Close
About
FAQ
Home
Collections
Login
USC Login
Register
0
Selected
Invert selection
Deselect all
Deselect all
Click here to refresh results
Click here to refresh results
USC
/
Digital Library
/
University of Southern California Dissertations and Theses
/
Patterns of EEG spectral power in 9-10 year old twins and their relationships with aggressive and nonaggressive antisocial behavior in childhood and adolescence
(USC Thesis Other)
Patterns of EEG spectral power in 9-10 year old twins and their relationships with aggressive and nonaggressive antisocial behavior in childhood and adolescence
PDF
Download
Share
Open document
Flip pages
Contact Us
Contact Us
Copy asset link
Request this asset
Transcript (if available)
Content
Patterns of EEG spectral power in 9-10 year old twins and their relationships with aggressive and nonaggressive antisocial behavior in childhood and adolescence Sharon Niv Department of Psychology, University of Southern California, Los Angeles CA Doctoral Dissertation August, 2013 Dissertation Committee Laura A. Baker, Ph.D., Chair Carol A. Prescott, Ph.D. Margaret Gatz, Ph.D. Maja Mataric, Ph.D. 2 Table of Contents Dissertation Abstract .................................................................................................................... 6 Chapter 1. General Introduction and Overview ........................................................................ 8 Background ......................................................................................................................... 8 The Present Study ............................................................................................................. 12 Table 1. Study Demographics ........................................................................................... 16 The overall aims of this research are the following: ......................................................... 17 Chapter 2. Aggressive and nonaggressive ASB: Heritability and Stability of Antisocial Behavior in Childhood and Adolescence .................................................................................. 18 Chapter 2: Introduction ..................................................................................................... 19 Chapter 2: Methods ........................................................................................................... 22 Figure 2.1 Cholesky Decomposition ACE Model Example Examining Longitudinal Aggressive and Nonaggressive ASB ................................................................................ 27 Figure 2.2 Independent Pathway ACE Model Example Examining Longitudinal Aggressive and Nonaggressive ASB ................................................................................ 28 Figure 2.3 Cholesky Decomposition ACE Model Example Examining Single Longitudinal Common Pathway for Aggressive and Nonaggressive ASB in Both Waves Combined .......................................................................................................................... 29 Figure 2.4 Cholesky Decomposition ACE Model Example Examining Longitudinal Common Pathways for Wave 1 and 3 Antisocial Behavior ............................................. 30 Figure 2.5 Cholesky Decomposition ACE Model Example Examining Longitudinal Common Pathways for Aggressive and Nonaggressive ASB .......................................... 31 Chapter 2: Results ............................................................................................................. 31 Table 2.1. Means, Standard Deviations, and Ns for Aggressive and Nonaggressive ASB Behavior by Sex ................................................................................................................ 32 Table 2.2. Phenotypic Correlations among Aggressive and Nonaggressive ASB Behavior Scores by Sex .................................................................................................................... 32 Table 2.3. Intraclass Correlations for Aggressive and Nonaggressive ASB Behavior, at Ages 9-10 and 14-15 Years............................................................................................... 33 Table 2.4 Cross-twin Cross-trait Correlations for Aggressive and Nonaggressive ASB by zygosity at Ages 9-10 and 14-15. ..................................................................................... 34 Table 2.5 Multivariate Model Fit Indices for Aggressive and Nonaggressive ASB Behavior in Waves 1 (9-10 years) and 3 (14-15 years) Combined ................................... 36 References ......................................................................................................................... 44 3 Chapter 3. Relationships of Alpha, Beta, and Theta EEG Spectral Properties with Aggressive and Nonaggressive ASB in Children and Adolescents ......................................... 48 Chapter 3: Abstract ........................................................................................................... 48 Chapter 3: Introduction ..................................................................................................... 49 Chapter 3: Methods ........................................................................................................... 52 Chapter 3: Results ............................................................................................................. 54 Table 3.1. Means, Standard Deviations, and Ns for Variables Used in Analyses by Sex 55 Table 3.2. Significant Main Effects found in Multilevel Modeling of EEG Power and Antisocial Behavior .......................................................................................................... 57 Chapter 3: Discussion ....................................................................................................... 58 References ......................................................................................................................... 62 Chapter 4. Childhood EEG Frontal Alpha Power as a Predictor of Adolescent Antisocial Behavior: A Twin Heritability Study ........................................................................................ 66 Chapter 4: Abstract ........................................................................................................... 66 Chapter 4: Introduction ..................................................................................................... 67 Chapter 4: Methods ........................................................................................................... 68 Chapter 4: Results ............................................................................................................. 72 Table 4.2 Twin and Cross-twin Cross-trait Correlations for FAP and Wave 3 Aggressive ASB by Zygosity............................................................................................................... 73 Table 4.2. Bivariate Model Fit Indices for FAP and Aggressive ASB in Wave 3 ........... 75 Figure 4.1. Bivariate Heritability Model of Wave 1 EEG Alpha Power with Wave 3 Aggressive ASB in Males ................................................................................................. 76 Chapter 4: Discussion ....................................................................................................... 76 References ......................................................................................................................... 82 Chapter 5. Additional Analyses ................................................................................................. 85 5.A. Spectral Power Heritability ....................................................................................... 87 5.A.1. Background ............................................................................................................ 87 5.A.2. Measures & Methods ............................................................................................. 88 5.A.3. Results .................................................................................................................... 90 Table 5.1. Theta Power AE Estimates by Site and Sex .................................................... 92 Table 5.2. Alpha Power AE Estimates by Site and Sex.................................................... 92 Table 5.3. Beta Power AE Estimates by Site in both Sexes ............................................. 94 Figure 5.1. Heritability of Male Full Brain Theta, Alpha, and Beta Power...................... 95 5.B. Additional Measures of Antisocial Behavior ............................................................ 97 4 5.B.1. Measures................................................................................................................. 97 5.B.2. Results .................................................................................................................... 98 Table 5.4. Means, Standard Deviations, and Ns for Measures of Antisocial Behavior by Sex..................................................................................................................................... 99 Tables 5.5a-5.5d. Multilevel Regression Coefficients (with Standard Error) for Frontal and Parietal EEG Spectral Power with Additional Measures of Antisocial Behavior, by Sex and Wave. ................................................................................................................ 101 5.C. Environmental Moderators in Relationships between Spectral Power and Antisocial Behavior .......................................................................................................................... 104 5.C.1. Background .......................................................................................................... 104 5.C.2. Measures............................................................................................................... 105 5.C.3. Results .................................................................................................................. 106 Table 5.6. Means, Standard Deviations, and Ns for Environmental Variables Used in Analyses by Sex. ............................................................................................................. 107 Table 5.7. Correlations between Aggressive and Nonaggressive ASB and Adverse Environmental Conditions in Waves 1 and 3 by Sex. .................................................... 107 Table 5.8. Unstandardized Regression Coefficients of Interaction Terms from Multilevel Models............................................................................................................................. 109 5D. Group Analysis based on Nonaggressive ASB Levels across Waves 1-4 ............... 110 5.D.A. Methods ............................................................................................................... 110 Table 5.9. Sample Size and Means of Latent Class Growth Analysis Groups for each Wave ............................................................................................................................... 111 5.D.2. Results .................................................................................................................. 111 Table 5.10. ANOVA Results for Nonaggressive ASB Group and EEG Spectral Power 113 5.E. Parental Antisocial Behavior ................................................................................... 114 5.E.1. Background........................................................................................................... 114 5.E.2. Measures. ............................................................................................................. 114 5.E.3. Results .................................................................................................................. 115 Table 5.11. Multilevel Regression Coefficients (with Standard Error) for Parental ASB Measures with Frontal and Parietal EEG Spectral Power, by Sex and Wave. ............... 117 5.F. Frontal Alpha Asymmetry and Antisocial Behavior ............................................... 118 5.F.1. Background ........................................................................................................... 118 5.F.2. Variables ............................................................................................................... 119 5.F.3. Results................................................................................................................... 120 Table 5.12. Multilevel Regression Coefficients (with Standard Error) for Measures of Asymmetry and Antisocial Behavior .............................................................................. 121 5 5. Summary ..................................................................................................................... 122 Chapter 6. General Discussion................................................................................................. 125 Summary of Findings ...................................................................................................... 125 Interpretations and Implications ..................................................................................... 127 Aim 1: Structure of influences on aggressive and nonaggressive ASB.......................... 127 Aim 2: Spectral power as a biomarker for ASB ............................................................. 127 Aim 3: Genetic overlap between frontal alpha power and ASB ..................................... 129 Developmental implications ........................................................................................... 130 Environmental Moderators ............................................................................................. 131 Intervention Implications ................................................................................................ 131 Strengths and Limitations ............................................................................................... 132 Future Directions ............................................................................................................ 133 Conclusions ..................................................................................................................... 135 References .................................................................................................................................. 137 Appendix A: CBCL Nonaggressive ASB and Aggressive ASB Items with Means, SDs, and Percentage Endorsement .......................................................................................................... 146 Table A.1. Nonaggressive ASB Items, Means, Standard Deviations, and Percentage Endorsement by Sex and Wave ...................................................................................... 146 Table A.2. Aggressive ASB Items, Means, Standard Deviations, and Percentage Endorsement by Sex and Wave ...................................................................................... 147 Appendix B. Univariate Fit Statistics and Estimates for Aggressive and Nonagressive ASB at Waves 1 and 3. ...................................................................................................................... 148 Table B.3 Model Aggressive ASB 1............................................................................... 148 Table B.4 Model Nonaggressive ASB 1 ......................................................................... 149 Table B.5 Model Aggressive ASB 3............................................................................... 149 Table B.6 Model Nonaggressive ASB 3 ......................................................................... 150 Acknowledgments ..................................................................................................................... 151 6 Dissertation Abstract This project examines commonalities and differences between aggressive and nonaggressive forms of antisocial behavior (ASB) using methodologies of behavioral genetics and EEG spectral power analysis. Using a longitudinal community sample of 1550 Los Angeles area twins, data were collected at two waves: Wave 1 (age 9-10 years) and Wave 3 (age 14-15 years). ASB was assessed using parent reports of the Child Behavioral Checklist (Achenbach, 1991), using the aggression and delinquency (non-aggressive antisocial behavior) scales. Biometric statistical modeling was applying to the twin variances and covariances in order to discern the pattern of genetic and environmental influences on aggressive and nonaggressive ASB across the two waves of data collection. Findings supported both the commonality and the distinction between these forms of ASB, as 44-79% of the influences were found to be shared through a latent common factor of ASB, and the rest were unique to each scale. Additive genetic, shared environmental and nonshared environmental effects were found to influence both time points; new influences in adolescence were exclusively genetic. EEG data from an open-eyed 3-minute resting period were used with principal components analysis to derive power scores in the frontal (channels: F8, F4, Fz, F3, F7, Af3, Af4, Fp1, Fp2) and parietal (channels: P8, P4, Pz, P3, P7) regions in the frequency ranges of theta (4-8 Hz), Alpha (8-10.5 Hz), and beta (10.5-30 Hz). Support was found for higher frontal alpha power, as well as higher frontal and parietal beta power, in association with Wave 3 aggressive ASB in the males. No significant relationships were found in the females, and none with nonaggressive ASB. Also, no significant concurrent relationships emerged. Bivariate genetic modeling found a small but significant genetic overlap (Rg=0.22) between frontal alpha power and aggressive ASB in the males, which entirely 7 accounted for the phenotypic correlation between them. This study provides some replication for past findings of low arousal and ASB, but suggests that: 1. These relationship may exist for aggressive but not nonaggressive ASB. 2. Females may not show identical arousal patterns in relation to ASB. 3. Relationships may be less pronounced in community rather than clinical or criminal samples. 8 Chapter 1. General Introduction and Overview Background Antisocial behavior (ASB) in children and adolescents is a topic of great interest for psychologists and sociologists, as it is characteristic of children at risk for developing criminal or violent behavior. This spectrum of behavior, which encompasses aggressive ASB, nonaggressive ASB, substance use, and other forms of disruptive behavior, has been associated with poor academic performance, poor employment records, and increased likelihood of adult violent and criminal behavior (Hinshaw, 1992; Farrington, 1989). In order to distinguish types of ASB and improve understanding of its etiology and trajectory, one distinction that has arisen is aggressive versus nonaggressive behavior. While aggressive behavior may include fighting, weapon use, and physical bullying, nonaggressive ASB encompasses rule-breaking behavior such as lying, cheating, theft, and other petty misbehaviors. Elucidating etiological and biological similarities and differences in aggressive and nonaggressive forms of ASB is the topic of the present research. Findings suggest that while aggressive and nonaggressive ASB are correlated, they are in fact distinct and have different developmental trajectories. For example, aggressive ASB shows more longitudinal stability than does nonaggressive ASB, and is considered more integrated within personality (Moffitt, 2003; Tremblay, 2003). In terms of development, aggressive ASB is also known to decrease over the period of childhood to adolescence, whereas nonaggressive ASB generally increases (Burt, 2009 & 2012; Rhee & Waldman, 2002). There also exists evidence that certain forms of nonaggressive ASB are adolescence-specific, emerging in puberty and ending in adulthood (Moffitt, 2003). These findings all shed light on development of different forms of dangerous behavior, and can direct researchers and clinicians to seek more targeted interventions. However, despite abundant research on these forms of ASB, many questions remain to be asked, 9 including whether there are different warning signs that predict aggressive and nonaggressive ASB, and whether they show different biological profiles. Early detection signs for individuals vulnerable to developing ASB are an important topic for research (White, et al., 1990). Some researchers posit that adolescence is critical period for trajectories of ASB, in which ‘downward spirals’ may begin, leading to academic failure and school drop-out (Reyes, et al. 2000), aggressive ASB, or nonaggressive ASB (Moffitt, et al. 2001). For these reasons, the period of late childhood to mid adolescence was selected for this research, as it may be more conducive to identifying developmental pathways. The goals of finding biological predictors of behavioral problems are two-fold: 1. Early detection; 2. Elucidation of underlying biological mechanism for potential intervention. In support of early detection, prominent theorists posit that early identification and treatment may increase the likelihood of benefit from evidence-based interventions (McNeil, et al., 1999; Liddle, et al. 2004). As for biological mechanisms, the identification of markers that precede development of behavior help to detect individuals at risk as well as potential neural targets for therapy. The biological markers selected in this research is EEG spectral power, due to past research suggesting it may be a marker for serious crime and psychopathy. EEG spectral power represents collective neuronal firing in a region. Ranges are regularly characterized in adults as delta (1-4 Hz, typical of slow wave sleep stage), theta (4-8 Hz, typical of drowsiness, deep relaxation, and pre-sleep stages – states of low arousal), alpha (8-13 Hz, typical of relaxation and sleepiness – state of low arousal), beta (13-30 Hz, typical of concentration and wakefulness), and gamma (30-100 Hz, typical of high arousal and alertness). In children, alpha frequencies are slowed, typically measured at 8-10.5 Hz (Baving, et al., 200 & 2003l Forbes, et al. 2006). Early research found that psychopaths and violent criminals showed 10 higher levels of slow-wave EEG cortical activity, comprising delta, alpha, and theta frequencies than did control subjects (Ellingson, 1954; Hill, 1952). These findings led Robert Hare to develop his slow arousal theory, which posits that a poorly aroused brain leads to impulsive, aggressive, and risk-taking behavior due to a ‘stimulus hunger’ (Hare, 1970). In children, increased cortical alpha activity (a marker of slow arousal) has been associated with later crime (Mednick, et al. 1981; Petersen, et al. 1982). In a group of 24-year-old male criminals, retrospective analysis found high alpha power at the age of 15 (Raine, et al. 1990). Slow wave activity related to antisocial behavior problems has primarily been linked to parietal and frontal regions, which were selected for the analyses in the present research. In addition, several other EEG markers have been associated with antisocial behavior problems, for example higher parietal and frontal theta power with attention deficient and hyperactivity disorder (ADHD), which often involves behaviors that are oppositional and disruptive (Chabot & Serfontein, 1996). A rise in frontal and parietal beta power has been found in relation to alcoholism (Ehlers & Schuckit, 1990; Rangaswamy, et al. 2002). Notably, most of these studies were performed exclusively with males, and with criminal or psychiatric populations, and it remains to be seen whether these findings would replicate in a community sample of both sexes, or whether different patterns might emerge for females or in community samples. In addition to investigating potential biological underpinnings of certain types of behavior, it is also crucial to examine the genetic and environmental etiology of those relationships. Twin studies are powerful tools for estimating genetic and environmental influences on any measurable trait. The twin design utilizes the assumption that monozygotic (MZ, “identical”) twins share 100% of their genes and dizygotic (DZ, “fraternal”) share on average 50% of their genes. Using this assumption, twin heritability models use variances and covariances of MZ and DZ twins to 11 partial out influences into additive genetic (A), shared environmental (C) and non-shared environmental (E) influences. Using tools such as twin studies or adoption studies, behavior geneticists may examine what proportion of behavior in a population is due to genetics versus to environment, thus leading the way for future research to find relevant genetic markers or appropriate interventions. The heritability of ASB on the whole ranges in estimates from 38-56% (Gjone & Stevenson, 1997; Pesenti-Gritti et al., 2008). Aggressive ASB has consistently shown more heritability than nonaggressive ASB (Rhee & Waldman, 2002; Burt, 2009; Tuvblad, et al. 2005). A cross-sectional metaanalysis on this topic concluded that aggressive ASB 65% heritable whereas nonaggressive ASB (operationalized as rule-breaking) is 48% heritable. Nonaggressive ASB also showed some shared environmental influences (18%) whereas aggressive ASB didn’t (Burt, 2009). An additional cross-sectional metaanalysis specifically examined the overlap between the two forms of ASB and found that 38% of genetic variance and 10% of shared environmental influence was shared between them (Burt, 2012). There have been a few studies of EEG power patterns, both in adults and in children. Power of the different spectral frequencies has emerged as highly heritable, at 70-80% across studies of different aged participants (Smit, et al. 2005; van Beijsterveldt, et al. 1996). However, these examinations all either examined single electrodes or utilized simple averaging methods of measuring heritability, whereas the present research utilizes more sophisticated principal components analysis (PCA) methodology of measuring power. Using PCA connectivity methods, this study should be able to discern more accurate relationships between resting state EEG patterns and parent-rated behavioral reporting of aggressive and nonaggressive ASB. 12 The Present Study The rationale of the present study is to further elucidate the distinctions and commonalities between aggressive and nonaggressive ASB, as it is important to understand how these harmful forms of behavior develop and the distinctions are between them, as well as to examine neurobiological correlates that may underlie either or both. First, the project investigated the roles of genes and environment in the development of aggressive and nonaggressive ASB in childhood and adolescence. While it is known that there is genetic and environmental overlap for aggressive and nonaggressive ASB, the longitudinal structure has yet to be explored. This information provides direction for future molecular genetic studies, as well as helps to clarify why these forms of behavior would develop in unison or separately. It was hypothesized that aggressive and nonaggressive ASB would share some but not all genetic and environmental influences across time. This analysis is presented in Chapter 2. Next, EEG spectral power was examined for potential biomarker relationships with ASB. Although several past studies exist that have found low arousal in relation to crime and violence, this project aims to expound upon the information in several ways. First, the majority of past studies examined male participants exclusively or with a significant majority – this research examines these potential relationships in both sexes. Second, past studies were conducted with criminal or clinical samples – this project investigates whether similar relationships would be found in the general population. Lastly, past research found that child/adolescent low arousal was predictive of later delinquency – this research aims to verify whether the pattern would replicate, with childhood arousal pattern predicting adolescent behavior. It was hypothesized that higher alpha and theta frequencies would be associated with ASB. No hypotheses were formed 13 regarding differences between aggressive and nonaggressive ASB. This investigation is presented in Chapter 3. After phenotypic relationships were investigated, attention was turned to the basis of the relationship between arousal and ASB. Although the phenotypic relationship between low arousal and ASB had been investigated previously, it was never done so in a genetically informative sample. Hence, these analyses were novel, and could be used to formulate hypotheses about biological mechanisms of ASB. Using the significant phenotypic relationship that arose between low arousal (frontal alpha power) and aggressive ASB in the males, bivariate genetic modeling was conducted to find the genetic and environmental nature of the relationship. This is presented in Chapter 4. Lastly, in Chapter 5, several additional investigations are pursued in order to fill in gaps in the patterns of findings in Chapters 2-4. Several hypotheses were formulated to explain why relationships that emerged in Chapters 3-4 were not as robust as expected. First, the quality of the EEG data was questioned, and so heritability analysis was conducted on spectral power across 30 channels to investigate this possibility. Next, the strength of the primary instrument used to gauge ASB – the Child Behavioral Checklist (CBCL) - was questioned, and so several additional instruments for ASB were analyzed for relationships