Symposium
Chairs: Tom S. Price and Kathryn S. Lemery
Discussant: H. Hill Goldsmith
Research using child and adult populations is establishing a consensus thatgenetic influences on behavioural measures are generally important.Behavioural genetic research has, however, a great deal more to offer thedevelopmental scientist than estimates of heritability.The study of developmental changes and continuities requires longitudinalresearch designs. All of the presentations in this symposium are drawnfrom projects with longitudinal, genetically-informative research designs,capable of investigating the aetiology of continuity and change inbehavioural measures. Multivariate genetic analyses can elucidate the genetic and environmentalinfluences on the overlap between different measures. The firstpresentation in this symposium (The Development of General CognitiveAbility in Pre-School Age Children) uses a longitudinal multivariategenetic analysis to investigate whether genetic influences on cognitiveabilities become more unitary between infancy and childhood, with the aimof identifying the developmental origins of a heritable general cognitiveability factor (g). The sample is drawn from the TEDS twin registry, whichprovides a very large population sample of young twins.Another application of multivariate genetic techniques is to refine ourunderstanding of the technology available to researchers for studyinginfant behaviours. The second presentation (Using Parental Characteristicsto Understand the Relationship between Parental Report and Laboratory-BasedMeasures of Infant Temperament) incorporates extensive observationalmeasures in a longitudinal genetic design to document the relationshipbetween parental report and laboratory-based measures of infanttemperament, and investigates the relationship between parentalcharacteristics such as stress, personality, and depression, and parentalreport measures of infant temperament.Measures of infant behaviour combined with measures of the environmentallows investigation of interactions between environmental risk factors andgenetic and environmental influences. This is the focus of the thirdpresentation (The Interaction of Prematurity with Genetic and EnvironmentalInfluences on Cognitive and Language Development in Twins During EarlyChildhood). Very pre-term infants are susceptible to delays in cognitiveand language development during infancy and early childhood. Whileprevious studies have identified early biomedical and social risk factors,this study throws light on the possible mediating pathways by investigatingthe interaction between the degree of prematurity and genetic andenvironmental influences at 2 and 3 years of age.The fourth presentation (Adolescent Outcomes of Infant Adoption) extendsthe range of research on the outcomes of adoption by exploring theemotional, scholastic and social adjustment of participants in the ColoradoAdoption Project during the transition from early to late adolescence,investigating possible genetic mediating mechanisms by utilising data fromthe birth parents of the adopted children.Twin studies that incorporate measurements of opposite-sex pairs provide anideal way to test hypotheses concerning the aetiology of sex differences.Our final presentation (Sex Differences in the Aetiology of Early Verbaland Non-verbal Development) investigates this issue for verbal andnon-verbal measures of cognitive development in a sample of over 30002-year-old twin pairs drawn from the TEDS twin registry.
Details of individual items:
paper
Multivariate genetic research suggests that from middle childhood onwardthe correlations among cognitive abilities are largely due to sharedgenetic influences, implying a heritable general cognitive ability factor(g). What little evidence there is from infancy suggests that geneticinfluences are modular, that is to say, they serve mainly to differentiatecognitive abilities. The current study uses a longitudinal design toinvestigate whether genetic influences on cognitive abilities do indeedbecome more unitary between infancy and childhood, with the aim ofidentifying the developmental origins of g.Verbal and non-verbal cognitive ability were assessed in an epidemiologicalsample of 2,745 twin pairs born in England and Wales in 1994 using parentalreport instruments administered at 2 and 3 years of age. Verbal abilitywas assessed using short forms of the MacArthur Communicative DevelopmentInventory (MCDI), non-verbal ability using a novel measure adapted for useat each age (PARCA). The verbal and non-verbal measures were moderatelyheritable and correlated .45 at both ages. The observed correlationbetween measures at ages 2 and 3 was .61 for verbal ability and .56 fornon-verbal ability. Small mean sex differences were evident for bothmeasures at each age.A multivariate genetic model provided an excellent fit to the data (X2(150) 120.14, p .48, AIC -119.86, RMSEA .000). It suggested thatthe genetic influences on cognitive abilities become increasingly molar:that is, it estimated that verbal and non-verbal cognitive abilities aremore highly correlated genetically at 3 years (rg .55 for males, rg .37for females) than at 2 years (rg .33 for males, rg .32 for females).These findings are consistent with the developmental emergence of aheritable general cognitive ability factor. In addition, geneticinfluences on verbal ability at 2 years are more strongly related togenetic influences on non-verbal ability at 3 years (rg .50 for males, rg .49 for females) than the converse (rg .25 for males, rg .22 forfemales), suggesting that verbal ability in infancy may be more predictivegenetically of later g.
