Skip to main content
PMC home page
Psychological Science logoLink to Psychological Science
. 2020 Jul 27;31(8):1025–1035. doi: 10.1177/0956797620938160

The Latent Genetic Structure of Impulsivity and Its Relation to Internalizing Psychopathology

Daniel E Gustavson 1,2,3,✉, Naomi P Friedman 4,5, Pierre Fontanillas 6, Sarah L Elson 6; the 23andMe Research Team6, Abraham A Palmer 3,7, Sandra Sanchez-Roige 3
PMCID: PMC7427138  PMID: 32716714

Abstract

Factor analyses suggest that impulsivity traits that capture tendencies to act prematurely or take risks tap partially distinct constructs. We applied genomic structure equation modeling to evaluate the genetic factor structure of two well-established impulsivity questionnaires, using published statistics from genome-wide association studies of up to 22,861 participants. We also tested the hypotheses that delay discounting would be genetically separable from other impulsivity factors and that emotionally triggered facets of impulsivity (urgency) would be those most strongly genetically correlated with an internalizing latent factor. A five-factor model best fitted the impulsivity data. Delay discounting was genetically distinct from these five factors. As expected, the two urgency subscales were most strongly related to an internalizing-psychopathology latent factor. These findings provide empirical genetic evidence that impulsivity can be broken down into distinct categories of differential relevance for internalizing psychopathology. They also demonstrate how measured genetic markers can be used to inform theories of psychology and personality.

Keywords: heritability, genomic structural equation modeling, self-control, UPPS-P Impulsive Behavior Scale, Barratt Impulsiveness Scale, open data, open materials

Impulsivity is a construct common to many theories of personality (Evenden, 1999; Eysenck & Eysenck, 1985; Tellegen, 1982). Impulsive personality traits (IPTs) typically refer to a tendency to act without planning or self-control (lack of premeditation), the inability to resist temptations while experiencing positive or negative affect (positive urgency and negative urgency), the inability to persist on difficult tasks (lack of perseverance), or the tendency to enjoy exciting situations (sensation seeking). These five IPTs are measured by the UPPS-P Impulsive Behavior Scale (Lynam, Smith, Whiteside, & Cyders, 2006; Whiteside & Lynam, 2001), arguably the most common questionnaire assessing impulsivity. Another widely used instrument is the Barratt Impulsiveness Scale (BIS), which also focuses on the tendency to act without premeditation (Barratt, 1993; Patton, Stanford, & Barratt, 1995).

IPTs are phenotypically and neurologically dissociable from one another (Dalley, Everitt, & Robbins, 2011; MacKillop, Weafer, Oshri, Palmer, & de Wit, 2016; Reynolds, Ortengren, Richards, & de Wit, 2006) and show divergent associations with psychiatric disorders, particularly substance use and internalizing psychopathology (Cyders & Smith, 2007; Johnson, Carver, & Joormann, 2013). For example, a meta-analysis of UPPS-P (115 studies, 40,432 individuals; Berg, Latzman, Bliwise, & Lilienfeld, 2015) revealed that only three of the five UPPS-P subscales were associated with substance use, and only UPPS-P Positive Urgency and Negative Urgency were associated with anxiety or depression symptoms. IPTs are also thought to be related to delay discounting (a tendency to devalue future events or rewards; Moreira & Barbosa, 2019), although these constructs sometimes show small or divergent associations (Murphy & Mackillop, 2012; Reynolds et al., 2006).

IPTs also appear to be genetically dissociable. The largest genome-wide association study (GWAS) of IPTs to date (Sanchez-Roige et al., 2019) showed only moderate genetic correlations between UPPS-P and BIS subscales. In addition, the UPPS-P Sensation Seeking subscale was only weakly genetically correlated with other IPTs and was instead more strongly genetically correlated with extraversion. Furthermore, the UPPS-P Lack of Perseverance subscale was weakly genetically correlated with other IPTs. These findings are consistent with results from twin studies (Gustavson et al., 2019).

