Using Dispositions to Understand Otherwise Intractable Causal Pathways to Psychological Problems During Childhood and Adolescence

ABSTRACT

Studies of the genetic and environmental factors that make children more or less likely to develop distressing and impairing psychological problems, and studies of the psychobiological pathways through which these causal factors operate, have the goal of improving our understanding of the basic nature of psychological problems to develop better methods of prevention and treatment. For this reason, we have long had our eye on the prize of discovering the causes and psychobiological mechanisms underlying each dimension of psychological problems. There are compelling reasons, however, to seek a different and more achievable prize to understand psychological problems. Dimensions of psychological problems are both far too heterogeneous and too highly correlated to line up with distinct causal pathways. In contrast, a small number of orthogonal cognitive and socioemotional dispositional dimensions are correlated with psychological problems in revealing cross-cutting patterns. Each of these dispositions shares its independent causal pathways with psychological problems and help us understand the complex shared and heterogeneous nature of their causal processes. I outline a strategy for understanding the causes and mechanisms of psychological problems using studies of independently measured dispositions.

Introduction

The manuscripts in the special section in this issue cogently address the critical importance of addressing individual differences in affect when attempting to understand disruptive behavior. In this paper, I address the same issue in a broader context. Because negative emotions are a topic of interest in the other papers in the section, I focus on that particular construct. In general, I posit that future studies of the causes and mechanisms of all psychological problems, including disruptive behavior, will be far more successful if they are based on an approach that integrates both socioemotional dispositions and psychological problems in the same model.¹ The gist of this view is not novel, as many theorists previously asserted that the structure of dispositions provides a framework for understanding psychological problems (Cattell, 1950; Clark, 2005; Cloninger, 2006; Eysenck, 1947; Klein et al., 2011; Krueger, 1999; Patrick et al., 2009; Watson et al., 1994). Based on this time-tested postulate, however, I outline a novel two-level approach to the integration of dispositions and psychological problems and offer a number of testable hypotheses. Although this model is stated in terms of child and adolescent behavior in this paper, it may well be applicable across the life span.

My goal is to address two issues that set the conditions for any attempt to understand the causes and psychobiological mechanisms of psychological problems:

1. Psychological problems are not discrete and independent from one another as assumed in the medical model of psychological problems instantiated in the various editions of the DSM and ICD. Rather, all dimensions of psychological problems are positively correlated with every other dimension (Lahey, 2021, Lahey et al., 2017). It is increasingly clear that these correlations reflect the sharing of many nonspecific causes and their related psychobiological mechanisms (Smoller et al., 2013, Zald & Lahey, 2017). The ubiquitous correlations among dimensions of psychological problems are not so high, however, that they rule out roles for more specific causes and mechanisms that differentiate finer-grain dimensions from one another. Rather, both shared and specific causes and mechanisms underlie the various dimensions of psychological problems (Lahey, 2021, Lahey et al., 2017).

2. It is very likely that the causes and mechanisms underlying each dimension of psychological problems are heterogeneous. That is, each dimension of psychological problems is hypothesized to be related to more than one causal process. This implies differences among different individuals who exhibit high levels of each problem dimension. For example, there is evidence that children who are impaired by attention-deficit/hyperactive problems exhibit mixes of elevations on dispositions that are not the same for each child (Nigg, 2022).

Recognizing the causal heterogeneity of each mental disorder, proponents of the Research Domain Criteria (RDoC) approach (Cuthbert & Insel, 2013) advocated abandoning studies of the psychobiological processes underlying mental disorders per se and instead discovering the causes and mechanisms of the multiple dispositional domains that are associated with each mental disorder in cross-cutting fashion. A number of broad domains, such as positive and negative valence, were recommended by RDoC committees as candidates for study (Cuthbert & Insel, 2013). The present approach is fully consistent with the RDoC approach, but uses a set of reliable and valid empirically developed and operationalized dispositional dimensions that should advance research based on the general RDoC approach.