with EEG spectral power. In addition, it was proposed that robust relationships did not emerge because clinical and criminal samples may have higher incidence of exposure to adverse environmental conditions, and so life adversity was examined for potential moderation relationships. Next, a high-risk model was investigated, in which parents with higher ASB were identified, and their children’s EEG profiles examined. This was done as it could provide further support for the genetic basis of any emerging relationships. Lastly, to further investigate the finding that childhood spectral power 14 predicted adolescent but not concurrent ASB, latent class growth analysis was applied in order to see if different trajectories of ASB associated with different spectral power profile. This chapter also included an investigation of frontal alpha asymmetry and ASB, in order to complete all proposed analyses for this project. Demographics for the participants within this project are presented in Table 1. Participants were drawn from the USC Twin Study Risk Factors for Antisocial Behavior (RFAB), an ongoing and longitudinal twin study of over 700 pairs of twins from the Los Angeles area. This study examines data from Waves 1 and 3 of the study, during which times the twins were 9-10 and 14-15 years of age. The total number of participants from both waves is 1550 (340 male MZ, 340 female MZ, 234 male DZ, 240 female DZ, 198 male DZO, 198 female DZO), with precise breakdowns from each wave presented in Table 1. Attrition analysis was conducted to discern differences between children who participated in both waves and children who participated in Wave 1 but not 3. This regression analysis found that scores on neither the CBCL aggression or non-aggression scales were predictive of discontinuation. Mean scores for Wave 1 CBCL scales and EEG PCA power variables did not show significant differences between children who returned and those who did not. Lastly, Wave 3 CBCL scores also were not found to significantly differ between children who participated in Wave 1 and those who joined as participants in Wave 2 or 3. Mean ages were 9.6 years in Wave 1 and 15.0 years in Wave 3. Ethnicity percentages were 25.6% Caucasian, 37.2% Latino/Hispanic, 14.1% Black, 5.0% Asian, 0.1% Native American, and 17.5% mixed or other. Throughout this work, aggressive and nonaggressive ASB are gauged through the Child Behavioral Checklist’s Aggression and Delinquency scales. Hence, it is important to discuss the items examined in these scales, as well as the endorsement patterns from this sample. Appendix 15 A presents items from both CBCL scales are provided, as well as means and percentage endorsement. From Appendix B, it can be seen that the more frequently endorsed nonaggressive ASB items are lying or cheating, not feeling guilty after misbehavior, swearing, and preferring the company of other children. The most commonly endorsed aggressive items were arguing, demanding attention, being disobedient at home, being disobedient at school, jealousy, screaming, showing off, talking too much, teasing others, and temper tantrums. For both sexes at both waves, frequency of endorsement of frequent fighting fell between 10-18%. Severe nonaggressive ASB items such as fire and vandalism were endorsed less then 3% of the time, and severe aggressive ASB items such as physically attacking others were endorsed less than 10% of the time. In all, these are profiles of aggressive and nonaggressive ASB within the general population, and do not represent clinical or criminal populations. All interpretations of findings are conducted in this context. Table 1. Study Demographics W1 EEG W1 CBCL W3 CBCL W1 & W3 overlap Age W1 Age W3 Child’s ethnicity % Zygosity N N N N Mean (SD) Mean (SD) Cauc. Lat. Black Asian Nat.A. Mixed or other MZM 204 269 249 182 9.59 (0.56) 15.02 (0.77) 18.1 44.1 7.4 10.8 0 19.6 MZF 234 288 229 177 9.62 (0.59) 14.93 (0.88) 22.2 36.8 13.7 5.1 0.4 21.8 DZM 120 170 175 111 9.62 (0.60) 15.07 (1.00) 27.5 32.5 12.5 2.5 0 25.0 DZF 129 184 212 156 9.61 (0.59) 15.01 (0.91) 34.9 32.6 18.6 2.3 0 11.6 DZO M 107 147 141 89 9.61 (0.61) 15.03 (0.85) 30.8 35.5 20.6 1.9 0 11.2 DZO F 106 146 142 91 9.60 (0.59) 15.00 (0.74) 29.2 38.7 18.9 2.8 0 10.4 Total 900 1204 1148 806 9.61 (0.59) 15.00 (0.86) 25.6 37.2 14.1 5.0 0.1 17.5 Note. W1 = Wave 1; W3= Wave 3; EEG = Electroencephalography; CBCL = Child Behavioral Checklist; SD = Standard Deviation; Cauc. = Caucasian; Lat. = Latino; Nat. A. = Native American; MZ = Monozygotic; DZ = Dizygotic; DZO = Dizygotic Opposite Sex. The overall aims of this research are the following: 1. To investigate the longitudinal structure of genetic and environmental influences on aggressive and nonaggressive ASB in order to gain more precise understanding of differences and similarities, as well as stability and change, on the development of these two forms of ASB. 2. To investigate the viability of EEG spectral properties as biomarkers for children at risk for development of antisocial behavioral problems in childhood and adolescence, and investigate potential differences in relationships with aggressive and nonaggressive ASB. 3. To investigate the genetic and environmental etiology of the relationships between EEG spectral power and antisocial behavior. 18 Chapter 2. Aggressive and nonaggressive ASB: Heritability and Stability of Antisocial Behavior in Childhood and Adolescence 1 Chapter 2: Abstract This longitudinal twin study examined the structure of genetic and environmental influences on aggressive and nonaggressive antisocial behavior (ASB) at two time points: age 9- 10 years and 14-15 years, in order to examine change and stability of these influences across the span of childhood to mid-adolescence. Aggressive and nonaggressive ASB were influenced by a latent common factor of ASB within each wave of data collection. The childhood-age common factor of ASB was influenced by 41% genetics, 40% shared environment and 19% nonshared environment. In adolescence, 41% of influences on the common factor were novel and entirely genetic, while the remainder of influences were stable across time. In addition, both aggressive and nonaggressive ASB within each wave were found to have unique influences not common across subscales or across waves, highlighting specificity of influences on different problem behaviors at both ages. This research sheds light on the commonality of influences on etiology of different forms of antisocial behavior, and suggests future directions for research into intervention for antisocial behavior problems in youth, such as investigation of pre-adolescence environmental influences on the development of antisocial behavior problems, and genetic influences related to age of pubertal onset. 1 Authors: Sharon Niv, Catherine Tuvblad , Adrian Raine, Laura A. Baker 19 Chapter 2: Introduction Concern about violence and crime within society is pervasive, as these forms of behavioral problems encompass broad antisocial behavior ranging from crime to drug use, violence to risky sexual behavior. These are broadly referred to as antisocial behavior problems, and are arguably a problem for society on the whole; high rates of crime, drug use, gang warfare, or risky sexual activity pose risks for the safety and well-being for all members of society, not only those who propagate such behavior. Hence, considerable research is aimed at understanding the etiology of this behavior in order to better prevent and treat it. However, within the broad category of antisocial behavior problems, the question of aggressive versus nonaggressive antisocial behavior is an important one. Nonaggressive antisocial behavior, or nonaggressive ASB, tends to emerge later developmentally than aggressive ASB, and is considered by some researchers to be less lifetime persistent and more likely to be adolescence-limited, although both sets of behaviors increase during the period of adolescence (Moffitt, 1993). These constructs have also been called overt (aggressive) and covert (nonaggressive) in past work (Loeber & Schmaling, 1985). These patterns of behavior are correlated and are known to co-occur at rates higher than would be expected by chance (Eley, et al. 1999). However, different developmental trajectories have emerged for these sets of behaviors, highlighting their distinctness. For example, lack of control in children of ages 3-5 years was found to predict aggressive ASB, but not nonaggressive ASB in adolescence, suggesting that nonaggressive ASB was more influenced by peers and aggressive ASB more innately influenced (Caspi et al. 1995). Over the course of adolescence, antisocial behavior problems may increase in inherent risk. Whereas younger children are unlikely to engage in significant substance use behavior, to 20 commit serious crimes, or to behave sexually, adolescents prone to antisocial behavior problems may become involved in these forms of activities. Additionally, forms of ASB in youth are strongly predictive of adult criminality and ASB (Loeber & Dishion, 1983), and hence it is important to investigate how influences on aggressive and nonaggressive ASB change over the course of childhood to adolescence. Additionally important to consider are sex differences - while males have consistently been found to show higher prevalence (or mean levels) of antisocial behavior, disagreements exist in the literature about whether the genetic and environmental influences on antisocial behavior are equal between the sexes: some metanalyses conclude equal genetic influence between the sexes (Rhee & Waldman, 2002), and some find higher heritability in males (Miles & Carey, 1997). Further investigating this discrepancy is an aim of this study. Antisocial behavior has been found to be heritable in past research. As measured by the Child Behavior Checklist (CBCL), the instrument used in this study, heritability estimates on antisocial behavior problems range from 38-56% (Gjone & Stevenson, 1997; Pesenti-Gritti et al., 2005). Specifically, aggressive ASB has been estimated to be heritable at a rate of 38-50% (Baker, et al. 2008) and nonaggressive ASB at approximately 48% (Burt, 2009). Two recent metaanalysis reviews approached the questions of the distinctions and shared etiologies of aggressive and nonaggressive ASB (referred to as “rule-breaking”). The first of these concluded that aggressive ASB showed higher heritability than rule-breaking, at 65% and 48%, respectively (Burt, 2009). Rule-breaking behavior was also found to be influenced by shared environmental factors, at 18%, further highlighting etiological distinctions between the two forms of antisocial behavior. 21 At least one previous study has attempted to examine the aggressive and nonaggressive ASB subscales of the CBCL longitudinally in late childhood and mid adolescence, similar to the present research. Correlated genetic factors were found among all four subscales (both subscales at both time points), as well as stronger genetic stability to aggressive than to nonaggressive ASB (Eley, et al. 2003). However, this study did not investigate different potential structures to explain influences on the data, such as latent pathways using sophisticated structural equation modeling, which is a strength of the present research. It is important to discern the structure of influences in order to understand their relationships more precisely and guide future research into molecular genetics or environmental conditions relevant to etiology. In general, there have been more longitudinal examinations of aggressive ASB or widely- defined antisocial behavior than of nonaggressive ASB on its own. A metaanalysis of overall antisocial behavior found that genetic and shared environmental influences decreased across childhood to adolescence, although differences for aggressive and nonaggressive ASB were not reported (Burt, 2009). In young childhood, ages 3-7 years, genetics and shared home environment were found to contribute to stability in overall antisocial behavior (van der Valk, et al. 2003). Stability in maternal ratings of aggressive ASB across the age span of 3-12 years was 65% accounted for by genetic factors and 25% by shared environmental factors. However, sex differences emerged with genetics as more influential on stability in males and shared environment in females (van Beijsterveldt, et al. 2003). The purpose of this study is threefold: 1. To examine the structure and proportion of genetic and environmental influences on aggressive ASB and nonaggressive ASB; 2. To investigate the longitudinal stability and change of genetic and environmental influences on aggressive ASB, nonaggressive ASB and the covariation between them from late childhood to 22 mid-adolescence; 3. To examine sex differences in these relationships in order to determine whether male and female antisocial behavior should be approached differently. Chapter 2: Methods Participants This study uses data collected through the University of Southern California (USC) Risk Factors for Antisocial Behavior (RFAB) twin study, a longitudinal study of over 750 participating families from the greater Los Angeles area. Currently, in its fifth wave of data collection, this study has followed the twins from the age of 9-10 years to their present age of 19- 20 years, and concentrates on biological and environmental risk factors for antisocial behavior. The sample is both ethnically diverse and representative of the Los Angeles population breakdown (44% Hispanic, 25% Caucasian, 16% African American, 3% Asian, and 12% mixed or other). Attrition analysis conducted with this sample found no demographic predictors of study continuation. To account for non-returning families from Wave 1, new families were recruited in the third wave of data collection. The analyses in this study utilize data collected in the first and third waves of collection, at which times the twins were 9-10 and 14-15 years of age, respectively. This study uses data from 1204 individuals (269 MZ male, 288 MZ female, 170 DZ male, 184 DZ female, 147 DZ opposite sex males and 146 DZ opposite sex females) in Wave 1 (mean age 9.61 years, SD=0.59) and 1148 individuals in Wave 3 (249 MZ male, 229 MZ female, 175 DZ male, 212 DZ female, 141 DZ opposite sex males and 142 DZ opposite sex females; mean age 15.0 years, SD=0.86). In this study, 73% of Wave 1 families had also participated in Wave 3. Regression analysis found that scores on neither scale were predictive of discontinuation. For full description of the project, see (Baker et al., in press & 2006). Procedure 23 The testing protocol was 6-8 hours long in Wave 1, and 4-6 hours long in Wave 3. The twins participated in clinical interviewing and neurocognitive testing, and also psychophysiological testing. Their accompanying parent (>92% biological mothers) participated in daylong clinical interviewing and questionnaire answering aimed at assessing home and school environment, behavior, personality, and psychopathology of both twins as well as of the parent. A portion of families in Wave 3 participated via mail (N=135), phone (N=15) or internet surveys (N=63), while the majority participated in laboratory visits. An analysis of variance found no significant differences in CBCL scores for different participation types. Zygosity Determination DNA microsatellite analysis determined zygosity in over 87% of cases. This analysis measures DNA marker concordance. Any discordance in the DNA marks the pairs DZ whereas more than 7 concordant markers marks the pair MZ. In cases of weak signal these analyses were considered inconclusive, and zygosity was determined by the Twin Similarity Questionnaire (Lykken, 1978), which is found to be over 90% concordant with DNA analysis in cases where both measures are available (Baker, et al. 2006). Measures CBCL. The CBCL is a widely used caregiver-response instrument for research and clinical work. It measures a wide range of behavior problems in children, both internalizing (consisting of scales for depression, anxiety, and social withdrawal) and antisocial (aggressive ASB, nonaggressive ASB) behavior problems (Achenbach, 1991). This instrument has 113 items that use a three-point scale (0 for not true, 1 for sometimes true, and 2 for very or often true). Parents are asked to consider their child’s behavior over the last six months. The Delinquency (nonaggressive ASB) subscale of the CBCL (20 items) examines such behavior tendencies as 24 lying, stealing, and destroying possessions. Internal consistencies of the nonaggressive ASB scale were 0.88 and 0.89 at Waves 1 and 3, respectively. The aggressive ASB subscale of the CBCL consists of 13 items and includes behaviors such as arguing, fighting with other children, and bullying others. The internal consistencies in Waves 1 and 3 were found to be 0.61 and 0.71, respectively. This study’s internal consistencies of the CBCL antisocial total subscale, which combines the 33 items from the aggressive and nonaggressive ASB subscales, – 0.88 and 0.91 for Waves 1 and 3, respectively – are consistent with estimates from past research (Pesentti-Gritt, 2005; Arseneault, et al. 2003). Descriptive Statistics and Correlations Descriptive statistics and phenotypic analyses were conducted with untransformed raw data. In twin analysis, intraclass correlations (correlations between the two twins) can serve as a first step to investigate heritability effects on any given observable. When MZ correlations are approximately twice as large as DZ correlations, it provides evidence of additive genetic effects (A); when MZ correlations are less than twice DZ correlations, it provides evidence for shared environmental influence (C); other variance is accounted for by nonshared environment (E) (Neale & Cardon, 1992). To achieve more accurate estimates of the proportions of these influences, more sophisticated genetic modeling is employed. Genetic Model Analyses Genetic modeling was performed with the software package Mx (Neale, et al. 2003). This program determines model fit by comparing observed and expected values in each model and yielding a likelihood ratio statistic (-2LL) on this basis. Fit is assessed with a log-likelihood ratio test statistic, which compares the difference between -2LLs of models, yielding a χ 2 value the distribution for which has degrees of freedom (df) equal to the difference between df of the two 25 models. Model fit was further assessed using two additional fit statistics, the Akaike Information Criterion (AIC) and the Bayesian Information Criterion (BIC). Lower AIC and BIC values are indicative of more parsimonious explanations of the data. Prior to genetic modeling, aggressive and nonaggressive ASB scores were transformed using Blom normalization. This was performed using a pooled mean and standard deviation from both waves simultaneously, in order to retain information about mean changes across waves. Each variable’s heritability was initially estimated within each wave using univariate genetic models that estimate A, C, and E individually without examining shared influences with any other variable in comparison to fully saturated models that allow means and variances to freely vary between zygosity groups. Models included: 1. ACE (estimates all three types of influences on the measure); 2. AE (examines only genetic and non-shared environmental influences); 3. CE (examines only shared and non-shared environmental influences); 4. and E (examines only non- shared environmental influences). Next, a series of multivariate models were employed to determine the structure of influences on the two scales at both waves of data. First, a fully saturated model was estimated (Model 0). This allows all means, variances and covariances to be estimated freely for each zygosity group. An additional saturated examined (Model 0a), which equated means across all males and all females. This model was found to provide a better fit to the data as compared to model 0, and was utilized for all subsequent comparisons. Next, an ACE model utilizing Cholesky-decomposition (Model 1 – Figure 2.1) was examined. The Cholesky decomposition estimates A, C, and E on each variable separately as well as influences of each type shared between variable by partitioning shared variance into genetic and environmental components at different times and on different subscales. This model uses cross-twin cross-trait covariances in 26 modeling how the twins correlate with one another on a given trait as opposed to across traits. Hence, the model estimates how much of each type of influence is unique to each variable as opposed to shared between variables. The next examined model is an independent pathway model (Figure 2.2), which is a more highly constrained model that supposes one source of A, C, and E influences that affect all variables. In addition, each variable may be influenced by specific a, c, and e that do not influence other variables. Next, three common pathway models were examined, in which A,C and E influences exert effect. The first of these modeled one latent factor that loads onto all four subscales (i.e., aggressive ASB waves 1 and 3, nonaggressive ASB waves 1 and 3) simultaneously (Figure 2.3). Next, two models with two latent variables were modeled. In the first, two different latent factors were modeled – one that loads onto nonaggressive ASB behavior in both waves, and one that loads onto aggressive ASB in both waves (Figure 2.4). In the second, one latent factor loaded onto both subscales in Wave 1, and one loaded onto both subscales in Wave 3 (Figure 2.5). 27 Figure 2.1 Cholesky Decomposition ACE Model Example Examining Longitudinal Aggressive and Nonaggressive ASB A1 Agg-1 NA-1 Agg-3 C1 E1 A2 C2 E2 C3 E3 A3 NA-3 C4 E4 A4 28 Figure 2.2 Independent Pathway ACE Model Example Examining Longitudinal Aggressive and Nonaggressive ASB Agg-1 NA-1 Agg-3 NA-3 c e a c e a c e a c e a C E A 29 Figure 2.3 Cholesky Decomposition ACE Model Example Examining Single Longitudinal Common Pathway for Aggressive and Nonaggressive ASB in Both Waves Combined ASB Agg-1 NA-1 Agg-3 NA-3 c e a c e a c e a c e a A C E 30 Figure 2.4 Cholesky Decomposition ACE Model Example Examining Longitudinal Common Pathways for Wave 1 and 3 Antisocial Behavior Non-agg Agg Agg-3 Agg-1 NA-3 NA-1 c e a c e a c e a c e a A1 C1 E1 C2 E2 A2 31 Figure 2.5 Cholesky Decomposition ACE Model Example Examining Longitudinal Common Pathways for Aggressive and Nonaggressive ASB Chapter 2: Results Descriptive Statistics Means, standard deviations, and Ns are presented in Table 2.1, by sex, for both CBCL subscales at each wave. Significant sex differences emerged with males showing higher aggressive ASB in both Waves 1 and 3 (t=4.69 df=1211, p<0.01; t=2.25 df=1151, p=0.02) and males showing higher nonaggressive ASB in Wave 1 (t=4.69 df=1151, p<0.01). Mixed model analyses of variance (ANOVA) using Time as a repeated measure and Sex as a between group effect found that Time significantly affected both CBCL subscales, with nonaggressive ASB significantly increasing and aggressive ASB significantly decreasing between childhood and ASB-3 ASB-1 NA-1 Agg-1 NA-3 Agg-3 c e a c e a c e a c e a A1 C1 E1 C2 E2 A2 32 mid-adolescence. However, interactions between Time and Sex were not significant, suggesting that developmental trajectories did not differ between the sexes. Table 2.1. Means, Standard Deviations, and Ns for Aggressive and Nonaggressive ASB Behavior by Sex Agg-1 Nonagg-1 Agg-3 Nonagg- 3 Males 6.36 (5.34) N=586 1.53 (1.85) N=586 5.33 (5.66) N=565 1.69 (2.38) N=565 Females 5.27 (5.19) N=618 1.08 (1.45) N=618 4.60 (5.26) N=583 1.47 (2.52) N=583 Note. Agg-1 = Aggressive ASB Wave 1; Nonagg-1 = Nonaggressive ASB Wave 1; Agg-3 = Aggressive ASB Wave 3; Nonagg-3 = Nonaggressive ASB Wave 3 Correlations Table 2.2 presents phenotypic correlations among the different measures, separately by sex. Correlations between Aggressive and nonaggressive ASB within wave were r=0.62 for both sexes in Wave 1 and 0.69 for males and 0.66 for females in Wave 3. Longitudinal stability of Nonaggressive ASB was r=0.41 for males and 0.59 for females, and longitudinal stability of Aggressive ASB was r=0.45 for males and r=0.44 for females. Table 2.2. Phenotypic Correlations among Aggressive and Nonaggressive ASB Behavior Scores by Sex Agg-1 Nonagg-1 Agg-3 Nonagg- 3 Agg-1 - .62* .45* .35* Nonagg-1 .62* - .29* .41* 33 Agg-3 .44* .40* - .69* Nonagg-3 .38* .59* .66* - Note. Females are above the diagonal. Agg-1 = Aggressive ASB Wave 1; Nonagg-1 = Nonaggressive ASB Wave 1; Agg-3 = Aggressive ASB Wave 3; Nonagg-3 = Nonaggressive ASB Wave 3; Starred* values indicated p<0.05. As presented in Table 2.3, MZ correlations are marginally higher than DZ correlations for aggressive and nonaggressive ASB scores in Wave 1, and considerably higher in Wave 3. This increasing difference between MZ and DZ twin similarity across age suggests emergence of additional genetic influence on aggressive and nonaggressive ASB in adolescence than in childhood. Univariate genetic analyses conducted that examined each subscale separately within each wave found that the best fitting models equated the influences on the sexes for all but nonaggressive ASB in Wave 1, in which females showed CE influences and males showed ACE influences. These are presented in Appendix B. Table 2.3. Intraclass Correlations for Aggressive and Nonaggressive ASB Behavior, at Ages 9- 10 and 14-15 Years MZ male DZ male MZ female DZ female DZ - OS Agg-1 0.62* 0.38* 0.53* 0.43* 0.41* Nonagg-1 0.63* 0.42* 0.63* 0.51* 0.30* Agg-3 0.74* 0.11 0.72* 0.42* 0.52* Nonagg-3 0.85* 0.39* 0.88* 0.46* 0.56* Note. MZ: monozygotic, DZ: dizygotic, DZ-OS: dizygotic opposite sex, * p< 0.05 Table 2.4 presents cross-twin cross-trait correlations for both Wave 1 (above the diagonal) and Wave 3 (below the diagonal). As can be seen, there is evidence for shared genetics 34 between aggressive and nonaggressive ASB in females and males for both forms of ASB, as Mz correlations are higher than Dz correlation Table 2.4 Cross-twin Cross-trait Correlations for Aggressive and Nonaggressive ASB by zygosity at Ages 9-10 and 14-15. Agg ASB Nonagg-ASB Agg ASB Nonagg-ASB Mz Male Dz Male Agg ASB -- 0.45* -- 0.32* Nonagg ASB 0.61* -- 0.24* -- Mz Female Dz Female Agg ASB -- 0.39* -- 0.33* Nonagg ASB 0.65* -- 0.36* -- Opposite sex Opposite sex Agg ASB -- 0.27* Nonagg ASB 0.43* -- Note. Agg ASB = aggressive antisocial behavior. Nonagg ASB = nonaggressive antisocial behavior. Mz = monozygotic twins. Dz = dizygotic twins. Wave 1 correlations presented above the diagonal, Wave 3 below the diagonal. Table 2.5 presents model fit statistics for the multivariate genetic models of aggressive and nonaggressive ASB across both waves of data collection. When compared to the full ACE model using Cholesky decomposition (Model 1), equating influences between the two sexes was found to produce a more parsimonious model (Model 1a) as assessed by AIC and BIC, which is consistent with univariate model results (Appendix B). Hence, all subsequent models (2-5) were tested with estimates equated between the sexes. Models 1a, 2, 3, 4, and 5 were compared using AIC and BIC values, as well as their relevance to research questions. While the lowest AIC is found in model 1a (ACE with Cholesky decomposition), the lowest BIC was found in model 5 (two factor common pathway, longitudinal), which better examines longitudinal change on the construct of antisocial behavior. In model 5, variances for the Wave 1 and Wave 3 latent factors 35 were freed while loadings were constrained, allowing for the consideration of change in behavior variance across time. When these variances were constrained (model 5a), the model comparison χ 2 difference was significant (χ 2 =21.45, df=2; P<0.01), suggesting that freed variance was a better explanation of the data. When loadings were equated across waves (model 5b), this more parsimonious model provided a better fit to the data (χ 2 =0.94, df=1; P=0.33), and the equated loadings constraint was used in models 5c and 5c*, as well. In model 5c, longitudinal influences on the latent factors were equated; that is, the sources of A, C, and E on the Wave 1 latent factor were constrained to be equal to those that stemmed from the same influences but affected the Wave 3 latent factor. In this manner, the influences that emerged exclusively for Wave 3 can be considered entirely new. While this model was not a better fit to the data, it provided stronger answers to questions about longitudinal change and stability of influences on antisocial behavior. Model 5c* uses model 5c as a basis but drops estimates that emerged as not significant (i.e. included zero in the 95% confidence interval), and was selected as the best explanation of the data. In this model, only A effects were found to be significant for new influences in adolescence. Additive genetics (A) accounted for 41% (0.645 2 ) of influences on the ASB latent factor in Wave 1, while C accounted for 40% (0.635 2 ) and E accounted for 19% (0.44 2 ) of influences. The Wave 3 latent factor was 42% (0.65 2 ) influenced by entirely new factors, all of which were additive genetic factors. The other 57.7% of influences on the Wave 3 latent factor were stable – that is, shared with those on the Wave 1 latent factor. Table 2.5 Multivariate Model Fit Indices for Aggressive and Nonaggressive ASB Behavior in Waves 1 (9-10 years) and 3 (14-15 years) Combined Overall fit Model Comparison Compared to model Model difference test Model -2LL Df AIC BIC χ 2 Δdf P χ 2 df P 0 Full Saturated 10116.16 4516 1084.16 -9958.25 0a Equate within sexes 10116.16 4516 1084.16 -9958.25 31.87 32 0.47 1 ACE Cholesky M ≠ F 10297.37 4636 1018.37 -10270.16 167.21 120 <0.01 1a ACE Cholesky M = F 10323.00 4654 1015.00 -10313.70 206.84 138 <0.01 1 25.63 18 0.11 2 1-Factor IP model M = F 10405.75 4672 1061.75 -10332.18 289.59 156 <0.01 1a 82.75 18 <0.01 3 1 CP M = F 10537.50 4678 1181.50 -10286.25 421.34 162 <0.01 1a 206.06 24 <0.01 4 2 CP Agg & Del M = F 10560.02 4675 1210.02 -10265.00 443.86 159 <0.01 1a 237.02 21 <0.01 5 2 CP Ext1 & Ext3 M=F Variances Freed 10397.55 4673 1051.55 -10339.60 281.39 157 <0.01 1a 96.00 21 <0.01 5a 2 CP Ext1 & Ext3 M = F variances constrained 10419.00 4675 1069.00 -10335.53 302.84 159 <0.01 5 21.45 2 <0.01 5b 2 CP Ext1 & Ext3 M=F Variances Freed, loadings equated 10398.49 4674 1050.49 -10342.46 282.33 158 <0.01 5 0.94 1 0.33 5c 2 CP Ext1 & Ext3 M=F Variances Freed, loadings equated, Equate longitudinal influences on latent factors 10422.15 4677 1068.15 -10340.60 305.99 161 <0.01 5b 23.66 3 <0.01 5c * Model 5c with nonsignificant estimates dropped 10437.38 4685 1067.38 -10359.59 321.22 169 <0.01 5c 15.15 8 0.06 37 In addition to the influences on the subscales from the latent factor, which comprised 61% (0.78 2 ) of influences on Wave 1 Aggressive ASB, 44% (0.66 2 ) on Wave 1 Nonaggressive ASV, 79% (0.89 2 ) on Wave 3 Aggressive ASB and 55% (0.74 2 ) of influences on Wave 3 Nonaggressive ASB, each subscale had additional E or A and E influences specific to it only. Notably, the only shared environmental influence on any subscale was common to both Aggressive and Nonaggressive ASB at both waves, as no specific C emerged. Nonaggressive ASB was found to have genetic influences not shared with Aggressive ASB in both waves. This full model is represented in Figure 2.6, in which rectangular variables were directly observed and circled variables represented latent variables. 