suffer less severe early adversity (late adoption, sociocultural deprivation) are unlikely to show recovery. The evidence for not-Q ranges from studies showing lack of correlation between early and late measures of IQ, shyness, and social deprivation to studies of Romanian orphans and late adoptees to the effects of early intervention programs like Head Start and the Abecedarian Project. All students of child development should have a critical understanding of the research the Clarkes survey in these chapters.
In the spirit of critical assessment, the Clarkes conclude with a discussion of what they believe is the strongest evidence against their thesis. This includes the research of Michael Butter and the English and Romanian study team and Elinor Ames and Kim Chisholm's Canadian-based study on globally deprived Romanian orphans. This work, especially O'Connor et al. (2000), prompt the Clarkes to modify their original thesis, granting that O'Connor et al.'s results leave open the possibility that there is a "sensitive period" of indeterminate duration begining after age six months, where further extreme privation results in "less, and in some cases far less, spectacular gains" (p. 101) after adoption.
Exploring New Terrain
On the basis of the evidence presented, should the Clarkes modify their thesis? Asking this question leads one through some interesting, new terrain on the logic, design, and interpretation of experiments to test sensitive period hypotheses. A sensitive period experiment requires that the experimental intervention vary in age of onset and duration (e.g., suturing kittens' right eyes shut at varying ages for varying durations), with a later measure of the behavior of interest (e.g., visual function). The measured values are then compared with those for a control group that did not experience the intervention. The age and shortest duration for which the results differ significantly between an experimental and the control group defines the sensitive period. We can do these experiments on rats, kittens, and monkeys, but we may not so intentionally manipulate experience during human development. Human studies rely on less well-controlled natural experiments, where groups of people have had unusual early experience, starting at different ages and for varying durations. For example, to test whether there is a sensitive period for second-language grammar acquisition, we can compare second-language performance of individuals who had their first exposure to the second language at varying ages with the performance of native speakers. The scientific interest, interest as distinct from the humanitarian in Romanian orphans is that they have been subjects in a tragic natural experiment, where the "intervention" was early global deprivation and neglect.
O'Connor et al. (2000) reported on one aspect of this natural experiment. In the O'Connor et al. study, the three "experimental groups" of Romanian orphans experienced up to six months, between six and 24 months, or between 24 and 42 months of institutionalization before adoption by families in the United Kingdom. The control group was infants from the United Kingdom adopted before age six months. At age four years, a comparison of the first two experimental groups (0-6 months and 6 24 months of institutionalization) on measures of cognitive competence, showed remarkable developmental recovery among the Romanian orphans, but also found that Romanian infants adopted after age six months, although scoring within the normal range, scored a standard deviation below Romanian infants adopted before age six months. The Clarkes take this result as suggesting the existence of a sensitive period beginning after age six months when children appear to suffer irreversible damage from institutionalization (p. 92).
However, this result itself does not support a sensitive period hypothesis. Although the period of institutionalization varied, the age of assessment did not. The Romanian children were evaluated at age four, six, or both four and six years. This design, as O'Connor et al. (2000) stated, confounds duration of institutional deprivation with duration of adoptive remediation. This is analogous to assessing English proficiency of nonnative speakers at age 18 where one group was first exposed to English before age three and the second group was first exposed to English at age 15. One would expect that speakers with at least 15 years of exposure to English to be more proficient than speakers who had at most three years exposure. Practice effects are not sensitive period effects.
To address this problem, O'Connor et al. (2000) performed a regression analysis that allowed them to compare infants who had experienced up to 18 months institutionalization and between 30 and 48 months adoption with infants who had experienced up to 42 months institutionalization and between 30 and 48 months in the adopted home. They found significant cognitive and developmental differences in favor of the early (0-18 month) adoptees and concluded that duration of deprivation is a more powerful predictor of individual differences than is time in the adoptive home. This analysis makes the O'Connor et al. study analogous to a second-language acquisition study where first exposure to the second language (i.e., time spent in the native country) is a better negative predictor of second-language proficiency than duration of exposure to the second language (i.e., time spent in the new language environment). Such studies are typically interpreted as supporting sensitive period claims.
But here the path can take a new turn. Another, often tacit, assumption in sensitive period experiments is that the experience after the intervention ends and before the assessment is made is the same for all experimental groups. For example, it assumed that 10 years of experience with English as a second language between the ages of 3 and 13 is the same as 10 years of experience with English between the ages of 15 and 25. Recent research on second-language acquisition suggests, however, that early age at immigration is correlated with increased formal schooling in the second language and late age at immigration is correlated with increased home use of the native language. If one corrects for these confounds, there no longer appears to be a sensitive period for second-language grammar acquisition (Flege, Yeni-Komshian, & Liu, 1999). Maybe the Clarkes conceded too much, too soon. Is three years experience in an adoptive home for children ranging in age from birth to 18 months the same as three years experience in an adopted home for children ranging in age from 24 to 42 months? If not, later experience may still be a significant factor affecting the eventual outcome. This is an open question, but one that research could answer. The work on second language acquisition suggests that maybe sensitive periods should be our hypotheses of last resort, rather than our default hypotheses. If studies of child development took this new path, rather than the popular well-trodden one, where might it lead?
Barton, M. A. (1998). Reiner: Justice starts in the high chair not electric chair. Retrieved August 10, 1998, from www.naco.org/pubs/cnews
Clarke, A. M., & Clarke, A. D. B. (Eds.). (1977). Early experience: Myth and evidence. New York: Free Press.
Flege, J. E., Yeni-Komshian, G. H., & Liu, S. (1999). Age constraints on second language acquisition. Memory and Language 41, 78-104.
O'Connor, T. G., Rutter, M., & Beckett, C. (2000). The effects of global severe privation on cognitive competence: Extension and longitudinal follow-up. Child Development, 71, 376-390.
This material was originally published in Contemporary Psychology, APA
Review of Books, June 2002, 268-70. © Copyright 2002 by the American
Psychological Association. Posted with permission. No further reproduction
or distribution is permitted without the written permission of the American
APA Journals http://www.apa.org/journals