nature, nurture, and development

8/14/2019 Nature, Nurture, And Development

1/21

Child Development, January/February 2002, Volume 73, Number 1, Pages 121

Nature, Nurture, and Development: From Evangelism through

Science toward Policy and Practice

Michael Rutter

During the second half of the 20th century there was an immense increase in both empirical findings on, andconceptual understanding of, the effects of nature, nurture, and developmental processes on psychologicalfunctioningboth normal and abnormal. Unfortunately, the good science has also been accompanied by ex-cessive polarizing claims and by unwarranted extrapolations. This article provides a summary review of thereal gains in knowledge, outlines some of the misleading claims, and notes the potential for research and forscience-led improvements in policies and practice. The need to bring about a better interpretation of genetic,psychosocial, and developmental research strategies and theoretical concepts is emphasized.

INTRODUCTION

Over the last half century, there has been an explosionof knowledge on the effects of nature, nurture, anddevelopmental processes. As a result, we have a muchimproved understanding of many of the mechanismsinvolved in normal and abnormal development,which carries with it a huge potential for improvingchildrens lives. Unfortunately, these advances have

been accompanied by as much misleading scientificevangelism and journalistic hype as by good science

and honest reporting. As a consequence, both the pagesof scientific journals and the media have been full of themost absurd confrontations and polarizations. Thesehave given rise to an unhelpful level of misunder-standing of the true scientific advances and, more es-pecially, about their meaning and the implications forpolicy and practice. Of course, there have also beennumerous examples of good reporting by scientistsand by journalists. There is every reason to be in-debted to both. The need is to avoid the twin dangersof destructive cynicism and gullible expectation.

Nature, nurture, and development are dealt with

in this article as separate topics (with the focus beingmainly on their effects on psychopathology). In eachcase, the real advances in knowledge are consideredfirst, some of the misleading claims are outlined sec-ond, and the potential for research and for improve-ments in policies and practice are noted third. Al-though these are considered as supposedly separateinfluences, the truth is that they are closely inter-twined. The separation is heuristically useful for test-ing causal hypotheses, but it is crucial that such hy-potheses deal with the different forms of interplay

that may be occurring. To a considerable extent, it isthe failure to do so that has led to many of the polar-izing battles and absurd claims.

NATURE: GENETIC RISK ANDPROTECTIVE MECHANISMS

With regard to the influences that reflect nature,quantitative genetics and molecular genetics are dis-cussed separately, because they have rather different

patterns of strengths and limitations.

Quantitative Genetics

Quantitative genetics uses various population de-signs (most particularly twin and adoptee studies) toquantify the relative strength of genetic and nonge-netic factors with respect to population variance; thatis, individual differences with respect to some trait ordisorder (McGuffin & Rutter, in press; Rutter et al.,1990; Rutter, Silberg, OConnor, & Simonoff, 1999a).Nongeneticists tend to assume that the only focus is

on the heritability quotient, but that is not the case.The available techniques are able to partition geneticinfluences into those that are additivethat is, due toa mixture of many genes without there being anyrequirement for a particular pattern or combinationand those that are nonadditivethat is, reflect interac-tions among different genes (epistasis) or among dif-ferent alleles of the same gene (dominance). Similarly,nongenetic influences can be subdivided into so-calledshared and nonshared effectsthose that tend tomake siblings similar and those that tend to make

2002 by the Society for Research in Child Development, Inc.All rights reserved. 0009-3920/2002/7301-0001

2001 Presidential Address to the Society for Research inChild Development.


2/21

2 Child Development

them unalike, respectively. Geneenvironment corre-lations and interactions may also be identified. Al-though traditional analyses have, in the past, tendedto assume that genetic and nongenetic influences areentirely separable, there is no need to make that as-

sumption. Quantitative genetic studies have increas-ingly tested for, and found, major interplay betweengenetic and non-genetic factors, such that the out-comes cannot sensibly be attributed to just one or theother, because they depend on both (Rutter & Silberg,in press).

Quantitative genetic designs rely on many as-sumptions, and it is crucial that these be put to the testin a rigorous fashion, making explicit just which arenecessary (Rutter, Pickles, Murray, & Eaves, 2001; Rut-ter et al., 1999a). Critics have rightly pointed to prob-lems, but any dispassionate reading of the evidence

leads to the inescapable conclusion that genetic fac-tors play a substantial role in the origins of individualdifferences with respect to all psychological traits, both normal and abnormal (McGuffin & Rutter, inpress; Rutter, Silberg, OConnor, & Simonoff, 1999b).In a few cases (such as with autism, schizophrenia, bi-polar disorder, and attention deficit disorder with hy-peractivity), genetic factors account for most of thevariance in populationsover 70%. For the great ma- jority of psychological characteristics, however, thegenetic effects are not as strong. Thus, the heritabili-

ties of unipolar depression, delinquency, and parent-ing qualities are in the 20% to 40% range.Two different answers can and should be given to

the question of how much faith we can place on theseestimates. First, there is every reason to believe thatthey are close to correct. This conclusion is based onthe fact that numerous studies have produced broadlycomparable findings; that substantial genetic effectsare evident from twin, adoptee, and family data; andcareful consideration of the findings indicates thatnongenetic explanations for the pattern of resultsfound lack plausibility.

The second answer, however, is that the estimatesmust be regarded as very approximate. This is becausethe precise figures produced by different investiga-tions are often rather different, and different methodsoften come up with somewhat discrepant findings.The use of parent reports and child reports, and oftwin and adoptee data, have made these points appar-ent. In the past, arguments have raged over whetherthe heritability of, for example, IQ is this figure or thatfigure, but the important thing to bear in mind is thatit does not really matter. With respect to some traits,there is real doubt whether the heritability in the pop-ulations studied is 20% or 40% or 60%. There are notheoretical or policy implications from such wide es-

timate variations (surprising though that may seem),because all indicate substantial genetic effects and thefigures are population specific; that is, they applyonly to the particular samples studied at a particulartime. Even a heritability as high as 90% does not mean

that changed environmental conditions could notmake a huge impact. This is not just speculation; thefindings with respect to height both in childhood andin adult life show the reality (Tizard, 1975). Height isone of the most strongly genetically influenced of allhuman characteristics, but it has increased enor-mously over the course of the 20th century (Kuh,Power, & Rodgers, 1991; van Wieringen, 1986), almostcertainly due to improved nutrition. Genetic influ-ences on the timing of the menarche are also strongbut, again, the age of menarche has fallen greatly overthe last 100 years or so (Tanner, 1962). It is clear that

environmental factors can bring about major changesin features that are strongly genetically influenced.

Despite this reservation, we may conclude thatthere can be no doubt about the importance of genetic(as well as experiential) influences on individual dif-ferences, especially with respect to persistent traitsand chronic or recurrent disorders. Genetic influencesare particularly strong with respect to several mentaldisorders, where the evidence from other biologicalstudies indicates the likelihood that brain abnormali-ties are implicated in the causal processes. This applies,

for example, to autism (Bailey, Phillips, & Rutter, 1996;Lord & Bailey, in press) and schizophrenia (Keshavan& Murray, 1997), and, in both cases, the quantitativegenetic findings point to the likelihood that severalinteracting genes are involvedthat is, synergisticinteraction among genes may be implicated.

Strengths and achievements. Several other, rather dif-ferent aspects of the genetic findings warrant empha-sis. To begin, it is necessary to note the pervasiveness ofgenetic influences across all psychological traits, eventhose involving attitudes or social behavior (Plomin,1994). Thus, genetic effects have been found for fea-

tures as diverse as divorce (Jockin, McGue, & Lykken,1996; McGue & Lykken, 1992), religiosity (Eaves,DOnofrio, & Russell, 1999), and various aspects ofparenting style (Kendler, 1996a). Critics have been quickto scorn such findings, arguing that it is ridiculous tosuppose that there could be a gene for divorce or crime(Rose, 1995, 1998). They are right, of course, that it isextremely unlikely that such a gene could ever occur.They also point to the indefinite boundaries of thephenotype, or behavior, being studied. Both argu-ments, however, miss the point completely. The mes-sage is that the workings of the mind are based on thefunctioning of the brain, and that genetic influencesapply to individual differences on all somatic features.


3/21

Rutter 3

It must be anticipated, therefore, that there will be a ge-netic effect on all behaviors. Biologically speaking, thisis exactly what one would expect, and what has beenfound. The challenge is to find out how these geneticeffects are mediated because, obviously, they are most

unlikely to operate directly on the social behavior asobserved.The second point is that the genetic evidence is

equally consistent in showing the major importanceof nongenetic influences (McGuffin & Rutter, in press;Rutter et al., 1999b). With many psychological charac-teristics, their influence is somewhat greater than thatof genetics and, even with the traits that are moststrongly genetically influenced, environmental ef-fects are far from trivial.

