Association between externalizing behavior (standardized within each sample) and IQ in 3 male samples, as a function of dopamine D4 receptor 7-repeat allele (DRD4-7) status. A, Sample 1: aggressive children (n = 48). B, Sample 2: adult attention-deficit/hyperactivity disorder (n = 42). C, Sample 3: low economic status community child sample (n = 87). All P values are 2-tailed.
Sample 3. Mean estimated IQ as a function of physical aggression category and dopamine D4 receptor 7-repeat allele (DRD4-7) status. Groups not sharing a superscript differ significantly (P<.05).
DeYoung CG, Peterson JB, Séguin JR, Mejia JM, Pihl RO, Beitchman JH, Jain U, Tremblay RE, Kennedy JL, Palmour RM. The Dopamine D4 Receptor Gene and Moderation of the Association Between Externalizing Behavior and IQ. Arch Gen Psychiatry. 2006;63(12):1410-1416. doi:10.1001/archpsyc.63.12.1410
Dopaminergic neurotransmission is implicated in externalizing behavior problems, such as aggression and hyperactivity. Externalizing behavior is known to be negatively associated with cognitive ability. Activation of dopamine D4 receptors appears to inhibit the functioning of the prefrontal cortex, a brain region implicated in cognitive ability. The 7-repeat allele of the dopamine D4 receptor gene produces less efficient receptors, relative to other alleles, and this may alter the effects of dopamine on cognitive function.
To examine the influence of a polymorphism in the third exon of the dopamine D4 receptor gene on the association between externalizing behavior and IQ.
In 1 community sample and 2 clinical samples, the presence or absence of the 7-repeat allele was examined as a moderator of the association between externalizing behavior and IQ; the strength of this effect across samples was estimated meta-analytically.
Eighty-seven boys from a longitudinal community study, 48 boys referred clinically for aggression, and 42 adult males diagnosed with attention-deficit/hyperactivity disorder.
Main Outcome Measures
IQ scores and observer ratings of externalizing behavior were taken from existing data sets.
Among individuals lacking the 7-repeat allele, externalizing behavior was negatively correlated with IQ (mean r = −0.43; P<.001). Among individuals having at least 1 copy of the 7-repeat allele, externalizing behavior and IQ were uncorrelated (mean r = 0.02; P = .45). The difference between these correlations was significant (z = −2.99; P<.01).
Allelic variation of the dopamine D4 receptor gene appears to be a genetic factor moderating the association between externalizing behavior and cognitive ability. This finding may help to elucidate the adaptive value of the 7-repeat allele.
Externalizing behavior refers to a broad category of behaviors involving disinhibition and approach, including aggression, antisocial behavior, hyperactivity, and impulsivity.1,2 Externalizing behavior problems are typically associated with cognitive impairment, and the negative association between externalizing behavior and IQ is particularly well established.3- 7 In attention-deficit/hyperactivity disorder (ADHD), this negative association appears to be primarily genetically based.5 The neuromodulator dopamine has been implicated in various forms of externalizing behavior8- 11 and is known to affect cognitive function.12 Localization of the dopamine D4 receptor in the prefrontal cortex13- 15 and findings that specific blockade of D4 receptors improves some cognitive impairments16- 18 prompted our investigation of the possibility that allelic variation in the dopamine D4 receptor gene (DRD4) might moderate the negative association between externalizing behavior and IQ.
DRD4 contains a functional polymorphism consisting of a variable number of tandem repeats of a 48–base pair sequence in the third exon of the gene.19 The 7-repeat allele (DRD4-7), which is present in about 20% of the population worldwide but varies widely in frequency geographically,20 has been associated with ADHD,21 novelty seeking,22,23 and alcohol and drug abuse.24,25 However, findings regarding these phenotypic associations with DRD4-7 have not been consistently replicated.26- 28 The best-established finding to date is the association between DRD4-7 and ADHD.29DRD4-7 is approximately twice as prevalent in ADHD probands and appears to be associated with 25% to 50% of the genetic risk for ADHD.29,30 Inconsistencies in past findings may reflect, in part, a complex relation between DRD4 variation and externalizing behavior problems. A review of recent research on the properties and functions of the dopamine D4 receptor may allow the generation of more sophisticated hypotheses regarding likely phenotypic associations with DRD4-7.