38 Figure 2.6. Best-fitting Two-Factor Common Pathway Model of A, C, and E Influences on Aggressive and Nonaggressive ASB Behavior at Ages 9-10 and 14-15 Years e a e .33* (.24-.38) e A E 0.65* 0.48* 0.78* 0.66* 0.89* 0.74* ASB-3 a e .46* (.33-.55) NA-1 Agg-1 NA-3 Agg-3 ASB-1 C 0.64* 0.49* 0.33* .46* (.41-.51) .33* (.29-.36) .49* (.43-.55) .32* (.28-.37) A 0.44* 0.65* 39 Chapter 2: Discussion The present study sought to examine the structure and longitudinal stability of influences on two prominent and concerning facets of antisocial behavior problems across childhood and adolescence – aggressive and nonaggressive ASB – as well as to investigate sex differences on these influences. Our primary finding is that aggressive and nonaggressive ASB are both influenced by a common factor of general antisocial behavior, through which they share genetic and environmental influences. Additionally, it emerged that in the latter time point, which took place during the age of mid-adolescence, new genetic influences emerged. Lastly, it emerged that males and females do not significantly differ in these influences, which is consistent with past findings (Burt, 2009; Rhee & Waldman, 2002). The latent antisocial behavior factor that emerged supports past research that found that aggressive and nonaggressive ASB share a proportion of their etiologies. The estimates of heritability on the latent factor are in line with past metaanalysis findings of shared heritability between aggressive and nonaggressive ASB (Burt, 2012). Past research found genetic correlation between aggressive and nonaggressive ASB concurrently and longitudinally, and our finding suggests that both sets of behaviors are influenced by some overlapping genetic and environmental influences. In addition, the phenotypic correlations as well as the subscale-and- time-specific influences on either set of behaviors demonstrate that some aspects of these behaviors develop from different influences and are not redundant. It is important that when constraints were applied to the model in order to discern the influences that were entirely novel, the only significant form of influence that emerged was genetic rather than environmental. This is notable, as it is consistent with past findings that change in overall antisocial behavior is attributable to genetic factors activated during puberty 40 (Jacobson et al 2002). These may represent genetic influences activated at the onset of hormonal puberty on the development of ASB. The structure of our findings highlights the significance of adolescent development, biologically and socially, on the etiology of antisocial behavior problems. In our analyses, both aggressive and nonaggressive ASB were found to be highly genetically influenced. This is somewhat counter to past research of nonaggressive ASB that found that nonaggressive ASB shows lower levels of heritability and higher levels of shared environment than aggressive ASB; notably, this metaanalysis combined studies with different instruments measuring ASB, and so the result may be pertinent to broader definitions of ASB (Burt, 2009). In our sample, MZ correlations for nonaggressive ASB were very high for both sexes, which accounts for the estimates of heritability. In their metaanalysis, Rhee and Waldman suggested that using parental reports may exaggerate estimates of familial influences (genetics and shared environment) because one reporter is reporting on both twins, and may have specific response bias patterns. While this may be a factor in our estimates, past studies also used parental ratings of nonaggressive ASB. One difference between correlations found here and correlations in past studies is that our DZ estimates were lower than those found in previous studies (e.g. 0.72; Cloninger & Gottesman, 1987). This may suggest differences in samples between our study and others. Phenotypically, our findings are consistent with past research, which found that nonaggressive ASB significantly increases during adolescence while aggressive ASB significantly decreases. Although males showed high levels of aggressive ASB in both waves and nonaggressive ASB in Wave 1, there were not significant interactions between time and sex, suggesting trajectories of phenotypic change in behavior across childhood to adolescence were 41 not different between the sexes. Some theories posit that increases in nonaggressive ASB are typically attributed to peer influence during adolescence, and decreases in aggressive ASB to the development of the frontal lobes. Past research has found that rates of impulsivity increase over the course of adolescence (Niv, et al. 2011), which may also account for increases in nonaggressive ASB as manifestation of behavioral disinhibition. While males showed higher phenotypic mean levels of aggressive ASB in both waves and nonaggressive ASB in Wave 1, the best fitting explanation of genetic and environmental influences on the pattern of observed results suggested equality between the sexes. This finding is consistent with some past literature reviews (Widom & Ames, 1988) but inconsistent with other reviews that found higher heritability of aggressive ASB in males (Miles & Carey, 1997). It is important to note the polygenic multiple threshold model, as elucidated by Rhee and Waldman (2002), which posits that the less affected sex needs higher liability in order to develop a phenotype, but not that the magnitude of influences is necessarily higher. That is, while females may need more liability to express antisocial behavior problems, this does not mean that different or more influences are necessarily involved than in males. These findings shed light on our understanding of etiology and developmental course of aggressive and nonaggressive ASB. While these forms of antisocial behavior problems are considered distinct, correlations exist between them both phenotypically and genetically. However, prior to this study, the structure of these influences had not been elucidated. The finding of a latent pathway supports the hypothesis that there exists a common thread between these different forms of antisocial behaviors, and that they are jointly influenced by a set of genetic and environmental circumstances and develop in unison. Differentiation between these behaviors can be attributed to symptom specific influences. Second, longitudinal change in 42 effects on ASB is entirely attributable to genetic influences, suggesting that potential targets for molecular genetic research may be genes related to pubertal development. There are a few limitations in this study. First, because parental reports were used, it is possible that underreporting is involved, either due to social desirability or to ignorance of children’s true behavior, especially during the adolescent age. Lastly, the twin design relies upon several assumptions, such as lack of assortative mating in the parent generation, which may slightly bias estimates (Plomin, 2001). Assortative mating would act to inflate DZ correlations because their proportion of shared genes would be higher than the estimated 50% if their parents share some genes. This increases estimates of shared environment (Krueger, et al. 1998). However, DZ correlations were found to be lower than past studies in our sample, suggesting assortative mating was not a concern. Future directions may include examinations of specific genes as well as environmental conditions leading to the development of aggressive or nonaggressive ASB, for the purposes of intervention. Specific genes involved in the etiology of both aggressive and nonaggressive ASB may be investigated in both childhood and adolescence, to better understand which genes affect these behaviors at which age, and especially more so in adolescence, when individuals gain more freedom and ability to become involved in dangerous activity. Our findings also suggest that the environmental conditions that influence ASB of both forms are present prior to adolescence, and may be identified early. By identifying the environmental conditions that most contribute to the development of antisocial behavior problems, appropriate clinical interventions can be directed specifically at these conditions, such as parenting, peer-interactions, exposure to violence, and others. 43 Chapter 2: Conclusions This study examined parent-reported aggressive and nonaggressive ASB at late childhood (age 9-10 years) and mid-adolescence (age 14-15 years) using the highly validated instrument of the Child Behavioral Checklist. These subscales were found to share genetic, shared environmental and nonshared environmental influences through latent common factors at each wave. These common factors were highly correlated longitudinally. In addition, however, exclusively genetic influences on this common factor at the second assessment provide explanation for observed changes across development. Influences on each subscale were 44-79% conveyed through this common factor, and the rest of influences through genetic and nonshared environmental influences on each subscale at each time point specifically, further explaining differentiation and change across age. 44 References Achenbach TM (1991a) Manual for the youth self-report and 1991 profile. University of Vermont, Department of Psychiatry, Burlington. Achenbach, T. M. (1991b). Manual for the Child Behavior Checklist/4-18 and 1991 Profile. Burlington, VT: University of Vermont Department of Psychiatry. Arsenault, L., Moffitt, T.E., Caspi, A., Taylor, A., Rijsdijk, F.V., Jaffee, S.R., Ablow, J.C., & Measelle, J.R. (2003). Strong genetic effects on cross-situational antisocial behaviour among 5-year-old children according to mothers, teachers, examiner-observers, and twin self-reports. Journal of Child Psychology and Psychiatry, 44, 832-848. Baker, L. A., Barton, M., Lozano, D. I., Raine, A. & Fowler, J. H. (2006). The Southern California Twin Register at the University of Southern California: II. Twin Research and Human Genetics 9, 933-40. Baker, L. A., Barton, M. & Raine, A. (2002). The Southern California Twin Register at the University of Southern California. Twin Research and Human Genetics 5(5), 456-9. Baker, L.A., Tuvblad, C., Wang, P., Gomez, K., Bezdjian, S., Niv, S., & Raine, A. (in press) The Southern California Twin Register at the University of Southern California: III. Twin Research. van Beijsterveldt, C.E.M., Bartler, M., Hudziak, J.J, & Boomsma, D.I., (2003). Causes of stability of aggression from early childhood to adolescence: A longitudinal genetic analysis in Dutch twins. Behavior Genetics, 33, 591-605. Caspi, A., Henry, B., McGee, R. O., Moffitt, T. E., & Silva, P. A. (1995). Temperamental origins of child and adolescent behavior problems: from age three to age fifteen. Child Development, 66, 55-68. 45 Cloninger, C.R., & Gottesman, I.I (1987). Genetic and environmental factors in antisocial behavior disorders. In S.A. Mednick, T.E. Moffitt, & S.A. Stack (Eds.), The causes of crime: New biological approaches (pp. 92-109). New York: Cambridge University Press. Edelbrock, C., Rende, R.D., Plomin, R., & Thompson, L.A. (1995). A twin study of competence and problem behavior in childhood and early adolescence. Journal of Child Psychology and Psychiatry, 36, 775-785. Eley, T.C., Lichtenstein, P., & Moffitt, T.E. (2003). A longitudinal behavioral genetic analysis of the etiology of aggressive and nonaggressive antisocial behavior. Development and Psychopathology, 15, 383-402. Eley, T.C., Lichtenstein, P., & Stevenson, J. (1999). The Etiology of aggressive and nonaggressive antisocial behavior: Results from two twin studies. Child Development, 70, 155-168. Gjone, H., & Stevenson, J., (1997). The association between internalizing and externalizing behavior in childhood and early adolescence: genetic or environmental influences? Journal of Abnormal Child Psychology 25, 277-286. Jacobson, K.C., Prescott, C.A., & Kendlar, K.S. (2002). Sex differences in the genetic and environmental influences on the development of antisocial behavior. Developmental Psychopathology, 14, 395-416. Krueger RF, Moffitt TE, Caspi A, Bleske A, Silva PA (1998) Assortative mating for antisocial behavior: developmental and methodological implications. Behav Genet 28, 173–186. Kuha, J. (2004). AIC and BIC comparisons of assumptions and performance. Sociological Methods & Research, 33, 188-229. Loeber, R. & Dishion, T.J. (1983). Early predictors of male delinquency: A review. 46 Psychological Bulletin, 94, 68-99. Loeber, R., & Schmaling, K.B. (1985). Empirical evidence for overt and covert patterns of antisocial conduct problems: a meta-analysis. Journal of Abnormal Child Psychology, 13, 337-352. Miles, D.R., & Carey, G. (1997). Genetic and environmental architecture of human aggression. Journal of Personality and Social Psychology, 72, 207-217. Moffitt, T.E. (1993). Adolescence-limited and life-course persistent antisocial behavior: A developmental taxonomy. Psychological Review, 100, 674-701. Neale, M. C., Boker, S. M., Xie, G. & Maes, H. (2003). Mx: Statistical modeling. Richmond, VA: Department of Psychiatry, Medical College of Virginia. Neale, M. C. & Cardon, L. R. (1992). Methodology for genetic studies of twins and families. Dordrecht, The Netherlands: Kluwer Academic Publications. Niv, S., Tuvblad, C., Raine, A., Baker, L. A. (2011) Heritability and longitudinal stability of impulsivity in adolescence. Behav Genet., 42, 378-392. Pesenti-Gritti, P., Spatola, C.A.M., Fagnani, C., Patriarca, V., Stazi, M.A., & Battaglia, M., (2005). The co-occurrence between internalizing and externalizing behaviors. A general population twin study. Journal of European Child and Adolescent Psychiatry 17, 82-92. Rhee, S. H. & Waldman, I. D. (2002). Genetic and Environmental Influences on Antisocial Behavior: A Meta-Analysis of Twin and Adoption Studies. Psychological Bulletin 128, 490-529. Rowe, D.C. (1983). Biometrical genetic models of self-reported delinquent behavior: A twin study. Behavior Genetics, 13, 473-489. 47 Tuvblad, C. Eley, T.C., & Lichtenstein, P. (2005). The development of antisocial behaviour from childhood to adolescence: A longitudinal twin study. European Child & Adolescent Psychiatry, 14, 216-225. Van der Valk, J. C., van den Oord, E., Verhulst, F.C., & Boomsma, D.I. (2003). Genetic and environmental contributions to stability and change in children’s internalizing and antisocial problems. J. Am. Acad. Child Adolesc. Psychiatry, 42, 1212-1220. Widow, C.S., & Ames, A. (1988). Biology and female crime. In T.E. Moffitt & S.A. Mednick (Eds.), Biological contributions to crime causation (pp. 308-331). Dordrect, the Netherlands: Martinus Nijhoff 48 Chapter 3. Relationships of Alpha, Beta, and Theta EEG Spectral Properties with Aggressive and Nonaggressive ASB in Children and Adolescents 2 Chapter 3: Abstract This research was aimed at investigating potential differences underlying EEG spectral power in predicting aggressive and nonaggressive antisocial behavior. Frontal and parietal EEG spectral properties were calculated using principal components analysis for three different ranges (theta: 4-8 Hz, alpha: 8-10.5 Hz, and beta: 10.5-30 Hz) in 900 twins of age 9-10 years in an open-eyed resting state. Using multilevel regression modeling, relationships emerged between EEG power properties and measures of antisocial behavior at age 14-15 years, but not at the concurrent age of 9-10 years, suggesting EEG power may represent vulnerability toward developing adolescent forms of antisocial behavior. Aggressive ASB was predicted by higher power values for alpha (frontal) and beta (frontal, parietal) for males. This emerged in the aggression data from Wave 3, and not from Wave 1. Nonaggressive ASB was not significantly predicted by cortical spectral power, and no significant relationships emerged in the females. These results support a distinction between aggressive and nonaggressive ASB, and raise questions about differing neurobiological profiles for antisocial behavior in males and females. 2 Authors: Sharon Niv, Syed Ashrafulla, Catherine Tuvblad, Anand Joshi, Adrian Raine, Richard Leahy, Laura A. Baker 49 Chapter 3: Introduction Antisocial behavior comprises a vast array of socially-negative behavior such as crime, violence, vandalism, substance use, and others. These behaviors in childhood and adolescence are correlated with academic failure, unemployment, incarceration, and other negative life outcomes (Hinshaw, 1992; Farrington, 1989). Research has drawn a distinction between violent/aggressive antisocial behavior, and nonaggressive antisocial behavior, which comprises rule-breaking, lying, cheating, and non-violent crime (Burt, et al. 2009 & 2012; Eley, et al. 1999). These behaviors are correlated but distinct, and show different developmental patterns. Aggressive antisocial behavior shows more longitudinal stability in children and adolescence, and is thought to be closer to a personality trait than does nonaggressive ASB (Moffitt, 2003; Tremblay, 2003). Nonaggressive ASB has been suggested to emerge in some cases as adolescent-limited, or at least adolescent-onset, suggesting that it is influenced more by conditions of pubertal development or adolescent-social development (Moffitt, 2003; Burt, 2012b). While nonaggressive ASB has been found to increase in adolescence (Moffitt, 2003), aggressive ASB is consistently found to decrease over the developmental course from childhood to adolescence (Tremblay, 2003). Both aggressive and nonaggressive ASB are found more prevalently in males than in females (Stanger, et al. 1997), but the disparity is greater for aggressive ASB. Many behavioral genetics examinations have produced findings that further support both the distinction and the relationship between aggressive and nonaggressive antisocial behavior. There is higher heritability and less peer influence in aggressive ASB than in nonaggressive ASB (Burt, 2009; Tuvblad, et al. 2005). Two recent meta-analyses of questionnaire and interview data were conducted to examine these two constructs. The first found that aggressive ASB was more 50 heritable (at 65% influences of genetics) than was nonaggressive ASB (at 48%), and that nonaggressive ASB showed 18% influence of shared environment between twins and family members while aggressive ASB did not (Burt, 2009). The second meta-analysis found that overlap of genetic influences between these two sets of behaviors is 38.4%, and the overlap of environmental influences only 10.2% (Burt, 2012a). These findings further support the distinction between aggressive and nonaggressive ASB. However, phenotypically, these behaviors are correlated typically between r=0.4 and 0.6, demonstrating that they are related forms of behavior (Burt, 2012b; Eley, et al. 1999). In addition, investigating the relationships between neural activity and behavior is a paramount goal to further our understanding of key biological markers of antisocial behavior. The harmless and inexpensive tool of electroencephalography (EEG), one of the oldest and most established neuroscience methods available, is used to investigate sleep, epilepsy, and evoked potentials. Different frequencies of brain activity (collective neuronal firing frequency) are typically characterized as such: delta (1-4 Hz, characteristic of slow wave sleep), theta (4-8 Hz, characteristic of deep relaxation and pre-sleep activity), alpha (8-12 Hz, characteristic of sleepiness, a “low arousal” brain), beta (15-30 Hz, characteristic of waking, alert activity), gamma (30-100+ Hz, characteristic of high alert activity). In children, alpha ranges have been measured as lower than in adults, at about 6-9 Hz in children of age 4 years (Marshall, et al. 2002) and showing profiles closer to adults’ around age 14 years (Gasser, et al. 1988). In accordance with these findings, many studies examining children of ages 8-11 years select the lower alpha range of approximately 8-10.5 Hz (Baving et al., 2000 & 2003; Forbes, et al. 2006), as does the present research. Specific EEG profiles have been associated with wide array of psychological symptoms. Criminality and violence have been associated with under-arousal in 51 the frontal brain regions (Raine, et al. 1990; Mednick, et al. 1981, Ellingson, 1954), potentially suggesting that individuals with low arousal require higher levels of risky or dangerous behaviors to attain normal levels of arousal (Hare, 1970). At least one study found that this was only true for theta and delta power ranges, and that frontal alpha power was in fact reduced in homicidal men (Lindberg, et al. 2005). Power values of beta and theta ranges have been found in relation to ADHD and inattentive, disinhibited behavior, which are also parts of the antisocial behavior problems spectrum (Barry, et al. 2003; Lazzaro, et al. 1998; Clarke, et al. 2001). Beta power in frontal and parietal regions have been associated with alcoholism and found in children of alcoholics, as well (Ehlers & Schukit, 1990, 1991). These studies were all performed using clinical samples, and our aim is to investigate relationships between these EEG measures and antisocial behavior in children from a community sample, to verify whether the constructs may be useful in identifying children at risk for developing antisocial behavior within the general population, as well. Hence, we used the instrument of the Child Behavioral Checklist (Achenbach, 1991), a gold standard within the field of psychology, aimed at discerning a wide array of behavioral problems in nonclinical samples. As antisocial behavior problems comprise a considerable concern for society, it is important to understand the biological underpinnings of its development, as well as identify potential neurobiological markers that may hint at vulnerability early in life. This study aims at contributing meaningful answers to these questions by examining relationships between EEG spectral power at frontal and parietal sites and antisocial behavior, both aggressive and nonaggressive ASB. It is hypothesized that higher power in alpha and theta ranges would be predictive of ASB, as it represents greater proportion of low arousal activation. 52 Chapter 3: Methods Participants. All EEG data utilized in this study were collected during the first wave of data collection of University of Southern California (USC) Risk Factors for Antisocial Behavior (RFAB) twin study, currently in its fifth wave of data collection. At the time of EEG data collection in the first wave, the twins were 9-10 years of age, a total of 605 families. This sample is socioeconomically and ethnically diverse and representative of Los Angeles’s population, with the following ethnicity percentages: 37.2% Hispanic, 25.6% Caucasian, 14.1% African American, 5% Asian, and 17.5% mixed or other. For a full project description and zygosity determination, see Baker, et al., 2002 & in press. This study uses data from 900 twins (47.9% male, 52.1% female) out of the total study sample of 1204, as some were removed for not being right handed (n=169) or poor EEG signal (n=135). The number of complete twin pairs, in which both twins provided good EEG data, was 376 (88 MZ male, 105 MZ female, 46 DZ male, 59 DZ female, 82 DZ opposite sex). EEG Collection and Processing. Twins underwent psychophysiological testing while their co-twins and caregivers were engaged in interviewing in nearby rooms. Before 1.5-2 hours of tasks, the children were instructed to sit quietly with their eyes open, looking at a cross on a computer screen. This 3-minute rest period was selected for these analyses, as prior cognitive activity carries effects on resting state networks, and may not represent baseline activity in its most naïve, natural state (Waites, et al. 2005). Recordings were collected using the James Long Inc. amplification system (Caroga Lake, NY) from 30 scalp sites using the 10-20 international system (Jasper, 1958). Children were fit with a lycra Electro-cap (Eaton, OH) after the use of a conductive abrasive on each site to increase conductivity. The ground electrode was the anterior midline site (AFz) and the reference 53 site was central midline site (Cz), and impedances were kept below 10 KOhms. Eye blink artifacts were removed through the use of a bipolar electrooculogram (EGO) channel above and below the supra and infra-orbital eye ridges of the left eye. Testers were instructed to keep detailed observations during testing, which were used to evaluate the validity of each participant’s data. The majority of children excluded were excluded for excessive fidgeting and head movements. All EEG data were pre-processed and processed in Matlab (Mathworks ®, R2011b). Data presented here uses absolute power, although relative power was examined and found to be consistent with open-eye resting data from past research (Fein, et al. 1984). Graphical representations of heritability throughout the brain were produced with Brainstorm 3.1. Derivation of PCA Power Values. To create frontal measures, recordings from nine frontal and prefrontal electrodes (Fp1, Af3, F7, F3, Fz, F4, F8, Af4, Fp2) were filtered into the low alpha band (8-10.5Hz), which is the alpha band range appropriate for use in children ages 9- 10, beta bands (10.5-25 Hz), and theta band (4-8 Hz). The weighted sum of this data across electrodes as used to calculate frontal alpha power (FAP), frontal beta power (FBP), and frontal theta power (FTP). Weights were derived through a principal components analysis (PCA), which attempted to maximize power in the weighted sum. Weights were normalized, so that weights squared summed to 1. This measure can thus be considered the strongest average power over all possible normalized weightings of the frontal electrode data. To derive parietal measures, similar procedures were undertaken for electrodes P7, P3, Pz, P4, and P8. Antisocial Behavior. The Child behavior Checklist (CBCL) is a widely used caregiver-response instrument for research and clinical work. It measures antisocial behavior problems using the Aggression scale for aggressive ASB and the Delinquency scale for nonaggressive ASB (Achenbach, 1991) over the prior 6 months. This instrument has 113 items that use a three-point 54 scale (0 for not true, 1 for sometimes true, and 2 for very or often true). The Delinquency subscale of the CBCL (20 items) examines behavioral tendencies such as lying, stealing, and destroying possessions. Internal consistencies of the Delinquency ASB scale were 0.88 and 0.89 at Waves 1 and 3, respectively. The Aggressive ASB subscale of the CBCL consists of 13 items and includes behaviors such as arguing, fighting with other children, and bullying others. The internal consistencies in Waves 1 and 3 were found to be 0.61 and 0.71, respectively. Our study’s internal consistencies are consistent with estimates from past research (Pesentti-Gritt, 2005; Arseneault, et al. 2003). Statistical Analyses. Multilevel modeling was used to perform regression analysis using PROC MIXED in the SAS (Statistical Analysis Software, SAS 9.1.3, 2005). These mixed models accounts for the dependency between twin 1 and twin 2 by having individual twins nested within family ID, with family ID as a grouping variable. In this manner, twin correlations are taken into account and effect sizes are not inflated due to non-independent observations (i.e., co-twins). Each regression model was performed for 4 outcomes variables: Waves 1 and 3 CBCL nonaggressive ASB and aggressive ASB scores. Chapter 3: Results Mean values for EEG, aggressive ASB, and nonaggressive ASB measures are presented in Table 3.1. In addition to boxplot outliers, EEG power values below 1.0 E-11 Voltage/second (V/s) were manually removed, as they were considered too low to rule out electrode failure. Significant sex differences emerged for all EEG measures except for frontal theta and parietal alpha power. Females showed higher values for frontal beta power (t (890) =-2.29, p=0.02), frontal alpha power (t (885) = -2.02, p=0.04), parietal beta power (t (890) =-2.89, p<0.01), and parietal theta power (t (889) =-2.45, p=0.02). Males showed higher aggressive ASB in both waves (Wave 1: 55 t (887) =3.09, p<0.01; Wave 3: t (596) =2.36, p=0.02) and higher nonaggressive ASB in Wave 1 (t (887) =3.85, t<0.01) but not Wave 3 (t (596) =1.21, p=0.23). This is consistent with past findings that the sex differences in nonaggressive ASB lessen in adolescence. Table 3.1. Means, Standard Deviations, and Ns for Variables Used in Analyses by Sex Males Females Frontal Alpha 1.02 E-10 V/s (0.86), N = 429 1.15 E-10 V/s (1.01) N = 468 Frontal Beta 1.34 E-10 V/s (0.86), N = 431 1.46 E-10 V/s (1.04) N = 467 Frontal Theta 1.53 E-10 V/s (0.60), N = 431 1.58 E-10 V/s (0.71) N = 467 Parietal Alpha Parietal Beta 0.75 E-10 V/s (0.44) N = 430 0.86 E-10 V/s (0.64) N – 468 Parietal Theta 0.58 E-10 V/s (0.26), N = 430 0.66 E-10 V/s (0.54) N = 467 Aggressive ASB Wave 1 6.12 (5.15) N = 428 5.10 (4.78) N = 467 Nonaggressive ASB Wave 1 1.49 (1.81) N = 428 1.07 (1.36) N = 467 Aggressive ASB Wave 3 5.36 (5.99) N = 281 4.31 (4.80) N = 322 Nonaggressive ASB Wave 3 1.79 (2.34) N = 281 1.55 (2.52) N = 322 Aggressive and nonaggressive ASB were significant correlated in both waves and for both genders (Wave 1 males r=0.60; Wave 1 females r=0.53; Wave 3 males r=0.52; Wave 3 females r=0.64; all p.