A third important finding in the realm of psychopa-thology is that it is quite common for the same genetic

factors to underlie supposedly different types of men-tal disorder. For example, with respect to anxiety anddepressive disorders, much of the shared variance isexplicable in terms of the temperamental or personal-ity trait of neuroticism (Kendler, 1996b). Similarly,there are shared genetic factors involved in the liabilityto oppositional/defiant, conduct, and hyperkinetic/attention deficit disorders, even though the prevail-ing psychiatric classification diagnostic systems clas-sify them as separate conditions (Rutter, 2001). Asomewhat related point is that often the genetic fac-

tors seem to operate across a broadly distributed con-tinuum, rather than just at the extreme disorder endof apparent abnormality (Plomin & Rutter, 1998). Aswith the rest of the internal medicine, it seems thatmany genetically influenced risk factors are dimen-sional rather than categorical a finding that has im-portant implications for our understanding of causalprocesses. In the past, there has often been the as-sumption that causal influences operate on disordersas such, and that these influences are different fromthose that underlie individual differences within thenormal range. It is now evident that often this is not

the case. Just as differences in cholesterol level acrossthe whole range from unusually low to unusuallyhigh are associated with variations in the risk for is-chaemic heart disease, so, too, are quantitative differ-ences in temperamental features associated with vari-ations in the risk for psychopathology.

A further important finding is the occurrence ofgeneenvironment interactions (Rutter & Silberg, inpress). The evidence is quite limited, but severalstudies have suggested that in the case of both antiso-cial behavior and depression, environmental risk fac-

tors operate most strongly with genetically vulnerableindividuals. In other words, adverse environmentsoften have the least impact on those who are not

genetically vulnerable and the most impact on the ge-netically susceptible.

The last example of a specific finding of importanceconcerns the differentiation between shared andnonshared environmental effects (Plomin & Daniels,

1987). It was argued that results suggested that formost psychological features, the strength of non-shared effects outweighed those of shared effects. Thisfinding was important in serving as a reminder that itcould not be assumed that environmental influencesimpinged equally on all individuals in the same fam-ily. As a consequence, environmental influences oftenserve to make children in the same family different,rather than the same.

The potential importance of the above findings isconsidered in conjunction with the findings on molec-ular genetics; however, it would seem evident that

there is a powerful case for the importance of geneticinfluences on psychological characteristics. Thus, it isnecessary to consider some of the problems associ-ated with the ways in which quantitative geneticshave been presented. These have taken several differ-ent forms.

Misleading claims. The first problem is that thereare misleading presentations of findings (see Rutter,Pickles, et al., 2001; Rutter et al., 1999a). For example,one study found that shared environmental effectswere substantially stronger than nonshared ones, but

the abstract made no mention whatsoever of shared ef-fects and emphasized only the rather minuscule non-shared effects (Pike, McGuire, Hetherington, Reiss, &Plomin, 1996; see Rutter, 2000e). Other examples in-clude failures to mention serious attrition biases insampling, downplaying of the environmental rangeproblems in adoptee samples, and the use of simplify-ing assumptions in modeling when it is known fromother evidence that the assumptions are wrong (e.g.,that supposedly there is no assortative mating, geneenvironment correlations, or geneenvironment inter-actions in the case of antisocial behavior). All too

often, claims have been maintained long after the em-pirical evidence has indicated that substantial modifi-cations were necessary. Thus, the original claim thatnonshared environmental effects far outweigh sharedones is no longer supportable as a general proposi-tion. Once continuities over time and measurementerror are taken into account, there is much more of a

balance between shared and nonshared effects (seeReiss, Neiderhiser, Hetherington, & Plomin, 2000; Rut-ter, Pickles, et al., 2001; Rutter et al., 1999a). That doesnot diminish the importance of the general point that

experiences frequently impinge differentially on dif-ferent children in the same family, but there are toomany exceptions to the claim that nonshared effects


4/21

4 Child Development

predominate for this supposed predominance to bemaintained as a rule. What remains of interest, how-ever, is the consistent finding that some maladaptivebehaviors with broadly similar heritabilities are usu-ally found in several children within the same family

(this would apply to antisocial behavior, for example),whereas others usually affect just one child (this wouldbe more frequent in the case of depression or anxiety).

A second problem has come from misleading ex-trapolations. For example, several commentators (see,e.g., Harris, 1998; Rowe, 1994; Scarr, 1992) have usedthe supposed findings on the nonimportance ofshared environmental effects to conclude that familyfeatures have little or no effect on childrens psycho-logical development. The claims on nonimportance ofshared environmental effects are themselves over-stated but, in this connection, the more important is-

sue is that the shared or nonshared environment ef-fects are inferences that derive from evidence thatenvironmental factors make siblings alike or differ-ent. They have nothing to do with whether the envi-ronmental influence is within or outside the family. Itis quite possible for familywide influences (such asdiscord or conflict) to have largely nonshared effectsjust because conflict impinges more on one child thananother and/or because some children are more vul-nerable to the effects of discord (as a result perhaps oftemperamental features). It is rare for behavior genet-

icists to make use of evidence deriving from nonge-netic strategies. As a consequence, they have fre-quently ignored well-based findings that run counterto their claims.

In addition, all too often there has been a cavalier ig-noring of evidence either on the importance of re-stricted samples (see Stoolmiller, 1999) or of violationof key assumptions of the twin design (Rutter, Pickles,et al., 2001; Rutter et al., 1999). Thus, for example, it isevident that with respect to both antisocial behaviorand depression, there is usually likely to be a violationof the equal environments assumption (EEA), and

that is indeed what has been found when it has beenlooked for in an appropriate manner: the environ-mental risk factors for both antisocial behavior anddepression are correlated with genetic susceptibility, but, in addition, these same risk factors are signifi-cantly associated with differences in psychopathol-ogy within monozygotic pairs who, of course, are ge-netically the same. The concern here is not that afailure to take this violation of EEA into account hasrather inflated heritability estimates (although that istrue), but rather that there is a failure to recognize theimportant implication that some genetic effects oper-ate indirectly via the environment, rather than directlyin a way that is separate from environmental risk. It is

usual in presenting quantitative genetic findings to in-corporate geneenvironment correlations and interac-tions within the genetic influence term. The rationalefor this, apart from convenience, is that the genetic fac-tors drive what is happening. In a sense, this is the

case, but the effects on the psychological features aredependent on the combination of genetic and environ-mental factors working togethernot just additively,but synergistically (see Rutter & Silberg, in press; Sil-berg, Rutter, Neale, & Eaves, 2001).

A similar problem has occurred with the extrapo-lations of the pervasive finding of geneenvironmentcorrelations (Plomin & Bergeman, 1991). This findinghas been used to argue that many of the risk effects at-tributed to environmental factors are actually geneti-cally mediated. The charge that most psychosocialresearch has failed to consider this possibility, and

hence that many claims regarding psychosocial influ-ences are unwarranted, is fair. Although downplayed by geneticists, however, the same genetic researchshows that there is environmental risk mediation (seeRutter, 2000a). Also, it is misleading to suppose thatjust because genetic factors influence the occurrenceof an environmental risk factor, this must mean thatthe risk process is genetically mediated. This assump-tion does not follow because there is no necessaryconnection between the causes of the origin of a riskfactor and its mode of risk mediationas the exam-

ple of smoking clearly illustrates (see Rutter, Silberg,& Simonoff, 1993).Perhaps, most crucially, what has been misleading

is the claim that quantitative behavior genetics consti-tutes a causal theory (Scarr, 1997). It constitutes noth-ing of the kind. Quantitative behavior genetics doesprovide a highly effective way to partition the vari-ance, and that is immensely useful for a variety ofpurposes (including the study of naturenurture in-terplay and the developmental processes underlyingcomorbidity). Knowing that a trait is genetically in-fluenced, however, is of zero use on its own in under-

standing causal mechanisms. The same, of course, ap-plies to parallel claims with respect to environmentalinfluences. To be of any use for policy or practice, it isnecessary to know much more with regard to the spe-cifics and how they work. The point is that geneswork in quite a diverse range of ways and the impli-cations are quite different according to the details.

Molecular Genetics

This last point is central to the topic of molecular ge-netics (see McGuffin & Rutter, in press; Rutter et al.,1999a). A key consideration here is that what is left unan-swered by the black box analyses of quantitative genetics


5/21

Rutter 5

can often be answered through molecular genetics.Molecular genetics does not involve the quantificationof genetic effects, but rather the identification of spe-cific individual genes that are involved in the suscep-tibility to particular features (either physical or mental).

Findings with respect to psychological characteris-tics and to psychopathology are still at a very earlystage, but already there have been some importantfindings. Quite a lot has been learned about unusual ge-netic mechanisms, the discovery of which has providedexplanations for what had hitherto been some verypuzzling genetic findings on patterns of inheritancethat seemed to contravene Mendelian expectations,and patterns of inheritance in which disorders arise atan earlier age in each succeeding generation. Thus, forexample, trinucleotide repeat sequences that expandthrough intergenerational transmission have been

shown to be responsible for a range of neuropsychiatricdisorders, including the fragile X anomaly (Skuse &Kuntsi, in press). The phenomenon of genomic imprint-ing (see Keverne, 1997; Ohlsson, Paldi, & Graves, 2001;Reik & Walter, 2001), in which gene expression is al-tered according to whether the gene is transmittedthrough the father or through the mother, is anotherfascinating phenomenon that has been shown to be im-plicated in two syndromesthe Prader-Willi syn-drome and Angelman syndromeboth of which areassociated with mental retardation. Mitochondrial in-

heritance has also been found to be involved in severalneurological disorders. This fact is of interest becausethe mitochondria, which are outside the nucleus, aretransmitted only through the mother and because mu-tations arise throughout life. Also, we have a betterunderstanding of Lyonization, the process by whichone of the X chromosomes in females is switched off(Avner & Heard, 2001). This process incidentally, cre-ates a mechanism that constitutes one of the ways inwhich identical twins occasionally differ with respectto wholly genetic single-gene disorders. Also, micro-deletions of chromosomes may, it now seems, be re-

sponsible for a proportion of cases of mental retarda-tion (Flint et al., 1995; Knight, Udalova, et al., 1999).