To study the multifaceted nature of impulsivity, we used a recently introduced method, genomic structural equation modeling (SEM; Grotzinger et al., 2019), which applies SEM methods to genetic correlations on the basis of GWAS results, using the same techniques as SEM based on phenotypic correlations. Because all GWASs use the same ancestral genome as a reference, their summary statistics across different traits and participants can be linked through this common reference. Thus, genetic covariances and correlations can be estimated between any pair of GWAS traits, provided that both samples were drawn from the same ancestral background (Bulik-Sullivan et al., 2015). Whereas modeling based on phenotypic or twin correlations requires all traits to be measured in the same sample, genomic SEM does not, greatly expanding the range of traits and models that can be examined. Moreover, because GWAS analyses do not rely on siblings, they are not subject to the same potential biases that arise from the assumptions of twin studies (e.g., the equal-environments assumption).

The aims of this study were threefold (Fig. 1). First, we leveraged data from our GWAS of UPPS-P and BIS subscales (Sanchez-Roige et al., 2019) to examine the latent genetic structure of impulsivity; namely, we tested multiple competing hypotheses about whether single or multiple genetic factors are needed to capture the genetic structure of IPTs. The UPPS-P Negative Urgency and Positive Urgency subscales are often considered two facets of a higher-order factor reflecting emotion-based rash action, given their phenotypic similarities (Cyders & Smith, 2007; Littlefield, Stevens, Ellingson, King, & Jackson, 2016). Similarly, there is some evidence that the UPPS-P Lack of Premeditation and Lack of Perseverance subscales might load on a common factor representing deficits in conscientiousness (Cyders & Smith, 2007; Whiteside & Lynam, 2001). Therefore, we compared models that had five unique IPT factors with models that had three or four IPT factors by collapsing (a) positive and negative urgency and (b) lack of premeditation and perseverance. We also considered whether all facets would load on a single factor. In these analyses, we used published GWAS results from an independent study of extraversion (van den Berg et al., 2016) as an additional indicator of sensation seeking. We did this for several reasons. First, UPPS-P Sensation Seeking was genetically correlated with extraversion in our earlier work (Sanchez-Roige et al., 2019). In addition, sensation seeking has long been conceptualized as a component of extraversion in theories of personality (Costa & McCrae, 1992; Eysenck, 1993), and both constructs are measured using questionnaires that contain some nearly identical items.

Fig. 1.

Fig. 1.

Open in a new tab

Visual representation of the study aims. In all models, summary statistics for individual impulsive personality traits (IPTs) are represented by squares, latent factors are represented by ovals, single-headed arrows indicate factor loadings, and double-headed arrows indicate correlations. Aim 1 was to evaluate whether the five IPTs were captured by separate genetic factors or whether certain facets could be collapsed together (e.g., positive and negative urgency, and lack of perseverance and lack of premeditation, as in Cyders & Smith, 2007). In these models, UPPS-P subscales were initially modeled as separate factors, with the Barratt Impulsiveness Scale (BIS) total score loading on the lack-of-premeditation factor. Extraversion was included as a second indicator of sensation seeking to aid in model fit (see the Supplemental Material available online). Aims 2 and 3 were evaluated simultaneously by adding delay discounting and internalizing psychopathology in the same model to provide the maximum information in the genetic-correlation matrix. Aim 2 evaluated whether genetic influences on delay discounting were best modeled as an independent genetic factor or a facet of lack of premeditation. Aim 3 was to evaluate the hypothesis that a latent factor capturing genetic influences on internalizing psychopathology would be most strongly genetically correlated with IPT genetic factors related to control over emotion-based rash action (positive urgency, negative urgency, or their combination, depending on Aim 1). Genetic correlations among IPTs are shown in gray for simplicity. Thicker black arrows indicate stronger hypothesized correlations. NU = UPPS-P Negative Urgency subscale; PU = UPPS-P Positive Urgency subscale; Prem = UPPS-P Lack of Premeditation subscale; Pers = UPPS-P Lack of Perseverance subscale; SS = UPPS-P Sensation Seeking subscale; Extr = extraversion; MDD = major depressive disorder; Neur = neuroticism; SWB = subjective well-being.

Second, we examined whether delay discounting could be modeled as a common or separate genetic factor. We hypothesized that delay discounting could be an indicator of lack of premeditation, on the basis of previous work showing strong genetic correlations between the two facets (Sanchez-Roige et al., 2019). Delay discounting captures the valuation of future versus present rewards, whereas lack of premeditation captures the tendency to act without thinking about future consequences. Because both delay discounting and lack of premeditation involve the consideration of future outcomes, they may reflect a common factor. We evaluated this possibility against a model in which delay discounting represents a distinct genetic factor that is simply more correlated with lack of premeditation than other IPTs.