<0.01). Longitudinally, aggressive ASB was correlated r=0.39 for males and 0.53 for females, and nonaggressive ASB was correlated r=0.39 for males and r=0.40 for females (all p.<0.01). Repeated measures analyses of variance showed that aggressive ASB decreased between Waves 1 and 3 (F (1,179) =8.16, p<0.01), and nonaggressive ASB increased (F (1,179) =22.12, p<0.01. However, no significant interactions emerged between wave and sex, suggesting the trajectories of behavioral change were not different between the sexes. Significant main effects emerged only in explaining Wave 3 aggressive ASB results from EEG PCA patterns, and not in Wave 1. Table 3.2 presents the unstandardized regression 56 coefficients from all main effects mixed models, with significant relationships bolded. An alpha value of p=0.05 was used as a threshold for hypothesis driven analyses, namely frontal and parietal alpha and theta power. An alpha power of p=0.025 was adopted for non-hypothesis driven results. Significant interactions arose between EEG and sex in these models, and so males and females were examined separately. As can be seen, Wave 1 EEG power was a predictor of male aggressive ASB, with frontal alpha, frontal beta, and parietal beta power predicting Wave 3 male aggressive ASB. No significant relationships in the females or with nonaggressive ASB emerged when lower alpha values were observed, but trends in the data (p<0.10) suggested that lower frontal alpha, parietal alpha, and parietal beta may be related to ASB in females. Table 3.2. Significant Main Effects found in Multilevel Modeling of EEG Power and Antisocial Behavior Wave Frontal alpha Frontal theta Frontal beta Parietal alpha Parietal theta Parietal beta 1 Females Aggressive ASB 0.17(0.22) p=0.44 0.44(0.31) p=0.15 0.24(0.22) p=0.27 0.18 (0.35) p=0.62 0.09(0.39) p=0.82 2.71(3.45) p=.43 Nonaggressive ASB 0.02(0.06) p=0.69 0.57(0.89) p=0.50 0.65(0.62) p=0.29 -0.16 (0.09) p=0.09 -0.19(0.11) p=0.08 -0.73(0.95) p=.44 Males Aggressive ASB -0.22(0.30) p=0.44 0.24(0.40) p=0.55 -0.08(0.29) p=0.78 -0.27 (0.21) p=0.20 0.33(0.94) p=0.72 -1.40(5.73) p=.81 Nonaggressive ASB -0.13(0.11) p=0.22 -0.17(0.15) p=0.25 -0.14(0.10) p=0.18 -0.44 (0.56) p=0.43 -0.39(0.33) p=0.25 -2.10(2.60) p=.31 3 Females Aggressive ASB -0.41(0.26) p=0.11 .01(0.41) p=0.97 -0.36(0.26) p=0.17 -1.41 (0.71) p=0.051 -0.67(0.56) p=0.24 -0.71 (0.39) p=.07 Nonaggressive ASB -0.18(0.12) p=0.09 0.23(0.18) p=0.21 -0.10(0.12) p=0.38 -0.55 (0.27) p=0.04 -0.28(0.28) p=0.32 -0.31(0.17) p=.054 Males Aggressive ASB 0.99 (0.44) p=0.03* 0.312 (0.55) p =0.82 1.27(0.44) p=0.004* -0.12 (0.68) p=0.87 -1.47(1.81) p=0.42 1.77(0.77) p= .02* Nonaggressive ASB 0.05(0.17) p=0.76 -0.10(0.20) p=0.63 0.19(0.16) p=0.25 0.08 (0.23) p=0.74 -1.04(0.64) p=0.10 0.13(0.28) p=.64 Chapter 3: Discussion The present research examined relationships between EEG spectral power and antisocial behavior of aggressive and nonaggressive nature in children and adolescent twins at two time points over the ages of 9-15 years. The primary findings of this study are that spectral power at age 9-10 years is not predictive of concurrent antisocial behavior in children, but that several significant relationships emerge in predicting male aggressive antisocial behavior from childhood EEG spectral power. One of the more immediately evident findings is that aggressive and nonaggressive forms of antisocial behavior were not predicted in unison. Mixed regression modeling analysis found that male aggressive behavior was predicted by higher frontal alpha and beta spectral power, as well as higher parietal theta power. Conversely, nonaggressive ASB was not significantly predicted in our results, although there was a non-significant trend in the data may suggesting that a negative relationship may exist between low arousal and nonaggressive ASB. Alpha and theta power both represent low arousal states, whereas beta power represents waking state activation. The emerging result that higher frontal alpha in males is related to higher aggressive ASB was consistent with our hypotheses and past findings, while patterns for nonaggressive ASB were not. This raises important questions about neural mechanisms underlying different forms of antisocial behavior. For example, this finding may suggest that the low arousal that leads to sensation and thrill seeking leads to impulsive or reactive behavior that is more prevalent in aggressive antisocial behavior. Conversely, rule-breaking behavior may involve less disinhibition. This hypothesis is consistent with past findings that aggressive ASB correlated more strongly with stress reaction than does nonaggressive ASB, perhaps suggesting that children with lower arousal are more likely to react aggressively to environmental stressors (Burt 59 & Donnellan, 2008). This may also be interpreted as consistent with findings that aggressive ASB shows more longitudinal stability than does nonaggressive ASB, with biological underpinnings showing more consistent predictive value. It should be noted that beta power has rarely been examined in relation to antisocial behavior, and so these results are difficult to contextualize in terms of the broader literature. Past examinations have focused largely on under arousal, and slow-wave activation in relation to antisocial behavior, stemming from Hare’s (1970) theories of low arousal leading to sensation and thrill seeking behavior. Higher beta power has been found in alcoholism and in relatives of alcoholics, suggesting perhaps a vulnerability toward addiction due to an imbalance in the reward system. However, there isn’t yet strong evidence that higher beta power would predict other forms of antisocial behavior directly. Our findings are novel in this respect, and merit further examination and replication. Another important pattern to consider in these results is that different findings emerged between the sexes. Potentially, another explanation to the results in contrast to differences in aggressive and nonaggressive ASB behavior, is that while male low arousal may contribute to antisocial behavior, a finding that has been supported at large in the literature, the same is not true for females. Because female antisocial behavior has consistently emerged as rarer than male antisocial behavior, it may be that more cognitive effort is required of females who develop patterns of antisocial behavior, hence explaining why low arousal did not predict ASB in females. In addition, because the base rates for aggressive behavior are much higher in males across species, it may be argued that antisocial behavior is more ‘default’ within males. Hence, a brain that is underaroused and relies upon predetermined behavioral instincts is more likely in 60 males to beget aggressive behavior, whereas the same is not true for females. This question should be addressed in future research. Longitudinally, the mean levels of aggressive and nonaggressive ASB were consistent with past findings, and showed an increase in aggressive ASB and a decrease in nonaggressive ASB (Eley, 2009 & 2012). It is noteworthy that mixed modeling suggests that EEG spectral power in children of age 9-10 years is a tool for predicting behavior in adolescence rather than a current measure of antisocial behavior in children. This pattern of results may suggest that children with particular spectral power may be more prone to antisocial behavior that accompanies pubertal changes. Conversely, adolescence-specific environmental factors, such as peer influence, may also be the mechanism for the biological vulnerability manifesting in behavior. The relationship between arousal levels and markers of puberty may be clarified in future studies combining EEG and hormonal properties, or other markers of pubertal change, as well as peer influence and other measures of adolescence-specific environment. The patterns that emerged in this research require further examination and replication in large studies of child and adolescent behavior, but provide important starting points for future research into how spectral power predicts behavior across development. That different relationships between spectral power and behavior emerged for aggressive and nonaggressive ASB further supports their distinction. The finding that lower arousal predicts aggressive antisocial behavior but not nonaggressive ASB may help clarify the neural mechanisms that underlie these different forms of behavior, suggesting that aggressive ASB may be a form of thrill seeking or stress response while nonaggressive ASB may stem from different neural mechanisms. Furthermore, that these relationships emerged only in examining later adolescent behavior, and not concurrent childhood behavior, suggests that spectral power may be marker of 61 propensity toward behavior that emerges specifically in adolescence, or that is influenced by pubertal or social factors. 62 References Achenbach, T. M. (1991b). Manual for the Child Behavior Checklist/4-18 and 1991 Profile. Burlington, VT: University of Vermont Department of Psychiatry. Arseneault, L., Moffitt, T. E., Caspi, A., Taylor, A., Rijsdijk, F. V., Jaffee, S. R., et al. (2003). Strong genetic effects on cross-situational antisocial behaviour among 5-year-old children according to mothers, teachers, examiner-observers, and twins' self-reports. J Child Psychol Psychiatry, 44(6), 832-848. Baker, L. A., Barton, M., Lozano, D. I., Raine, A. & Fowler, J. H. (2006). The Southern California Twin Register at the University of Southern California: II. Twin Research and Human Genetics 9, 933-40. Baker, L. A., Barton, M. & Raine, A. (2002). The Southern California Twin Register at the University of Southern California. Twin Research and Human Genetics 5(5), 456-9. Barry RJ, Johnstone SJ, Clarke AR. (2003) A review of electrophysiology in attention- deficit/hyperactivity disorder: II. Event-related potentials. Clin Neurophysiol, 114, 184– 198. Burt SA (2009). Are there meaningful etiological differences within antisocial behavior? Results of a meta-analysis. Clinical Psychology Review, 29, 163-178. Burt SA (2012). Do etiological influences on aggression overlap with those on rule breaking? A meta-analysis. Psychological Medicine, 1.1, 1-12. Burt SA (2012). How do we optimally conceptualize the heterogeneity within antisocial behavior ? An argument for aggressive versus nonaggressive behavioral dimensions. Clinical Psychology Review 32, 263–279. Burt SA, Donnellan MB (2008). Personality correlates of aggressive and nonaggressive 63 antisocial behavior. Personality and Individual Differences 44, 53–63. Clarke A, Barry R, McCarthy R, Selikowitz M. (2001). Age and sex effects in the EEG: differences in two subtypes of attention-deficit/hyperactivity disorder. Clin Neurophysiol, 112, 806–814. Ehlers, C. L., & Schuckit, M. A. (1990). EEG fast frequency activity in the sons of alcoholics. Biological Psychiatry, 27, 631–641. Ehlers, C. L., & Schuckit, M. A. (1991). Evaluation of EEG alpha activity in sons of alcoholics. Neuropsychopharmacology, 4, 199–205. Eley, T.C., Lichtenstein, P., & Stevenson, J. (1999). The Etiology of aggressive and nonaggressive antisocial behavior: Results from two twin studies. Child Development, 70, 155-168. Ellingson, R. J. (1954). The incidence of EEG abnormality among patients with mental disorders of apparently nonorganic origin: a critical review. Am J Psychiatry, 111(4), 263-275. Hare, R.D. (1970). Psychopathy: Theory and Research. Wiley, New York. Farrington, D. P. (1989). Early predictors of adolescent aggression and adult violence. Violence Vict, 4(2), 79-100. Fein, G., Galin, D., Yingling, C. D., Johnstone, J., & Nelson, M. A. (1984). EEG spectra in 9–13 year-old boys are stable over 1–3 years. Electroencephalography and clinical neurophysiology, 58(6), 517-518. Hinshaw, S. P. (1992). Externalizing behavior problems and academic underachievement in childhood and adolescence: causal relationships and underlying mechanisms. Psychol Bull, 111(1), 127-155. Krueger RF, Moffitt TE, Caspi A, Bleske A, Silva PA (1998) Assortative mating for antisocial 64 behavior: developmental and methodological implications. Behav Genet 28, 173–186. Lazzaro I, Gordon E, Whitmont S, Plahn M, Li W, Clarke S, Dosen A, Meares R. (1998). Quantitative EEG activity in adolescent attention deficit hyperactivity disorder. Clin Electroencephalogr 29, 37–42. Lindberg, N., Tani, P., Virkkunen, M., Porkka-Heiskanen, T., Appelberg, B., Naukkarinen,H., & Salmi, T. (2005). Quantitative electroencephalographic measures in homicidal men with antisocial personality disorder. Psychiatry Research, 136, 7-15. Mednick, S.A., Volavka, J., Gabrielli, W.F., Itil, T.M., 1981. EEG as a predictor of antisocial behavior. Criminology 19, 219–229. Moffitt TE (2003). Life-course persistent and adolescence limited antisocial behavior : a research review and a research agenda. In The Causes of Conduct Disorder and Serious Juvenile Delinquency (ed. B. Lahey, T. E. Moffitt and Caspi), pp. 49–75. Guilford Press : New York. Pesenti-Gritti, P., Spatola, C. A., Fagnani, C., Ogliari, A., Patriarca, V., Stazi, M. A., et al. (2008). The co-occurrence between internalizing and externalizing behaviors. A general population twin study. Eur Child Adolesc Psychiatry, 17(2), 82-92. Raine, A., Venables, P. H., & Williams, M. (1990). Relationships between central and autonomic measures of arousal at age 15 years and criminality at age 24 years. Arch Gen Psychiatry, 47(11), 1003-1007. Stanger, C., Achenbach, T. M. & Verhulst, F. C. (1997). Accelerated longitudinal comparisons of aggressive versus delinquent syndromes. Development and Psychopathology, 9, 43-58. Tremblay RE (2003). Why socialization fails: the case of chronic physical aggression. In The Causes of Conduct Disorder and Serious Juvenile Delinquency (ed. B. Lahey, T. E. 65 Moffitt and A. Caspi), pp. 182–224. Guilford Press : New York. Tuvblad, C., Eley, T. C., & Lichtenstein, P. (2005). The development of antisocial behavior from childhood to adolescence: A longitudinal twin study. European Child & Adolescent Psychiatry, 14, 216−225. Waites, A.B., Stanislavsky, A., Abbott, D.F., & Jackson, G.D. (2005). Effect of prior cognitive state on resting state networks measured with functional connectivity. Human Brain Mapping, 24, 59-68. 66 Chapter 4. Childhood EEG Frontal Alpha Power as a Predictor of Adolescent Antisocial Behavior: A Twin Heritability Study 3 Chapter 4: Abstract High EEG frontal alpha power (FAP) is thought to represent a state of low arousal in the brain, which has been related in past research to antisocial behavior (ASB). We investigated a longitudinal sample of 900 twins in two assessments in late childhood and mid-adolescence to verify whether relationships exist between FAP and both aggressive and nonaggressive (nonaggressive ASB) ASB. Significant positive predictive relationships emerged in males between childhood FAP and adolescence aggressive behavior using multilevel mixed modeling. No concurrent relationships were found. ASB was measured by the Child Behavioral Checklist, and FAP was calculated using connectivity analysis methods that used principal components analysis to derive power of the most dominant frontal activation. Using bivariate biometric twin modeling analysis, the relationship between childhood FAP and adolescent aggressive ASB in males was found to be entirely due to genetic factors, which were correlated r = 0.22. 3 Authors: Sharon Niv, Catherine Tuvblad, Syed Ashrafulla, Anand Joshi, Richard Leahy, Adrian Raine, Laura A. Baker, 67 Chapter 4: Introduction In the 1970s, a theory of antisocial behavior (ASB) emerged that is referred to as the Slow Arousal Theory. Posited by Robert Hare, a researcher of psychopathy, this theory sought to explain several findings of low arousal levels in individuals prone to crime and violence. Early research witnessed increased levels of slow-wave brain waves, including theta (4-8 Hz) and delta (1-4 Hz) in the adult brains of incarcerated psychopaths (Ellingson, 1954) and violent criminals (Hill, 1952). Since these early studies, lower prefrontal activation was found in males with past aggressive behavior (Volkow, et al. 1995), in 9 males and 1 female with repetitive violent behavior (Critchley, et all. 2000), and in violent psychiatric inpatients (Kuruoglu, et al. 1996). In examining these findings, the Slow Arousal Theory suggests a ‘stimulus hunger’ in brains marked by slow-wave activity. A stimulus hungry and poorly aroused brain, it is suggested, may require risky, impulsive, or other ‘high excitement’ external stimuli to achieve the arousal levels that a normally-aroused brain typically experiences. This may explain higher propensity toward violent behavior and delinquent or criminal activity. In children, slow arousal as marked by high alpha power, has also been predictive of criminal activity. EEG alpha power, typically measured as 8-13 Hz in adults and 8-10.5 Hz in children of the age examined in the present research (Gasser, et al., 1988), is representative of a sleepy quality. In children, increased cortical alpha power (a marker of slow arousal) has been associated with later crime (Mednick, et al. 1981; Petersen, et al. 1982). In a group of 24-year- old male criminals, retrospective analysis found high alpha power at the age of 15 (Raine, et al. 1990). Reduced frontal activation was also found in children with oppositional defiant disorder, using single-photon emission computerized topography (SPECT; Amen & Carmichael, 1997). 68 The majority of this research, both in adults and in children, was conducted with male subjects, and an aim of the present research is to investigate these relationships in females, as well. One important question that has arisen in research of ASB is that of the differences between aggressive and nonaggressive ASB, sometimes referred to as delinquency or rule- breaking, and captured by the Delinquency scale of the CBCL, which is used in this research. Aggressive ASB shows heritability of approximately 44% across studies (Rhee & Waldman, 2002), and 38-50% in our study (Baker, et al. 2008). Alpha power has shown high heritability, from 63-89% in children and adolescents (van Baal, et al. 1996; van Beijsterveldt, et al. 1996). In our sample, full range frontal alpha power (FAP) at 8-13 Hz was found to have 71- 85% genetic influence, with the rest of the variance accounted for by non-shared environmental factors (Gao, et al. 2009). However, no past study has examined potential genetic correlations between FAP and any form of ASB, of either aggressive or nonaggressive nature. This is a crucial gap the present research aims to fill. Chapter 4: Methods USC Study of Risk Factors for ASB (RFAB). The data in the present research are drawn from the USC Twin Study, a longitudinal project currently in its fifth wave of data collection, which focuses on genetic, environmental, social, and biological influences on the development of ASB. The twins and their families were recruited from the greater Los Angeles area, and are ethnically and socioeconomically representative of the region (Baker, et al. in press). The present research uses data from Waves 1 and 3, when twins were 9-10 and 14-15 years old, respectively. EEG data were only collected in Wave 1 (605 families; N=1,219 twins/triplets). Caregivers (91.4% biological mothers) were administered the child behavior checklist (CBCL) in both 69 Waves 1 and 3. In accordance with past research of EEG (Baving, et al. 2003; Deckel, et al. 1996), only right-handed twins were included (i.e., excluding 169 left-handed participants). In total, 431 males and 469 females provided EEG data, all of whom also provided Wave 1 aggressive ASB data, and 311 and 378 of whom, respectively, also provided Wave 3 aggressive ASB data. The number of complete twin pairs used in longitudinal analyses, not counting single twins, was 383 (90 MZ male, 105 MZ female, 47 DZ males, 54 DZ females, 87 Opposite sex). Aggressive and nonaggressive ASB, Waves 1 and 3, parent-ratings. The Child behavior Checklist (CBCL) is a widely used caregiver-response instrument for research and clinical work. It measures a wide range of behavior problems in children over the prior 6 months (Achenbach, 1991). This instrument has 113 items that use a three-point scale (0 for not true, 1 for sometimes true, and 2 for very or often true). The Aggressive ASB subscale consists of 13 items and includes behaviors such as arguing, fighting with other children, and bullying others. The internal consistencies in Waves 1 and 3 were found to be 0.61 and 0.71, respectively. The nonaggressive ASB subscale (CBCL Delinquency) consists of 20 items gauging cheating, lying, petty crimes, and other nonaggressive ASB activities, and showed internal consistencies of 0.88 and 0.89 at Waves 1 and 3, respectively. EEG Collection and Processing, Wave 1. First, a 3-minute rest period baseline of activity was obtained prior to any task completion after which the twin participated in approximately 1.5- 2 hours of tasks. In this condition, the children kept their eyes open and paid attention to a blue cross on a computer screen, as it was observed that 9-10 year old participants had trouble keeping eyes closed for three full minutes. The rest period was selected as it likely represented the most naïve brain activity before a series of tasks. 70 EEG data were collected from 30 scalp sites and left and right mastoids based on the 10- 20 international system (Jasper, 1958) using James Long Inc. amplification system (Caroga Lake, NY). A lycra Electro-Cap (Eaton, OH) was used in the data collection. Anterior midline site (AFz) was the ground electrode whereas the central midline site (Cz) served as the reference site. The scalp recording sites were prepared with conductive abrasive, which aids in increasing conductivity. Impedances were kept below 10 KOhms. To account for eye blink artifacts, a bipolar electrooculogram (EOG) channel was recorded from surrounding (above and below the supra and infra-orbital eye ridges of the left eye). Tester comments and observations, as well as unusual data values, were taken into account in evaluating the validity of each participant’s data for inclusion in data analysis. The majority of manually excluded cases were excluded for tester- reported excessive movements and fidgetiness. Derivation of FAP. Data from selected electrodes, nine frontal and prefrontal electrodes: Fp1, Af3, F7, F3, Fz, F4, F8, Af4, Fp2, was filtered into the low alpha band (8-10.5Hz), appropriate for analysis of resting-state EEG recordings from children aged 9-10 years. The FAP is estimated via a weighted sum of this data across electrodes. The weights were calculated via principal components analysis (PCA), which attempts to maximize the power in the weighted sum, given that the weights squared sum to 1 (that is, they are normalized). Thus, the FAP analyzed is the strongest average power over all possible normalized weightings of the frontal electrode data. Statistical Analyses. Data were visualized, cleaned and analyzed for descriptive statistics and phenotypic correlations in Statistical Analysis Software (SAS 9.1.3, 2005). Mixed modeling regression was also conducted in SAS, using family ID to nest twins, taking into account non- 71 independent observations between family members. Outcome variables were aggressive and nonaggressive ASB in Waves 1 and 3. Genetic Analyses. ASB and EEG data were transformed using Blom Normalization for genetic analyses to reduce skewness in the ASB data. Analyses were conducted using Mx GUI (Neale, 2003), a structural equation-modeling program specialized for twin data. Genetic model comparison in Mx relies on raw maximum-likelihood estimation procedures, which yield a likelihood ratio statistic (-2LL) based on differences between observed and expected data values. Calculating the difference between two models’ -2LL values yields a chi-square (χ 2 ) test of significance with degrees of freedom (df) equal to the differences in df between the two compared models (Neale & Cardon, 1992). All models were compared to a baseline saturated model which freely estimates means, variances and covariance for each zygosity group. Model fit was also assessed by comparing the Akaike’s Information Criterion (AIC). Lower values of AIC indicate more parsimonious fit and a better model. Bivariate Cholesky decomposition models were used to determine proportions of A (additive genetic), C (shared environmental), and E (nonshared environmental) influences on FAP and aggressive ASB or nonaggressive ASB separately as well as the shared A, C, and E influences between them. This model uses covariances between twins for given traits and across traits, hence partitioning the influences that are unique or shared between variables. Different models were examined to verify whether sexes could be equated, and whether types of influences (e.g. shared environment) could be dropped to explain observed results most parsimoniously. 72 Chapter 4: Results Males showed higher aggressive ASB levels than females in both Wave 1 (t=5.08, df =887, p=0.02) and Wave 3 (t=2.36, df = 596, p=0.02). Nonaggressive ASB was higher in males than in females in Wave 1 (t=3.975, df=887, p<0.01), but not Wave 3 (t=1.21, df=596, p=0.23). No significant sex differences emerged in FAP. Mean levels of aggressive ASB in Wave 1 were 6.12 (SD=5.16) in males and 5.10 (SD=4.78) in females, and in Wave 3 were 5.35 (SD=6.00) in males and 4.32 (SD=4.80) in females. Means levels of nonaggressive ASB were 1.49 (SD=1.81) in males and 1.06 (SD=1.36) in females in Wave 1, and 1.79 (SD=2.34) in males and 1.55 (SD=2.52) in females in Wave 3. Repeated measures analyses of variance revealed that aggressive ASB decreased between Waves 1 and 3 (F (1,179) =8.16, p<0.01), and nonaggressive ASB increased (F (1,179) =22.12, p<0.01, with no significant interactions with sex. Mixed modeling conducted with both sexes did not yield significant betas in Wave 1. In Wave 3, significant main effects of FAP were deemed not interpretable due to significant interactions of sex with FAP. Consistent with phenotypic correlations, when analyzed separately, FAP was not significantly predictive of aggressive ASB in females (p=0.26). In males, main effect of FAP was Beta = 0.99 (SE=0.44), p=0.03. FAP did not significantly predict Wave 3 nonaggressive ASB in males (Beta = 0.05, SE = 0.17, p=0.76), nor Wave 3 nonaggressive ASB in females (Beta = -0.18, SE=0.12, p=0.09). Twin and cross-twin cross-trait correlations are presented in Table 4.1 for the two variables that were found to be phenotypically related – FAP and Wave 3 aggressive ASB. Twin correlations are consistently higher in the Mz than in the Dz twins, suggesting heritability. Conversely, the cross-trait correlations show slightly higher correlations in the male Mz twins 73 than in the male Dz twins, with no differences found between Mz and Dz females, which suggests genetic overlap in the males but not in the females. Table 4.2 Twin and Cross-twin Cross-trait Correlations for FAP and Wave 3 Aggressive ASB by Zygosity Mz male Mz Female Dz Male Dz female Dz Opposite FAP Agg- 3 FAP Agg- 3 FAP Agg- 3 FAP Agg- 3 FAP Agg- 3 FAP 0.73* -- 0.64* -- 0.60* -- 0.54* -- 0.37* -- Agg- 3 0.22* 0.74* -0.07 0.72* -.0.06 0.11 -0.10 0.42* 0.05 0.52* Note. Bolded values represent twin correlations, and non-bolded values cross-twin cross-trait correlations Genetic models were only explored for the relationship that demonstrated significant phenotypic relationships – namely, FAP with Wave 3 male aggressive ASB. Model fits and comparisons are shown in Table 4.2. Comparing models 1 and 1a demonstrates that equating the estimates on the two sexes significantly worsens model fit, and so sexes were estimated separately. Comparing Models 1 and 2 demonstrates that dropping C improved model fit, suggesting that shared environment was not significant in any of these relationships. Finally, dropping E correlations provided the best fit to the data (Models 9). Bivariate genetic analysis found that influences on FAP are 78% A (raw variance 0.88 2 ) and 22% E (raw variance 0.47 2 ) in males. Influences on Wave 3 aggressive ASB were found to be 65% A (raw variance 0.81 2 ) and 35% E (raw variance 0.59 2 ) in the males. The covariation between FAP and male Wave 3 74 aggressive ASB was accounted for exclusively by significant genetic correlation at Rg=0.22 (95% CI 0.16-0.30). Although the models investigated included both males and females, and estimated influences on them separately, Figure 4.1 depicts only the relationships that demonstrated significant bivariate relationships between FAP and ASB, for which the 95% confidence intervals (CI) did not include zero. Female FAP was not significantly related to aggressive ASB in either wave through bivariate genetic or environmental influences. These findings are consistent with phenotypic mixed modeling results. This figure presents squared standardized estimates of the influences. 