Over the last few years, a few specific individualgenes have been identified that are involved in the li-ability to mental disorders. The ApoE4 allele that isinvolved in the liability to Alzheimers disease is the

best known in the psychopathological arena (Plass-man & Breitner, 1996; Rubinsztein, 1995), but thereare also genes that affect dopamine metabolism thathave been implicated in attention deficit disorderwith hyperactivity and other child psychiatric condi-

tions (Levy & Hay, 2001).In the field of internal medicine, it has been particu-larly important that genes have been associated with

very marked individual variations in response to envi-ronmental hazards (Rutter, 2000a). This phenomenonhas become evident, for example, with respect to smok-ing and coronary artery disease, vulnerability to thesequelae of head injury, and responses to malaria and

other infections (Hill, 1998; Knight, Regan, et al.,1999). In addition, research is beginning to identifygenes that influence individual responses to therapeu-tic medicationthe field of pharmacogenetics (Evans& Relling, 1999). In other words, the black-box con-cept of geneenvironment interaction, which derivedout of quantitative genetics, is now being shown tohave meaningful substance through advances in mo-lecular genetics.

Finally, research in the field of reading difficultieshas shown that there is the possibility that differentgenes may be involved in different aspects of syn-

dromes that had hitherto been thought to be rela-tively unitary (Grigorenko et al., 1997; Grigorenko,Wood, Meyer, & Pauls, 2000). It is too early to knowwhether these findings will prove to be solid and,more specifically, whether this genetic effect on specificfeatures within syndromes will be found to apply toother conditions. Nevertheless, already the findingsserve as a reminder that in understanding causal pro-cesses, it is necessary to appreciate that mental disor-ders may involve several different causal pathwaysthat are responsible for different aspects of syndromes.

Two very important publications occurred in Feb-ruary 2001: both Science and Nature announced thedraft sequence of almost the entire human genome.There is no doubt that this was quite a remarkable sci-entific achievement, as well as one that has produceda few surprises. I consider below why and how thisachievement is going to make a real difference to thediscovery of susceptibility genes associated with psy-chological features, as well as those with any other as-pect of the human condition.

Misleading claims. The downside of molecular ge-netics has involved both what has been said and what

has not been brought out. With respect to the former,the most obvious hype and scientific evangelism haveconcerned claims both about the speed with whichsusceptibility genes (genes that affect the liability ofsome trait or disorder, but which do not determinethe trait or disorder directly) are likely to be discov-ered and the extent to which this is going to have clin-ical utility. There has been much talk about how it willsoon be possible to have genetic profiles at birth thatwill enable us to know all about our propensities andsusceptibilities, and the diseases that we are going to

develop. This talk is highly misleading for several dif-ferent reasons. As Weatherall (1999, p. 2008) put it in astraightforward summary conclusion: Many state-


6/21

6 Child Development

ments being made about the imminence of accuratepredictive genetics . . . are simply not true.

What is true is that the sequencing of the human ge-nome will make it much easier than before to identifydisease genes, because candidate genes can readily be

detected using computers to question public sequencedatabases, which can then be followed up with muta-tion screening of plausible candidates. This is wherethe tool of bioinformatics really comes into its own.Databases can be used to search the human genomefor proteins that are similar in other organisms, or toidentify all the proteins of a particular family (such asthose affecting a particular set of disease processes).Already, this strategy has facilitated the identification ofgenes implicated in the causation of important humandiseases (such as complete color blindness or early-onset Alzheimers disease; see International Human

Genome Sequencing Consortium, 2001). The first cau-tion, however, is that it is going to be much more dif-ficult to discover susceptibility genes for multifacto-rial disorders, especially when genes play only acontributory role in the context of major environmen-tal influences. The difficulties that have attended thesearch for susceptibility genes for a well-defined dis-order such as juvenile diabetes (Todd, 1999) provide awarning of how it is likely to be even more difficult inthe field of mental disorders, in which definition of aphenotype is so much more problematic. It may turn

out, as enthusiasts have claimed (Plomin & Crabbe,2000), that we will soon be awash with susceptibilitygenes for psychological characteristics, but I ratherdoubt claims on the speed with which this will hap-pen. The more important caveats, however, are of adifferent kind.

Three main points need to be stressed. First, find-ing the susceptibility genes is the relatively easy part.What is likely to prove much more difficult is deter-mining what these genes do; that is, their effects onproteins and the ways in which these protein effectslead to particular psychological outcomes (see Rutter,

2000b). Research will be needed in three broad areas:transcriptomics (the study of which genes are switchedon in particular cells), proteomics (which looks at theinterplay among proteins in cells), and structuralgenomics (which tackles the question of the three-dimensional structures of all the proteins encoded bygenes). Although all of this is potentially doable, solv-ing all the research problems in this overall field offunctional genomics is going to be quite difficult andwill undoubtedly take a long time. Second, whendealing with multifactorial traits, and that means vir-tually the whole of those involved with psychologicalfeatures, there is the further challenge of understand-ing how genes are involved in the interplay with en-

vironmental risk factors. This requires advances inthe field of molecular epidemiology and, for this tosucceed, there will have to be advances in the mea-surement of environmental risk factors on very largesamples, at least as big a challenge as in the field of

molecular genetics itself. Genetic enthusiasts seem tohave paid almost no attention to this need. Third,there has been a great underplaying of the extent ofgeographic and ethnic variability in genetic effects(see Rutter, 2001). All too often, enthusiasts write orspeak about genetics as if, because genes are a fixedpart of the constitution, their effects should be univer-sal across populations. However, they are not. Forexample, there is huge geographical variation inwhich particular alleles are responsible for thalas-saemia (sickle cell disease; see Weatherall & Clegg,2001); and, for reasons that remain ill understood, the

apolipoprotein E4 seems to carry less of a risk forAlzheimers disease in individuals of African or His-panic heritage (Farrer et al., 1997).

Psychiatric molecular genetics got off to a thor-oughly bad start with premature claims that subse-quently had to be withdrawn (see Rutter, 1994). It is tothe credit of the researchers involved in the study ofaffective disorder that they were quick to note theproblems and to withdraw their own earlier claims(Kelsoe et al., 1989). Withdrawal of the claims withrespect to schizophrenia took quite a bit longer. It

continues to prove incredibly difficult to replicatefindings in the identification of susceptibility genesfor multifactorial psychiatric disorders. Surprisingly,however, distinguished researchers still allow them-selves to get carried away with the excitement ofunconfirmed findings. For example, a few years agothere was a report that one of the genes responsiblefor individual variations in intelligence had been dis-covered (Chorney et al., 1998), and the media went totown with academics making astonishing statementsthat this revolutionized our thinking about the im-portance of genetic factors in intelligence. The re-

search, however, had not taken into account stratifi-cation biases, and some years later it still remainsunconfirmed by independent investigators. Indeed,the report of DNA pooling as used with a genomewide scan by the same research team did not includethe same finding (Plomin & Craig, 2001). Identifica-tion of genes that play a role in individual differencesin intelligence may well come, but it is dubiouswhether this will provide an understanding of the ba-sis of intelligence (Rutter, 2000d).

Potential of molecular genetics. How do we come upwith an appropriate balance with respect to the po-tential of molecular genetics? To begin, there is theinitial question of whether molecular genetics will


7/21

Rutter 7

live up to its promise in identifying susceptibility(and protective) genes for psychopathology and psy-chological traits. The identification of multiple genesof small effect, particularly when the phenotypes aredifficult to define and lack strong validity, will be

quite difficult. Nevertheless, through the use of mul-tiple research strategies, it is likely that delivery willcome (Evans, Muir, Blackwood, & Porkus, 2001;Owen, Cardno, & ODonovan, 2000) even if it takeslonger than some expect. It may be, however, thatgreater attention will need to be paid to epigeneticmisregulation of genes, as well as DNA sequencevariation (Petronis, 2001). In essence, epigenetics re-fers to the genetic processes involved in the expres-sion of particular genes in individual cells. Although,ordinarily, the genotype is the same across cells, itsstructural and functional consequences are not. Both

genomic imprinting and X chromosome inactivation(see above) are good examples of this kind, but restruc-turing of DNA methylation during gametogenesisprovides the opportunity for de novo epigenetic er-rors; hormone-induced epigenetic changes also mayplay a role in the differential susceptibility of malesand females to complex diseases.