The D4 receptor is heavily expressed in the prefrontal cortex,13- 15 where it appears to modulate excitatory signaling.16 This suggests that D4 receptors may be involved in the dopaminergic modulation of the cognitive functions of the dorsolateral prefrontal cortex, which include working memory and have been linked to general cognitive ability and IQ.12,31- 33 Dopamine has a generally salutary effect on these cognitive functions, but the effect exhibits an inverted U–shaped function, with impairments evident at high, as well as low, levels of dopamine.12 Some of the negative effects of excess dopamine on cognition may be produced by the action of D4 receptors.12 Clozapine, an atypical antipsychotic with a much stronger affinity for D4 than for other dopamine receptors, appears to improve the cognitive symptoms of schizophrenia, a disorder that involves irregularities of dopaminergic transmission.16 In contrast, traditional antipsychotics, which do not preferentially target D4 receptors, do not improve these cognitive symptoms.16 Additionally, selective D4 receptor antagonists appear capable of reversing pharmacologically induced cognitive deficits in monkeys.17,18 Further, it appears that D4 blockade is only effective in producing cognitive benefits when other dopamine receptors are not blocked, suggesting a unique role for D4 among dopamine receptors in inhibiting cognitive processes.16,18
Relative to the 2 other most common DRD4 alleles (2- and 4-repeat) and to the 10-repeat allele, DRD4-7 produces less efficient receptors, with decreased potency for coupling with adenylate cyclase, part of the receptor's second messenger system.34,35DRD4-7 also appears to decrease gene expression, which would further diminish the effects of D4 receptors in the brain.36 Because of these reductions in D4 function and expression, DRD4-7 may act as an endogenous D4 suppressor; carriers of this allele seem likely to exhibit lower levels of the processes associated with D4 receptors. One might expect some similarity, therefore, between the effects of D4 antagonists and the DRD4-7 phenotype. Because D4 antagonists appear to alleviate some cognitive impairments, DRD4-7 could conceivably attenuate negative associations between cognitive ability and behaviors associated with increased dopaminergic activity, such as externalizing behavior.8- 11 (Although dopamine agonists, such as methylphenidate, are used in relatively low doses to reduce externalizing behavior, they appear to produce this effect by decreasing net dopaminergic activity through activation of presynaptic inhibitory autoreceptors.11) Consistent with this possibility, Swanson and colleagues37 found that among children diagnosed with ADHD, those who had DRD4-7 did not show deficits relative to controls on 3 neuropsychological tests of the attentional network involving dorsolateral prefrontal cortex, whereas those who did not have DRD4-7 did show deficits.
We tested the hypothesis that DRD4 variation might moderate the commonly reported negative association between externalizing behavior and IQ,3- 6 using 3 male samples with high mean levels of externalizing behavior. We then employed meta-analysis to obtain an estimate of effect size across all 3 samples.
For all 3 samples, DNA was isolated from peripheral leukocytes using standard procedures and genotyped following amplification of the region in the third exon of the DRD4 gene containing the 48–base pair variable number of tandem repeats, using a modification of the methods of Lichter and colleagues.38 The forward and backward primers were D4-3 (5′-GCGACTACGTGGTCTACTCG-3′) and D4-42 (5′-AGGACCCTCATGGCCTTG-3′). The polymerase chain reaction was performed in 25 μL (final volume) of Taq polymerase buffer (50 mmol/L potassium chloride, 10 mmol/L Tris-chloride, pH = 9, 1 mmol/L magnesium chloride, and 1% Triton X-100), containing 10% dimethyl sulfoxide; 200 μm each of deoxyadenosine, deoxycytidine, and deoxythymidine triphosphates; 100 μm deoxycyguanosine triphosphate; 100 μmol/L deaza-deoxyguanosine triphosphate; 500 ng of each primer; and 1 ng of DNA. Conditions for amplification were 40 cycles of 20 seconds at 95°C, 20 seconds at 54°C, and 40 seconds at 72°C, followed by a final extension of 4 minutes at 72°C, using an MJ Research Inc PT-100 thermocycler (Waltham, Mass). Polymerase chain reaction products were analyzed after electrophoresis (10% nondenaturing polyacrylamide or 3.5% agarose) and ethidium bromide staining. Participants were considered DRD4-7 positive if they had at least 1 copy of DRD4-7 and DRD4-7 negative if they did not.
Genotypes were available for 50 male children who were referred clinically to participate in a study of aggression at the Centre for Addiction and Mental Health in Toronto, Canada. They ranged in age from 5 to 15 years (mean, 9.89 ± 2.53 years). Forty were white (80%), 8 were African American (16%), 1 was East Asian (2%), and 1 was Native American (2%). The Asian and Native American participants were excluded from analysis because of population stratification; DRD4-7 frequencies are very different in East Asian and Native American populations than in white and African American populations.20 Because the frequency of DRD4 alleles is very similar in white and African American populations, inclusion of African Americans does not present a significant risk of population stratification. Nonetheless, we note how our results would have changed if African American participants had been excluded. Twenty-eight boys were DRD4-7 negative (58%) and 20 were DRD4-7 positive (42%). Twelve of these boys were being treated with a psychostimulant (eg, methylphenidate) at the time of assessment, but treatment status was not significantly related to DRD4-7 status (χ21 = 1.87; P = .17).