Using Dispositions to Parse the Nature of Psychological Problems

A potentially powerful way to study the complex causal processes underlying psychological problems is to use an integrated model that includes both psychological problems and dispositions. One argument for such an integrated approach is that both terms simply refer to individual differences in behavior. This is not to say, however, that no distinction should be made between dispositions and psychological problems. Indeed, there is a fundamental difference in the correlational structure of dispositional and problem items that opens a window to understanding the heterogeneous causes and psychobiological mechanisms of psychological problems. Whereas the dimensions of psychological problems are all positively correlated, some dispositional dimensions, including the ones described in the present paper are essentially orthogonal-–weakly correlated, if at all (Lahey et al., 2008, 2010). The lack of correlations among dispositions is fundamentally important. This is because, if they are reliably measured, it is unlikely that uncorrelated dimensions of behavior share the same causes and mechanisms. If they did, they would be correlated.

Eye on a Different Prize

An important line of research seeks to help children with distressing and impairing psychological problems by understanding causal and mechanistic processes well enough to develop new methods of prevention and treatment. In this spirit, we have long had our eye on the prize of discovering the causes and psychobiological mechanisms that give rise to each dimension of psychological problems. We must strongly consider the possibility that we should be seeking a different prize to accomplish the long-term goal of helping persons with psychological problems. Instead, we should seek to discover the causes and mechanisms underlying the small number of orthogonal dispositions that are related to the full range of psychological problems (Clark, 2005; Cuthbert & Insel, 2013). This is because it is likely that each uncorrelated dispositional dimension lines up with different causes and psychobiological mechanisms, but this is very unlikely for the substantially correlated dimensions of psychological problems (Lahey, 2021; Lahey et al., 2017). Crucially, the orthogonality of dispositions allows us to use patterns of associations between dispositions and dimensions of psychological problems to understand why problem dimensions both (a) share their causes and mechanisms across dimensions, (b) and are heterogeneous within dimensions in their causes and mechanisms. Thus, two high priority goals for research on the nature of psychological problems are to determine (a) if each uncorrelated dispositional dimension lines up with different causal risk factors, and (b) to determine if dispositions reflect the biological mechanistic links between causes and behavior.

We have further hypothesized that dispositional dimensions are associated with the correlated problem dimensions in cross-cutting fashion (Lahey et al., 2017). This means that each disposition appears to be related to more than one problem dimension and each problem dimension is associated with more than one dispositional dimension. The former would point to processes shared by multiple dimensions of psychological problems, whereas the latter would provide a starting place for understanding the heterogeneity within each problem dimension.

Distinguishing Dispositions from Psychological Problems

If the behaviors that define the constructs of dispositions and psychological problems are just that – behaviors—how do we distinguish them enough to study the differences in their correlational structures and the patterns of associations of dispositions with problem dimensions? The field has long distinguished dispositions and problems using fuzzy boundaries that refer to specificity or temporal course (DeYoung et al., 2022; Lahey, 2004; Tackett, 2006), but there is no hard and fast distinction. In the research described below, my colleagues and I began with the traditional distinction between dispositions and psychological problems. Critically, however, we strictly followed two rules for studying the separate correlational structures of dispositions and problems that we believe to be necessary for using dispositions to understand the nature of psychological problems:

The first rule is that the pool of items used to define dispositions cannot include synonyms or antonyms of the items that define psychological problems to avoid spurious correlations due to item overlap (Lahey, 2004). In DSM terms, symptoms of psychopathology cannot be used to define the dispositional dimensions. For example, items like “fearful” and “angry,” which are included in many general-use measures of dispositions, cannot be used in disposition scales because they are too similar to symptoms of specific phobia and oppositional defiant disorder. There is some evidence that removing items from temperament scales that are symptoms of psychopathology may not completely eliminate their correlations with psychopathology dimensions (Lemery et al., 2002). Nonetheless, because dropping items changes the meaning of the temperament dimension in unknown ways and means that all psychometric evidence on the temperament scale is no longer applicable, it seems sounder to develop temperament scales without overlapping items in the first place.

Crucially, when symptoms of psychopathology are not included in the item pool of a dispositional scale, not all dispositional scales included in general purpose temperament scales can be identified. For examples, this rule eliminates the potential dispositional trait of “effortful control” because it is defined almost entirely by antonyms of ADHD symptoms, and excludes the trait of “positive emotionality,” which is defined by antonyms for depression, particularly anhedonia. This will be viewed as a shortcoming to researchers who use these dispositions in their work, but it creates an opportunity to reconsider the distinction between dimensions of dispositions and psychological problems. Should dimensions like effortful control and positive emotionality be thought of as dispositional traits or as (part of) dimensions of psychological problems? There is evidence, for example, that low levels of effortful control are associated with impaired school functioning (Berger et al., 2017). Because of the item overlap, this could equally be interpreted as evidence that ADHD is associated with poor school functioning. Is it possible to measure self-regulation in a way that is not contaminated by ADHD behaviors? This would allow us to avoid circularity and perhaps discover that self-regulation is one of multiple dispositions related to ADHD behaviors.