Assuming that the relevant genes are found, there is aconsiderable potential for advances in knowledge thatshould bring worthwhile human benefits (McGuffin& Rutter, in press; Plomin & Rutter, 1998; Rutter, 2001;

Rutter & Plomin, 1997). First, and perhaps most cru-cially, genetic advances should foster research thatwill lead to a much better biological understanding ofthe causal processes involved in such serious disordersas autism or schizophrenia. Up until now, biological re-search has proved frustratingly inconclusive on thespecifics of the underlying neuropathophysiology forsuch conditions. Genetic findings, in and of them-selves, will not provide any understanding of suchcausal processes, but what they should do is provideinvaluable leads as to how complementary biologicalresearch can identify what is involved. Second, pro-

vided that genetic research moves ahead in harnesswith environmental research, genetic advances shouldenable us to gain a much greater appreciation of the in-terplay between nature and nurture. In other words,one of the really important potential gains is that ge-netic findings should greatly facilitate the study of en-vironmental risk mechanisms. That is not going to beat all easy, both because the effects of single genes arequite minor, and because the same applies to individ-ual environmental risk factors. The cumulative effectof genetic risks, of environmental risks, and espe-

cially of their interplay, are very great, but it is clearthat the effective study of gene environment correla-tions and interactions is going to require attention to

many genes, and not just one; and attention to manyenvironmental risk factors, and not just one.

Third, there is substantial potential in the field ofpharmacogenetics (Evans & Relling, 1999; Wolf, Smith,& Smith, 2000). With psychopathology, just as in the

whole of internal medicine, it is obvious that there arehuge individual variations in how people respond totherapeutic medication. Clearly, genetic factors willprove to play an important role in that individualvariation. It follows that an understanding of howthese operate will be hugely helpful in allowing ther-apeutic interventions to be tailored in ways that arespecific to individuals, and will aid in the under-standing of pharmacological actions and how they

bring therapeutic benefits.Fourth, there is a limited potential for using ge-

netic findings to improve the classification and diag-

nosis of mental disorders. For example, genetic find-ings could help in sorting out which social deficits arepart of the autism broader phenotype and which, al-though superficially similar, are not, because they donot involve the same susceptibility genes (Rutter,2000b). The reason why this constitutes a more lim-ited potential is that most diagnoses are based onpathophysiology and not on causal factors. For exam-ple, the diagnosis of diabetes is based on laboratoryfindings with respect to glucose metabolism, and noton whether the patient has a specific susceptibility

gene. Similarly, coronary artery disease is diagnosedon the basis of atheroma of the coronary arteries, andnot on the role of smoking, raised cholesterol levels,or clotting factors in etiology.

In theory, gene therapy could play some role in thetreatment of mental disorders, but it is likely that thisrole will be quite small; not because of the difficultiesat the moment in gene delivery, but rather because ofthe problems of applying this technique to multifacto-rial disorders. A further potential benefit will comefrom the individualization of genetic risks, an advancethat clearly will be of benefit in genetic counseling

(McGuffin & Rutter, in press; Rutter & Plomin, 1997).The huge potential benefits are obvious, even

though it is clear that it will take quite some time formolecular genetic findings to lead to major improve-ments in clinical practice. In recognizing the reality ofthese great benefits, it is equally important that atten-tion is paid to the attendant ethical risks involvingdiscrimination in its many various forms, and thepossibility of relative neglect of public health issuesand of the need to study, and take action on, environ-mental risks (see Buchanan, Brock, Daniels, & Wikler,

2000; Rutter, 1999a). It should be possible to avoidthese disadvantages, but only if we accept their real-ity and act accordingly.


8/21

8 Child Development

NURTURE: ENVIRONMENTAL RISKAND PROTECTIVE MECHANISMS

Strengths and Achievements

In considering next the advances in understanding

of environmental risk and protective processes, thefirst point that stands out is that there is good evidenceof the environmentally mediated effects of specific en-vironments (Rutter, 2000a). It has been necessary for re-search to address two serious issues: the need to checkthat the effects are truly environmentallyrather thangeneticallymediated, and that the direction of causalinfluence is from the environment to the child, ratherthan the reverse. A considerable range of effective re-search strategies is available for these purposes (Rutter,Pickles, et al., 2001), including multiple variants of thetwin design, several varieties of adoptee design, natu-

ral experiments, migration strategies, and studies ofintervention effects. The research findings have beenconsistent in showing the psychopathological risksassociated with (1) persistent discord and conflictparticularly when it involves scapegoating or otherforms of focused negativity directed toward an indi-vidual child, (2) a lack of individualized personal care-giving involving continuity over time (as is usuallythe case with an institutional upbringing), (3) a lack ofreciprocal conversation and play, and (4) a negativesocial ethos or social group that fosters maladaptive

behavior of one kind or another. The risk and protec-tive factors involve not only the immediate family, butalso the peer group (Rutter, Giller, & Hagell, 1998), theschool (Maughan, 1994; Mortimore, 1995, 1998), andthe broader social community (Leventhal & Brooks-Gunn, 2000). It is also important that research findingshave indicated that various factors once thought tocarry serious risks for mental disorder, in actuality donot. Thus, for example, it is clear that parental loss orseparation carries quite mild risks unless the loss leadsto impaired parenting or other forms of family mal-adaptation. Similarly, it has long been clear that it mat-

ters little when weaning takes place, when children aretoilet trained, or the type of disciplinary techniqueused (Maccoby & Martin, 1983).

An important distinction that increasingly has beendrawn, as a result of empirical research findings, is thatbetween proximal and distal risk processes (see Rutteret al., 1998). Thus, for example, although parental losscarries with it little direct (or proximal) psychopatho-logical risk, it is important because, in certain circum-stances, it predisposes to other psychosocial risks andmakes adaptive parenting more difficult. Poor parent-ing does predispose to mental disorder, whether or notit is associated with parental loss; whereas parentalloss does not predispose to disorder, if poor parenting

does not follow. In the same way, poverty has a quitelimited role as a proximal risk factor, but is rathermore important as a distal risk factor that makes co-hesive and harmonious family functioning more dif-ficult. The same seems to apply to inner city life: It is

statistically associated with increased rates of childpsychiatric disorder, but these risks are mediated, not by the effects of city life directly on children, butrather through their effects on family functioning andtheir associations with less positive schooling (Rutter,1979a; Rutter & Quinton, 1977).

A major shift in studies of psychosocial risk factorshas come with the awareness of the major individualdifferences in response, and the huge heterogeneity inoutcome. The findings have focused attention on thephenomenon of resilience, meaning relatively goodpsychological functioning despite the experience of

serious psychosocial adversities (Rutter, 1999b, 2000c).There have been important methodological chal-lenges to overcome in studying resilience, but there isnow evidence to demonstrate the reality of the phe-nomenon. Some useful leads on the factors that pro-mote resilience have been obtained; but, so far, theyare just thatleads, rather than established knowl-edge on mechanisms.

For a long time, one of the findings that made manypeople reluctant to take seriously the possibility thatenvironmental risks played a major role in the causa-

tion of mental disorder was the apparent lack of speci-ficity of effects (Steinberg & Avenevoli, 2000). The viewwas that if negative experiences predisposed to all theills of humankind, they might have a contributory pre-disposing role, but it was unlikely that they constituteda key causal influence. Two things have changed thatsituation somewhat. First, there is now some evidencesupporting a degree of specificity of effects. For exam-ple, an institutional rearing has been found to predis-pose to so-called disinhibited attachment problemsand patterns of inattention and overactivity (see Rut-ter, in press-a; Rutter, Kreppner, OConnor, & the ERA

Study Team, 2001). When the institutional rearinghas been accompanied by severe global deprivation,but not otherwise, it seems to predispose to atypicalquasiautistic patterns and cognitive impairment. Se-vere and unusual stress experiences (such as exempli-fied by shipwrecks) are associated with a range ofphenomena that has come to be termed posttraumaticstress disorder (Yule, in press). This is by no meansthe only form of psychopathology associated with se-vere and unusual stress, but it is a characteristic pat-tern. Family disorganization and discord are particu-larly associated with antisocial behavior (Rutter et al.,1998). In adults, stresses involving the threat of futuredanger tend to be associated with anxiety, whereas


9/21

Rutter 9

those involving the feeling of psychological loss seemto be particularly likely to predispose to the onset of adepressive disorder. The same most likely applies inchildhood (Eley & Stevenson, 2000). It should be notedthat it is not the physical loss that seems important, but

rather the long-term threat that is implicit in the loss ofa love relationship, or from a public humiliation. Thus,although there are many nonspecific effects of psycho-social adversity (Steinberg & Avenevoli, 2000), it hasalso become evident that there is more specificity thanhad been apparent some years ago.