Genotypes were available for 67 male adults diagnosed with ADHD in adulthood who participated in a study of pharmacological treatments for ADHD at the Centre for Addiction and Mental Health. Of these, 42 had observer ratings from which an index of externalizing behavior could be calculated. They ranged in age from 18 to 56 years (mean, 35.17 ± 10.22 years). Forty-one (98%) were white and 1 (2%) was African American. Twenty-nine (69%) were DRD4-7 negative and 13 (31%) were DRD4-7 positive. Externalizing behavior and IQ were assessed prior to all pharmacological treatment.
Genotypes were available for 87 boys participating in a longitudinal study of French-speaking white boys who started kindergarten in 1983 in the 53 lowest socioeconomic status schools of the Catholic School Commission of Montreal, Montreal, Canada.39 Genotyping was done at age 17 years, recruiting from a sample of 203 boys who had been selected from the longitudinal cohort, primarily on the basis of teacher ratings of physical aggression at ages 6, 10, 11, and 12 years, to participate in laboratory studies at 15 years.40 Nonaggressive boys had teacher-rated aggression scores below the 70th percentile at all assessment points and constituted 35% of the complete longitudinal sample. Unstable-aggressive boys exceeded the 70th percentile for aggression at 1 or 2 assessment points and constituted 46% of the complete longitudinal sample. Stable-aggressive boys had aggression scores above the 70th percentile at age 6 years and at least twice more from ages 10 to 12 years and constituted 19% of the complete longitudinal sample. In the genotyped sample, 29 boys (33%) were nonaggressive, 29 were unstable aggressive, and 29 were stable aggressive. Fifty-seven boys (66%) were DRD4-7 negative and 30 (34%) were DRD4-7 positive. DRD4-7 status was unrelated to physical aggression category (χ22 = 1.22; P = .53). Six boys were being treated with a psychostimulant at 1 or more assessment points (an additional 6 did not have treatment status reported). Treatment status was unrelated to DRD4-7 status (χ21 = 0.00; P = .95). Not surprisingly, treatment status was significantly linearly related to aggression category; 2 of the 6 boys undergoing treatment were unstable aggressive and 4 were stable aggressive (χ21 = 4.11; P<.05).
Assessments of externalizing behavior and IQ were made using the instruments described in Table 1. Assessments took place at time of entry to the studies, unless otherwise noted. In all cases where multiple assessments of either construct were available, they were standardized and averaged to yield a single score. Full-scale IQ scores in samples 1 and 3 were estimated based on scores from the administered subtests. IQ estimates of this sort typically correlate at approximately r = 0.90 with full-scale IQ, even when based on a single administration; this is a common method for assessing cognitive ability while conserving time and resources.5,48 IQ is highly heritable and stable over time, so minor differences in time of administration relative to assessments of externalizing behavior should not reduce the sensitivity of our analyses.5,49
Figure 1 shows the main finding of interest, confirming in all 3 samples our hypothesis that possession of the DRD4-7 allele attenuates the association between externalizing behavior and IQ. Only in the DRD4-7–negative groups was there a significant negative correlation between these 2 variables. Because different instruments and assessment techniques were used in the 3 samples, the Schmidt-Hunter method of meta-analysis was used to calculate pooled effect sizes.50 Correlations between externalizing behavior and IQ were weighted by sample size and combined for each genotype group. The mean n-weighted (population-weighted) r for the DRD4-7–negative group (n = 114) was −0.43 (P<.001), while for the DRD4-7–positive group (n = 63) it was 0.02 (P = .45). These effect sizes differed significantly between the genotype groups (z = −2.99; P<.01). (All P values are 2 tailed, unless noted otherwise.)
The mean estimated IQ across both genotype groups was 99.29 (SD = 12.97; range, 70-127). Boys who were being treated with psychostimulants did not differ significantly in IQ or externalizing behavior from those who were not: IQ, t48 = −0.17, P<.86; externalizing behavior, t48 = −0.35, P = .73. The genotype groups did not differ significantly in IQ or externalizing behavior (Table 2), and the correlation between IQ and externalizing behavior when both genotype groups were combined was not significant (r = −0.20; P = .18). If the 8 African American participants were excluded from the primary analysis, the correlation in the DRD4-7–positive group remained the same as that reported in Figure 1A; in the DRD4-7–negative group, it changed only slightly, from −0.38 to −0.34.