75 Table 4.2. Bivariate Model Fit Indices for FAP and Aggressive ASB in Wave 3 Likelihood Values Overall Fit Model Comparison Model -2LL DF AIC BIC χ 2 (∆df) P Compared to Model ∆χ 2 (∆df) P 0 Saturated model (means constrained) 3281.56 1509 263.56 -3082.07 1 ACE 3345.97 1557 231.97 -3299.10 64.41 (48) 0.06 1a. ACE M=F 3364.39 1566 232.39 -3219.06 83.83 (57) 0.02 1 7.61 (9) 0.57 2 Drop shared environmental effects (AE) 3351.39 1563 225.39 -3216.17 69.83 (54) 0.07 1 6.14 (6) 0.41 3 Drop genetic effects (CE) 3405.03 1563 279.03 -3189.35 123.47 (54) P<0.01 1 60.03 (6) P<0.01 4 Drop genetic and shared environmental effects (E) 3636.89 1569 498.89 -3092.20 355.33 (60) P<0.01 1 371.32 (12) P<0.01 5 No genetic covariance 3350.88 1559 232.88 -3203.91 69.32 (50) 0.04 1 5.25(2) 0.07 6 No common environmental covariance 3349.59 1559 231.59 -3204.55 68.03 (50) 0.05 1 3.98 (2) 0.14 7 No A or C covariance 3355.34 1561 233.34 -3207.94 73.78 (52) 0.03 1 11.34 (4) 0.02 8 No nonshared env. covariance 3348.71 1559 230.71 -3204.99 67.15 (50) 0.05 1 2.86 (2) 0.24 9 AE, no nonshared env. covariance 3353.39 1565 223.39 -3221.43 71.83 (56) 0.08 2 6.38 (8) 0.38 76 Figure 4.1. Bivariate Heritability Model of Wave 1 EEG Alpha Power with Wave 3 Aggressive ASB in Males Chapter 4: Discussion This study sought to investigate the phenotypic and genetic relationships between FAP and ASB, both aggressive and nonaggressive ASB, in childhood and adolescence. It emerged that in males but not females, FAP at the age of 9-10 years was predictive of aggressive ASB at the ages of 14-15 years. No significant concurrent relationships emerged in either sex, and no A A E E Aggression Wave 3 (14-15 years old) 0.65* (0.51-0.75) 0.22* (0.16-0.30) 0.35* (0.25-0.49) Frontal Alpha Power Wave 1 (9-10 years old) 0.78*(0.70-0.84) 0.22* (0.16-0.30) 77 relationships emerged with nonaggressive ASB. Frontal power and aggressive ASB both showed high heritability and no shared environmental influences. Significant genetic correlations emerged in the males but not female between aggressive ASB and FAP.. Although these genetic correlations were modest, they constituted 100% of the influences accounting for the phenotypic relationship between FAP and ASB. These findings are novel in the literature of developmental ASB. While past studies found links between alpha power and criminality (Kuruoglu, et al. 1996; Raine, et al. 1990; Mednick, et al. 1981), this relationship had never been demonstrated in a community sample. These findings suggest that high FAP, representing slow wave activity, is predictive of male aggressive ASB in the general population. Very importantly, the majority of past research examined male participants exclusively, usually using clinical or criminal populations, or sons of offenders. This is one of the first examinations of female neural arousal and its relationship to ASB, and hence the results are of considerable interest as they suggest that past findings may be significant exclusively within males. Lastly, this was the first demonstration of joint genetic etiology to FAP and ASB. The genetic findings that emerged in this research are novel. While the heritability of ASB and the heritability of alpha power have been demonstrated in past research, as well as the phenotypic relationship between them, the genetics of that relationship had not been explored. Our findings suggest very moderate but significant heritability overlap to these two variables. This suggests that while some of the same genes that contribute to FAP are those that determine ASB in adolescents, there are many genetic influences that do not overlap. It would be beneficial for researchers to continue exploring biological markers that could account for a greater proportion of genetic influence on ASB. 78 One notable finding is that childhood FAP was not correlated with concurrent childhood ASB, but was significantly correlated with adolescent ASB. This may imply that FAP is a pre- dispositional marker of children likely to develop more aggressive ASB response patterns in the future. It may also imply that FAP is more strongly related to a qualitatively different form of ASB that emerges in adolescence over childhood. As EEG data from Wave 3 were not collected, it is not possible to say whether concurrent relationships in adolescence would have emerged. Inarguably, the barrier to ASB behavior is higher for females than for males. Female aggressive ASB is both rarer, uniformly and consistently across studies, and may be regarded with less social acceptance (Bjorkvist, 1994). This may explain why a low-arousal brain in the males may default to thrilled or risk-seeking behavior but not so in the females. It is also theorized by some that female aggressive ASB regularly takes alternative forms. In recent years, attention has been paid to the concept of ‘relational aggressive ASB, which comprises excluding others from social groups, gossiping, purposely withdrawing friendship, and other forms of indirect and non-physical attacks (Crick & Grotpeter, 1996). It is possible that the absence of correlation with aggressive ASB that is observed in these results fails to capture aggressive ASB as it is most pertinent to females in mid-adolescence. Future research may benefit from exploring relationships between FAP and relational aggressive ASB in females. While aggressive behavior was associated with lower arousal, the same was not true for nonaggressive ASB. This pattern of results may provide further support to the distinction between aggressive and nonaggressive ASB. Much past research concerning slow-arousal has focused on violent offending, which is a subset of aggressive ASB. Nonaggressive ASB, or rule- breaking behavior, may in fact stem from neural mechanisms that are distinct from low arousal. This finding should be further investigated in other samples for the purpose of replication. 79 Two additional configurations of electrodes were tested: five frontal electrodes (F7, F3, Fz, F4, F8) and three frontal regions (F3, Fz, F4) without prefrontal electrodes. Genetic correlations between FAP and male Wave 3 aggressive ASB were significant using five but not three electrodes, but were slightly lower (Rg=0.17*). Neither of the other configurations showed significant correlations with female ASB in either wave. This suggests that that inclusion of prefrontal electrodes is beneficial for prediction of ASB. It should be noted that when activation was simply averaged across electrodes rather than calculated using PCA weighting, correlations were significant and higher than those reported here (r=0.30, p<0.01 in males; r=-0.19, p=0.02 in females). However, using this evenly weighted sum may not be as able to find the strongest activations in the frontal area. Hence, we believe that the PCA method of calculating alpha power is more accurately reflective of frontal activation. Also, the weighting procedure derived from PCA may be less likely to inflate correlations between EEG and behavior as opposed to an even weighting, because the PCA procedure tries to reduce the effects of noise from other regions or the EEG equipment. In addition, frontal asymmetry was also tested for relationship with ASB, and was not found to correlate significantly with behavior. Despite great interest in frontal asymmetry over the recent decades, our findings suggest that overall frontal power may be more predictive of ASB. Slow Arousal Theory is further informed by another prominent and influential theory of the past few decades, the somatic marker hypothesis (Damasio, et al. 1991). This theory stemmed from a set of findings that showed that in most individuals, risky actions are accompanied by a somatic nervousness, typically measured with skin conductance response. In psychopathic individuals, however, this somatic response was absent, leading to the suggestion that psychopathic individuals experience lower arousal, as well as an absence of discomfort that 80 prevents most individuals from taking risky actions. In a recent project conducted with the present twin sample, it was found that males with reduced orienting response to novelty, as measured by skin conductance, showed higher levels of psychopathy (Tuvblad, et al., under review). In these analyses, children who had persistently high or persistently low skin conductance response over the course three waves of data collection were compared; boys with persistently low skin conductance orienting response also showed significantly higher psychopathy scores in adolescence, suggesting some longitudinal stability. This analogous finding lends support to the gender difference explanation rather than necessarily differences between aggressive and nonaggressive ASB, as the ASB instrument in these analyses was the Child Psychopathy Scale (Lynam, et al. 1997) a measure of personality rather than aggressive ASB. However, because both of these findings emerged in the same sample, it is crucially important for large twin studies other than the present study to replicate these promising findings. Lastly, this research has mechanistic implications for ASB. A question to consider in future research is whether FAP constitutes an endophenotype or a mediator for ASB. While the Slow Arousal Theory suggests that a stimulus hunger is at the core of the tendency toward stimulation seeking, in this case through violence, it is also possible that both FAP and male aggressive ASB are in fact affected by a third variable. For example, it may be that insufficient maternal attention causes both low arousal due to lack of stimulation, as well as aggression. A potential manner in which to investigate this may be to temporarily manipulate frontal activation in the laboratory, such as through transcranial magnetic stimulation, and investigate ASB tendencies while in this altered brain state. If the mechanism of this relationship were discerned, treatment implications would emerge for methods such as neurofeedback, meditation, or 81 medication. These findings would help the field to build a clearer picture of the relationships between genes, brain states, and behavior. 82 References Achenbach, T. M. (1991). Manual for the Child Behavior Checklist/4-18 and 1991 Profile. Burlington, VT: University of Vermont Department of Psychiatry. Amen DG, Carmichael BD (1997) High-resolution brain SPECT imaging in ADHD. Annals of Clinical Psychiatry, 9, 81-86. American Psychiatric Association. (1994). Diagnostic and Statistical Manual of Mental Disorders (4 th edition). Washington, DC. Baker, L. A., Barton, M., Lozano, D. I., Raine, A. & Fowler, J. H. (2006). The Southern California Twin Register at the University of Southern California: II. Twin Research and Human Genetics 9, 933-40. Baker, L. A., Barton, M. & Raine, A. (2002). The Southern California Twin Register at the University of Southern California. Twin Research and Human Genetics 5(5), 456-9. Baker, L.A., Tuvblad, C., Wang, P., Gomez, K., Bezdjian, S., Niv, S., & Raine, A. (in press) The Southern California Twin Register at the University of Southern California: III. Twin Research. Bjorkvist, K. (1994) Sex differences in physical, verbal, and indirect Aggression: A review of recent research. Sex Roles, 30, 177-188. Crick, N.R. & Grotpeter, J.K. (1996). Children’s treatment by peers: Victims of relational and over Aggression. Developmental Psychopathology, 8, 267-280. Critchley HD, Simmons A, Daly EM, Russell A, van Amelsvoort T, Robertson DM, Glover A, Murphy DG (2000). Prefrontal and medial temporal correlates of repetitive violence to self and others. Biol Psychiatry, 47, 928–934. Damasio, A.R., Tranel, D., & Damasio, H. (1991). Somatic markers and the guidance of 83 behavior. In H. Levin, H. Eisenberg, & A. Benton (Eds.), Frontal lobe function and dysfunction (pp. 217-228). New York: Oxford University Press. Ellingson, R. J. (1954). The incidence of EEG abnormality among patients with mental disorders of apparently nonorganic origin: a critical review. Am J Psychiatry, 111(4), 263-275. Gao, Y., Tuvblad, C., Raine, A., Lozano, D. I., & Baker, L. A. (2009). Genetic and environmental influences on frontal EEG asymmetry and alpha power in 9-10-year-old twins. Psychophysiology, 46(4), 787-796. Gasser T, Verleger R, Bacher P, Sroka L. Development of the EEG of school age children and adolescents. I. Analysis of band power. Electroenceph clin Neurophysiol 1988b;69:91– 99. Hill, D. (1952). EEG in episodic psychotic and psychopathic behaviour; a classification of data. Electroencephalogr Clin Neurophysiol, 4(4), 419-442. Jasper, H. H. (1958). The 10-20 electrode system of the International Federation. Electroencephalography and Clinical Neurophysiology, 10, 371-37. Kuruoglu AC, Arikan Z, Karatas M, Arac M, Isik E. (1996). Single photon emission computerized tomography in chronic alcoholism: antisocial personality disorder may be associated with decreased frontal perfusion. Br J Psychiatry 16, 348–354. Lynam, D. R., (1997). Pursuing the psychopath: Capturing the fledgling psychopath in a nomological net. Journal of Abnormal Psychology, 106, 425-438. Mednick, S.A., Volavka, J., Gabrielli, W.F., Itil, T.M., 1981. EEG as a predictor of ASB. Criminology 19, 219–229. Neale, M. C., Boker, S. M., Xie, G. & Maes, H. (2003). Mx: Statistical modeling. Richmond, VA: Department of Psychiatry, Medical College of Virginia. 84 Neale, M. C. & Cardon, L. R. (1992). Methodology for genetic studies of twins and families. Dordrecht, The Netherlands: Kluwer Academic Publications. Niv. S., (2013). Clinical efficacy and potential mechanisms of neurofeedback. Personality and Individual Differences, 54, 676-686. Petersén, K. G., Matousek, M., Mednick, S. A., Volavka, J., & Pollock, V. (1982). EEG antecedents of thievery. Acta Psychiatr Scand, 65(5), 331-338. Raine, A., Venables, P. H., & Williams, M. (1990). Relationships between central and autonomic measures of arousal at age 15 years and criminality at age 24 years. Arch Gen Psychiatry, 47(11), 1003-1007. Rhee, S. H. & Waldman, I. D. (2002). Genetic and Environmental Influences on ASB: A Meta- Analysis of Twin and Adoption Studies. Psychological Bulletin 128, 490-529. Tuvblad, C., Gao, Y., Niv, S., Raine, A., & Baker, L.A. The Skin Conductance Orienting Response and Psychopathic Personality: A Longitudinal Twin Study. Van Baal, G. C., De Geus, E. J., & Boomsma, D. I. (1996). Genetic architecture of EEG power spectra in early life. Electroencephalogr Clin Neurophysiol, 98(6), 502-514. van Beijsterveldt, C. E., Molenaar, P. C., de Geus, E. J., & Boomsma, D. I. (1996). Heritability of human brain functioning as assessed by electroencephalography. Am J Hum Genet, 58(3), 562-573. Volkow ND, Tancredi LR, Grant C, Gillespie H, Valentine A, Mullani N, Wang GJ, Hollister L (1995). Brain glucose metabolism in violent psychiatric patients: a preliminary study. Psychiatry Res 61, 243–253 85 Chapter 5. Additional Analyses The purpose of this section is to address several additional topics that provide a fuller understanding of findings but are outside the scope of Chapters 2-4. Several questions arose as to why a robust pattern of findings supporting low arousal and ASB did not emerge in this study. Potential reasons identified were weakness of EEG data, weakness of the CBCL instrument, or differences between clinical/criminal samples and our community sample. It was also asked whether high-risk models would find additional support for genetic underpinnings of the power/ASB relationship, and whether there would be further support to the longitudinal finding that childhood EEG spectral power was associated with adolescent but not concurrent (childhood) ASB. Lastly, this chapter presents results of initially-proposed analyses with frontal alpha asymmetry that did not yield significant findings. The first question arose in the course of Chapters 3 and 4 analyses, and concerned the quality and validity of the EEG data. It was considered a possibility that poor data quality may be a reason why there emerged no robust patterns of low arousal in relation to ASB, unlike past studies. In order to answer this question, the full-brain heritability of theta, alpha, and beta frequencies was explored. The aim of this approach was to reproduce past heritability assessments, which found high heritability in the rages of 50-70% for full brain EEG power in both children and adults. These analyses are presented in section 5A. Next, it was considered possible that the CBCL did not fully capture ASB, and questions were raised about whether other instruments would yield similar or different results. Presented in section 5B, other measures of ASB were substituted for CBCL variables to verify whether relationships would be upheld. This was also used to examine generalizability of the findings that did emerge in Chapter 3. 86 One more hypothesis raised as to why robust patterns of low arousal and ASB were not found is that there may be fundamental differences between the present sample and the clinical/criminal samples previously examined. It is possible that relationships between EEG spectral power and ASB may be dependent upon third variables, such as exacerbating environmental conditions. It is known that stress produces effects on the brain and behavior both, and the majority of past findings emerged in clinical or criminal samples, which may be pre- selected for greater stress. This was proposed as a reason why robust relationships did not emerge in our sample, as community samples likely include fewer cases of significant environmental stress than clinical samples. It is hence possible that if moderators were considered, suppressed relationships would have emerged. Mixed models were texted for significant interaction terms for several identified environmental conditions examined in the twin study: stressful life events, socioeconomic status, neighborhood adversity, and peer victimization. These analyses are presented in section 5C. The finding that Wave 1 spectral power related to Wave 3 but not Wave 1 ASB was also further explored by examining ASB trajectories previously derived using four waves of data collection; in section 5D, group analysis of nonaggressive ASB trajectories across ages 9-18 were applied to the EEG data to test for mean differences. In section 5E, parental ASB values from Wave 1 were used to predict EEG scores in order to investigate whether genetic risk conferred by parental behavior may affect neural patterns in the children- this was conducted in order to investigate whether further support could be found for the genetic relationship that emerged in Chapter 4. 87 Lastly, to address an initially-hypothesized relationship between frontal alpha asymmetry and ASB, analyses were performed using mixed modeling. Results from these analyses are presented in section 5F. Below, literature relevant to these questions is summarized and methods and results are explained. This is followed by a summary of Chapter 5 findings. Discussion of these results is integrated into the Chapter 6 General Discussion. 5.A. Spectral Power Heritability The purpose of conducting these analyses is twofold: 1. To investigate the heritability of late childhood spectral power using a large sample in open-eyed condition; 2. To verify the quality of the data collected in this study. 5.A.1. Background Understanding the relationships between neural activity and behavior is a paramount goal of neuroscience and psychology both. The harmless and inexpensive tool of electroencephalography (EEG), one of the oldest and most established neuroscience methods available, is used to investigate sleep, epilepsy, and emotional response. Additionally, different EEG profiles have been associated with wide array of psychological symptoms, such as depression (Davidson, et al. 1998, 1992), criminality (Raine, et al. 1990; Mednick, et al. 1981, Ellingson, 1954), hyperactivity and disinhibition (Barry, et al. 2003), and alcoholism (Ehlers & Schukit, 1990, Rangaswamy, et al. 2002). Hence, studying the heritability of EEG activation of different frequencies during childhood is an important step in understanding developmental influences on the brain and on behavior. 88 Past research has asked similar questions using adults and adolescents. In closed-eyed adults, heritability of theta, alpha, and beta bands was uniformly high, at approximately 70-80% (Smit, et al. 2005). Averaged over 14 electrodes in 213 16-year-old twins with eyes closed, theta, alpha, and beta frequencies were found to be heritable at 76-89% with few sex differences (van Beijsterveldt, et al. 1996). In 119 15 and 17-year-old closed-eyed adolescents, four sites were found to be heritable between 28-79% for theta, alpha and beta power (McGuire, et al. 1998). In 209 5-year-old children with closed eyes, high heritability and no sex differences were found in theta, alpha and beta ranges for 14 sites (van Baal, et al. 1996). However, in the present study, 9- 10-year-old children grew restless when asked to close their eyes quietly for 3 minutes, and this suggests that open-eyed measurement may in fact represent a more accurate resting state, as it involved less agitation for the children. To the best of our knowledge, there have been no studies conducted with a sample of this size (N=900), with this age-range, or using an open-eyed rest period in children. 5.A.2. Measures & Methods Participants. All EEG data utilized in this study were collected during the first wave of data collection of the Southern California’s Twin Project, when twins were 9-10 years old. This study uses data from 900 twins (47.9% male, 52.1% female), as some were removed for left- handedness or poor EEG signal. The number of complete twin pairs, in which both twins provided good EEG data, was 376 (88 MZ male, 105 MZ female, 46 DZ male, 5 DZ female, 82 DZ opposite sex). Zygosity Determination. To determine zygosity, DNA microsatellite analysis was performed, which tests for concordance of DNA markers. Any discordance deems a pair DZ, whereas >7 concordant markers deems a pair MZ. This was effective in 87% of cases, and a 90% 89 reliable twin similarity questionnaire was used in the other 13% of cases where DNA signal was too weak for use (Lykken, 1978). EEG Collection and Processing. Twins underwent psychophysiological testing while their co-twins and caregivers were engaged in interviewing in nearby rooms. Before 1.5-2 hours of tasks, the children were instructed to sit quietly with their eyes open for three minutes, looking at a cross on a computer screen. This rest period was selected for these analyses, as prior cognitive activity carries effects on resting state networks, and may not represent baseline activity in its most naïve, natural state (Waites, et al. 2005). Recordings were collected using the James Long Inc. amplification system (Caroga Lake, NY) from 30 scalp sites using the 10-20 international system (Jasper, 1958). Children were fit with a lycra Electro-cap (Eaton, OH) after the use of a conductive abrasive on each site to increase conductivity. The ground electrode was the anterior midline site (AFz) and the reference site was central midline site (Cz), and impedances were kept below 10 KOhms. Eye blink artifacts were removed through the use of a bipolar electrooculogram (EGO) channel above and below the supra and infra-orbital eye ridges of the left eye. Testers were instructed to keep detailed observations during testing, which were used to evaluate the validity of each participant’s data. The majority of children excluded were excluded for excessive fidgeting and head movements. All EEG data were pre-processed and processed in Matlab (Mathworks ®, R2011b). Data presented here use absolute power, although relative power was examined and found to be consistent with open-eye resting data from past research (Fein, et al. 1984). Graphical representations of heritability throughout the brain were produced with Brainstorm 3.1. Analyses. All data were Blom Normalized before genetic analysis. The statistical software package Mx (Neale, et al. 2003) was used for all genetic analyses. This program 90 compared observed and expected values in each specified model and yields a likelihood ratio statistic to represent this difference (-2LL). The -2LL scores of two models can be compared, yielding a χ 2 distribution, the degrees of freedom (df) for which is equal to the difference in df between the two models. The resulting p-value is used to estimate fit likelihood, with p<0.05 consider poor and unlikely explanations of the observed data. Model fit was also assessed using the Akaike Information Criterion (AIC) and the Bayesian Information Criterion (BIC) fit statistics, both of which indicate more parsimonious fits at lower values. Because of the volume of analyses conducted in the present study, these figures are not presented and are available upon request. Univariate analyses were employed for all genetic models in the present research. These models partition variation and covariation between co-twins to estimate three kinds of influences on an observed variable: Additive genetics (A), shared environment between the twins (C), and nonshared environment unique to each twin (E). When correlations between MZ twins are higher than correlations between DZ twins, it provides evidence for genetic influence, as MZ twins share 100% of their genes and DZ twins on average 50%. The magnitude of covariation is also used to estimate common effects that equally influence twins regardless of zygosity, or shared environmental factors. E estimates include error in addition to unique environment. 5.A.3. Results Heritability of EEG Spectral Power In addition to boxplot outliers, power values below 1.0 E-11 Voltage/second (V/s) were manually removed, as they were considered too low to rule out electrode failure. Overall theta power per site ranged from 1.0 E-11 V/s to 1.03 E-8 V/s. T- tests revealed significant sex differences in theta power at sites CP5, P7, P8, PO3, PO4, and O2, with females showing higher theta power in all cases. Alpha power ranged from 1.91 E-11 V/s to 91 1.03 E-10 V/s. Significant sex differences emerged in C3, T8, and PO4, with females showing higher alpha power. Beta power ranged from 1.09 E-11 V/s to 3.16 E-8 V/s. Significant beta power sex differences emerged in AF3, F7, and CP1, with females showing higher beta power. Figures 5.1 and 5.2 show heritability for males and females, respectively, for the selected frequency ranges across the full scalp. Tables 5.1-5.3 display the A and E estimates at every site in each of the three frequency ranges. Male and Female influences showed some differences for heritability of the alpha and theta power ranges, as significant sex differences emerged in genetic modeling. Conversely, heritability of beta power was found to be equitable between the sexes. For theta power, genetic models found equal estimates of genetic and environmental influences for males and females in the entire frontal region (Fp, F, AF), as well as in T7, C3, and CP6. Other than these sites, and in the entire posterior brain (P, PO, O), females showed higher heritability estimates than males. For alpha power, the pattern of sites with significantly different sex patterns was less distinct, with the following sites showing sex difference: Fp1, F7, F4, FC2, T8, CP6, P7, Pz, P8, P4, O1, PO4, and O2. In each of these, females showed higher heritability than males. Males and females showed equal heritability of beta power across the brain. Heritability ranged between 56 and 72%, with the remaining influence accounted for by nonshared environment. In eight Univariate analyses, estimates of common environment emerged as significant, but due to the large number of analyses conducted (30 univariate analyses for each of three frequency ranges), it can be expected that C may emerge as significant in some analyses. AE models were selected as best fitting in these cases as well, to remain consistent with the majority of our results, as well as past findings. 