The second point is that many adverse experiencesinvolve a range of disparate elements, each of whichmay carry a relatively specific risk effect. Becausethese are multiple, however, there is a false impres-sion of nonspecificity. Cigarette smoking provides anobvious example: It predisposes to an apparently het-

erogeneous range of medical disordersincludingosteoporosis, lung cancer, coronary artery disease, em-physema, and wrinkling of the skinwhich wouldseem to suggest a lack of specificity. This assumption,however, is incorrect (see Rutter, 1997). With severalof these outcomes, it is known that the effects arespecificinvolving features such as carcinogenic tars,carbon monoxide, or nicotinic effects on blood ves-sels. It is just that cigarette smoking involves quite acollection of different risk processes. Comparable ev-idence is lacking with respect to psychosocial risks,

but it is highly likely that the same applies. For exam-ple, parental depression involves genetic risk (i.e., it ismore likely that offspring will develop depressivedisorders themselves), but parental depression alsopredisposes to family breakdown and family discord,which, in turn carry risks for antisocial behavior.

At one time, almost all the focus was on adverserearing experiences, but it has become apparent that itis also necessary to pay attention to the possibility ofprenatal risks of various kinds. It is now clear that amothers ingestion of large amounts of alcohol duringthe early months of pregnancy carries risks of damag-

ing effects on the development of the fetus, effects thatare evident later in relation to both somatic abnormal-ities and behavioral disturbanceparticularly in theform of inattention and overactivity (Stratton, Howe,& Battaglia, 1996; Streissguth & Kanter, 1997). Thereare probably comparable risks associated with otherforms of substance abuse in pregnancy (Mayes, 1999),and there is some indication that prenatal damage ofother kinds is also associated with increased psycho-pathological risks (Munk-Jrgensen & Ewald, 2001).

Finally, there has been the successful development

of various forms of psychosocial interventionboth toprevent psychopathology (Offord & Bennett, in press)and to alleviate disorder (Brent, in press). Knowledge

on the mechanisms involved in therapeutic efficacy isdecidedly limited still (but see Forgatch & De Garmo,1999; Vitaro, Brendgen, Pagani, Tremblay, & McDuff,1999); nevertheless, there is good evidence that cer-tain forms of intervention do provide real and worth-

while benefits.

Misleading Claims

The period extending from the 1950s to the early1970s may be viewed as one of rampant environmen-talism. There was an uncritical acceptance of the lastingand irreversible effects of early childhood experiencesand of the extent to which social disadvantage consti-tuted a major cause of mental disorder. The initialclaims with respect to maternal deprivation (Bowlby,1951) constitute one example. The extrapolation to the

supposed permanently damaging effects of day-care(World Health Organization Expert Committee onMental Health, 1951) constitutes an even more strikingexample, as does the naive expectations of somepeople with regard to how much could be achieved

by brief interventions in the preschool years, such asthose initiated by Head Start (see Clarke & Clarke,1976; for balanced reviews see Zigler & Styfco, 1997;Zigler & Valentine, 1997). The background was one of awell-justified concern to better the lives of young chil-dren, together with an awareness of the many things

that needed to be righted in the care of such children.The defenders of the field would undoubtedly arguethat it was necessary to overstate claims of environ-mental effects to bring about political action. That maywell have been true, but, from a scientific perspective,there was a serious neglect of the need to provide rig-orous tests of environmental mediation hypotheses,and a comparable ignoring of the need to differentiate

between person effects on the environment and envi-ronmental effects on the individual (see Bell, 1968; Bell& Chapman, 1986). In addition, there was a failure toappreciate the substantial continuities in environmental

disadvantage and, therefore, an exaggeration of the ex-tent to which persistent sequelae derived from theearly environment, rather than from continuing psy-chosocial adversity (see Clarke & Clarke, 1976, 2000).

The late 1970s to early 1980s saw substantial crit-icisms of the exaggerated psychosocial influencesclaims, and in the late 1980s to early 1990s there wasan excessive swing of the pendulum in the oppositedirection in terms of a denial of any substantial envi-ronmental effects within the normal range. Unfortu-nately, this polarization between nature and nurture

has remained all too widespread, and has been ac-companied by a considerable reluctance among someresearchers to accept the need to take seriously the


10/21

10 Child Development

possibility of genetic mediation (see, e.g., Baumrind,1993; Brown, 1996). Also, most psychosocial researchcontinues to use designs that provide inadequatetests of environmental mediation.

Far too much psychosocial research, even today, is

involved with the demonstration of statistical asso-ciations between some hypothesized risk factor andsome postulated outcome variable, without any atten-tion paid to the necessity of differentiating betweenrisk indicators and risk mechanisms. As a conse-quence, remarkably little is known about psychosocialrisk processes; and even less is known about the effectsof such risk processes on the organism (and, therefore,why and how effects persist when they do). In addi-tion, little is known about individual differences in re-sponse to psychosocial stress and adversity; and thereis almost total ignorance with regard to the environ-

mental factors that are responsible for the major secu-lar trends that have been evident over the course of the20th century (see Rutter & Smith, 1995). There is goodevidence of a major rise in the level of antisocial behav-ior, of substance abuse problems, and of suicidal ratesamong young malesto mention but a few examples.The speed of the rise indicates that some environmen-tal influence must have been responsible (althoughpossibly enhanced by the multiplying effect that couldcome from geneenvironment correlations; Dickens &Flynn, 2001), but there has been little systematic re-

search into possible causes.The dismissal by some commentators of the impor-tance of psychosocial influences on psychological de-velopment and on psychopathology was clearly mis-guided. There is good evidence that there are importanteffects, but the knowledge as to how these risks arebrought about, and how effects sometimes persist tomuch later stages in development, is much less thanpsychosocial enthusiasts would have us believe.

Potential of Psychosocial Research

The potential of psychosocial research is consider-able. If it is to be actualized, however, much more at-tention will need to be paid to both the conceptual andmethodological challenges. Without doubt, one of themajor growth areas is going to be the study of geneenvironment interplay (Rutter & Silberg, in press). Onthe whole, the empirical research findings suggestthat genetic vulnerabilities operate, in part, throughtheir role in bringing about an increased susceptibil-ity to environmental hazards. It seems that ill effectsfollowing psychosocial stress and adversity are rela-tively minor in those who are not genetically at risk(Rutter, 2000a; Rutter et al., 1997; Rutter, Pickles, et al.,2001). This is not a universal tendency, and there may

well be circumstances in which the reverse is the case(see Rowe, Jacobson, & van den Oord, 1999). Neverthe-less, what is evident is that psychosocial researchersneed to make greater use of genetically sensitive de-signs and that if they do, there should be a substantial

payoff in understanding environmental risk and pro-tective mechanisms. In that connection, more usecould be made of studies with animals, as well asstudies with humans.

Several phenomena warrant research attention.First, both animal and human studies have shown thereality of sensitization and steeling effects (Rutter,1981a; Wachs, 2000). That is, stress experiences makeindividuals either more resistant or more vulnerableto later psychosocial hazards. The question then iswhat is it about the individual, or the experience, thatleads to one outcome rather than the other. There is

some suggestion that milder stresses, or, more likely,ones that are accompanied by successful coping andadaptation, tend to foster steeling, whereas over-whelming stresses that bring about maladaptationand unsuccessful coping lead to sensitization (seeRutter, 2000c). This matter, however, has been subjectto remarkably little systematic investigation.

The related phenomenon of resilience warrants sim-ilar attention. It is clear that many different features arelikely to be involved in resilience (Luthar, Cicchetti, &Becker, 2000; Rutter, 1999b, 2000c), including prior ex-

periences, how the individual deals with stress at thetime, inherent qualities of the individual, and subse-quent experiences. At one time, writers on the topictended to imply that vulnerability and invulnerabil-ity were general characteristics of the individual, butthat is most unlikely to prove to be the case. Peoplemay be resilient with respect to some types of experi-ences and yet very vulnerable with respect to others.

A further phenomenon is that of so-called kin-dling effects (Kendler, Thornton, & Gardner, 2000,2001; Post, 1992). This term refers to the phenomenonof individuals becoming less responsive to environ-

mental stressors as a result of having developed adisorder. It appears that in some circumstances theexperience of disorder brings about changes in the or-ganism that predispose it to perpetuation that is rela-tively independent of the environment. Thus far, re-search has scarcely begun to chart the qualities andfrequency of the phenomenon, let alone the mecha-nisms that are implicated. Future systematic empiri-cal study is warranted.

Finally, there is the important research priority ofdetermining the changes in the organism that havebeen brought about by psychosocial experiences andof the ways in which such changes predispose the or-ganism to the continuation or occurrence of psycho-


11/21

Rutter 11

pathology. Many possibilities exist (Rutter, 1989a;Rutter, OConnor, & the ERA Study Team, 2001). Dothe mechanisms involve cognitive and affective sets,self-concepts, and internal working models? Do theyinvolve changes in the neuroendocrine system? Do

they come about through effects on styles of interper-sonal interaction? Are they brought about through ef-fects on individual behavior that predispose peopleto act in ways that engender later stresses or adver-sity? Or, are the effects a consequence of changes in

brain structure or function? Research is only just be-ginning to tackle these questions, and it is importantthat more be done. If psychosocial research is to de-liver effectively on its very considerable potential, it isessential that psychosocial research be a part of biol-ogy, and not separate from it.