The second rule for defining dispositional dimensions that are useful for disentangling the complex causes and mechanisms of psychological problems concerns the number of factors extracted in factor analyses of dispositional items. The number of extracted factors is guided by the usual criteria for factor analysis, of course, but for dispositional dimensions to be useful in understanding psychological problems, the extracted factors of dispositional dimensions should be limited to factors that are as close to orthogonal as possible (Lahey, 2004). This is because it is far more likely that dispositional dimensions that are not correlated with one another align with distinct causal and mechanistic processes. Thus, it may be technically allowable to extract either N uncorrelated dimensions or N + 1 dimensions, with the latter solution including two correlated factors split from a single dimension in the N factor solution. Such correlated dispositional dimensions are unlikely to line up with entirely distinct causal processes, however.

Not including synonyms and antonyms of symptoms in measures of disposition may help reduce correlations among dispositional dimensions. Because psychological problems are robustly positively correlated with one another, including items that refer to psychological problems in item pools for dispositional scales makes it difficult to extract orthogonal dispositional scales.

CADS Model of Dispositions

I will state concrete and testable hypotheses here that were induced from studies conducted to develop parent and youth report versions of the Child and Adolescent Dispositions Scale (CADS) using data from the large and representative Tennessee Twins Study (TTS) (Lahey et al., 2008, 2010). These hypotheses are based on analyses of data from the samples described in Supplemental Table S1.

The CADS does not necessarily provide a completely satisfactory description of dispositional dimensions, but unlike other measures of dispositions, the CADS was developed following the two rules stated above to create a dispositional scale that is useful for studying relationships between dispositions and psychological problems. The CADS dispositional dimensions are highly reliable and the validity of the CADS dispositional dimensions has been demonstrated in a series of studies linking them to dimensions of psychological problems and external criteria in a number of cross-sectional and longitudinal samples of different ages (Class et al., 2019; Del Giudice et al., 2020; Lahey et al., 2018; Mathesius et al., 2017; Mikolajewski et al., 2013, 2019; Olino et al., 2014; Trentacosta et al., 2009).

The items scored on the three factors of negative emotionality, prosociality, and daring extracted in the developmental studies of the CADS are presented in Supplemental Table S2 (Lahey et al., 2008, 2010). Supplemental Table S2 also provides the CADS items in oral administration format, shows the 4-point response scales, and gives the scoring keys for both the parent and youth respondents. As detailed next, these three dispositional dimensions are not novel, but have long been identified by other researchers as playing important roles in child development (Nigg, 2006; Rettew & McKee, 2005). The novelty and utility of the CADS derives from developing a scale for studies of relations between these dispositions and psychological problems that does not use overlapping symptoms of mental disorders.

Negative Emotionality

Children who are high on the disposition of negative emotionality respond frequently and intensely with negative emotions to loss, threat, and frustration, and recover more slowly after the stressor ends. The particular negative emotions evoked in each instance depend on the nature of the stressor, the circumstances, and other characteristics of the individual (Compas et al., 2004; Servaas et al., 2013). Negative emotionality can be thought of as essentially the important trait of neuroticism, which plays a central role in many models of personality (Barbaranelli et al., 2003; Eysenck, 1947; Lahey, 2009; McCrae et al., 2002), but defined without the many contaminating synonyms for symptoms of mental disorders that most measures of neuroticism contain.

Across the life span, persons who are high on negative emotionality are at increased risk for essentially every dimension of psychological problems (Jeronimus et al., 2016; Lahey, 2009). For many reasons, persons who are disposed to respond strongly to aversive events with negative emotions have an increased likelihood of developing psychological problems. Persons who are higher in negative emotionality have a bias for paying attention to negative information, giving themselves more opportunities to respond with negative emotions (Servaas et al., 2013) and they are particularly sensitive to signs of rejection by others (Brookings et al., 2003; Cain et al., 2017). Furthermore, negative emotionality influences how easily we learn psychological problems from our experiences in two ways. First, negative emotional behavior provide the “raw material” that our experiences shape into specific psychological problems such as oppositional and aggressive behaviors through coercive transactions with others (Snyder et al., 2003). In addition, persons high in negative emotionality develop conditioned fears more easily (Servaas et al., 2013).