Finally, we leveraged published GWASs of phenotypes related to internalizing psychopathology to further inform the genetic structure of IPTs. Previous work has shown that IPTs were associated with internalizing psychopathology, including self-report and diagnostic assessment of major depressive disorder and generalized anxiety disorder (Berg et al., 2015), and that these correlations were driven by shared genetic influences (Gustavson et al., 2019). Thus, we evaluated whether genetic separability among IPTs was accompanied by differential relations to an internalizing psychopathology factor based on data from well-powered published GWASs of major depressive disorder (Howard et al., 2018), neuroticism (Luciano et al., 2018), and subjective well-being (Okbay et al., 2016). Building on previous phenotypic analyses (Berg et al., 2015; Carver & Johnson, 2018), we hypothesized that specific IPTs, particularly those pertaining to emotion regulation (i.e., negative and positive urgency), would be more strongly genetically correlated with internalizing psychopathology than other IPTs, supporting their distinction from other IPTs.

Method

Genome-wide association studies

Impulsive personality traits

We used GWAS summary statistics for IPTs from our previously published work (Sanchez-Roige et al., 2019); these association results included measures from the UPPS-P Impulsive Behavior Scale (Cyders, Littlefield, Coffey, & Karyadi, 2014; Whiteside & Lynam, 2001) and the BIS (Patton et al., 1995). The 20-item brief version UPPS-P Impulsive Behavior Scale includes 4 items for each subscale (Lack of Premeditation, Lack of Perseverance, Positive Urgency, Negative Urgency, and Sensation Seeking). Although the 30-item BIS is composed of three subscales (Attentional, Motor, and Nonplanning), genetic correlations among the subscales were essentially 1.0 (Sanchez-Roige et al., 2019), suggesting that the BIS subscales largely capture a single set of genetic influences related to lack of premeditation. Therefore, we limited our analyses to the BIS total score. All research participants included in the analyses were of European ancestry and were from the 23andMe website. The final number of research participants included in the analyses ranged from 21,495 to 22,861. These data sets have been extensively described elsewhere (Sanchez-Roige et al., 2019).

Extraversion

Publicly available GWAS summary statistics for extraversion were obtained from a recent meta-analysis of 63,030 individuals of European ancestry (van den Berg et al., 2016). Extraversion was assessed with a harmonized measure across 29 cohorts using common measures of extraversion, including the NEO Personality Inventory, the NEO Five-Factor Inventory, the Eysenck Personality Questionnaire (EPQ), the Eysenck Personality Inventory (EPI), the Reward Dependence scale of Cloninger’s Tridimensional Personality Questionnaire, and the Positive Emotionality scale of the Multidimensional Personality Questionnaire. Multiple items from these questionnaires refer to the tendency to enjoy and seek out exciting situations—for example, “Do you often long for excitement?” (EPI), “Would you do almost anything for a dare?” (EPI), and “Do you like plenty of action and excitement around you?” (EPQ). Although we included extraversion in our initial model of IPTs to aid in model identification, similar models were evaluated without extraversion, and its inclusion did not alter the pattern of results (see the Supplemental Material available online).

Delay discounting

We used GWAS summary statistics for delay discounting from our previous study (Sanchez-Roige et al., 2018), which included 23,127 European research participants from 23andMe. Participants completed the 27-item Monetary Choice Questionnaire, a widely used measure of delay discounting (Kirby, Petry, & Bickel, 1999).

Internalizing psychopathology

We used publicly available summary statistics for internalizing psychopathology from three independent GWASs: depression (170,756 cases and 329,443 controls; Howard et al., 2018), neuroticism (N = 390,278; Luciano et al., 2018), and subjective well-being (N = 298,420; Okbay et al., 2016). All individuals were of European ancestry. Depression was assessed in the UK Biobank on the basis of whether an individual had a diagnosis of a depressed mood disorder from linked hospital records or had answered “yes” to either of the following questions at any assessment: “Have you ever seen a general practitioner (GP) for nerves, anxiety, tension or depression?” or “Have you ever seen a psychiatrist for nerves, anxiety, tension or depression?” Neuroticism was also assessed in the UK Biobank with a 12-item version of the EPI Revised Short Form (Luciano et al., 2018). Subjective well-being was assessed with multiple study-specific measures, although the majority used validated life-satisfaction scales, such as the Satisfaction with Life Scale or the Geriatric Depression Scale (Okbay et al., 2016). Subjective well-being was reverse scored in these analyses, so higher scores indicated lower well-being (i.e., more internalizing problems).