DEVELOPMENTAL PROCESSES

Strengths and Achievements

An immense amount has been achieved with re-spect to increasing knowledge on the course of devel-opment, and on some of the key processes that areinvolved. A few examples serve to illustrate the ad-vances. It has become clear that brain developmentinvolves initial proliferation of neurons and synapses,with extensive neuronal migration. This overproduc-

tion of nerve cells and connections is then followed bya selective pruning, which serves to fine-tune brain de-velopment with respect to both structure and function(Goldman-Rakic, Bourgeois, & Rakic, 1997; Greenough& Black, 1992; Greenough, Black, & Wallace, 1987;Nelson & Bloom, 1997). In other words, the biology of

brain development is probabilistic, such that there is agenetic programming of the general pattern andcourse, but extensive opportunities to correct the pro-cess of development in accord with both environmen-tal input and the workings of the brain, in terms ofcellcell interactions. Much of this development takes

place during the early years of life, but it is also appar-ent that development continues for much longer, thetiming varying across different parts of the brain.

Similarly, much has been learned about the courseof psychological development (see Rutter & Rutter,1993; Shonkoff & Phillips, 2000) with respect, for ex-ample, to the development of mentalizing skills, self-concept, social attachments and relationships, andemotional expression. Long-term longitudinal studieshave also been crucially important in showing the ex-tent to which peoples behavior in childhood predicts

their stressful experiences in adult life (Champion,Goodall, & Rutter, 1995; Quinton, Pickles, Maughan,& Rutter, 1993; Robins, 1966; van Os, Park, & Jones,

2001). The findings have highlighted the need to con-sider the origin of individual differences in experi-ences of stress and adversity (Rutter, Champion,Quinton, Maughan, & Pickles, 1995); such origins in-clude peoples actions in shaping and selecting envi-

ronments and societal influences, as reflected, for ex-ample, in housing policies or racial discrimination.There also has been documentation of important

gender differences, for example, in relation to the riseof depressive disorders (Bebbington, 1996; Silberg etal., 1999) and eating problems (Lucas, Beard, OFallon,& Kurland, 1991; Pawluck & Gorey, 1998) in femalesduring late adolescence; the ebb and flow of genderdifferences in antisocial behavior as they are evidentacross the lifespan (Moffitt, Caspi, Rutter, & Silva,2001); and the timing of the onset of schizophrenia(Castle, Wessely, van Os, & Murray, 1998; Tarrant &

Jones, 2000). Remarkably little, however, is knownabout the mechanisms involved in the causes of thesegender differences.

One of the important elements that has derived outof research into biological development has been theawareness of the likely importance of epigenetic and ofchance effects (Jensen, 1997; Molenaar, Boomsma, &Dolan, 1993). The probabilistic nature of biological de-velopment means that some of the variations will be aconsequence of perturbations of a quasirandom na-ture, rather than the effects of specific environments or

genetic programming. At a group level, these follow ameaningful pattern; but at an individual level, they areunpredictable (see Rutter, in press-b). Thus, for exam-ple, minor congenital anomalies are much more likelyto occur in infants born to elderly mothers and aremore common in twins than in singletons (Vogel &Motulsky, 1997). There is probably no specific cause ofwhy one particular anomaly is found in any specific in-dividual, however. Similarly, there is a universal pat-tern of one of the two X chromosomes possessed byfemales to be suppressed, but which one seems to bedetermined largely by chance. It might be assumed

that such epigenetic effects cannot be the subject of sys-tematic investigation, but that is not entirely the case.Researchers have sought to index developmental per-turbations through the study of minor congenitalanomalies and so-called fluctuating asymmetry of der-matoglyphic patterns (Naugler & Ludman, 1996). Theissue, then, is not the functional consequences of theseanomalies or asymmetries (because there are not likelyto be any), but rather their use as indices of develop-mental perturbations that may play a role in deficits ordisorders of psychological development.

During recent years, concepts of developmental pro-gramming have come to the fore (Bateson & Martin,1999; Greenough & Black, 1992; OBrien, Wheeler, &


12/21


Barker, 1999). There are at least two different types thatneed to be considered. First, biological maturation, in-cluding brain development, is dependent on the indi-vidual having experiences within a broad expectablerange (Greenough & Black, 1992; Greenough et al.,

1987). Thus, Hubel and Weisel (1965; Hubel, Wiesel, &Le Vay, 1977) showed that the structural and functionaldevelopment of the visual cortex was dependent on in-dividuals having appropriate visual experiences duringa sensitive period of development, during which thestructure and function of the visual cortex were estab-lished (see Blakemore, 1991; Mitchell, 1989). The pro-gramming of development is thereby experience ex-pectant, to use Greenough et al.s (1987) term. It needsto be emphasized, however, that a wide range of experi-ences is adequate for normal development to take place.

Development is also shaped to provide optimal ad-

aptation to the specific environments experienced at thetime (Bateson & Martin, 1999; OBrien et al., 1999)what may be termed experience-adaptive program-ming (Rutter, OConnor, & the ERA Study Team, 2001).This is different in the sense that it is concerned withvariations within, as well as outside of, the normalrange, and it is concerned not with normal developmentin an absolute sense, but rather with development that istailored to the specific environments experienced dur-ing the relevant sensitive phase. Thus, for example, inthe psychological arena, this sort of programming prob-

ably occurs in relation to the ways in which infants abil-ity to make phonological discriminations is influenced by the language environment that they experience inearly life (Kuhl, 1994; Kuhl et al., 1997). The effects arelong lasting, although not completely immutable(Werker & Tees, 1992). In the broader field of biologyand internal medicine, other examples are evident in thedevelopment of immune responses and in metabolic re-sponses to diet (Bock & Whelan, 1991). Thus, for exam-ple, babies who are poorly nourished at birth and inearly life have been found to be more vulnerable tocoronary heart disease, hypertension, and diabetes in

midlife (Barker, 1997). This finding is interesting be-cause the association is the reverse of what one finds inadult life; that is, low weight constitutes a risk factorin infancy, but being overweight constitutes a risk factorin middle age. The physiological mechanisms have yetto be properly worked out, but what is hypothesized isthat individuals are programmed to deal with poor-quality diets and that they then are at risk if, later on,they are exposed to rich diets. The implication that fol-lows is a challenging one: if there is an attempt, throughgood feeding, to try to make up in middle childhood forsubnutrition in early life, it may actually make thingsworse. The psychological query is whether there is anyequivalent to that phenomenon in relation to psychoso-

cial experiences. Although the answer to this question isnot known, the counter-intuitive possibility raises chal-lenging issues that call out for serious investigation.What is clear, however, is that some forms of seriousdeprivation do lead to persistent sequelae that continue

long after there has been restoration of a normal rearingenvironment, as the findings on Romanian adoptees in-dicate (see Rutter, in press-b; Rutter, OConnor, & theERA Study Team, 2001).

Research has provided some knowledge on age-related progressions in psychopathology (see Rutter,in press-b). Thus, for example, it is clear that early hy-peractivity predisposes to later antisocial behavior(Rutter et al., 1998), early conduct problems predis-pose to substance use and abuse (Rutter, in press-c),substance abuse predisposes to later depression (Rutter,in press-c), and early anxiety problems are often fol-

lowed by depressive symptomatology (Silberg, Rutter,& Eaves, 2001). It is also well documented that neu-rodevelopmental abnormalities in childhood are as-sociated with an increased risk of schizophrenia de-velopment in early adult life (Keshavan & Murray,1997). Much less is known, however, about the under-lying processes that these progressions reflect.

Developmental research has highlighted severalimportant age-indexed effects of environmental haz-ards. For example, the effects of unilateral brain in-jury in infancy are quite different to those seen in later

childhood or adult life (Vargha-Khadem, Isaacs, vander Werf, Robb, & Wilson, 1992). Its not that the ef-fects in infancy are greater or lesser than those in laterlife, but rather that the pattern is different. In adultlife, there is a clear lateralization of psychological ef-fects, but this is not found in early life. It is obviousthat, in some way, these differences reflect changingpatterns of brain plasticity (with respect to the take upor transfer of mental functions), but less is knownabout what this means with respect to physiologicalprocesses (but see Neville & Bavelier, 1998). In relationto psychosocial experiences, there is some evidence

that the negative effects of hospital admission tend tobe less in infancy or in middle childhood than they arein the toddler age period (Rutter, 1979b), and the ef-fects of an institutional rearing on social relationshipsseems to be a feature of adverse rearing in early life,rather than adverse experiences later in life (Rutter,1981b; although this is much less well documented).

Finally, some information has been obtained with re-gard to the effects of individual differences on the tim-ing of developmental transitions. For example, girlswho experience an unusually early menarche tend tohave an increase in disruptive behavior (Caspi, Lynam,Moffitt, & Silva, 1993; Stattin & Magnusson, 1990). Thestimulus is biological but, in this case, the mediation of


13/21

Rutter 13

psychological effects seems to be social, with the mainmediation being provided by the peer group.