92 Table 5.1. Theta Power AE Estimates by Site and Sex Site Males Females A E A E Fp1 0.40 0.60 0.40 0.60 Fp2 0.43 0.57 0.43 0.57 Af3 0.51 0.49 0.51 0.49 Af4 0.52 0.48 0.52 0.48 F7 0.51 0.49 0.51 0.49 F3 0.61 0.38 0.61 0.38 Fz 0.58 0.42 0.58 0.42 F4 0.62 0.38 0.62 0.38 F8 0.55 0.45 0.55 0.45 FC5* 0.56 0.44 0.59 0.41 FC1* 0.59 0.31 0.61 0.39 FC6* 0.52 0.48 0.63 0.37 FC2 0.64 0.36 0.64 0.36 T7 0.58 0.42 0.58 0.42 C3 0.60 0.40 0.60 0.40 T8 0.65 0.35 0.65 0.35 C4 0.75 0.25 0.75 0.25 CP5* 0.55 0.34 0.67 0.33 CP1* 0.55 0.45 0.62 0.38 CP2* 0.51 0.49 0.61 0.39 CP6* 0.53 0.47 0.70 0.30 P7* 0.55 0.45 0.71 0.29 P3* 0.61 0.39 0.69 0.31 Pz* 0.46 0.54 0.67 0.33 P8* 0.59 0.41 0.68 0.32 P4* 0.53 0.47 0.79 0.21 PO3* 0.64 0.46 0.73 0.27 O1* 0.46 0.54 0.75 0.25 PO4* 0.49 0.51 0.56 0.44 O2* 0.40 0.60 0.67 0.33 Note. Starred (*) values indicate higher heritability in females than in males. Table 5.2. Alpha Power AE Estimates by Site and Sex Site Males Females A E A E Fp1 0.39* 0.61 0.68 0.32 Fp2 0.56 0.44 0.56 0.44 Af3 0.66 0.34 0.66 0.34 93 Af4 0.56 0.44 0.56 0.44 F7 0.51* 0.49 0.64 0.36 F3 0.68 0.32 0.68 0.32 Fz 0.77 0.23 0.77 0.23 F4 0.58 0.42 0.58 0.42 F8 0.50* 0.50 0.62 0.38 FC5 0.61 039 0.61 0.39 FC1 0.71 0.29 0.71 0.29 FC6 0.70 0.30 0.70 0.30 FC2 0.51 0.49 0.51 0.49 T7 0.60 0.40 0.60 0.40 C3 0.72 0.28 0.72 0.28 T8 0.67* 0.33 0.71 0.29 C4 0.65 0.35 0.65 0.35 CP5 0.74 0.26 0.74 0.26 CP1 0.75 0.25 0.75 0.25 CP2 0.78 0.21 0.78 0.21 CP6 0.72 0.28 0.72 0.28 P7 0.65* 0.35 0.81 0.19 P3 0.78 0.22 0.78 0.22 Pz 0.66* 0.44 0.77 0.23 P8 0.74 0.26 0.74 0.26 P4 0.62* 0.38 0.83 0.17 PO3 0.76 0.24 0.76 0.24 O1 0.54* 0.46 0.73 0.27 PO4 0.45 0.55 0.45 0.55 O2 0.45* 0.55 0.71 0.29 Note. Starred (*) values indicate higher heritability in females than in males. 94 Table 5.3. Beta Power AE Estimates by Site in both Sexes Site A E Fp1 0.61 0.39 Fp2 0.56 0.44 Af3 0.64 0.36 Af4 0.63 0.37 F7 0.71 0.29 F3 0.69 0.31 Fz 0.65 0.35 F4 0.66 0.34 F8 0.63 0.37 FC5 0.65 0.35 FC1 0.72 0.28 FC6 0.67 0.33 FC2 0.66 0.34 T7 0.62 0.38 C3 0.64 0.36 T8 0.65 0.35 C4 0.68 0.32 CP5 0.68 0.32 CP1 0.66 0.34 CP2 0.69 0.31 CP6 0.65 0.35 P7 0.69 0.31 P3 0.64 0.36 Pz 0.67 0.33 P8 0.69 0.31 P4 0.69 0.31 PO3 0.70 0.30 O1 0.67 0.33 PO4 0.56 0.44 O2 0.62 0.38 Figure 5.1. Heritability of Male Full Brain Theta, Alpha, and Beta Power Heritability of Male Full Brain Theta, Alpha, and Beta Power 95 Figure 5.2. Heritability of Female Full Brain Theta, Alpha, and Beta Power 2. Heritability of Female Full Brain Theta, Alpha, and Beta Power 96 97 5.B. Additional Measures of Antisocial Behavior The purpose of these analyses is to investigate the generalizability of EEG findings presented in Chapter 3 through multilevel modeling with additional measures of ASB. For consistency, parental data are used from both waves, with the exception of the Barratt Impulsiveness Scale, which was available only in Wave 3, and the Delinquency Interview, which was available only in Wave 1. Both of these instruments are self-report. 5.B.1. Measures Diagnostic Interview Schedule for Children – Version IV (DISC-IV) – Conduct Disorder (CD) and Attention Deficit and Hyperactivity Disorder (ADHD) Modules. This clinical diagnostic interview measure was created by the National Institute of Mental Health in order to assess diagnostic criteria of different disorders in the Diagnostic and Statistical Manual, IV edition (DSM-IV; American Psychiatric Association, 1994). Within the interview, participants are asked to provide information about severity and frequency of symptoms over the past month, year, and lifetime. These analyses utilized past-year symptom count data of ADHD and CD. Parental data was available from both Waves 1 and 3. Child Aggressive Questionnaire (CAQ). The CAQ was developed as a 23-item questionnaire gauging reactive (12 items) and proactive (11 items) aggression in children (Raine, et al. 2006). Reactive aggression is measured as anger and violence that is a response to provocation, whereas proactive aggression consists of damage for fun, bullying, or other instrumental actions of aggression. Cronbach alphas ranged from 0.76-0.83 in both waves, suggesting good internal consistency. The structure of the CAQ has been validated in this sample (Baker, et al. 2008). Parental data were available from both Wave 1 and 3. 98 Child Psychopathy Scale (CPS). This 58-item instrument is a well-validated measure of psychopathic personality traits in children (Lynam, et al. 1997). The CPS was developed on the basis of the youth version (PCL-YV) of the Psychopathy Checklist-Revised interview (PCL-R), the gold standard psychopathy-assessment instrument (Hare, 1999). This yes-or-no scale is designed to measure psychopathic personality traits such as grandiosity, glibness, manipulation, untruthfulness, and lack of guilt. Wave 1 and 3 parental reports were used in these analyses. Barratt Impulsiveness Scale (BIS). This 30-item Likert-type questionnaire gauges impulsive behavior, and was adapted for adolescent participants in this study from the adult version of the survey (Patton, et al. 1995). This instrument is considered the gold standard of impulsiveness measures, and has high reliability (Cronbach alpha = 0.85 for total impulsivity in Wave 3). This is a self-report instrument, and is available from Wave 3. Delinquency Interview. This is a 71-item interview measure administered to the twins, which assessed truancy, minor violations, property damage, theft, and animal violence, and is generally considered to be a measure of nonaggressive ASB. It was adopted for the Twin Study from a measure used in the Pittsburgh Youth Study (Loeber & Farrington, 1998). It assesses lifetime prevalence of these activities, as well as prevalence over the previous year, in addition to whether or not the twin was alone, with a friend, or with his/her co-twin. The variable used in these analyses was number of Delinquency acts endorsed over the entire lifetime. This questionnaire was available from Wave 1 only. 5.B.2. Results Means, standard deviations, and Ns of additional ASB measures can be found in Table 5.4. Significant sex differences emerged for all variables except for BIS (t=0.35, df=511, p=0.72) 99 and Wave 3 reactive aggression (t=0.26, df=572, p=0.80) and proactive aggression (t=0.79, df=572, p=0.43). Males showed higher scores for all other measures. Table 5.4. Means, Standard Deviations, and Ns for Measures of Antisocial Behavior by Sex. Wave Measure Males Females 1 DISC ADHD 6.27(5.25) N=428 4.21(4.38) N=455 DISC CD 1.52(2.09) N=428 0.82(1.35) N=455 CPS 1.28(0.13) N=431 1.24 (0.11) N=462 CAQ-Reactive 1.70 (0.33) N=431 1.63 (0.32) N=463 CAQ-Proactive 1.11 (0.16) N=431 1.08 (0.13) N=463 Delinquency Interview 6.12 (6.05) N=430 4.70 (5.22) N = 461 3 DISC ADHD 3.96 (4.43) N=183 2.77 (3.74) N=236 DISC CD 4.40 (3.54) N=194 3.15 (3.28) N=242 CPS 1.55 (1.12) N = 262 1.32 (1.07) N = 300 CAQ-Reactive 1.56 (0.35) N =269 1.57 (0.34) n = 305 CAQ-Proactive 1.10 (0.16) N = 269 1.08 (0.16) N = 305 BIS 2.17 (0.38) N = 235 2.18(0.39) N = 278 Within wave, CBCL Nonaggressive ASB and Aggressive ASB were significantly correlated with each of the other ASB measures, ranging in Wave 1 from r=0.18 to 0.65 for the males, and 0.33 to 0.63 in the females. In Wave 1, the additional ASB measures correlated with one another between r=0.31 to 0.60. In Wave 3, correlations with CBCL Aggressive and nonaggressive ASB ranged from r=0.26 to 0.60, all P<0.05. Correlations among the additional measures were also all significant in Wave 3, ranging from r=0.16 to 0.58. Higher correlations occurred between analogous instruments (e.g. Wave 1 CBCL aggression and Wave 1 CAQ- Proactive Aggression r=0.65) than between instruments measuring different ASB aspects (e.g. Wave 1 CBCL Aggression and Wave 1 Delinquency Interview r=0.18). Multilevel regression analysis main effects are presented in Tables 5.5a-5.5d by sex and wave. Similarly to analyses with the CBCL, Wave 1 ASB was not significantly associated with 100 EEG spectral power, with the exception of a significant positive relationship between parietal theta and male proactive aggression (p<0.01). In Wave 3, positive relationships emerged in the females between frontal theta and both CPS and CD symptoms, which did not mirror any of the Chapter 3 results. In the males, positive relationships emerged between proactive aggression and frontal alpha, frontal beta, and parietal beta, as well as between parietal beta and reactive aggression. The positive relationships between spectral power and aggression are consistent with those that emerged using the CBCL. However, the results in the females are more difficult to interpret. Although CPS and CD were expected to follow the patterns of aggression, CPS is a measure of personality rather than of aggressive or nonaggressive ASB specifically, and CD comprises aspects of both aggressive and nonaggressive ASB, and so it may be that the original prediction was incorrect. These results provide some support for generalizability. 101 Tables 5.5a-5.5d. Multilevel Regression Coefficients (with Standard Error) for Frontal and Parietal EEG Spectral Power with Additional Measures of Antisocial Behavior, by Sex and Wave. 5.5a. Males Wave 1 Frontal alpha Frontal theta Frontal beta Parietal alpha Parietal theta Parietal beta DISC CD -6.86(12.15) p=.57 -.20(.17) p=0.21 -0.64(1.21) p=0.60 0.08(0.34) p=0.81 -40.76(38.45) p=.29 -19.57 (23.46) p=.40 DISC ADHD 8.09(30.96) p=.79 -0.37(0.43) p=0.39 0.13(0.31) p=0.67 -0.22(0.23) p=0.34 -80.20(97.69) p=.41 -12.31(60.19) p=.83 CPS -1.61(7.07) p=0.82 -0.95(9.89) p=0.92 -0.56(7.00) p=0.94 -5.35(14.64) p=0.72 -7.05(22.31) p=0.75 -1.89(13.75) p=0.89 AGG- Reactive -0.30(0.19) p=0.12 0.45(0.27) p=0.09 -0.22(0.19) p=0.26 0.25(0.38) p=0.51 0.39(0.61) p=0.52 -0.21(0.37) p=0.57 AGG- proactive -0.53(0.92) p=0.56 -2.14(01.28) p=0.10 -0.09(0.91) p=0.92 0.53(0.19) p=0.005* -0.67(2.88) p=0.82 0.34(1.78) p=0.85 Del-Int -0.13(0.35) p=0.71 0.02 (0.49) p=0.97 -0.06 (0.35) p=0.86 -0.59 (0.72) p=0.41 0.76 (1.11) p=0.50 -0.30 (0.69) p=0.67 5.5b. Females Wave 1 Frontal alpha Frontal theta Frontal beta Parietal alpha Parietal theta Parietal beta DISC CD -2.09(6.16) p=0.73 0.77(0.83) p=0.36 0.06(0.61) p=0.93 -0.65(0.62) p=0.31 -11.95(10.68) p=0.26 -8.23(9.34) p=0.38 DISC ADHD 11.27(20.83) p=0.59 1.64(2.93) p=0.57 2.45(2.04) p=0.22 0.11(0.09) p=0.22 9.11(37.72) p=0.81 1.37 (32.34) p=0.97 CPS 4.72(5.02) p=0.34 -1.02(7.03) p=0.88 5.12 (4.94) p=0.30 -2.86(8.16) p=0.73 -7.21(9.02) p=0.42 6.22(7.75) p=0.42 102 AGG-Reactive -0.08(1.51) p=0.96 -0.29(2.10) p=0.89 0.44(1.49) p=0.77 -2.22(2.41) p=0.36 -1.67(2.70) p=0.54 0.23(2.33) p=0.92 AGG- proactive 0.31(0.59) p=0.60 0.81(0.83) p=0.33 0.59(0.58) p=0.31 -0.58(0.97) p=0.55 0.09(1.07) p=0.93 0.11(0.92) p=0.91 Del-Int 0.28 (0.25) p=0.25 0.07 (0.35) p=0.83 0.26(0.24) p=0.28 -0.14 (0.40) p=0.73 0.09 (0.45) p=0.84 0.12 (0.38) p=0.76 5.5c. Males Wave 3 Frontal alpha Frontal theta Frontal beta Parietal alpha- -0.20 (0.77) p=0.80 Parietal beta DISC CD 0.23 (0.18) p=0.21 -0.06 (0.22) p=0.77 0.29 (0.18) p=0.11 0.27(0.20) p=0.17 -0.12(0.15) p=0.40 0.23 (0.32) p=0.47 DISC ADHD 30.00(40.14) p=0.46 8.19(5.47) p=0.14 5.12(4.11) p=0.21 -0.33(0.58) p=0.57 23.93 (18.88) p=0.20 9.94(7.39) p=0.18 CPS 0.04(0.87) p=0.96 -0.43(1.01) p=0.66 0.06(8.6) p=0.99 -0.61(1.19) p=0.61 -1.95(3.61) p=0.59 0.08 (0.30) P=0.69 AGG- Reactive 0.32(0.25) p=0.22 -0.15(0.35) p=0.67 0.41(0.27) p=0.12 -0.85(4.56) p=0.85 0.98(1.26) p=0.44 1.08(0.49) p=0.03* AGG- proactive 0.24(0.12) p=0.04* 0.15(0.16) p=0.35 0.34(0.12) p=0.005* 0.03(0.20) p=0.86 0.89(0.57) p=0.12 0.54(0.22) p=0.02* BIS 0.65(3.24) p=0.84 -4.91(4.04) p=0.23 -0.16(3.22) p=0.96 -3.47(5.14) p=0.50 -2.82(1.47) p=0.06 -1.12(5.73) p=0.83 5.5d. Females Wave 3 Frontal alpha Frontal theta Frontal beta Parietal alpha Parietal theta Parietal beta DISC CD 0.14(0.12) p=0.26 0.48 (0.19) p=0.02* 0.08 (0.14) p=0.58 0.24 (0.29) p=0.41 0.13 (0.28) p=0.65 0.18 (0.18) p=0.33 DISC ADHD 1.49(23.56) p=0.95 -.78(3.66) p=0.83 2.89(4.24) p=0.50 -0.17 (0.64) p=0.79 -27.06(45.33) p=0.55 -9.53(36.00) p=0.79 CPS -0.64(0.58) p=0.27 1.75(0.84) p=0.04* -0.44(0.60) p=0.46 -0.10(1.40) p=0.94 0.63(1.23) p=0.61 -0.11(0.83) p=0.89 AGG- 0.06(0.18) 0.11(0.29) -0.05(1.84) 0.38(0.52) p=0.46 0.18(0.40) p=0.65 0.25(0.28) 103 Reactive p=0.73 p=0.70 p=0.98 p=0.37 AGG- proactive 0.59(0.89) p=0.51 0.26(0.14) p=0.06 0.08(0.08) p=0.32 3.03(3.63) p=0.25 0.12(0.19) p=0.52 0.15(0.14) p=0.26 BIS -0.24(2.39) p=0.92 -2.33(3.44) p=0.50 -0.97(2.39) p=0.68 -3.48(6.51) p=0.59 1.12(4.62) p=0.81 0.91(3.59) p=0.80 104 5.C. Environmental Moderators in Relationships between Spectral Power and Antisocial Behavior These analyses were conducted to investigate potential interactions between spectral power and adverse environmental conditions in relation to ASB. 5.C.1. Background Adverse environmental conditions are known to interact with genetic vulnerability in influencing antisocial behavior problems (Tuvblad & Baker, 2011). Given genetic vulnerability, for example, it has been shown that exposure to violence and other adverse home conditions are strong predictors of whether antisocial behavior will develop, both in behavioral and molecular genetic studies (Cadoret, et al. 1983; Mednick, et al. 1984; Caspi, et al. 2002; Roth & Sweatt 2011). It is the aim of the present research to extend this investigation to the question of neurobiological vulnerability, as well. One reason to examine neural pathways is that maltreatment has been found to affect the child and adolescent brain. For example, neglected children showed reduced corpus callosum volume (Teicher, at al. 2004), and abused children showed reduced cortical volumes (Teicher, et al. 1997). In addition, life stress affects hippocampal development in children (Andersen, et al. 2004). This is consistent with broader findings that stress is harmful to the brain (Sapolsky, 1996). In the present research, environmental conditions previously found to relate to antisocial behavior – socioeconomic status, stressful life events, peer victimization, and home neighborhood quality – have been selected, and were examined for potential interactions with EEG patterns in predicting aggressive and nonaggressive ASB. 105 5.C.2. Measures Socioeconomic Status (SES) Family SES was determined in each wave using caregiver interviews, based on the Hollingshead’s Four-Factor Index (Hollingshead, 1975). This index assesses parental educational levels and occupations, as well as family income. SES is rated on a scale of 8-66. In this sample, mean SES was 42.29 (SD = 12.07; Skewness = -0.18), suggesting slightly higher SES than in the general population and a normal distribution. Victimization Index The Child Friendship Questionnaire, adapted for this study from an established measure for child peer-interaction (Bierman & McCauley, 1987), is a self-report measure consisting of questions about a child’s social experiences with peers, friends, and classmates. In this questionnaire, there are eight questions that gauge victimization, both nonviolent (e.g. how often does someone at school tease you or make fun of you?) and violent (e.g. how often does someone at school hit, push, or start fights with you). This scale uses a 5-point Likert scale (Never, rarely, sometimes, often, almost always), and shows high internal consistency (Cronbach alpha were 0.85 in Wave 1 and 0.81 in Wave 3). Stressful Life Events This scale examined life events over the course of the past year, examining events ranging in severity from difficulty with homework to moving houses to the death of a relative, and uses a 5-point Likert-scale to measure frequency of the experience. In Waves 3 several age-appropriate items were revised into the LSE, such as events concerning dating, body image, or substance use. Cronbach alphas were 0.72 in Wave 1 and 0.83 in Wave 3. Neighborhood Quality 106 The scale consisted of 17 questions answered by caregivers concerning frequency of crime, gang warfare, drug use, and other adverse circumstances in the neighborhoods in which the families live. This measure was developed in the Pittsburg Youth Study (Loeber, et al. 1998). Cronbach alpha was high in both waves (0.94). Statistical Analyses Multilevel modeling was used to perform regression analysis in these samples using Statistical Analysis Software (SAS 9.1.3, 2005). Each regression model was performed for 4 outcomes variables: Waves 1 and 3 CBCL Nonaggressive ASB and Aggressive ASB scores. Moderation was tested in these models using interaction variables of EEG measures and adversity measures. 5.C.3. Results Tables 5.6 and 5.7 present descriptive statistics and phenotypic correlations among environmental variables. Adverse environmental variables were all significantly correlated with one another within wave except for Wave 3 neighborhood quality and peer victimization. SES was significantly negatively correlated with the three other variables; all other variables were significantly positively correlated at value ranges of 0.10<r<0.42. Several correlations between adverse conditions in Wave 1 and EEG power values were found significant at the alpha=0.05 level, but when a Bonferroni correction was applied, these relationships were no longer considered significant, suggesting that these adverse conditions did not influence brain activation patterns at the time of data collection. Significant sex difference emerged for Wave 3 stressful life events (t(385) = -2.25, p=.0.03) and Wave 3 victimization (t(449) = 3.28, p<0.01), with females reporting more stressful events and males reporting more peer victimization. 107 Table 5.6. Means, Standard Deviations, and Ns for Environmental Variables Used in Analyses by Sex. Males Females Socioeconomic status Wave 1 42.18 (12.06) N = 423 42.37 (12.24) N = 457 Peer Victimization Wave 1 1.88 (0.87) N = 404 1.89 (0.88) N = 444 Stressful Life Events Wave 1 1.45 (0.43) N = 419 1.53 (0.43) N = 455 Neighborhood Wave 3 1.60 (11.24) N = 196 1.67 (0.69) N = 246 Socioeconomic status Wave 3 40.80 (11.25) N = 199 42.84 (10.87) N = 244 Peer Victimization Wave 3 1.48 (0.51) N = 198 1.34 (0.38) N = 253 Stressful Life Events Wave 3 0.61 (0.73) N = 163 0.79 (0.80) N = 224 Two interactions between EEG power variables and adverse environmental conditions arose in explaining male antisocial behavior, presented in Table 5.8. (This table does not present SD and P-values because of the number of models examined. Starred values represent models with P<0.025, to correct for multiple comparisons). No significant interactions arose for the females. In Wave 3, male Aggressive ASB was significantly predicted by an interaction between neighborhood quality and both frontal alpha power (Main effect of frontal alpha power: -2.69 (SE=1.57), p=0.09; main effect of neighborhood: -0.47 (SE=1.27), p=0.71; interaction effect: 2.18 (SE=0.82), p<0.01); and frontal beta power (Main effect of frontal beta power:-2.29 (SE=1.60), p=0.15; main effect of neighborhood: -1.07 (SE=1.49), p=0.47; interaction effect: 2.14 (SE=0.85), p=0.01). Table 5.7. Correlations between Aggressive and Nonaggressive ASB and Adverse Environmental Conditions in Waves 1 and 3 by Sex. Aggressive ASB Nonaggressive ASB Males Females Males Females Wave 1 Nbd 0.09 0.16** 0.15** 0.14** SES -0.13** -0.17** -0.13** -0.11* SLE 0.17** 0.17** 0.24** 0.06 Vict 0.14** 0.12* 0.15** 0.01 Wave 3 Nbd 0.16* 0.18** 0.32** 0.15* SES -0.12 -0.18** -0.21** -0.21** SLE 0.21** 0.29** 0.26** 0.34** 108 Vict 0.03 0.10 0.08 0.05 Note. Nbd = Neighborhood Quality; SES = Socioeconomic Status; SLE = Stressful Life Events; Vict = Peer victimization; * = p<0.05; **=p<0.01 109 Table 5.8. Unstandardized Regression Coefficients of Interaction Terms from Multilevel Models Wave Frontal alpha Frontal theta Frontal beta Parietal alpha Parietal theta Parietal beta 1 Females AGG NBD SES SLE Vict 0.31 -0.02 -0.07 -0.23 0.69 -.02 0.81 -.37 0.10 -0.01 0.05 -0.26 1.01 .01 -.86 -.97 2.22 0.04 2.10 -0.80 2.21 0.04 2.10 -0.80 NA NBD SES SLE Vict 0.13 -0.002 -0.10 -0.03 0.01 0.01 -0.19 -0.06 0.09 -0.001 0.02 0.02 0.17 -0.001 -0.19 -0.02 0.35 -0.001 -0.99 -0.15 0.35 -0.001 -0.99 -0.15 Males AGG NBD SES SLE Vict 0.69 0.01 0.13 0.09 0.70 -0.05 0.28 -0.09 0.53 0.03 0.02 0.12 -4.27 0.11 -0.09 -0.32 -1.88 -0.03 -0.36 -1.03 -1.88 -0.03 -0.36 -1.03 NA NBD SES SLE Vict 0.16 0.009 -0.36 0.02 0.28 -0.01 -0.26 .01 0.07 0.02 -0.32 0.01 -1.67 -0.03 -0.60 -0.38 -0.75 -0.02 -0.32 -0.40 -0.75 -0.02 -0.32 -0.40 3 Females AGG NBD SES SLE Vict -0.60 0.05 0.004 -0.69 -0.50 0.03 0.05 0.35 -0.92 0.04 0.004 -0.61 0.96 -0.09 -0.04 -4.41 -0.31 0.07 0.01 -1.45 -0.28 0.02 -0.03 -1.38 NA NBD SES SLE Vict -0.12 -0.01 0.002 -0.04 0.03 -0.0002 0.01 0.18 -.010 -0.01 0.005 0.01 0.38 -0.06 -0.02 -1.08 0.36 -0.01 -0.001 -0.21 -0.04 -0.02 -0.01 0.08 Males AGG NBD SES SLE Vict 2.10* -0.04 0.04 -0.43 0.96 -0.08 0.01 1.77 2.07* -0.03 0.03 -0.83 1.17 -0.08 -0.08 6.23 -5.70 -0.25 -0.26 6.82 0.20 0.02 -0.04 0.27 NA NBD SES SLE Vict -0.02 0.01 -0.002 -.010 -0.33 0.01 0.02 -0.15 -0.16 0.02 -0.01 -0.12 -1.47 -0.04 -0.18 0.51 -1.59 -0.04 -0.05 0.54 -0.81 0.03 -0.03 0.05 Note. AGG = Aggressive ASB. NA = Nonaggressive ASB. NBD = Neighborhood. SES = Socioeconomic Status. SLE = Stressful Life Events. Vict = Peer Victimization. Starred terms are significant at P<0.05. 110 5D. Group Analysis based on Nonaggressive ASB Levels across Waves 1-4 The purpose of these analyses is to use groups derived in latent class growth analysis methodology (Wang, et al. in preparation) in order to discern whether different trajectories of nonaggressive ASB behavior show different profiles of EEG spectral power. Although Chapters 3-4 found relationships with aggressive rather than nonaggressive ASB, group analysis of aggressive ASB trajectories were not readily available for this analysis. 5.D.A. Methods This analysis was based on four waves of CBCL nonaggressive ASB (Delinquency Scale) data using a randomly selected twin from each pair (N=767). Latent class growth analysis (LCGA) was used in the statistical program Mplus using these data to investigate how many different trajectories best explain the pattern of results obtained. Variables were specified as count variables, which are assumed to have a Poisson rather than a normal distribution, because they are highly skewed (that is, nonaggressive ASB is not a common behavior, and many children report little or no nonaggressive ASB across the four waves of data collection). Five potential models were explored, ranging from 2-6 groups. Models were compared using Bayesian Information Criteria (BIC), with the lowest BIC representing the best-fitting model, which emerged as five groups. Latent basis model representing a non-linear change pattern was fitted to each group. This model can be written in equation: Y ij = g0 + g1*α j Where i= 1…765; j=1, 2, 3, 4; In this model, α 1 and α 4 are fixed at values 0 and 3 respectively while α 2 and α 3 are freely estimated and set to be equal across each group. The obtained five groups thus only differed on g0 and g1. 111 The expected values at each time point within each group were thus calculated and displayed in Table 5.9. Examination of these five groups’ mean scores at each wave led to the following classification: 1. Low-nonaggressive ASB (N=276); 2. Nonaggressive ASB-persistent (N=23); 3. Childhood-limited (N=67); 4. Non-offenders (N=343); and 5. Adolescent-onset (N=58). Of these individuals, 353 also provided valid Wave 1 EEG data. Using this group variable, analyses of variance (ANOVAs) were performed to check for group differences with the six EEG spectral power variables (frontal and parietal alpha, theta, and beta) as outcomes. First, group and gender were entered as factors simultaneously. If interaction between gender and group was significant, then each gender was examined in a separate ANOVA with only group as an examined factor. If the interaction was not significant, then the main effect of group was examined for significance. Pairwise comparisons between the different groups were conducted between the Non- offender and Adolescent-Onset groups, as well as the Non-offender and Nonaggressive ASB-persistent groups. In order to avoid the problem of spurious significance due to multiple comparisons, only theoretically-driven pairwise comparisons were examined. Table 5.9. Sample Size and Means of Latent Class Growth Analysis Groups for each Wave Group Wave 1 Wave 2 Wave 3 Wave 4 N 1 Low-nonaggressive ASB 0.89 1.13 1.78 1.75 276 2 Nonaggressive ASB-persistent 4.87 5.9 8.52 8.38 23 3 Childhood-limited 2.66 2.13 1.39 1.42 67 4 Non-offenders 0.37 0.31 0.22 0.22 343 5 Adolescent-onset 1.86 2.59 4.87 4.74 58 5.D.2. Results Table 5.10 presents Univariate ANOVA results from these analyses. Frontal alpha, frontal beta, and parietal beta showed significant group by sex interactions, and hence were examined separately by 112 sex. Frontal theta showed no significant interaction of group and sex, but a significant group main effect. Parietal alpha and theta showed no significant group differences. Main effect of group was also significant in the males for frontal alpha and frontal beta power, as well as in females for parietal beta power. For main effects that emerged as significant, pairwise comparisons were conducted between groups hypothesized to differ: Non-offenders with both Adolescent-onset and Nonaggressive ASB- persistent. For frontal theta, which showed a main effect without sex interaction, no significant pairwise comparisons emerged. For frontal alpha and frontal beta, comparisons for differences between Non- offenders and Adolescent-onset or Nonaggressive ASB-persistent were also non-significant. In the females for parietal beta, significant differences arose between the Non-offender (mean=0.84 E-10 V/s, SD = 0.51) and the Adolescent-onset groups (mean=0.59 E-10 V/s, SD=0.25) (t=3.47, df=51 P<0.01). 113 Table 5.10. ANOVA Results for Nonaggressive ASB Group and EEG Spectral Power EEG Variable Group by sex interaction Full sample group main effect Male group main effect Female group main effect Frontal alpha F=3.96, P<0.01* -- F=2.97, P=0.02* F=2.15, P=0.08 Frontal theta F=0.64, P=0.63 F=2.94, P=0.02* -- -- Frontal beta F=3.59, P<0.01* -- F=3.72, P<0.01* F=1.73, P=0.14 Parietal alpha F=2.26, P=0.06 F=1.27, P=0.28 -- -- Parietal theta F=0.79, P=0.53 F=1.36, P=0.25 -- -- Parietal beta F=4.04, P<0.01* -- F=2.23, P=0.07 F=2.88, P=0.02* 114 5.E. Parental Antisocial Behavior The purpose of these analyses is to investigate whether children of parents who have engaged in criminal activity show different EEG power spectral profiles than children whose parents have no criminal background. 5.E.1. Background Antisocial behavior is known to be highly heritable, as demonstrated through behavioral and molecular genetic studies both, and as is highlighted in Chapter 2 and 4. Another research method sometimes explored to examine genetic influence of a phenotype is a genetic high-risk design. In these designs, high-risk subjects are identified as those who are offspring of individuals with the phenotype of interest. Examples of this already exist in the field of alcoholism. Parental alcoholism has been found to predict higher frontal and parietal EEG beta power in young adult sons (Ehlers & Schuckit, 1990 & 1991), as well as lower occipital and frontal alpha power (Finn, et al. 1999). Additionally, reduced amplitudes of the P300 response, an EEG event-related potential measure of orientation, have been found in children of individuals with antisocial personality disorder (O’Conner, et al. 1994). 5.E.2. Measures. Maternal ASB was assessed through interviewing with the caregiver who accompanied the twins to the laboratory (883 biological mothers – other caregivers were excluded from these analyses for consistency). Questions were asked about the other biological parent (Paternal), as 115 well. The primary caregiver answered 31 questions about her own lifetime ASB. These questions (Maternal) included 7 items assessing lifetime aggressive ASB (Cronbach alpha=0.48), such as threatening, pushing, verbally assaulting, and weapon usage, as well as 24 items assessing nonaggressive ASB (Cronbach alpha=0.78), such as lying, property damage, using someone else’s credit card or checkbook, and petty theft. The Paternal ASB questionnaire, also answered by the primary caregiver accompanying the twins, consisted of questions about paternal behavior before and after the age of 18. The Paternal questionnaire included 4 aggressive ASB items (Cronbach alpha=0.59) and 8 nonaggressive ASB items (Cronbach alpha=0.77). Each positive answer from the Maternal questionnaire after the age of 18 was tallied as one instance of that behavior. Responses were tallied (that is, it was taken into account only whether a behavior occurred after the age of 18, not how frequently it occurred), and measures of total ASB, aggressive ASB, and nonaggressive ASB were formulated for Maternal and Paternal count based on these tallies exclusively for behavior after the age of 18 years. A second analysis examined only whether either parent had been arrested past the age of 18 years, creating three groups: children whose parents had never been arrested (N=522), children who had one parent arrested (N=271), and children who had both parents arrested (N=48), with 59 children not providing data for either or both parents. Only parents identified as biological mother for Maternal and biological father for Paternal were entered into these analyses, for consistency. This excluded 50 reports for the “Maternal” questionnaire and 30 reports for the “Paternal” questionnaire of caregivers whose relationships to the twins were adoptive, step, grand-parental, or otherwise. 5.E.3. Results The mean Maternal aggressive ASB was 0.16 (SD=0.60, skewness = 4.05, kurtosis =16.69), and Paternal aggressive ASB was 0.17 (SD=0.52, skewness = 4.01, kurtosis = 19.86). 