Prosociality

Individual differences in caring about the wellbeing of others, attempting to help and please others, enjoying the company of others, and experiencing guilt over misbehaviors is an extensively studied aspect of socioemotional behavior (Flook et al., 2019; Pfattheicher et al., 2022; Spinrad & Eisenberg, 2014). Furthermore, the widely studied construct of callousness, which is part of callous-unemotional traits, is essentially the opposite of prosociality (Frick et al., 2014; Lahey, 2014; Waldman et al., 2011). Children who are high on dispositional prosociality are friendly, helpful, caring, and obedient (Lahey et al., 2008, 2010), and, therefore, are likely to evoke mostly positive social behavior from adults and peers. In addition, when children who are high in prosociality behave in ways that upset others, the natural consequences of their behaviors are likely punishing. Seeing that others have been upset by their words or actions would likely punish those behaviors in children who are high in prosociality. Furthermore, children high in prosociality care about the opinions of other people and are hypothesized to be more responsive to social rewards, fostering the learning of adaptive behavior (Lahey & Waldman, 2012). At the other end of the prosociality dimension, children low in prosociality are not motivated to please others and likely find the adverse consequences that their misbehaviors have on others – such as making another child cry – to be neutral or even reinforcing (Frick & Viding, 2009).

Daring

Children who are low on daring stay clear of risky situations and potential dangers. In contrast, persons who are high on daring find intense and risky situations to be attractive and rewarding. Thus, the rough-and-tumble of fighting, the intense stimulation of vandalism, and the risk of being caught when shoplifting are attractive and positively reinforcing to highly daring children and likely teach them to be antisocial, whereas these same consequences would be aversive to children who are high in daring. Daring overlaps with the well-studied concepts of sensation seeking (Russo et al., 1993; Zuckerman & Aluja, 2015) at one end of the the continuum, and behavioral inhibition (Biederman et al., 2001; Schwartz et al., 1996) and low risk tolerance (Karlsson Linnér et al., 2019) at the other end.

Cognitive Dispositions to Psychological Problems

Individual differences in cognitive skills constitute another dispositional dimension that is related to many if not all dimensions of psychological problems (Craig et al., 2016; Harden et al., 2019; Nuno et al., 2021; Vilgis et al., 2015; Zainal & Newman, 2022). More research is needed on the optimal way to define such cognitive disposition, as it is still not clear how to parse general intelligence, language development, and executive functioning (Lahey et al., 2014; Raine et al., 2005). The CADS does not measure individual differences in cognitive skills based on the assumption that they are better measured by standardized tests. Cognitive dispositions are not included in the present new exemplary analyses because limited cognitive measures were included in the TTS, but cognitive dispositions must be included in future research.

Joint Modeling of Dispositions and Psychological Problems

Correlations among the three dimensions of the parallel parent and youth versions of the CADS in the TTS are shown in Table 1 (Lahey et al., 2008, 2010). The youth-rated dimensions are essentially orthogonal, with the highest correlation between factors being less than 0.10. This degree of orthogonality is greater than in other measures, such as measures based on the five-factor model. For example, not counting openness to experience which addresses a dimension not included in the CADS, correlations among high-school students between the five-factor model dimensions were: neuroticism and extraversion: −0.28; neuroticism and agreeableness −0.18; neuroticism and conscientiousness −0.18; extraversion and agreeableness and 0.25; extraversion and conscientiousness 0.16; agreeableness and conscientiousness 0.17 (McCrae & Costa, 2004).

Table 1. Pearson correlations among dimensions within the parent and youth versions of the Child and Adolescent Dispositions Scales (CADS) in the Tennessee twins study (N = 4,049) (Lahey et al., 2008, 2010).