Misleading Claims

The misleading evangelism with respect to exces-sive claims in the field of development has been oftwo different kinds. First, in the mid-twentieth cen-tury there was a period in which some researcherssought to explain much of development in terms ofstraightforward biological maturation (Gesell, 1946;McGraw, 1946). During the second half of the twenti-eth century, there were also many psychological re-searchers who saw age-related progressions as pro-viding an explanation (Wohlwill, 1970, 1973). Therewas resistance to unpacking age changes into sepa-rate differentiated processes. Critics noted the extent

to which so many of the findings relied on cross-sectional studies, rather than longitudinal research(De Ribaupierre, 1989), and they also noted the factthat age was an ambiguous variable (Rutter, 1989b).Not only did the different aspects of biological matu-ration not necessarily go together (thus, sexual devel-opment and intellectual development did not runclosely in parallel), but also age reflected experiencesas well as biological maturation. It has not proved atall easy to determine which age-indexed change is re-sponsible for altering psychological functions, but

clearly that is the research need.The last decade or so has been accompanied by adifferent type of evangelismnamely, claims on theextent to which early experiences determine braindevelopment (see, e.g., Kotulak, 1996). There has

been a misleading extrapolation of the findings on ex-perience-expectant development to the entirely dif-ferent notion that higher quality psychosocial experi-ences in the first 2 or 3 years of life will have a muchgreater effect than similar experiences later on, be-cause the early experiences bring about a lastingchange in brain structure. As several commentators

have pointed out, the claims (which come frompeople outside the field of neuroscience research) aremisleading and fallacious for several different rea-sons (Bruer, 1999). To begin with, brain developmentis far from over by age 3 years. On the contrary, impor-tant changes continue to occur through adolescence(Giedd et al., 1999; Huttenlocher, 1979; Keshavan &Murray, 1997; Sowell, Thompson, Holmes, Jernigan,& Toga, 1999). Also, it is not the case that neuronalgrowth stops in early life or that plasticity is lost afterthe infancy years. Animal studies have shown that

neuronal growth and increase in number of synapsescan and does take place in later life, at least with re-spect to certain parts of the brain, such as the hippo-

campus and cerebellum (Diamond, 1991; Erikssonet al., 1998; Gould, Reeves, Graziano, & Gross, 1999;Kempermann, Kuhn, & Gage, 1998; Kleim et al., 1998;Klintsova, Matthews, Goodlett, Napper, & Greenough,1997; Lowenstein & Parent, 1999).

Moreover, research in humans using structural im-aging has shown the effects of later experiences. Theincreased size of the posterior hippocampus in Lon-don taxi drivers (who have to memorize the locationsof all London streets and all routesa process knownas the knowledgewhich usually requires severalyears of study) provides a striking example (Maguireet al., 2000), as do findings with Braille readers (Sterr etal., 1998) and violinists (Elbert, Pantev, Wienbruch,Rockstroh, & Taub, 1995). It is not known how farthere is brain plasticity after the first part of child-hood, nor the extent to which it varies across different

parts of the brain or brain systems. There is some ev-idence that later learning may be mediated in differ-ent ways than in early learning. For example, onestudy showed that the parts of the brain used in learn-ing a second language after the postinfancy yearswere different than those parts of the brain used withinitial language learning (Kim, Relkin, Lee, & Hirsch,1997). Nevertheless, what is clear is that the assump-tion that later experiences necessarily have only minoreffects is clearly wrong. Experience-dependent learn-ing (Greenough et al., 1987), in which individualized

experiences have neural effects, goes on throughoutlife. This learning is different from developmentalprogramming, but also involves experiential effectson the brain. It should be added that although it is ob-vious that the workings of the mind must be based onthe functioning of the brain, remarkably little isknown about structurefunction links.

In addition, little is known about the mechanismsinvolved in the risks that stem from pre- and peri-natal problems, once the effects leading to grosshandicap are put aside. Why, for example, are therereplicated associations between pre- and perinatal

abnormalities and later schizophrenia (McDonald,Fearon, & Murray, 2000) or suicide (Jacobson et al.,1987; Salk, Lipsitt, Sturner, Reilly, & Levat, 1985)?Some of the mediators that seem to be obviously rele-vant have not been proven to be so. For example, verylow birthweight is associated with a marked in-creased in brain scan abnormalities, and is also asso-ciated with an increased risk of psychological impair-ment. It would seem reasonable to suppose that the

brain scan abnormalities would be associated with,and perhaps responsible for, the cognitive impair-

ment, but several studies have failed to find this(Cooke & Abernethy, 1999; Stewart et al., 1999). Whynot? The puzzle remains.


14/21


Potential for Developmental Research

The potential for developmental research is obvi-ous, as indicated by the range of questions already out-lined. It is important to note several areas of particular

promise, however, starting with functional brain imag-ing as used in relation either to specific psychologicaltasks or to the administration of particular pharmaco-logical substances (Ernst & Rumsey, 2000). One of themajor technological developments in recent years hasbeen the establishment of magnetic resonance imag-ing (MRI). The spatial resolution is better than withpositron emission tomography, which preceded it, andMRI does not have the disadvantage of radiation sideeffects. During the last few years there have been somestriking successes in showing which areas of the brainsubsume particular psychological functions (see, e.g.,

Fletcher et al., 1995; Schultz et al., 2000). Functionalimaging has provided a new way of testing hypothe-ses about differences between particular psychologi-cal processes and also differences among clinicalgroups in the ways in which they deal with particularpsychological tasks. There is no doubt that much willbe learned through functional imaging, provided thata rigorous hypothesis-testing approach is followedand the expertise in physics is accompanied by equiv-alent expertise in psychology. Nevertheless, certaincautions are necessary. The technique is often sold

as a means of actually seeing the brain in action, butthis is not quite so. What MRI does show are changesin blood flow and oxygen take-up, which reflect met-abolic activity, but provide only a very indirect meansof investigating brain physiology and neurochemis-try. It is important, also, to appreciate that just be-cause changes in brain function can be seen, does notnecessarily mean that the biology of the brain hascaused whatever changes in psychological function-ing are being investigated. Thus, it has been foundthat the changes in brain function that can follow psy-chological treatments of obsessive-compulsive disor-

der closely parallel those that are brought about bytherapeutic medication (Baxter et al., 1992). There is atwo-way interplay between soma and psyche, and itis important to be careful to not make false assump-tions about the direction of effects. In addition, it isnecessary to get beyond the crudities of changes inblood flow. Magnetic resonance spectroscopy35 cantake things further, and this is likely to be a techniqueof increasing application in experimental studies, al-though there are many problems still to be overcome.

MRI does not constitute the only form of functionalbrain imaging. Further exploration of the use of electro-physiological methods such as magnetoencephalog-raphy (MEG)which has a better time resolution, but

a worse spatial resolution than MRI, is likely. It is notclear as yet whether MEG will deliver on its promise.Cautions are required; again, as a result of an aware-ness that the comparable field of neurometrics failedto deliver what had been thought to be its potential.

Currently, there has been a recrudescence of interestin neuroendocrinology (Carlson & Earls, 1997; Gunnar,2000; Heim, Ehlert, & Hellhammer, 2000; Heim, New-port, et al., 2000). Both animal and human studies haveshown important neuroendocrine effects of stress ex-periences. Most of this research has been concernedwith acute stresses, but neuroendocrinology is begin-ning to be applied to chronic psychosocial adversi-ties. Its not clear how far this field of research will aidin understanding normal and abnormal developmentalprocesses. The fact that there are neuroendocrine corre-lates is not in doubt; the question, however, is whether

learning more about those correlates will increase un-derstanding of the mechanisms underlying psycholog-ical functions. Maybe it will, but I am not sure.

Cognitive psychologists have wanted to claim thewhole of mental functioning as their domain (Morton& Frith, 1995). Such presumptuousness needs to beresisted. There is much to be learned from the studyof the interconnections between different facets ofcognitive functioning. For example, in relation to au-tism, it will be important to find out how joint atten-tion, theory-of-mind skills, central coherence, and

executive planning do or do not reflect the same un-derlying cognitive skills. There is also much to belearned about the interconnections between thosecognitive skills and both emotional and social devel-opment. Why, for example, does profoundly depriv-ing institutional rearing seem to lead to both so-calleddisinhibited attachment patterns and also quasi-autisticbehavior (Rutter, Andersen-Wood, et al., 1999; Rutter,Kreppner, et al., 2001)? Why, too, are such quasi-autisticfeatures also found in children with congenital blind-ness (Hobson, Lee, & Brown, 1999)? What does thistell us about developmental processes, and what are

the implications for the syndrome of autism? An-swering these questions requires the skills of cogni-tive psychology, but these must be brought togetherwith parallel skills in the study of socioemotional de-velopment and in the investigation of psychosocialexperiences.