paper
Our study has three main goals: (1) To document the relationship betweenparental report and lab-based measures of infant temperament and determinetheir continuity from 6 to 12 months. (2) To establish the relationshipbetween parental characteristics such as stress, personality, anddepression, and infant temperament. Lastly, (3) to use the power of agenetically-informative design to consider whether or not parentalcharacteristics account for the subjective aspect of parent report oftemperament. Infant twins were assessed in Goldsmith and Rothbart's LaboratoryTemperament Assessment Battery (Lab-TAB) longitudinally at 6 and 12 monthsof age as part of an ongoing project. N's ranged from 50 to 200, dependingon the analysis. Anger, Fear, and Pleasure composites were formed fromcoding the latency, peak, and duration of facial, bodily, and vocalexpressions of discrete emotions in 6 episodes. Fear was coded in StrangerApproach and Nonsocial Mask Display, Anger in Arm Restraint and Car Seat,and Pleasure in Peek-a-boo and Puppets. At both 6 and 12months, mothersand fathers completed Rothbart's Infant Behavior Questionnaire (IBQ),Abidin's Parenting Stress Index, Watson et al.'s Positive and NegativeAffect Schedule (PANAS), and Halberstadt's Family ExpressivenessQuestionnaire. Mothers completed the CES-D at 3 months. For analyses ofphenotypic correlations between measures, one twin was selected randomlyfrom each pair to create an independent sample, and the cotwin was used ina replication sample. Average correlations from the two samples arereported, and a (*) indicates significance at p < .05.Parental report and lab measures of anger (r .25* for mother and .43* forfather report) and fear (.26* and .21, respectively) were moderatelyassociated, whereas measures of pleasure were not related at 12 months.Correlations at 6 months were lower in magnitude and did not reachsignificance. All six IBQ scales yielded significant longitudinalcorrelations, ranging from .40* to .69*. Lab measures showed lesscontinuity, r .13 for lab anger, .30* for lab fear during StrangerApproach, .10 for lab fear during Masks, and .26* for lab pleasure.At 12 months, mother and father PANAS personality correlated in .30*s withrelevant IBQ scales. Infant anger in the car seat was associated withfathers' negative affect (r .28*), and fear in the lab was associatedwith mothers' negative affect (r .32*). Mothers' CES-D depression wasassociated with infant fear in the lab (r .37*), IBQ Distress toLimitations (r .34*) and Smiling and Laughter (r -.33*). FEQ familyexpressiveness was related to relevant IBQ scales, r's in the .30*s, andnegative expressiveness was correlated .32* with infant anger during armrestraint and .30* with lab fear. Similar associations are present at 6months, although they are typically weaker in magnitude.We found associations between parental characteristics and both parent andlab measures of temperament, suggesting that rater bias cannot entirelyaccount for the associations. Parents and children share genes, thus theyshould be somewhat similar on genetically-influenced traits. Furtheranalyses explore whether or not parent and lab measures share their geneticinfluence, concurrently and longitudinally.
paper
Very preterm infants are susceptible to a variety of neonatal medicalcomplications and often face developmental problems during infancy andearly childhood. Most prominent are cognitive deficits, learningdifficulties and delays in language development. Although previous studieshave identified early biomedical and social risk factors, the mediatingpathways remain elusive. The objective of this study was to determine howdegree of prematurity interacts with genetic and environmental influenceson outcomes of verbal and non-verbal cognitive development during earlychildhood.The target sample consisted of over 2000 pairs of same sex dizygotic (DZss)and monozygotic (MZ) twins drawn from the 1994 cohort of the Twins EarlyDevelopment Study (TEDS), a study of twins born in England and Wales from1994-1996. The twins were assessed on measures of verbal and non-verbalcognitive development when they were aged 2 years and again at 3 years.Assessment measures were based on parent reports and parent-administeredtests. The sample was divided into three groups according to degree ofprematurity: very preterm or high risk (< 32 weeks), moderately preterm ormedium risk (32-33 weeks), and mildly preterm/fullterm or low risk (< 34weeks). Using quantitative genetic model fitting analyses, the contributions ofgenetic and environmental influences on vocabulary and cognitivedevelopment were assessed. At 2 years, the results indicated agene-environment interaction (GxE). For the high risk group, geneticeffects on both verbal and non-verbal cognitive ability were completelyovershadowed by shared environmental factors whereas for both medium andlow risk groups, additive genetic effects explained 18-33% of the variance.At 3 years the GxE remained stable for the verbal measure pointing towardsa similar trend of genetic and environmental effects between the three riskgroups. In contrast, for non-verbal cognitive development gene-environmentinteractional effects disappeared with genetic factors becoming morepronounced within the high risk group.Our findings indicate that genetic factors are not responsible forcognitive outcomes in very preterm children assessed at 2 years and suggestthat early environmental effects of neonatal intensive care predominate toaffect both verbal and non-verbal cognitive development. Thegene-environment interactional pattern emerged again at 3 years for verbaldevelopment, but disappeared for non-verbal cognitive development. Wepredict that GxE will also disappear for verbal development as the effectsof prematurity lose their grip and allow genetic sources of individualdifferences to emerge.