	Negative emotionality	Prosociality	Daring
Parent CADS
Negative emotionality		−.297	0.140
Prosociality			−0.149
Youth CADS
Negative emotionality		−.027	0.063
Prosociality			−0.084

In contrast, Table 1 shows that the parent-rated CADS dimensions exhibit small, but significant correlations that are comparable in magnitude to the correlations among five-factor dimensions assessed by a parent rating scale (Barbaranelli et al., 2003). Therefore, variables of interest must be regressed on all three parent-rated CADS dimensions simultaneously to quantify the independent association of each disposition. For example, when the number of symptoms of adult antisocial personality assessed 12 years later was regressed on the three CADS dispositional dimensions measured by youth reports in the TTS at 9–17 years of age, each CADS disposition independently predicted adult antisocial behavior. Using the parent CADS, negative emotionality and prosociality independently predicted adult antisocial behavior (Lahey et al., 2018). Similar statistical strategies can be used to identify variables of etiologic and mechanistic significance that line up with the three dispositional constructs.

Patterns of Phenotypic Associations

To illustrate the hypothesized model, cross-cutting associations were estimated in new analyses for this paper by regressing 10 illustrative fine-grain dimensions of psychological problems on the three CADS dimensions in separate multiple regression analyses using data from the TTS adjusting for age, sex, and race-ethnicity and taking the clustering within twin pairs into account (see Figure 1). The 10 dimensions of psychological problems used in the analyses are based on previous factor analyses of data collected using the Child and Adolescent Psychopathology Scale (CAPS) in the TTS (Lahey et al., 2008). Demographic characteristics of the TTS sample are presented in Table S1. The CAPS was developed to measure DSM-IV symptoms of mental disorders and other psychological problems over the past 12 months in dimensional terms, with each individual item rated on the same 4-point scale as the CADS.² Aggressive and nonaggressive conduct disorder were distinguished in the present illustrative analyses owing to previous evidence of etiologic and developmental differences between them (Burt, 2013; Lahey et al., 2000).

Figure 1. Cross-cutting associations between 10 dimensions of parent-rated psychological problems and three parent-rated CADS dispositional dimensions, adjusting for age, sex, family income, and race-ethnicity in 9–17 year old participants in the Tennessee twin study. Dimensions of dispositions were standardized to mean of 0 and standard deviation of 1. Arrows between dispositions and problem dimensions are quantified as standardized regression coefficients that were significant after 5% FDR correction for the 30 tests of associations (dashed lines represent significant inverse associations).

Figure 1 shows that each of the 10 parent-rated psychological problem dimensions measured by the CAPS was significantly associated with at least one parent-rated CADS dispositional dimension after controlling for multiple testing in the TTS, but every dimension of psychological problems differed from the others in terms of the magnitudes (and sometimes the directions) of their associations with the CADS dispositions. Figure 2 shows that the same pattern of associations was found even across informants (i.e., between youth self-ratings on the dispositions and parent ratings of psychological problems).

Figure 2. Cross-cutting associations between 10 dimensions of parent-rated psychological problems and three youth self-rated CADS dispositional dimensions, adjusting for age, sex, family income, and race-ethnicity in 9–17 year old participants in the Tennessee twin study. Dimensions of dispositions were standardized to mean of 0 and standard deviation of 1. Arrows between dispositions and problem dimensions are quantified as standardized regression coefficients that were significant after 5% FDR correction for the 30 tests of associations (dashed lines represent significant inverse associations).

Associations with Negative Emotionality

Notably, Figures 1 and 2 show that every dimension of psychological problems is positively associated with CADS negative emotionality. This implies that the causes of individual differences in negative emotionality also increase the likelihood of every dimension of psychological problems and contribute to the ubiquitous positive correlations among every dimension of psychological problems. Note that these findings complement our previous analyses of the associations between dispositions and psychological problems using bifactor models (Moore et al., 2020). For example, correlations of CADS negative emotionality with each of the correlated dimensions of psychological problems is consistent with our previous finding that negative emotionality is robustly associated with the general factor of psychological problems, which is defined by the phenotypic variance shared by all dimensions of psychological problems (Class et al., 2019).

It is equally important to note, however, that negative emotionality is associated with each dimension of psychological problems in these analyses to varying degrees in the TTS. For examples, parent-rated negative emotionality accounted for 36.0% of the variance in aggressive conduct problems, 11.4% of the variance in oppositional defiant problems, 10.2% of the variance in depression, and 4.8% of the variance in specific fears. These marked variations in the magnitudes of associations of negative emotionality with each individual dimension of psychological problems suggest that the various dimensions of psychological problems are differentiated to a degree partly because they are associated with negative emotionality to varying extents.