The field of animal studies has, for the most part,remained rather separate from the study of child de-velopment. Of course, there have been pioneers suchas Hinde (Hinde & McGinnis, 1977) and Suomi (1997),who have sought to bring the two together but, de-spite such important exceptions, the two arenas haveremained separate from one another. It is importantto ask in what areas could animal studies be informa-


15/21

Rutter 15

tive. There are many, including the investigation ofthe different forms of developmental programming,the study of sensitizing and steeling effects of stress(and the parallel field of resilience), and the delinea-tion of the effects of psychosocial experiences on the

organism.Attention has been drawn already to the importantand puzzling field of gender differences in psycho-logical development. Elucidation of the underlyingprocesses remains a considerable challenge and onethat will require the use of research designs differentfrom those ordinarily employed to investigate indi-vidual differences. For the most part, it has beenfound that the risk and protective processes withinmales and females are broadly comparable; this doesnot explain, however, why the levels of psychologicaltraits or the frequency of particular disorders are so

markedly different between males and females.

FUTURE: RESEARCH, POLICY, AND PRACTICE

The findings on nature, nurture, and developmentmay be used to tentatively look into the future. Atten-tion has been drawn already to the very considerableresearch challenges that remain ahead. It is apparentthat a diversity of causal processes needs to be con-sidered. We must recognize, accept, and seek to un-derstand the anomalies and apparent paradoxes in

the findings thus far available, which will require in-dividual creativity and innovation with respect toboth concepts and research strategies. Most important,it will require a bringing together of genetics, environ-mental studies, and developmental investigations. Thethree fields have remained, for the most part, distress-ingly separate up until to now, and it is crucially im-portant that they become much better integrated. Forexample, both genetic and psychosocial research will

benefit from a focus on geneenvironment correla-tions and interactions. Similarly, the developmentalstudy of psychopathological progressions needs to

use genetic designs to investigate nature nurture in-terplay. In these connections, there is much to begained by bringing together studies of normal andabnormal development. The field of developmentalpsychopathology is one that has much to offer withrespect to the investigation of naturenurture inter-play in developmental processes and in the two-wayflow of understanding from the normal to the abnor-mal and vice versa (Rutter & Sroufe, 2000).

What about implications for policy and practice?In seeking to answer this question, it is necessary that

the magnitude of the challenge is accepted. Over thecourse of the last 50 years, there have been tremen-dous improvements in the physical health of children

and in the life expectancy of adults. It is chastening torealize that there have not been parallel improve-ments in psychological functioning or mental health(Rutter & Smith, 1995). On the contrary, psychosocialdisorders in young people have tended to increase in

frequency over the last half century. Why has thisbeen so? I would argue that this has to be an answer-able question. If we had a proper understanding ofwhy society has been so spectacularly successful inmaking things psychologically worse for childrenand young people, we might have a better idea as tohow we can make things better in the future. To suc-ceed in that gargantuan task, use of a diverse range ofresearch strategies is necessary. The answers will notcome from genetic research on its own, or from envi-ronmental studies, or developmental investigations;the combination of the three might do much, how-

ever. It will be necessary to recognize the range of dif-ferent causal questions that have to be considered.The explanation for individual differences may not bethe same as that for differences in the level of a trait orthe frequency of disorder.

Effective interventions with respect to either pre-vention or treatment are not necessarily dependenton understanding basic causal processes but, clearly,an understanding of causal mechanisms is likely to behelpful. Although it is true that all societies have beenslow in taking effective action to put into practice

knowledge that is already available, it is important thatwe are realistic about how limited our knowledge is.One of my favorite American sayings is: It aint igno-rance that does the harm, its knowing so many thingsthat aint so. This concern is most relevant with re-gard to seeking policy and practice implications thatderive from basic science. Of course, it is important toact promptly and expeditiously when what is neededis apparent; however, caution should be taken in

jumping too readily onto the bandwagons of what-ever happens to be the prevailing enthusiasm of themoment. Psychology as a whole, and child develop-

ment in particular, gains its strength from its impor-tance as an applied science. We must never lose sightof that (Rutter & Yule, in press), which means retain-ing a broad view of the types of research that are go-ing to be rewarding, and ensuring that we do notavoid directing attention to policy and practice impli-cations simply because it is so difficult to be sure ofwhat is needed. Let us question, but let us study in rig-orous fashion, using a creative approach to researchstrategies; and let us throughout make sure that atten-tion is paid to questions of policy and practice, but

also that practical applications of research findingsare studied with the same rigorous scrutiny as thatused in basic science itself.


16/21


ADDRESS AND AFFILIATION

Corresponding author: Michael Rutter, Institute ofPsychiatry, Social, Genetic and Developmental Psy-chiatry Research Centre, De Crespigny Park, Den-mark Hill, London SE5 8AF, United Kingdom; e-mail:[email protected].

REFERENCES

Avner, P., & Heard, E. (2001). X-chromosome inactivation:counting, choice and initiation. Nature Reviews Genetics,2

, 5967.Bailey, A., Phillips, W., & Rutter, M. (1996). Autism: Towards

an integration of clinical, genetic, neuropsychological,and neurobiological perspectives.Journal of Child Psychol-ogy and Psychiatry Annual Research Review, 37

, 89126.Barker, D. J. (1997). Fetal nutrition and cardiovascular dis-

ease in later life. British Medical Bulletin, 53

, 96108.Bateson, P., & Martin, P. (1999). Design for a life: How behav-

iour develops.

London: Jonathan Cape.Baumrind, D. (1993). The average expectable environment

is not good enough: A response to Scarr. Child Develop-ment, 64

, 12991317.Baxter, L. R., Schwartz, J. M., Bergman, K. S., Szuba, M. P.,

Guze, B. H., Massiotla, J. C., Alazraki, A., Selin, C. E.,Ferug, H.-K., Munford, P., & Phelps, M. E. (1992). Cau-date glucose metabolic rate changes with both drug and behavior therapy for obsessive-compulsive disorder.

Archives of General Psychiatry, 49

, 681689.Bebbington, P. (1996). The origins of sex differences in de-

pressive disorders: Bridging the gap. International Reviewof Psychiatry, 8

, 295332.Bell, R. Q. (1968). A reinterpretation of the direction of effects

in studies of socialization. Psychological Review, 75

, 8195.Bell, R. Q., & Chapman, M. (1986). Child effects in studies

using experimental or brief longitudinal approaches tosocialization. Developmental Psychology, 22

, 595603.Blakemore, C. (1991). Sensitive and vulnerable periods in

the development of the visual system. In G. R. Bock &J. Whelan (Eds.), The childhood environment and adult dis-ease. Ciba Foundation Symposium 156

(pp. 129146). Chich-ester, U.K.: Wiley.

Bock, G. R., & Whelan, J. (1991). The childhood environmentand adult disease.

Chichester, U.K.: Wiley.Bowlby, J. (1951). Maternal care and mental health.

Geneva,Switzerland: World Health Organization.

Brent, D. A. (in press). Cognitive behavioral approaches tothe treatment of depression and anxiety. In M. Rutter &E. Taylor (Eds.), Child and adolescent psychiatry

(4th ed.).Oxford, U.K.: Blackwell Scientific.

Brown, G. W. (1996). Genetics of depression: A social sci-ence perspective. International Review of Psychiatry, 8

,387401.

Bruer, J. T. (1999). The myth of the first three years.

New York:Free Press.

Buchanan, A., Brock, D. W., Daniels, N., & Wikler, D. (2000).

From chance to choice: Genetics and justice.

Cambridge,U.K.: Cambridge University Press.

Carlson, M., & Earls, F. (1997). Psychological and neuro-endocrinological sequelae of early social deprivation ininstitutionalized children in Romania. Annals of the NewYork Academy of Sciences, 807

, 419428.Caspi, A., Lynam, D., Moffitt, T. E., & Silva, P. A. (1993).

Unraveling girls delinquency: Biological, dispositional,

and contextual contributions to adolescent misbehavior.

Developmental Psychology, 29

, 1930.Castle, D. J., Wessely, S., van Os, J., & Murray, R. M. (1998).

The effect of gender in age at onset of psychosis. In D. J.Castle (Ed.), Psychosis in the inner city: The CamberwellFirst Episode Study. Maudsley Monograph No. 40

(pp. 2736). Hove, U.K.: Psychology Press.

Champion, L. A., Goodall, G. M., & Rutter, M. (1995). Be-havioural problems in childhood and stressors in earlyadult life: A 20-year follow-up of London school chil-dren. Psychological Medicine, 25

, 231246.Chorney, M. J., Chorney, K., Seese, N., Owen, M. J., Daniels,

J., McGuffin, P., Thompson, L. A., Detterman, D. K., Ben-

bow, C., Lubinski, D., Eley, T., & Plomin, R. (1998). Aquantitative trait locus associated with cognitive abilityin children. Psychological Science, 9

, 159166.Clarke, A. M., & Clarke, A. D. B. (1976). Early experience:

Myth and evidence.

London: Open Books.Clarke, A. M., & Clarke, A. D. B. (2000). Early experience and

the life path.

London: Jessica Kingsley.Cooke, R. W. I., & Abernethy, L. J. (1999). Cranial magnetic

resonance imaging and school performance in very lowbirthweight infants in adolescence. Archives of Disease inChildhood: Fetal Neonatal Edition, 81

, F116F121.De Ribaupierre, A. (1989). Transition mechanisms in child de-

velopment: The longitudinal perspective.

Cambridge, U.K.:Cambridge University Press.

Diamond, M. C. (1991). Environ

nature, nurture, and development

Documents