paper
Previous research on the outcomes of adoption has suggested elevated ratesof problem behaviour for adoptees. Although comparisons of adopted andnon-adopted persons indicated comparable indices of adjustment in infancyand early childhood, externalising and conduct problems became morepronounced among adoptees in middle childhood. Aside from a fewlongitudinal studies that have addressed this issue, little is known as towhether adopted children continue to experience these problems in theiradolescent years. The present study extended this research by exploringadoptees' emotional, scholastic and social adjustment in the transitionfrom early to late adolescence. Possible genetic mediating mechanisms werealso explored by utilising data from the birth parents of the adoptedchildren.The target sample consisted of 450 adopted and non-adopted adolescents fromthe Colorado Adoption Project (CAP), a longitudinal study of childdevelopment from infancy through adolescence. All of the adopted childrenwere relinquished at birth and placed in foster care for an average of 29days prior to being placed in permanent adoptive homes. Consequently,psychological adjustment was unlikely to have been influenced bypre-adoption factors, other than possible prenatal factors. The ethnicbackgrounds of adopted adolescents, and the social circumstances of theiradoptive families closely matched those of non-adopted adolescents.Emotional adjustment was assessed yearly across early (9-12 years) and late(13-16 years) adolescence for both adopted adolescents and theirnon-adopted comparisons. Scholastic achievement, also assessed yearly, wasbased on teachers' reports. Comparisons of adopted and non-adopted adolescents in CAP indicated smallbut significant differences for measures of problem behavior and scholasticadjustment. Although for each of these measures adopted adolescents showedslightly poorer indices of adjustment, adoptive status seldom accounted formore than 4% of the overall variance of these measures, and adoptedadolescents' scores fell within the normal variation range. Surprisingly,in areas of social adjustment and social competence, the findings from thisstudy further indicated better adjustment among adoptees. Longitudinal comparisons, examining developmental change in the transitionfrom early to late adolescence, indicated increasingly diverging patternsof adjustment for adopted and non-adopted adolescents. While a significantincrease in conduct, externalizing, and internalizing behaviors wasobserved among adoptees with increasing age, a significant decrease for thesame behaviors was observed in non-adopted adolescents. This increasedsusceptibility to problem behaviors observed for adoptees in lateadolescence suggests that the developmental transition from early to lateadolescence might be somewhat more stressful for adopted, as compared tonon-adopted adolescents.Comparisons of adopted and non-adopted adolescents' biological parents'adjustment indicated somewhat greater antisocial personality among thebiological parents of adopted adolescents. The differences in mean werehowever slight and largely accounted for by differences in age. Theseresults suggest that heritable factors are unlikely to account for theincreased vulnerability to problem behavior in adolescent adoptees ascompared to controls.
paper
Attempts to describe and explain sex differences in measures of abilityhave largely focused on mean differences. Previous work suggests thatalthough there are usually only trivial sex differences in means forgeneral intelligence measures, there are more notable and consistent meandifferences in several specific abilities. One important difference is thatfemales are consistently superior to males in a wide range of verbal tasks,even from a very early age. Socialisation theorists suggested that sexdifferences in verbal ability occur because parents talk to girls more thanboys Differential brain development has also been suggested as anexplanation for sex differences in verbal ability, although thesedifferences too could be an effect rather than a cause of other sexdifferences. Although research on sex differences has focused on mean differencesbetween the sexes, sex differences in the description and explanation ofindividual differences may be more powerful and more informative. The usualtwin method decomposes observed variance into genetic and environmentalcomponents of variance. Using model-fitting techniques, this decompositionof variance can be analysed separately for males and females to findquantitative genetic (heritability) differences, and also allowsexploration of qualitative differences - i.e. whether boys and girls aresensitive to all the same developmental influences. The present study used a sex-limitation model to examine genetic andenvironmental origins of sex differences in verbal and non-verbal cognitiveability. The sample included 476 male and 563 female identical twin pairs,496 male and 497 female same-sex non-identical twin pairs, and 974 pairs ofopposite-sex twins. The twins and their families are participants in theTwins' Early Development Study (TEDS), which sampled from all twins born in1994 in England and Wales.Girls scored significantly higher on both measures, although genderaccounted for approximately 3% of the variance in verbal ability and 1% ofthe variance in non-verbal cognitive ability. For the verbal measure boysshowed greater heritability than girls. Also the twin-pair correlation issignificantly lower for opposite-sex twins than for non-identical same-sextwins. This indicates that individual differences in verbal ability includesome sex-specific factors. Non-verbal cognitive ability did not differ inaetiology for boys and girls. Our findings indicate that genetic and environmental influences differ forgirls and boys for early verbal but not non-verbal development. There aresmall but clear sex differences in verbal development that are distinctfrom general cognitive ability, and the elucidation of what sex specificfactors are responsible should provide more information into the nature oflanguage development and, potentially, language problems.