Associations with Prosociality and Daring

Figures 1 and 2 also show that every dimension of psychological problems is distinct in being associated to varying degrees with at least one other CADS disposition in addition to negative emotionality in the present analyses. For examples, inverse associations with parent-rated prosociality accounted for 31.4% of the variance in aggressive conduct problems and 2.9% of the variance in inattention, whereas the positive association of prosociality explained 0.5% of the variance in separation anxiety. The positive association with daring accounted for 4.8% of the variance in aggressive conduct problems, whereas inverse associations with daring accounted for 3.6% of the variance in social anxiety and 4.0% of the variance in fears.

The patterns of associations shown in Figures 1 and 2 suggest the hypothesis that individual differences in prosociality and daring reflect causal processes that influence that structure of correlations among dimensions of psychological problem in two opposing ways: On the one hand, the processes underlying variations daring and prosociality cause some psychological problems to be even more correlated than would be expected on the basis of their shared associations with negative emotionality. For example, the externalizing dimensions of inattention, hyperactivity, oppositional-defiant behavior, aggressive conduct problems, and nonaggressive conduct problems are substantially correlated partly because they all share positive correlations with daring.

On the other hand, the processes underlying variations in daring and prosociality could give rise to differentiations among the dimensions of psychological problems. This is seen in the variations in magnitudes of associations of prosociality and daring with dimensions of psychological problems. Strikingly, daring is positively associated with the externalizing dimensions and inversely associated with the fear-based dimensions.

Non-Linear Associations and Interactions Between CADS Dimensions

It is important to consider the linearity of associations between the CADS dimensions and psychological problems presented in Figures 1 and 2. Mapping variations in the magnitudes of risk for psychological problems onto the full range of rated dispositions is useful both for theory and possible clinical applications of ratings of dispositions. In addition, it is necessary to identify nonlinear associations to fully evaluate the degree of independence of the associations of each three CADS dimensions with psychological problems. This is because unusually sharp departures from linearity, such as U-shaped associations, could make independence difficult to assess. For these reasons, the regression analyses TTS data were repeated including quadratic terms for each CADS dimension. As show in Table S3, the quadratic components of the associations of CADS negative emotionality with 9 of the 10 CAPS dimensions of psychological problems were significant after adjusting for multiple testing, reflecting accelerating strengths of association in each case. That is, for each dimension of problems, the same difference in ratings of negative emotionality was associated with a progressively greater risk for problems at higher levels of negative emotionality. To illustrate these findings, the two associations of negative emotionality with problems that deviated most from linearity (aggressive and nonaggressive conduct problems) are presented on the left side of Figure 3. There were also significant decelerating curvilinear associations between CADS daring with generalized anxiety and significant decelerating curvilinear associations between CADS prosociality and inattention and both aggressive and nonaggressive conduct problems (Table S3). The latter two associations are plotted on the right side of Figure 3.

Figure 3. Plots of parent ratings of aggressive (top) and nonaggressive (bottom) DSM-IV conduct disorder behaviors (means and standard errors of ratings of child and adolescent psychopathology scale items transformed to a 0–3 scale) against sample deciles of parent-rated negative emotionality and prosociality among 6–17 year olds in the Tennessee Twins study, adjusted for clustering within twin pairs. Dotted lines are best-fit exponential lines.

It is also important to estimate interactions among each of the CADS dispositions in the associations with psychological problems to fully understand the degree of independence of their associations. As shown in Table S4, the only significant interactions detected among the three CADS dimensions after correcting for multiple testing were between negative emotionality and prosociality in their associations with the externalizing dimensions of inattention, oppositional-defiant problems, and aggressive conduct problems. In each case, the association of the problem dimension with negative emotionality was weaker when levels of prosociality were greater, and vice-versa. No interactions of dispositions with the child’s sex were detected at even nominally levels of significance for either parent- or youth-rated CADS dimensions in their associations with any dimension of psychological problems.

These significant interactions suggest that the extent to which greater negative emotionality is associated with more externalizing problems is lessened in the presence of greater prosociality reflects more than an additive combination of the two dispositions. The interpretation of this interaction is ambiguous, however, because of the curvilinear associations of negative emotionality and prosociality with externalizing problems and the small inverse correlation between these dimensions when rated by parents. This limitation does not detract from the fundamentally important overall conclusion, however, that the likelihood of exhibiting psychological problems reflects the combination of variations in all three CADS dispositions.