The mean full-scale IQ across both genotype groups was 113.48 (SD = 14.00; range, 79-143). The genotype groups did not differ significantly in IQ or externalizing behavior (Table 2). The correlation between IQ and externalizing behavior in the full sample was significant (r = −0.32; P<.05), but weaker than the correlation in the DRD4-7–negative group alone, which is reported in Figure 1B. If the 1 African American participant was excluded, the only change was in the DRD4-7–negative group, in which the correlation rose from −0.44 to −0.45.
The mean estimated IQ across both genotype groups was 101.89 (SD = 12.85; range, 59-126). The 6 physically aggressive boys who were treated with psychostimulants did not differ significantly in IQ from the physically aggressive boys who were not treated (t51 = 1.26; P = .21). The genotype groups differed significantly in IQ but did not differ in externalizing behavior (Table 2). The correlation between IQ and externalizing behavior when both genotype groups were combined was significant (r = −0.34; P<.05) but was weaker than the correlation in the DRD4-7–negative group alone, which is reported in Figure 1C.
The physical aggression, opposition, and hyperactivity scales, which were combined to yield a single externalizing behavior score for sample 3, were also examined separately to determine whether the moderating effect of DRD4-7 was similar for different types of externalizing behavior. All 3 scales showed the same pattern as the composite score: significant negative correlations with IQ in the DRD4-7–negative group (r = −0.42, −0.44, and −0.37, respectively; P<.01) and no correlations with IQ in the DRD4-7–positive group (r = −0.04, 0.04, and −0.09, respectively; P>.60).
To examine more closely the significant difference in IQ between genotypes in this sample, boys were grouped according to the physical aggression categories by which they were originally selected for laboratory studies. (As these categories were based on the stability of aggression longitudinally, they were not applicable in samples 1 and 2.) The regression lines in Figure 1C suggest that the 2 genotype groups are more likely to differ significantly in IQ among boys higher in externalizing behavior, and this hypothesis is supported in Figure 2, which shows mean IQ for each of the 3 aggression categories, broken down by genotype group. An analysis of variance comparing 3 aggression categories with 2 DRD4-7 status categories revealed significant main effects of genotype and aggression category on IQ: genotype, F1 = 5.42, P<.05; aggression category, F2 = 5.06, P<.01. The effect of the interaction between aggression category and DRD4-7 status neared 1-tailed significance: F2 = 2.16, P = .06 (a 1-tailed test is appropriate because we could predict this effect based on the results in Figure 1C). Planned comparisons revealed that the significant difference in IQ between DRD4-7–negative and DRD4-7–positive groups was found only in the stable-aggressive group (Figure 2).
In 3 male samples, the association between externalizing behavior and IQ was moderated by the presence or absence of the DRD4-7 allele. Specifically, the negative association between externalizing behavior and IQ was completely attenuated by DRD4-7. Although the samples were fairly small, they were independent replications, and meta-analysis provided strong evidence that this finding is likely to be robust and reliable. Variation in the DRD4 gene thus appears to differentiate 2 distinct male phenotypes. In males without DRD4-7, externalizing behavior is negatively associated with IQ. In males with DRD4-7, however, there is no association between externalizing behavior and IQ. This effect was evident in both adults and children and it remained similar in magnitude across various types of externalizing behavior, suggesting that it may be related to a common feature of externalizing behaviors, such as increased dopaminergic neurotransmission.8- 11
The attenuation seen in the DRD4-7–positive group seems likely to be the result of the fewer, less efficient D4 receptors produced by the DRD4-7 allele.34- 36 Decreased D4 receptor function should produce a DRD4-7 phenotype in which the cognitive disruption associated with D4 receptor activation is diminished. Males with the DRD4-7 allele thus appear to be protected from the decrement in cognitive ability associated with externalizing behavior. Further research is necessary to determine whether this finding holds true for measures of cognitive ability other than IQ.