116 The mean Maternal nonaggressive ASB was 0.53 (SD=1.18, skewness =3.15, kurtosis=11.59), and Paternal nonaggressive ASB was 2.21 (SD=2.70, skewness =1.85, kurtosis=4.08). Univariate analyses of variance (ANOVAs) revealed no significant main effects of the arrest report variable for any of the EEG variables, as well as no significant interactions between sex of the twin and parental arrest reports. Maternal ASB count did not significantly correlate with EEG measures for either sex, whereas Paternal ASB did for both aggressive and nonaggressive ASB. For the Paternal count, aggressive ASB was correlated with frontal beta r=0.10 (p<0.01), with frontal alpha r=0.07 (p=0.04), and with parietal beta r=0.08 (p=0.02). Nonaggressive Paternal ASB was correlated with frontal beta r=0.08 (p=0.02), with frontal alpha r=0.08 (p=0.01), and parietal beta r=0.08 (p=0.01). When twin sexes were examined separately, relationships were not significant for males, but remained significant for females: Paternal aggressive ASB was correlated with frontal beta r=0.17 (p<0.01), with frontal alpha r=0.13 (p<0.01), and with parietal beta r=0.08 (p=0.02). Paternal Nonaggressive was correlated with frontal beta r=0.10 (p=0.04), with frontal alpha r=0.10 (p=0.03), and parietal beta r=0.12 (p=0.02). Table 5.11 presents results of multilevel regression model coefficients of untransformed parental ASB in association with twin EEG spectral power. It emerged that female spectral power (frontal alpha, frontal beta, parietal beta) was significantly associated with Paternal aggressive ASB. The relationships were positive, and mirrored the spectral properties that positively associated with male twin aggressive behavior in Chapter 3. No other significant relationships emerged. 117 Table 5.11. Multilevel Regression Coefficients (with Standard Error) for Parental ASB Measures with Frontal and Parietal EEG Spectral Power, by Sex and Wave. Twin sex Parental measure Frontal alpha Frontal theta Frontal beta Parietal alpha Parietal theta Parietal beta Male Maternal Agg 0.09 (0.09) p=0.31 0.03 (0.06) p=0.63 0.04 (0.09) p=0.68 0.04 (0.03) p=0.28 -0.0005 (0.03) p=0.99 -0.02 (0.04) p=0.61 Maternal Non- Agg 0.01 (0.02) p=0.43 0.004 (0.01) p=0.74 0.01 (0.02) p=0.52 0.01 (0.01) p=0.12 -0.004 (0.01) p=0.50 0.003 (0.009) p=0.78 Paternal Agg -0.03 (0.08) p=0.68 -0.04 (0.06) p=0.48 0.01 (0.09) p=0.92 0.01 (0.03) p=0.80 0.001 (0.03) p=0.97 0.01 (0.04) p=0.84 Paternal Non- Agg 0.03 (0.04) p=0.44 -0.02 (0.03) p=0.39 0.04 (0.04) p=0.37 -0.005 (0.02) p=0.77 0.001 (0.01) p=0.92 0.02 (0.02) p=0.30 Female Maternal Agg -0.03 (0.11) p=0.78 -0.001 (0.07) p=0.99 0.04 (0.11) p=0.75 -0.03 (0.05) p=0.57 -0.05(0.05) p=0.35 -0.02 (0.07) p=0.76 Maternal Non- Agg 0.001 (0.02) p=0.98 0.002 (0.01) p=0.85 0.01 (0.02) p=0.75 -0.002 (0.01) p=0.81 0.001 (0.01) p=0.94 -0.002 (0.01) p=0.89 Paternal Agg 0.18 (0.08) p=0.02* 0.04 (0.05) p=0.42 0.23 (0.08) p=0.006* 0.01 (0.04) p=0.78 0.01 (0.04) p=0.89 0.10 (0.05) p=0.04* Paternal Non- Agg 0.07 (0.04) p=0.10 -0.01 (0.03) p=0.81 0.06(0.04) p=0.13 0.01 (0.02) p=0.48 0.04 (0.02) p=0.09 0.05 (0.03) p=0.07 118 5.F. Frontal Alpha Asymmetry and Antisocial Behavior The purpose of these analyses was to examine an initial hypothesis set forth: that frontal alpha asymmetry with higher relative left alpha power would predict aggressive behavior in children. 5.F.1. Background Frontal alpha asymmetry is considered a marker of affect dysregulation (Fox, 1994), with relatively higher right frontal activation (right asymmetry) predicting more negative affect and poorer emotional regulation (Davidson, et al. 1992, 1998). However, there have been mixed findings with regard to ASB. Relative left frontal asymmetry has been found to predict externalizing behavior problems, a finding that largely emerged through studies of anger. Some evidence suggests that left frontal activity actually generates aggressive impulses, for example because transcranial magnetic stimulation of the left prefrontal region increases attention to angry face (d’Alfonso, et al. 2000) and because aggressive responses to insults increased after a contraction of the right hand, which is sensed in the left hemisphere (Peterson, et al. 2008). Another theory is that right frontal activity corresponds to affective regulation, inhibiting aggressive tendencies (Raine, 1997; Anderson, et al. 1999). Evidence arose that both trait anger (Harmon-Jones, 2000; Harmon-Jones & Allen, 1998), and state anger (Peterson, et al. 2008) correlate with left lateral frontal activation. The relationship between state anger and left frontal asymmetry was found as early as infancy (Dawson, et al. 1992; Fox & Davidson, 1988). On the other hand, some findings suggest a role of right frontal asymmetry in ASB, instead. While boys who were diagnosed with oppositional defiant disorder were found to show higher left frontal activation, girls were found to show higher right frontal activation (Baving, et 119 al. 2000). Also, preschool-aged children of depressed mothers showed correlations between relatively higher right frontal activation and decreased empathy (Jones, et al. 2000). Left asymmetry was found to predict lower levels of anger control, as well as higher levels of outward-direction-motivated anger, both representative of trait anger-proneness (Hewig, et al. 2004). This highly powered study (>0.8) found only moderate effect sizes, for example R 2 = 0.05 for laterality x region x anger. Correlations between left frontal asymmetry and anger were estimated at r=0.43 in middle school children for trait anger (Harmon-Jones, et al. 1998), and r =0.35-0.40 for state anger (Peterson, et al. 2008). From these past findings, it was hypothesized that left frontal asymmetry would relate to aggressive, not nonaggressive ASB, ASB. 5.F.2. Variables Several asymmetry variables were explored in the course of these analyses. First, simple difference scores were created using methods previously published in frontal alpha asymmetry research. Simple medial frontal alpha asymmetry was calculated as the natural log of F4 alpha power subtracted from the natural log of F3 (left minus right). Similarly, simple lateral frontal alpha asymmetry was calculated as LnF7-LnF8. These are consistent with past work (Forbes, et al. 2006; Harmon-Jones & Allen, 1998; Peterson, et al. 2008). Another established measure calculated is (F4-F3)/(F4+F3) (Fox, 1994). Lastly, PCA methods were utilized. PCA were conducted on the right prefrontal region (F8, F4, AF4, Fp2), and left frontal region (F3, F7, AF3, Fp1). One measure was derived that subtracts the first principal component of the right prefrontal region from the left. A second measure subtracted the second component of the right from the second component of the left. The reasoning behind this second PCA measure is the potential that the first component represents an overall prefrontal activity that is common to both hemispheres. 120 5.F.3. Results Table 5.12 presents multilevel modeling unstandardized regression coefficients results (with standard error) for aggressive and nonaggressive ASB scales of the CBCL. The expected direction of association was positive for all asymmetry measures except for Fox. As is evident, frontal alpha asymmetry in any derived formulation did not significantly predict antisocial behavior except for simple lateral asymmetry in Wave 1 females’ scores of nonaggressive ASB. is consistent with past findings that lateral right frontal asymmetry predicts female Oppositional Defiant Disorder (Baving, et al. 2000). However, when accounting for multiple comparisons and adopting an alpha level of p=0.025, this results is no longer consistent. 121 Table 5.12. Multilevel Regression Coefficients (with Standard Error) for Measures of Asymmetry and Antisocial Behavior Wave Sex Measure PCA-1 PCA-2 F3-F4(ln) F7-F8(ln) Fox (F4- F3)/(F4+F3) 1 Male Agg -1.76(1.29), p=0.18 1.34(0.90) p=0.14 -7.21(4.96) p=0.15 0.50(4.50) p=0.91 2.66(2.22) p=0.23 Del -0.03(0.48) p=0.95 0.98(3.30) p=0.77 -2.42(1.80) p=0.18 2.47(1.63) p=0.13 0.96(0.80) p=0.23 Female Agg 2.31(1.72) p=0.18 -0.54(0.81) p=0.51 -2.92(4.29) p=0.50 6.52(4.22) p=0.12 0.58(1.82) p=0.75 Del -0.02 (0.46) p=0.97 0.68(2.23) p=0.76 -1.02(1.13) p=0.37 2.32 (1.11) p=0.04 0.28(0.47) p=0.55 3 Male Agg -0.38(2.18) p=0.86 1.71(1.22) p=0.16 5.05(7.36) p=0.49 -4.81(6.25) p=0.45 -1.85(3.15) p=0.55 Del 0.18(0.92) p=0.85 0.37(0.51) p=0.46 0.57(3.13) p=0.86 0.45(2.69) p=0.87 0.49(1.34) p=0.71 Female Agg 1.92 (1.94) p=0.32 0.35(0.98) p=0.72 -4.50(5.15) p=0.38 3.94(5.07)p=0.44 2.62(2.11) p=0.21 Del 0.56 (0.82) p=0.49 0.52(0.42) p=0.22 -1.87(2.27) p=0.41 1.59(2.24) p=0.48 1.05(0.90) p=0.24 122 5. Summary This chapter explores several additional concerns raised by analyses in Chapters 2-4, as well as questions initially proposed for this project, which did not fit into the preceding chapters. The validity of the EEG data was explored through heritability analysis of full brain theta, alpha, and beta frequencies. The results were in line with past findings of full brain heritability, both in children and in adults, and suggest that the EEG data collected in this sample are valid and not random. Generalizability of findings and appropriateness of the CBCL instrument were also explored. First, additional measures of ASB were tested for relationships with EEG variables. Phenotypically, these other measures all showed higher male than female ASB, consistent with CBCL results. Similarly to results with CBCL, relationships were not significant in Wave 1, except for parietal alpha power in males, which positively predicted Wave 1 proactive aggression. The results in the males emerged as more consistent with CBCL results. Wave 3 proactive aggression showed similar significant positive relationships with frontal alpha, frontal beta, and parietal beta power, and Wave 3 reactive aggression showed positive relationship with parietal beta power. In the females, results were not fully consistent with CBCL findings. The only significant results in the females using CBCL data emerged with Wave 3 nonaggressive ASB data, which was not directly available in Wave 3, and thus could not be replicated precisely. However, conduct disorder DISC symptoms, arguably the closest measure to CBCL Nonaggressive ASB, showed neither of the significant negative relationships that emerged using CBCL data: frontal alpha, and parietal beta power. Both DISC CD and CPS showed positive significant relationships with frontal theta, which is more consistent with slow arousal theory. 123 Environmental moderators were also explored, to investigate whether patterns in this data did not replicate past findings because of differences in samples. While it is not possible to draw resounding conclusions, several interesting patterns should be noted. Males showed two interactions between power and neighborhood quality, and females showed none. Future examinations may focus on male EEG and ASB relationships exclusively. Neighborhood quality was found to interact with frontal alpha and beta power in males, suggesting that males with higher EEG power and more adverse neighborhoods were more likely to behave aggressively. However, this pattern of results did not lend strong support to the hypothesis that differences in levels of adversity explains why past work with criminal/clinical samples showed stronger association between spectral power and ASB than did the present findings. Generalizability of the longitudinal finding that spectral power predicted Wave 3 but not Wave 1 ASB was also explored using previously derived group analysis based on CBCL nonaggressive ASB data. Analyses of variance suggested that the groups did show different profiles for four of the six EEG power measures: frontal alpha, frontal theta, frontal beta, and parietal beta. For frontal theta, sex was not a significant factor, and group effect was significant. Sex differences emerged for frontal alpha, frontal beta, and parietal beta. Because no significant associations arose in Chapter 3 between spectral power and nonaggressive ASB, no specific hypotheses were formed for these analyses. Hence, while the omnibus group differences suggest that these group analyses may be an avenue for future research, it is not clear how these results relate to Chapter 3. It should be noted that all of these groups were derived using Wave 2 and 4 data in addition to Wave 1 and 3, and so patterns differ between these examinations and those conducted in Chapter 3. 124 To provide further support for genetic relationships between spectral power and ASB, parental ASB was examined as a predictor of twin EEG. It was found that paternal ASB, but not maternal, was slightly but significantly predictive of female EEG profiles. Paternal aggressive ASB was predictive of higher EEG spectral power in female frontal alpha, frontal beta, and parietal beta, mimicking the patterns observed in male twins with higher ASB in Chapter 3. Lastly, frontal alpha asymmetry was also explored, as it was initially proposed to be related to aggressive ASB but not nonaggressive ASB. Results suggested that frontal alpha asymmetry is not related to ASB, aggressive or nonaggressive ASB, in our sample. 125 Chapter 6. General Discussion There is great interest in understanding the developmental trajectories and mechanisms of antisocial behavior (ASB), two topics to which this dissertation aims to contribute. In three main studies and several additional analyses, this research investigated biological markers and heritability pertaining to the constructs of aggressive and nonaggressive ASB antisocial behavior in children and adolescents. This section provides a summary of findings, interpretations and implication, limitations of the study, and directions for future research. Summary of Findings Chapter 2 • Aggressive and nonaggressive ASB are jointly influenced by a common latent factor in Wave 1 and Wave 3, which is influenced by genetics, shared environment and nonshared environment. • Longitudinal stability of the latent common factor was due to all three forms of influence, with new genetic influences emerging in Wave 3. • At each time point, both aggressive and nonaggressive ASB were also influenced by unique genetic and nonshared environmental influence. • Males and females showed equal estimates in these analyses. Chapter 3 • No significant relationships emerged between Wave 1 EEG spectral power and Wave 1 aggressive ASB or nonaggressive ASB. 126 • Higher Wave 1 frontal alpha power, frontal beta power, and parietal beta power was associated with Wave 3 male aggressive ASB. • No significant associations emerged in the females or with nonaggressive ASB. Chapter4 • In males, Wave 1 frontal alpha power showed 78% influence of genetics and 22% influences of nonshared environment, and Wave 3 aggressive ASB showed 67% influence of genetics and 33% influence of nonshared environment. The genetic correlation between the two was significant at rg=0.22. Chapter 5 • Alpha, theta, and beta spectral power across 30 electrodes was found to be influenced by genetic and nonshared environmental influences. Heritability ranged between 0.39-0.78 for alpha 0.40-0.75 for theta, and 0,56-0.72 for beta power. Females showed some higher heritability for alpha and theta power in the posterior brain regions. • Other measures of ASB replicated findings of higher frontal alpha, frontal beta, and parietal theta power for Wave 3 male aggressive ASB. • Differences emerged between adolescent-onset and non-offending nonaggressive ASB groups for female parietal beta power. ANOVA analyses found significant omnibus group differences between groups for frontal alpha, theta, and beta, as well. • Paternal aggressive ASB was significantly associated with higher frontal alpha, frontal beta, and parietal theta in daughters. This mirrored male twin patterns relating to higher CBCL aggressive ASB. • Trends suggested potential moderating roles for poor neighborhood quality for EEG spectral power in association with male ASB. • Frontal alpha asymmetry did not emerge as a significant predictor of ASB in this sample. 127 Interpretations and Implications Aim 1: Structure of influences on aggressive and nonaggressive ASB The first aim set forth in the introduction of this work was to investigate the longitudinal structure of influences on the constructs of aggressive and nonaggressive ASB. Past research has found support for both phenotypic and genetic distinctions and overlaps between these two constructs, and these results further refine past findings. Across time, the latent common factors of Waves 1 and 3 shared genetic and environmental influences, which accounted for high stability. Additionally, influences on the latent factor that emerged in Wave 3 specifically were entirely comprised of genetics, consistent with the theory that adolescent ASB is influenced by puberty-related hormonal and genetic influences (Jacobson, et al. 2002). Lastly, the fact that this study found no sex differences on the structure of longitudinal influences is consistent with several major examinations of ASB (Widom & Ames, 1988; Rhee & Waldman, 2002). This may be explained through the polygenic multiple threshold model, which states that the less affected sex needs higher liability to develop a phenotype, but does not necessarily have less genetic involvement. Aim 2: Spectral power as a biomarker for ASB The second aim set forth for this project was to investigate the viability of EEG spectral power as a biomarker for ASB in children and adolescents. Analyses were conducted to verify the quality of the EEG data, and both mean levels and heritability was comparable to past studies, suggesting adequate data quality. The decision to pursue PCA data derivation rather than averaging relies on four important rationales for using this method. First, the first eigenvariate 128 reflects the strongest single source of activation whereas averaging can in fact combine several multiple sources, diluting any one single source of activation (Thomas, et al. 2011). Second, this method also reduces the weight of recording from channels that produce a lower signal to noise ratio, hence ensuring that the clearest signal is derived. Third, due to the wavelike nature of recording, PCA methods prevent a situation whereby the summation of waves produces a cancelation due to opposite phases. Lastly, this method accounts for potential hardware-derived differences in recording strength – that is, if one channel is persistently producing a stronger recording regardless of activation, it is assigned lower weighting in the PCA derivation of power. However, the validity of the EEG data in general is not a clear indication of the appropriateness of using EEG spectral power as a biomarker for ASB. Findings of Chapters 3 and 4 suggest that EEG is not a strong tool for such purposes within the general population. While past studies examining clinical and criminal populations have been consistent in findings that lower arousal relates to ASB, the results that emerged in this study were less consistent, both between the sexes and between the forms of ASB. Some of the hypothesized relationships emerged, but their small magnitude does not suggest this method would be a robust screening tool in the general population. On the other hand, these findings may provide further support for the distinction between aggressive and nonaggressive ASB forms of ASB. While the positive relationships with alpha power and aggressive ASB are consistent with past findings of ASB, the other findings are novel. Past research had not drawn distinction between aggressive and nonaggressive ASB with relation to spectral power, so this is the first study to suggest the nonaggressive ASB may not be related to low arousal levels. There are several potential explanations for this: first, most past research has in fact focused on violent offending specifically, and hence this may be an entirely 129 valid and novel distinction between aggressive and nonaggressive ASB neural profiles. Second, the behaviors that were endorsed in the nonaggressive ASB scale, such as lying or cheating, preferring the company of older children, or swearing, may not be behaviors related to low arousal, unlike more severe nonaggressive ASB that may emerge in criminal samples. Third, these results may emerge as not replicable in other community samples, and hence it is very important to examine these relationships in different samples. If our results are confirmed by further replication, they would constitute an important shift in understanding about the underlying mechanisms of different forms of ASB. There are potential clinical implications to these findings, such as risk-taking and impulsivity playing a bigger role in aggressive ASB than in nonaggressive ASB due to low arousal. Additionally, proactive rather than reactive aggressive ASB showed relationships more consistent with the CBCL in Chapter 5 analyses, which may suggest that lower arousal leads to proactive seeking of risk and stimulation. With replication, this could also provide meaningful clinical and biological distinction between these two forms of aggressive ASB, which have been debated in the literature. It is notable that alpha asymmetry, as elaborated upon in Chapter 5, did not emerge as predictive of ASB in this community sample. Past findings showed some support for asymmetry, and stronger support for low arousal, as markers of ASB. Because some of the predicted relationships did not emerge at all (asymmetry), it is significant that some did in fact emerge (spectral power). This suggests that despite small relationships, results are not merely spurious. Aim 3: Genetic overlap between frontal alpha power and ASB 130 The third aim of this project was to investigate genetic relationships between EEG spectral power and ASB, in order to formulate hypotheses about mechanism. Frontal alpha power was selected for these analyses as a frequently studied measure of past low arousal in children. The genetic relationships that were found in Chapter 4 were consistent with phenotypic relationships that emerged in Chapter 3. That the magnitude of these relationships is small suggests that frontal alpha power is not an exceptional tool for identifying individuals at a genetic risk of developing ASB. In another examination of genetic risk, parental ASB was explored as a potential predictor of twin EEG spectral power. Paternal aggressive behavior was found to predict females’ EEG profiles in the same pattern as male twin aggressive behavior was originally correlated with EEG profiles. However, the relationships between these spectral properties and CBCL aggressive ASB were not significant in the females in Chapter 3. This interesting finding raises the question of whether these fathers would indeed show similar EEG profiles to their daughters and to the male twins with higher CBCL aggressive ASB scores. Another question is why these EEG patterns are associated with aggressive ASB only in the males, and not in the females. In all, because the magnitude of these relationships was small, conclusions about mechanism are not strongly informative. To the extent that relationships emerged, they were explained entirely by genetics, but it is possible that results would have appeared different in a clinical sample with stronger phenotypic relationships. Developmental implications Another important consideration is the developmental and longitudinal aspect of these relationships. The pattern that emerged in Chapters 3 and 4 is that late childhood EEG power predicts mid-adolescent, but not late childhood, ASB to varying degrees. Chapter 5 analyses utilizing different measures of ASB were consistent with this, finding that Wave 3, but not Wave 131 1 (with one exception), measures of conduct disorder, and psychopathy in females, and measures of aggressive ASB in males were predicted by Wave 1 EEG spectral power. The consistency of this finding across measures suggests it is not spurious, and hints at a childhood biological vulnerability toward later ASB. One clinical interpretation of this is that spectral power in childhood suggests vulnerability toward specifically adolescent or adult specific ASB that may not yet have emerged in childhood. Environmental Moderators EEG did not emerge as a robust tool for identification of children at risk for developing ASB in the general population, as opposed to past findings in clinical populations. Environmental moderators were hence explored. The theory behind this was that clinical and criminal participants, who were used similar past research, might be more likely to have experienced adverse conditions that would have influenced their trajectory toward higher ASB. In order to attempt some replication of these conditions, moderation relationships were examined in this sample. Although the moderator analysis did not find resounding relationships across all forms of adversity and in all EEG measures, trends in the results suggest potential involvement of harsh neighborhood conditions in males. The direction of these trends is that high frontal alpha and beta interacted with worse neighborhood quality in association with aggressive ASB. However, this did not provide strong support for the environmental adversity hypothesis. Intervention Implications There are intervention implications to these findings, as well. For one, as Chapter 2 demonstrates, the nature of influences on ASB undergoes some changes across development, with new shared genetic influences emerging during adolescence. Hence, this finding supports 132 more the notion that socially-integrated interventions such as youth groups or community centers, are appropriate both in childhood and adolescence. Regarding findings of EEG, if results are replicated in further examinations, these may suggest different directions of research and intervention for aggressive and nonaggressive ASB. If indeed aggressive ASB is preceded by low arousal and nonaggressive ASB is not, there are implications for both behavioral treatments such as relaxation or meditation treatment for nonaggressive ASB children, or more biological treatments, such as biofeedback or medication to enhance arousal for aggressive males. More research would be needed to discern what integrated approaches are best for children who show both forms of behavior. Strengths and Limitations A significant strength of this project is the community nature of this sample, which allows for the exploration of biomarker relationships in the general population. While the findings from this research are in part negative findings, it is important to know that EEG is more predictive of behavior in specific populations and less so for the general public. Additionally, this highly diverse sample allows for conclusions that extend beyond Caucasian populations, which are generally overrepresented in psychological research. We also examined female participants, who were vastly underrepresented in past research on these topics, and found that ASB relationships to EEG are not identical to those in males. However, a corresponding weakness of the study is that due to the community nature of the sample, the relationships that emerged were small in magnitude and possibly showed lesser heritability than a clinical sample would have. Parental report, which was selected for the ASB measures utilized in this project for purposes of continuity, are also both a strength and a limitation. Parents have the benefit of 133 continuous, mature perspective on the child’s behavior, and reliability of the CBCL instrument supports this claim. Additionally, children’s test-retest reliability within our sample was found to be low to moderate over the course of six months, suggesting children’s insight may be compromised. However, by mid-adolescence, it is undeniable that children have some degree of privacy, and that parents may not be entirely privy to their children’s behavior. Hence, it is possible that using both forms of report may have led to less error in assessing child ASB. An additional limitation of the project involves EEG as a neuroimaging method. Although this method is inexpensive, temporally quick, and well established, there are two main problems with EEG. First, EEG is a method with relatively low signal to noise ratios, and is prone to high rates of error. It is well known that EEG profiles can change profoundly even over the course of several hours. Past research of how resting states change after cognitive tasks led to the selection of the first rest period of data collection in the aim of minimizing error, but even so, this cross-sectional data likely contains a great deal of error variance. Second, EEG shows poor spatial resolution, as there are many individual differences in brain structure. In order to correct for this, PCAs were conducted over larger anatomical regions, but this may dilute the specificity of any one anatomically distinct source of activation. Future Directions The most important future direction necessitated by these findings is further replication of the arousal distinctions that arose between aggressive and nonaggressive ASB. In this study, trends suggested that low arousal is a marker of aggressive behavior, which is a replication of numerous past studies with mostly violent criminal populations. However, nonaggressive ASB was not associated with spectral power. While some past studies included both violent and nonviolent offenders, this is the first to specifically draw the distinction and investigate potential 134 differences between them. These results necessitate further replication in order to verify whether they have spurious elements, or whether they reveal true underlying neurobiological differences. For example, it may be informative to conduct EEG analysis on purely non-violent or even non- physical offenders, such as white-collar criminals or computer hackers. Several additional interesting questions arose over the course of these analyses. For example, the group analyses conducted suggested differences between individuals with persistently low nonaggressive ASB and individuals whose nonaggressive ASB emerged only in adolescents, which is consistent with the Wave 1-3 phenotypic relationships between EEG and CBCL or other ASB measures. It would be a worthwhile future direction to replicate and further explore these findings. In addition, it would be important to repeat these analyses with a latent class growth analysis based on other measures, such as the Child Aggression Questionnaire or the Delinquency Interview. An additional avenue for future directions beyond group analysis would be further examination of potential environmental moderators. Although trends in the data suggest that neighborhood quality may play a role in moderating relationships between EEG power and ASB, the results are too sporadic to draw firm conclusions. However, it is possible that examining more severe environmental conditions would reveal clearer relationships. Future research may examine measures of abuse and neglect in order to verify whether severe adversity would moderate or even mediate the relationships between arousal and ASB. Another question to explore in future research is one of sex differences in the clinical presentation of ASB. Whereas no significant relationships emerged between spectral power and aggressive ASB in the females, the finding that daughters of aggressive fathers showed profiles more similar to their aggressive male counterparts raises potential questions about presentation 135 of aggressive ASB in the females. There are theories that female aggressive ASB presents in relational (e.g. gossip, deliberate social isolation) rather than physical ways. It is possible that if strong measures of relational aggressive ASB and other social behaviors were examined, female EEG profiles for forms of relational or social aggressive ASB would mirror the male profiles for aggressive ASB. On the other hand, it may also be that aggressive ASB is more of a ‘default’ behavior in males, and so while low arousal in males leads to levels of aggressive ASB, it does not in females. It is possible that females who are more highly aroused are more prone to ASB through reactive mechanisms – these are topics for future research, as well. Conclusions In conclusion, this research finds support for both differences and commonalities between aggressive and nonaggressive ASB forms of antisocial behavior in children and adolescents, and helps to elucidate their developmental properties. Distinctions arose phenotypically, genetically, and biologically. Aggressive and nonaggressive ASB were commonly influenced by a latent common factor of antisocial behavior, which was stable due to genetic, and shared environmental, and nonshared environmental influences. In addition, each form of behavior at each time point showed unique influences. EEG spectral power profiles were found to have small but significant associations with male aggressive ASB, and not with female ASB or with nonaggressive ASB. These findings suggest neurobiological differences between the two forms of behavior, with aggressive ASB showing stronger relation to cortical arousal. Genetically, small but significant correlations emerged between childhood frontal alpha power and adolescent male aggressive ASB. This research contributes to our understanding of the etiological and 136 biological properties underlying different forms of ASB, and provides future directions for further investigation of the role cortical arousal plays in this distinction. 