Using Dispositions to Understand the Heterogeneity of Psychological Problems

The greatest roadblock to research on causes and mechanisms is the heterogeneity of each dimension. Fortunately, the findings presented in Figure 1 suggest an important framework for understanding the heterogeneity of each dimension of psychological problems. This refers to the widely suspected possibility that different individuals may display high levels of the same dimension of psychological problems (or meet criteria for the same mental disorder) for different reasons (Insel et al., 2010; Lahey et al., 2017; Nigg, 2022). This issue is critically important because it suggests that different children and adolescents with similar psychological problems may respond best to different treatments.

The present analyses suggest the hypothesis that the heterogeneity among dimensions of psychological problems arises in part because every dimension of psychological problems is associated with more than one dispositional dimension. For example, highly inattentive children who only exhibit higher levels of negative emotionality might experience biases in attention away from tasks to emotion-related cues and they may be inattentive because they disengage from challenging tasks before they are completed (Marco et al., 2009). In contrast, inattentive children who exhibit only lower prosociality might be said to be inattentive because they are not motivated to persevere on challenging or uninteresting tasks to please adults. Furthermore, other inattentive children who are high only in daring might find school tasks to be boring and seek more exciting stimulation instead. Additionally, of course, children who exhibit inattention could do so for any combination of elevations on the three CADS dispositions or because of lower cognitive/executive functioning.

Note that I am not proposing the distinction of categorical subtypes of categorical mental disorder based on individual differences in these dispositions, such as have been advocated for ADHD (Nigg, 2022). That approach is reasonable, but dichotomizing continua results in the loss of information and creates unreliable categories that change over time (Lahey et al., 2005). Rather, I believe the heterogeneity among dimensions of psychological problem is better understood by studying quantitative variations in dispositional dimensions.

Taken together, these marked differences in the associations of dispositions with problem dimensions are one reason that I disagree with the hypothesis that a single “one for all” general dimension (Caspi & Moffitt, 2018) should replace the multiple dimensions of psychological problems. In my view, there are meaningful specific differences among the multiple correlated dimensions of psychological problems that should not be de-emphasized or lost entirely (Lahey, 2021).

Evidence of Causal Links Between CADS Dispositions and Problem Dimensions

The hypothesized role of dispositions in the heterogeneity among children with the same problems implies that dimensions of psychological problems are poor targets for genetic and brain research. In contrast, orthogonal dispositional dimensions are more tractable targets for study. Therefore, to understand psychological problems, we need a substantial investment in the neuroscience of dispositions (DeYoung, 2010). For the reasons stated above, well-powered studies are needed using measures of dispositions that are reliable and valid and do not yield circular findings because they use synonyms and antonyms of symptoms of psychopathology to define the dispositions. This rules out the use of most current measures of personality and temperament (Lahey, 2004; Lahey et al., 2010).

This change in the focus of research only makes sense, of course, if there is reason to believe that the etiology and mechanisms of dispositions are actually related to the origins of psychopathology. That requires the existence of shared causal pathways between dispositional dimensions and psychological problems. Fortunately, there is already strong evidence that they exist. One finding relevant to this question has been replicated in two independent studies of child and adolescent twins. In these studies, the CADS dispositional dimensions were not only found to be moderately heritable, but the phenotypic associations of externalizing problems with CADS negative emotionality, prosociality, and daring were largely explained by shared genetic and environmental influences (Taylor et al., 2013; Waldman et al., 2011). Furthermore, shared causal influences between negative emotionality and internalizing problems were identified (Mikolajewski et al., 2013; Tackett et al., 2011). Indeed, there is evidence of substantial shared causal variance between negative emotionality and the variance common to all problem dimensions (Tackett et al., 2013). Additional studies using dispositional measures similar to negative emotionality and daring also have found considerable overlap between the genetic influences on both dispositions and internalizing and externalizing problems (Harden et al., 2017; Hink et al., 2013). Thus, there is solid evidence that the phenotypic correlations of each CADS disposition with every dimension of psychological problems reflect shared genetic and environmental influences (Tackett et al., 2013).