DRD4-7 status does not appear to affect IQ directly, despite the significant difference in IQ between genotype groups in sample 3, but rather to moderate its association with externalizing behavior. Comparisons of the 3 physical aggression groups in sample 3 revealed that significant differences in IQ between DRD4-7–positive and DRD4-7–negative groups emerged only in the most consistently aggressive group (a group that was intentionally overrepresented in this sample, relative to the general population). The pattern of increasing differences in IQ between genotype groups (Figure 2), as severity of physical aggression increases, is paralleled by the increasing gap between the 2 regression lines in Figure 1C as a function of increasing levels of externalizing behavior. Group differences in the strength of association between 2 variables can lead to differences in group means in the extremes of the distributions of those variables. The significant main effect of genotype on IQ in sample 3 seems likely to be an anomaly resulting from the selection process for that sample, rather than a genuine indicator of any direct effect of DRD4 variation on IQ. In keeping with this hypothesis, a study by Ball and colleagues,49 which compared individuals of high IQ with others of average IQ, failed to find any difference in the frequency of DRD4 alleles.
Many theoretical models view intelligence as an important element in behavioral self-regulation, and both low IQ and deficits in working memory have been described as risk factors for externalizing behavior.6,7,51 Researchers interested in antisocial behavior and delinquency have usually argued that low IQ contributes causally to externalizing behavior, rather than the reverse.6 However, the possibility that a causal pathway in the opposite direction (from externalizing behavior to IQ) might be mediated by neurophysiological processes, such as D4 receptor activation, has not previously been considered. In this case, D4 receptor activation resulting from the increased dopaminergic activity associated with externalizing behavior might lead (both immediately and developmentally) to decreased cognitive ability. It is also possible that externalizing behavior and cognitive ability are not directly causally linked, but that they vary together because both are affected by high levels of dopamine resulting from some other genetic or environmental factor. Whatever the causal pathway, our results indicate that it is likely to involve D4 receptors, as the negative association between externalizing behavior and IQ is completely attenuated by the presence of the DRD4-7 allele. The causal possibilities discussed here, therefore, apply only to the DRD4-7–negative group. Externalizing behavior may be associated with dopaminergic activity in both genotype groups, but DRD4-7 appears to prevent this dopaminergic activity from affecting cognitive ability.
Given the meta-analysis by Faraone and colleagues,29 which indicates that DRD4-7 is associated with a minor increase in risk for ADHD, it is important to consider how DRD4-7 could lead both to the decoupling of cognitive ability from externalizing behavior and to increased risk for a disorder that involves externalizing behavior. One possible explanation may be developed based on findings that DRD4-7 is associated with faster reaction times,30,37,52 which might lead to certain specific forms of impulsive or inattentive behavior without an associated impairment of general cognitive ability. The fact that the present study did not show an association between DRD4-7 and externalizing behavior is not incompatible with the evidence for an association of DRD4-7 with ADHD, because we were not testing for association with ADHD specifically. Additionally, our results in sample 2, which was composed entirely of adults diagnosed with ADHD, indicate that DRD4 variation can moderate the association of externalizing behavior and cognitive ability even within an ADHD population.
Recent studies indicating that DRD4-7 is a relatively new allele that has increased in frequency because of positive selection support the argument that DRD4-7 must have some adaptive value.30,53 Our results suggest a novel hypothesis for what that value might be. DRD4-7 appears to produce a phenotype in which cognitive ability is decoupled from behaviors like hyperactivity, impulsivity, and aggression (at least in some male populations). Some environments may favor or even demand such externalizing behaviors, and in these environments, the ability to manifest these behaviors without associated decrement in cognitive ability could be highly advantageous to individuals with DRD4-7, outweighing any associated drawbacks. Externalizing behavior, especially aggression, must have been important in human evolution, given the probability that intraspecies conflict between human groups has constituted a strong selection pressure.54,55 Externalizing behavior may have been particularly advantageous in the unstable or resource-depleted environments that are hypothesized to have driven selection for DRD4-7.30,53 The other DRD4 alleles, in contrast, may be more adaptive in stable environments, in which externalizing behavior is less likely to be advantageous.
Correspondence: Jordan B. Peterson, PhD, Department of Psychology, University of Toronto, Toronto, ON M5S 3G3, Canada (firstname.lastname@example.org), or Colin G. DeYoung, PhD (email@example.com).
Accepted for Publication: July 28, 2005.
Financial Disclosure: None reported.
Funding/Support: This work has been supported by grants from the Fonds de Recherche en Santé du Québec (Drs Tremblay and Palmour); grants from Québec's Conseil Québécois de la Recherche Sociale and Canada's Social Sciences and Human Research Council (Dr Tremblay); grants from the Alva Foundation (Drs Beitchman and Kennedy); and fellowships to Dr Séguin from the Canadian Institutes for Health Research and the National Science and Engineering Research Council.
Acknowledgment: We thank Chawki Benkelfat for helpful discussions and review of a previous version of the manuscript. Drs Tremblay, Kennedy, and Palmour share senior authorship of this work.