137 References Achenbach, T. M. (1991b). Manual for the Child Behavior Checklist/4-18 and 1991 Profile. Burlington, VT: University of Vermont Department of Psychiatry. d’Alfonso, A. A. L., van Honk, J., Hermans, E., Postma, A., & de Haan, E. H. F. (2000). Laterality effects in selective attention to threat after repetitive trancranial magnetic stimulation at the prefrontal cortex in female subjects. Neuroscience Letters, 280, 195– 198. American Psychiatric Association, (1994). Diagnostic and statistical manual of mental disorders, (4th ed.) Washington, DC: Author. Andersen, S. L., & Teicher, M. H. (2004). Delayed effects of early stress on hippocampal development. Neuropsychopharmacology, 29(11), 1988-1993. Anderson, S.W., Bechara,A.,Damasio,H., Tranel,D.,& Damasio,A.R. (1999). Impairment of social and moral behavior related to early damage in human prefrontal cortex. Nature Neuroscience, 2, 1032–1037. van Baal, G.C., de Geus, E.J., & Boomsma, D.I. (1996) Genetic architecture of EEG power spectra in early life. Electroencephalography and Clinical Neurophysiology, 98, 502-514. Baker, L. A., Barton, M., Lozano, D. I., Raine, A. & Fowler, J. H. (2006). The Southern California Twin Register at the University of Southern California: II. Twin Research and Human Genetics 9, 933-40. Baker, L. A., Barton, M. & Raine, A. (2002). The Southern California Twin Register at the University of Southern California. Twin Research and Human Genetics 5(5), 456-9. 138 Baker, L. A., Raine, A., Liu, J., & Jacobson, K. C., (2008). Differential genetic and environmental influences on reactive and proactive aggression in children. Journal of Abnormal Child Psychology, 36, (8), 1265-1278. Barry, R. J., Clarke, A. R., & Johnstone, S. J. (2003). A review of electrophysiology in attention- deficit/hyperactivity disorder: I. Qualitative and quantitative electroencephalography. Clinical Neurophysiology, 114, 171–183. Baving, L., Laucht, M., & Schmidt, M. H. (2000). Oppositional children differ from healthy children in frontal brain activation. J Abnorm Child Psychol, 28(3), 267-275. van Beijsterveldt, C. E., Molenaar, P. C., de Geus, E. J., & Boomsma, D. I. (1996). Heritability of human brain functioning as assessed by electroencephalography. Am J Hum Genet, 58(3), 562-573. Bierman, K. L., & McCauley, E. (1987). Children's descriptions of their peer interactions: Useful information for clinical child assessment. Journal of Clinical Child Psychology, 16(1), 9- 18. Burt SA (2009). Are there meaningful etiological differences within antisocial behavior? Results of a meta-analysis. Clinical Psychology Review, 29, 163-178. Burt SA (2012). Do etiological influences on aggression overlap with those on rule breaking? A meta-analysis. Psychological Medicine, 1.1, 1-12. Burt SA (2012). How do we optimally conceptualize the heterogeneity within antisocial behavior? An argument for aggressive versus nonaggressive behavioral dimensions. Clinical Psychology Review 32, 263–279. Chabot, R. J., & Serfontein, G. (1996). Quantitative electroencephalographic profiles of children with attention deficit disorder. Biological Psychiatry, 40, 951–963. 139 Cadoret, R. J., Cain, C. A., & Crowe, R. R. (1983). Evidence for gene-environment interaction in the development of adolescent antisocial behavior. Behavior Genetics, 13, 301-310. Caspi, A., McClay, J., Moffitt, T., Mill, J., Martin, J., Craig, I. W., et al. (2002). Role of genotype in the cycle of violence in maltreated children. Science, 297, 851–854. Coan, J. A. (2003). The heritability of trait frontal EEG asymmetry and negative emotionality: Sex differences and genetic nonadditivity. PhD diss., University of Arizona. Davidson, R. J. (1992). Anterior cerebral asymmetry and the nature of emotion. Brain and Cognition, 25, 125-151. Davidson, R. J., (1998). Anterior electrophysiological asymmetries, emotion and depression: Conceptual and methodological conundrums. Psychophysiology35, 607-614. Dawson, G., Panagiotides, H., Klinger, L. G., & Spieker, S. (1997). Infants of depressed and nondepressed mothers exhibit differences in frontal brain electrical activity during the expression of negative emotions. Dev Psychol, 33(4), 650-656. Downey, G., & Coyne, J.C. (1990). Children of depressed parents: An integrative review. Psychological Bulletin, 108, 50–76. Ehlers, C. L., & Schuckit, M. A. (1990). EEG fast frequency activity in the sons of alcoholics. Biological Psychiatry, 27, 631–641. Ehlers, C. L., & Schuckit, M. A. (1991). Evaluation of EEG alpha activity in sons of alcoholics. Neuropsychopharmacology, 4, 199–205. Ellingson, R. J. (1954). The incidence of EEG abnormality among patients with mental disorders of apparently nonorganic origin: a critical review. Am J Psychiatry, 111(4), 263-275. Farrington, D. P. (1989). Early predictors of adolescent aggression and adult violence. Violence Vict, 4(2), 79-100. 140 Finn PR, Justus A, Mazas C, Steinmetz JE (1999): Working memory, executive processes and the effects of alcohol on Go/No-Go learning: Testing a model of behavioral regulation and impulsivity. Psychopharmacology (Berl) 146:465–472. Forbes, E.E., Shaw, D.S., Fox, N.A., Cohn, J.F., Silk, J.S., & Kovacs, M. (2006). Maternal depression, child frontal asymmetry, and child affective behaviors as factors in child behavior problems. Journal of Child Psychiatry and Psychology, 47, 79-87. Fox, N.A. (1994). Dynamic cerebral processes underlying emotion regulation. Soc Res Child Dev, 59, 152-166. Fox, N. A., & Davidson, R. J. (1988). Patterns of brain electrical activity during facial signs of emotion in 10-monthold infants. Developmental Psychology, 24, 230–236 Hare, R. D. (1999). The Hare Psychopathy Checklist-Revised: PLC-R. MHS, Multi-Health Systems. Fein, G., Galin, D., Yingling, C. D., Johnstone, J., & Nelson, M. A. (1984). EEG spectra in 9–13 year-old boys are stable over 1–3 years. Electroencephalography and clinical neurophysiology, 58(6), 517-518. Gjone, H., & Stevenson, J. (1997). The association between internalizing and externalizing behavior in childhood and early adolescence: genetic of environmental common influences? J Abnorm Child Psychol, 25(4), 277-286. Harmon-Jones, E. (2003). Early Career Award. Clarifying the emotive functions of asymmetrical frontal cortical activity. Psychophysiology, 40(6), 838-848. Harmon-Jones, E., & Sigelman, J. (2001). State anger and prefrontal brain activity: evidence that insult-related relative left-prefrontal activation is associated with experienced anger and aggression. J Pers Soc Psychol, 80(5), 797-803. 141 Hewig, J., Hagemann, D., Seifert, J., Naumann, E., & Bartussek, D. (2004). On the selective relation of frontal cortical asymmetry and anger-out versus anger-control. J Pers Soc Psychol, 87(6), 926-939. Hill, D. (1952). EEG in episodic psychotic and psychopathic behaviour; a classification of data. Electroencephalogr Clin Neurophysiol, 4(4), 419-442. Hinshaw, S. P. (1992). Externalizing behavior problems and academic underachievement in childhood and adolescence: causal relationships and underlying mechanisms. Psychol Bull, 111(1), 127-155. Hollingshead, A. B. (1979). Four-Factor Index of Social Status. Unpublished manuscript, Yale University. Jones, N. A., Field, T., & Davalos, M. (2000). Right frontal EEG asymmety and lack of empathy in preschool children of depressed mothers. Child Psychiatry Hum Dev, 30(3), 189-204. Krueger RF, Moffitt TE, Caspi A, Bleske A, Silva PA (1998) Assortative mating for antisocial behavior: developmental and methodological implications. Behav Genet 28, 173–186. Loeber, R., Farrington, D. P., Stouthamer-Loeber, M., Moffitt, T. E., & Caspi, A. (1998). The development of male offending: Key findings from the first decade of the Pittsburgh youth study. Studies on Crime and Crime Prevention, 7, 141-171. Liddle, H. A., Rowe, C. L., Dakof, G. A., Ungaro, R. A., & Henderson, C. E. (2004). Early intervention for adolescent substance abuse: pretreatment to posttreatment outcomes of a randomized clinical trial comparing multidimensional family therapy and peer group treatment. J Psychoactive Drugs, 36(1), 49-63. Lykken, D. T. (1978). The diagnosis of zygosity in twins. Behav Genet, 8(5), 437-473. 142 Lynam, D. R., (1997). Pursuing the psychopath: Capturing the fledgling psychopath in a nomological net. Journal of Abnormal Psychology, 106, 425-438. McGuire, K.A., Katsanis, J., Iacono, W.G., & McGue, M. (1998). Genetic influences on the spontaneous EEG: An examination of 15-year-old and 17-year-old twins. Developmental Neuropsychology, 14, 7-18. McNeil, J. J., & O'Brien, P. E. (1999). Screening, case finding and evidence-based guidelines. Med J Aust, 171(7), 344-345. Mednick, S. A., Gabrielli, W. F., & Hutchings, B. (1984). Genetic influences in criminal convictions: evidence from an adoption cohort. Science. Mednick, S.A., Volavka, J., Gabrielli, W.F., Itil, T.M., 1981. EEG as a predictor of antisocial behavior. Criminology 19, 219–229. Moffitt TE (2003). Life-course persistent and adolescence limited antisocial behavior : a research review and a research agenda. In The Causes of Conduct Disorder and Serious Juvenile Delinquency (ed. B. Lahey, T. E. Moffitt and Caspi), pp. 49–75. Guilford Press : New York. Moffitt, T. E., & Caspi, A. (2001). Childhood predictors differentiate life-course persistent and adolescence-limited antisocial pathways among males and females. Dev Psychopathol, 13(2), 355-375. Neale, M. C., Boker, S. M., Xie, G. & Maes, H. (2003). Mx: Statistical modeling. Richmond, VA: Department of Psychiatry, Medical College of Virginia. O'Connor, S., Bauer, L., Tasman, A., & Hesselbrock, V. (1994). Reduced P3 amplitudes are associated with both a family history of alcoholism and antisocial personality 143 disorder. Progress in Neuro-Psychopharmacology and Biological Psychiatry, 18(8), 1307-1321. Patton, J.H., Stanford, M.S., & Barratt, E.S., (1995) Factor Structure of the Barratt Impulsiveness Scale. Pesenti-Gritti, P., Spatola, C. A., Fagnani, C., Ogliari, A., Patriarca, V., Stazi, M. A., et al. (2008). The co-occurrence between internalizing and externalizing behaviors. A general population twin study. Eur Child Adolesc Psychiatry, 17(2), 82-92. Petersén, K. G., Matousek, M., Mednick, S. A., Volavka, J., & Pollock, V. (1982). EEG antecedents of thievery. Acta Psychiatr Scand, 65(5), 331-338. Peterson, C. K., Shackman, A. J., & Harmon-Jones, E. (2008). The role of asymmetrical frontal cortical activity in aggression. Psychophysiology, 45(1), 86-92. Poznanski, E., & Zrull, J. P. (1970). Childhood depression. Clinical characteristics of overtly depressed children. Arch Gen Psychiatry, 23(1), 8-15. Raine, A. (1997). Psychophysiology and antisocial behavior: A biosocial perspective and a prefrontal dysfunction hypothesis. In D. Stoff, J. Breiling, & J.D. Maser (Eds.), Handbook of antisocial behavior (pp. 289–304). New York: Wiley. Raine, A., Lencz, T., Bihrle, S., LaCasse, L., & Colletti, P. (2000). Reduced prefrontal gray matter volume and reduced autonomic activity in antisocial personality disorder. Archives of general psychiatry, 57(2), 119. Raine, A., Venables, P. H., & Williams, M. (1990). Relationships between central and autonomic measures of arousal at age 15 years and criminality at age 24 years. Arch Gen Psychiatry, 47(11), 1003-1007. Rangaswamy, M., Porjesz, B., Chorlian, D. B., Wang, K., Jones, K. A.,Bauer, L. O., et al. 144 (2002). Beta power in the EEG of alcoholics. Biological Psychiatry, 52, 831–842. Reyes, O., Gillock, K. L., Kobus, K., & Sanchez, B. (2000). A longitudinal examination of the transition into senior high school for adolescents from urban, low-income status, and predominantly minority backgrounds. Am J Community Psychol, 28(4), 519-544. Rhee, S. H. & Waldman, I. D. (2002). Genetic and Environmental Influences on Antisocial Behavior: A Meta-Analysis of Twin and Adoption Studies. Psychological Bulletin 128, 490-529. Roth, T. L., & Sweatt, J. D. (2011). Epigenetic marking of the BDNF gene by early-life adverse experiences. Hormones and behavior, 59(3), 315-320. Rybak, M., Crayton, J. W., Young, I. J., Herba, E., & Konopka, L. M. (2006). Frontal alpha power asymmetry in aggressive children and adolescents with mood and disruptive behavior disorders. Clin EEG Neurosci, 37(1), 16-24. Sapolsky, R. M. (1996). Why stress is bad for your brain. Science, 273(5276), 749-750 Smit, D.J.A., Posthuman, D.I, & De Geus, E.J.C., (2005). Heritability of background EEG across the power spectrum. Psychophysiology, 42, 691-697. Smit, D. J. A., Posthuma, D., Boomsma, D. I., & de Geus, E. J. C. (2007). The relation between frontal EEG asymmetry and the risk for anxiety and depression. Biological Psychology, 74, 26–33. Teicher, M. H., Dumont, N. L., Ito, Y., Vaituzis, C., Giedd, J. N., & Andersen, S. L. (2004). Childhood neglect is associated with reduced corpus callosum area. Biol Psychiatry, 56(2), 80-85. Teicher, M. H., Ito, Y., Glod, C. A., Andersen, S. L., Dumont, N., & Ackerman, E. (1997). Preliminary evidence for abnormal cortical development in physically and sexually abused 145 children using EEG coherence and MRI. Ann N Y Acad Sci, 821, 160-175. Tremblay RE (2003). Why socialization fails: the case of chronic physical aggression. In The Causes of Conduct Disorder and Serious Juvenile Delinquency (ed. B. Lahey, T. E. Moffitt and A. Caspi), pp. 182–224. Guilford Press : New York. Tuvblad, C., Eley, T. C., & Lichtenstein, P. (2005). The development of antisocial behavior from childhood to adolescence: A longitudinal twin study. European Child & Adolescent Psychiatry, 14, 216−225. Tuvblad, C., & Baker, L. A. (2011). 8 Human Aggression Across the Lifespan: Genetic Propensities and Environmental Moderators. Advances in genetics,75, 171. Waites, A.B., Stanislavsky, A., Abbott, D.F., & Jackson, G.D. (2005). Effect of prior cognitive state on resting state networks measured with functional connectivity. Human Brain Mapping, 24, 59-68. White, J. Moffitt, T.E., Earls, F., Robins, L., & Silva, P.A. (1990). How early can we tell? Predictors of childhood conduct disorder and adolescent delinquency. Criminology, 28, 507-535. 146 Appendix A: CBCL Nonaggressive ASB and Aggressive ASB Items with Means, SDs, and Percentage Endorsement Table A.1. Nonaggressive ASB Items, Means, Standard Deviations, and Percentage Endorsement by Sex and Wave # Item Wave 1 Means (SD), % endorsements Wave 3 Means (SD), % endorsements Males Females Males Females 26 Doesn’t seem to feel guilty after misbehaving 0.30 (0.53), 26.1 0.24 (0.48), 21.4 0.29 (0.57), 23.2 0.27 (0.55), 21.6 39 Hangs around with others who get in trouble 0.12 (0.35), 11.8 0.05 (0.25), 4.6 0.17 (0.41), 15.5 0.13 (0.38), 11.7 43 Lies or cheats 0.22 (0.44), 20.8 0.20 (0.43), 18.9 0.28 (0.51), 25.1 0.20 (0.46), 18.2 63 Prefers being with old children or youths 0.44 (0.59), 38.8 0.34 (0.56), 29.6 0.28 (0.52), 23.4 0.31 (0.58), 25.1 67 Runs away from home 0.01 (0.09), 0.8 0.00 (0.04), 0.2 0.02 (0.17), 1.8 0.03 (0.21), 2.7 72 Sets fires 0.02 (0.14), 1.8 0.00 (0.07) 0.5 0.03 (0.17), 2.8 0.01 (0.14), 1.0 81 Steals at home 0.10 (0.32), 9.8 0.08 (0.29), 7.7 0.06 (0.27), 4.8 0.06 (0.28), 4.9 82 Steals outside the home 0.03 (0.20), 2.8 0.03 (0.18), 2.4 0.02 (0.17), 1.8 0.05 (0.26), 4.2 90 Swears or uses obscene language 0.26 (0.48), 23.5 0.11 (0.34), 10.7 0.35 (0.57), 29.7 0.26 (0.51), 22.6 96 Seems preoccupied with sex 0.02 (0.16), 2.2 0.01 (0.13), 0.9 0.06 (0.27), 4.9 0.05 (0.25), 3.7 101 Truancy, cuts classes 0.01 (0.08), 0.7 0.00 (0.07), 0.5 0.09 (0.33), 6.9 0.04 (0.23), 3.9 105 Uses alcohol or drugs or non- medical purposes 0.00 (0), 0 0.00 (0), 0 0.04 (0.23), 3.7 0.03 (0.23), 2.5 106 vandalizes 0.01 (0.07), 0.5 0.00 (0.07), 0.2 0.02 (0.15), 1.6 0.02 (0.17), 1.2 147 Table A.2. Aggressive ASB Items, Means, Standard Deviations, and Percentage Endorsement by Sex and Wave # Item Wave 1 Means (SD), % endorsements Wave 3 Means (SD), % endorsements Males Females Males Females 3 Argues a lot 0.74 (0.65), 61.6 0.73 (0.66), 60.8 0.71 (0.70), 56.1 0.68 (0.68), 55.7 7 Bragging, boasting 0.44 (0.58), 39.0 0.27 (0.49), 24.9 0.35 (0.57), 29.3 0.20 (0.44), 18.6 16 Cruelty, bullying, or meanness to others 0.12 (0.34), 11.8 0.09 (0.31), 7.9 0.11 (0.36), 9.8 0.08 (0.31), 6.9 19 Demands a lot of attention 0.41 (0.59), 35.6 0.46 (0.61), 39.6 0.27 (0.51), 24.3 0.32 (0.57), 27.3 20 Destroys his/her own things 0.13 (0.37), 12.0 0.05 (0.23), 4.2 0.09 (0.32), 7.4 0.05 (0.25), 3.5 21 Destroys property belonging to others 0.19 (0.41), 17.5 0.10 (0.33), 9.3 0.09 (0.31) 7.8 0.05 (0.23), 4.5 22 Disobedient at home 0.53 (0.54), 50.7 0.46 (0.55), 43.7 0.47 (0.51), 42.0 0.37 (0.56), 32.0 23 Disobedient at school 0.28 (0.47), 26.8 0.10 (0.33), 9.6 0.28 (0.51), 25.3 0.14 (0.37), 12.7 27 Easily jealous 0.40 (0.56), 35.5 0.51 (0.62) , 45.9 0.26 (0.51), 22.8 0.29 (0.53), 24.8 37 Gets in many fights 0.17 (0.38), 17.1 0.13 (0.37), 11.2 0.14 (0.38), 13.2 0.11 (0.33), 10.5 57 Physically attacks people 0.09 (0.31), 8.6 0.07 (0.27), 6.4 0.05 (0.23), 4.6 0.04 (0.22), 3.8 68 Screams a lot 0.27 (0.50), 24.0 0.23 (0.50), 19.5 0.20 (0.46), 16.6 0.25 (0.51), 21.3 74 Showing off or clowning 0.55(0.64), 46.4 0.29 (0.53), 25.3 0.46 (0.65), 38.3 0.24 (0.47), 22.6 86 Stubborn, sullen or irritable 0.43 (0.56), 38.8 0.44 (0.58), 39.2 0.45 (0.60), 39.0 0.41 (0.56), 36.7 87 Sudden changes in mood or feelings 0.22 (0.45), 20.8 0.24 (0.49), 20.6 0.28 (0.53), 23.6 0.38 (0.57), 33.5 93 Talks too much 0.48 (0.64), 40.2 0.42 (0.61), 36.3 0.34 (0.58), 27.8 0.36 (0.61),29.2 94 Teases a lot 0.24 (0.46), 22.4 0.14 (0.39), 13.0 0.26 (0.53), 21.9 0.15 (0.38), 13.5 95 Temper tantrums or hot temper 0.41 (0.58), 35.2 0.36 (0.57), 31.0 0.31 (0.54), 27.1 0.32 (0.55), 27.2 97 Threatens people 0.06 (0.25), 5.2 0.03 (0.20), 3.0 0.05 (0.26), 4.6 0.04 (0.23), 3.2 104 Unusually loud 0.22 (0.46), 19.4 0.17 (0.45), 14.1 0.16 (0.43), 13.4 0.15 (0.42), 12.9 148 Appendix B. Univariate Fit Statistics and Estimates for Aggressive and Nonaggressive ASB at Waves 1 and 3. Overall fit Model difference test Table B.3 Model Aggressive ASB 1 -2LL Df AIC χ 2 Δdf P 1 Saturated (means constrained) 3001.79 1183 786.73 2 ACE 3031.43 1196 639.43 29.64 13 0.01 3 ACE M=F 3031.72 1199 633.71 29.93 16 0.02 4 AE M=F 3039.76 1200 639.76 37.97 17 <0.01 5 AE M≠F 3039.38 1198 643.38 37.59 15 <0.01 6 CE M=F 3038.24 1200 638.24 36.45 17 <0.01 7 CE M≠F 3038.14 1198 642.14 36.35 15 0.02 8 E 3193.76 1201 791.76 191.98 18 <0.01 Estimates: A 0.29 (0.07-0.51); C 0.27 (0.09-0.44); E0.44 (0.37-0.52) 149 Overall fit Model difference test Table B.4 Model Nonaggressive ASB 1 -2LL Df AIC χ 2 Δdf P 1 Saturated (means constrained) 2646.51 1183 280.51 2 ACE 2661.99 1196 269.99 15.48 13 0.28 3 ACE M=F 2680.19 1199 282.19 33.68 16 0.01 4 AE M=F 2684.34 1200 284.34 37.83 17 <0.01 5 AE M≠F 2674.26 1198 278.26 27.75 15 0.02 6 CE M=F 2700.12 1200 300.12 53.61 17 <0.01 7 CE M≠F 2685.62 1198 289.62 39.11 15 <0.01 8 E 2886.13 1201 484.13 239.62 18 <0.01 9 Male ACE, female CE 2662.40 1197 268.40 15.89 14 0.32 Estimates: male - A 0.48 (0.30-0.63); C 0.17 (0.06-0.31); E 0.35 (0.27-0.45); female - C 0.62 (0.54-0.69); E:0.38 (0.31-0.46) Overall fit Model difference test Table B.5 Model Aggressive ASB 3 -2LL Df AIC χ 2 Δdf P 1 Saturated (means constrained) 2997.05 1127 743.05 2 ACE 3008.44 1140 728.44 11.39 13 0.58 3 ACE M=F 3011.94 1143 725.94 14.89 16 0.53 4 AE M=F 3012.13 1144 724.13 15.08 17 0.59 5 AE M≠F 3008.65 1142 724.65 11.60 15 0.71 6 CE M=F 3045.21 1144 757.21 48.16 17 <0.01 150 Estimates: A 0.48 (0.33-0.65); C 0.31 (0.14-0.44); E 0.21 (0.18-0.26) 7 CE M≠F 3043.65 1142 759.65 46.60 15 <0.01 8 E 3230.69 1145 913.69 206.64 18 <0.01 Estimates: A 0.68 (0.61-0.73); E 0.32 (0.27-0.39) Overall fit Model difference test Table B.6 Model Nonaggressive ASB 3 -2LL Df AIC χ 2 Δdf P 1 Saturated (means constrained) 2708.86 1127 454.86 2 ACE 2720.05 1140 440.05 11.19 13 0.60 3 ACE M=F 2720.35 1143 434.35 11.49 16 0.78 4 AE M=F 2732.66 1144 444.66 23.80 17 0.13 5 AE M≠F 2732.52 1142 448.52 23.66 15 0.07 6 CE M=F 2759.30 1144 471.30 50.44 17 <0.01 7 CE M≠F 2758.59 1142 474.59 49.73 15 <0.01 8 E 3079.11 1145 789.11 360.25 18 <0.01 151 Acknowledgments I want to thank Dr. Laura Baker for immense support academically, professionally, and personally over the course of my graduate schooling. I also wish to thank Dr. Catherine Tuvblad for superb guidance and scientific training. In addition, I want to express gratitude to Drs. Carol Prescott, Margaret Gatz and Adrian Raine for mentorship, as well as to thank my final committee member, Dr. Maja Mataric. I am also grateful for Dr. Pan Wang, Syed Ashrafulla, M.A., and Nick Jackson, M.A. for guidance. In addition, I would like to thank Karina Gomez, Gemma Quick, Surya Pasupuleti, Theo Botwick, and the staff, students, and volunteers of the USC twin project for their hard work collecting and processing this data. Lastly, I wish to thank my loved ones for supporting me unconditionally throughout the years.
Abstract (if available)
Abstract
This project examines commonalities and differences between aggressive and nonaggressive forms of antisocial behavior (ASB) using methodologies of behavioral genetics and EEG spectral power analysis. Using a longitudinal community sample of 1550 Los Angeles area twins, data were collected at two waves: Wave 1 (age 9-10 years) and Wave 3 (age 14-15 years). ASB was assessed using parent reports of the Child Behavioral Checklist (Achenbach, 1991), using the aggression and delinquency (non-aggressive antisocial behavior) scales. Biometric statistical modeling was applying to the twin variances and covariances in order to discern the pattern of genetic and environmental influences on aggressive and nonaggressive ASB across the two waves of data collection. Findings supported both the commonality and the distinction between these forms of ASB, as 44-79% of the influences were found to be shared through a latent common factor of ASB, and the rest were unique to each scale. Additive genetic, shared environmental and nonshared environmental effects were found to influence both time points
Linked assets
University of Southern California Dissertations and Theses
Conceptually similar
PDF
Different genetic and environmental structures for the overlap of three antisocial behavior factors with alcohol initiation
PDF
Heritability and longitudinal stability of impulsivity in adolescence
PDF
The development of electrodermal fear conditioning from ages 3 to 8 years and relationships with antisocial behavior at age 8 years
PDF
Developmental etiology of reactive and proactive aggression, electrodermal reactivity and their relationships: a twin study
PDF
Multivariate biometric modeling among multiple traits across different raters in child externalizing behavior studies
PDF
Psychometric and biometric evaluation of psychopathic personality traits in 9 and 10 year old twins
PDF
Heritability of schizotypal traits during adolescence
PDF
Psychophysiological reactivity to stress, psychopathic traits, and developmental trajectories of antisocial behavior: a prospective, longitudinal study
PDF
Sources of stability and change in the trajectory of openness to experience across the lifespan
PDF
Attrition in longitudinal twin studies: a comparative study of SEM estimation methods
PDF
Examining the longitudinal relationships between community violence exposure and aggressive behavior among a sample of maltreated and non-maltreated adolescents
PDF
Biometric models of psychopathic traits in adolescence: a comparison of item-level and sum-score approaches
PDF
Family aggression exposure and community violence exposure associated with brain volume in late adolescence: a comparison of automated versus manual segmentation
PDF
Family aggression, prosocial friends, and the risk of dating and friend victimization in late adolescence
PDF
Examining the effect of fertility on lifespan using data from the Swedish Twin Registry
PDF
Risky behaviors, interpersonal conflict, and their relation to fluctuations in adolescents’ diurnal HPA rhythms
PDF
Antecedents of marijuana initiation
PDF
Dementia, co-morbid depressive symptoms, and mortality in older adults: a twin study
PDF
Underlying neural mechanisms of depression and dementia
PDF
Social dynamics of intragroup aggression and conflict resolution in wild chimpanzees (Pan troglodytes) at Kanyawara, Kibale National Park, Uganda
Asset Metadata
Creator
Niv, Sharon
(author)
Core Title
Patterns of EEG spectral power in 9-10 year old twins and their relationships with aggressive and nonaggressive antisocial behavior in childhood and adolescence
School
College of Letters, Arts and Sciences
Degree
Doctor of Philosophy
Degree Program
Psychology
Publication Date
07/30/2013
Defense Date
03/13/2013
Publisher
University of Southern California
(original),
University of Southern California. Libraries
(digital)
Tag
adolescence,aggression,antisocial behavior,behavioral genetics,cortical arousal,EEG,OAI-PMH Harvest,Twins
Format
application/pdf
(imt)
Language
English
Contributor
Electronically uploaded by the author
(provenance)
Advisor
Baker, Laura A. (
committee chair
), Gatz, Margaret (
committee member
), Mataric, Maja J. (
committee member
), Prescott, Carol A. (
committee member
)
Creator Email
sharon.niv@gmail.com,sniv@usc.edu
Permanent Link (DOI)
https://doi.org/10.25549/usctheses-c3-308257
Unique identifier
UC11295140
Identifier
etd-NivSharon-1900.pdf (filename),usctheses-c3-308257 (legacy record id)
Legacy Identifier
etd-NivSharon-1900.pdf
Dmrecord
308257
Document Type
Dissertation
Format
application/pdf (imt)
Rights
Niv, Sharon
Type
texts
Source
University of Southern California
(contributing entity),
University of Southern California Dissertations and Theses
(collection)
Access Conditions
The author retains rights to his/her dissertation, thesis or other graduate work according to U.S. copyright law. Electronic access is being provided by the USC Libraries in agreement with the a...
Repository Name
University of Southern California Digital Library
Repository Location
USC Digital Library, University of Southern California, University Park Campus MC 2810, 3434 South Grand Avenue, 2nd Floor, Los Angeles, California 90089-2810, USA
Tags
aggression
antisocial behavior
behavioral genetics
cortical arousal
EEG