The present proposal to use dispositions to understand the nature of correlated dimensions of psychological problems has potentially profound implications for understanding the heterogeneity of each problem dimension. The levels of each dimension of psychological problems can be influenced by more than one causal pathway, and can be influenced by different combinations of pathways that work together additively and sometimes interactively. The cross-cutting associations of dispositional dimensions with problem dimensions provide an essential key to understanding the casual complexity of psychological problems. Sorting out the causes and psychobiological mechanisms of dimensions of psychological problems without using dispositional construct would be daunting at best and may be impossible at worst (Zald & Lahey, 2017). Consistent with the RDoC perspective (Beauchaine & Hinshaw, 2020; Cuthbert & Insel, 2013), we posit that focusing on the causes and psychobiological mechanisms associated with each orthogonal disposition will be far more tractable and informative than studies of the problem dimensions themselves. There may well be, of course, additional influences on problem dimensions that are not shared with any set of measured dispositions. If so, however, these will be far easier to identify after the causal processes associated with dispositions have been discovered. Notably, the approach outlined in this paper is very similar to, and has benefitted from an earlier paper on dispositions and “transdiagnostic neural vulnerabilities to psychopathology” (Beauchaine & Hinshaw, 2020). I monistically view the these vulnerabilities as dispositions at the behavioral level of analysis and mechanisms at the biological level of analysis.

Dispositions and Transactions

Dispositions do not influence risk for psychopathology on their own, of course. Following many others (Anderson et al., 1986; Beauchaine & McNulty, 2013; Klein et al., 2018; Sameroff, 2009; Shiner & Masten, 2002), I posit that dispositions transact with the environment to give rise to the dimensions of psychological problems (Lahey, 2021). Environments do not passively shape our psychological problems. We actively transact with our environments, meaning that our environments influence our behavior, our behavior and other characteristics influence our environments, and our characteristics moderate the extent to which our experiences influence our behavior. Many child characteristics influence our transactions with the environment, including age, sex, and race and ethnicity, but individual differences in cognitive and socioemotional dispositions almost certainly play key roles in transactions with the environment that result in psychological problems. No two people live in the same world and, if they did, they would not react to it, or be shaped by it, in the same ways.

Summary

We will be able to develop more effective methods of preventing and treating psychological problems when we have a better understanding the genetic and environmental factors that make some children more likely to develop distressing and impairing psychological problems, and of the biological pathways through which these causal factors operate. I posit that we have long been asking the wrong questions about these casual pathways, however. It is simply not possible to identify the causes and mechanisms underlying each dimensions of psychological problems because they are far too heterogeneous and too highly correlated with one another to line up with distinct causal pathways. Fortunately, it is quite feasible to identify the causes and mechanisms underlying the small number of orthogonal dispositional dimensions such as negative emotionality that are correlated with psychological problems in cross-cutting patterns. These dispositions are associated with psychological problems and are already known to share causal pathways with psychological problems, but they have been studied far less than dimensions of psychopathology. I have outlined a strategy for studying the heterogeneous causes and mechanisms of psychological problems by defining dispositions independently and studying their causes and mechanisms. This will require a continued shift in the allocation of resources away from studies of the causes and mechanisms of mental disorders to well-powered studies of the biology of dispositions and their transactions over time with the environment.

Disclosure Statement

No potential conflict of interest was reported by the author(s).

Supplementary Data

Supplemental material for this article can be accessed online at https://doi.org/10.1080/15374416.2023.2292050

Additional information

Funding

This work was supported by the National Institute of Mental Health [R01 MH59111].

Notes

¹ I use the term psychological problems as a denotative synonym for psychopathology to avoid the stigmatizing connotations that it reflects pathology – a sick mind (Lahey, 2021). Disposition is a synonym for temperament or personality traits, but without the theoretical baggage regarding age of onset, heritability, and other issues often attached to those terms (Mischel, 2004).

² The 10 unit-weighted dimensions of psychological problems used in the present regression analyses were chosen to illustrate the hypothesized variations of patterns of associations of the three CADS dimensions with psychological problems at a fine-grain level. That is, a structural equation model with a confirmatory factor analytic measurement model of these 10 dimensions was not conducted. Because the 10 dimensions are correlated, other published findings have shown that either a correlated factors model with only three broad factors (externalizing, internalizing, and attention-deficit hyperactivity problems) or a bifactor model with a general factor and the same three specific factors would have provided better model fits. Associations of the CADS with these less fine-grain broad dimensions of psychological problems have been reported by Class et al. (2019).