Data analyses

Analyses were conducted using the genomicSEM package for R (Grotzinger et al., 2019), a novel statistical method that applies SEM methods to GWAS results. Genomic SEM is an extension of linkage-disequilibrium-score regression (Bulik-Sullivan et al., 2015), which calculates genetic correlations with any two traits for which summary statistics are available, provided the samples were drawn from the same ancestral background. Using linkage-disequilibrium-score regression, genomic SEM computes a full genetic-correlation matrix across the set of traits for which GWAS summary statistics are provided and then estimates the model with this correlation matrix using the lavaan package (Version 0.5–12) in R. Table S1 (in the Supplemental Material) and Figure 2 display the final genetic-correlation matrix for analyses of Aims 2 and 3 that includes all study variables.

Fig. 2.

Fig. 2.

Open in a new tab

Genetic-correlation matrix generated by genomic structural equation modeling (SEM) for Aim 2 and Aim 3 analyses involving all study measures. Matrices for Aim 1 (impulsive personality traits, or IPTs, only) were similar but not identical because each matrix was generated separately in genomic SEM. See Table S1 in the Supplemental Material for exact r values. Subjective well-being was reverse scored. UPPS-P = UPPS-P Impulsive Behavior Scale; BIS = Barratt Impulsiveness Scale.

Most of the summary statistics included in the analyses were based on overlapping samples (e.g., 23andMe, UK Biobank); however, this method adjusts for sample overlap by estimating a sampling covariance matrix that indexes the extent to which sampling errors of the estimates are associated (Grotzinger et al., 2019). The R data files containing these genomic-SEM matrices for all analyses, along with R analysis scripts, are displayed at https://osf.io/5x3ft/. Table S1 also displays the genomic-SEM matrix for Aim 3 (which includes all measures examined here).

We applied genomic SEM to test confirmatory factor models that were informed by psychology and psychometric theories (Carver & Johnson, 2018; Cyders & Smith, 2007). We used the default diagonally weighted least-squares estimation method. We used a series of metrics to evaluate the best-fitting confirmatory factor model. Specifically, model fit was determined using chi-square tests, the comparative fit index (CFI), and the Akaike information criterion (AIC). Models with good fit are expected to have CFIs greater than .95 and smaller AIC values than competing nested models (Hu & Bentler, 1998). Good-fitting models also traditionally have nonsignificant χ2 statistics, but because GWAS sample sizes are extremely large, and this statistic is sensitive to sample size, we focused on other fit indexes. Significance of individual parameter estimates were established with 95% confidence intervals (CIs) and with χ2 difference tests (Δχ2). When fitting models with only two indicators, we equated standardized factor loadings to help identify the model. In some cases, dummy latent factors were created when we had only a single indicator (e.g., lack of perseverance), with fixed factor loading at 1.0 on their single indicators (and no residual variance). We refer to these as “factors” in the Results section, but they should be interpreted as the single indicators that they represent.

Our recent GWAS of IPTs (Sanchez-Roige et al., 2019) indicated that we had sufficient power to observe genetic correlations between IPTs and the other traits examined here. Power to detect correlations is typically larger for latent constructs.

Results

Genomic SEM of impulsivity facets

We first fitted a genomic structural equation model using GWAS data from the UPPS-P subscales, the BIS total score, and extraversion. Table 1 displays model comparisons, and Figure 3 shows the best-fitting model (Model 1; see Table S2 in the Supplemental Material for 95% CIs), χ2(9) = 12.52, p = .185, CFI = .959, AIC = 50.52. This model included all five factors: negative urgency (UPPS-P Negative Urgency), positive urgency (UPPS-P Positive Urgency), lack of premeditation (UPPS-P Lack of Premeditation and BIS total score), sensation seeking (UPPS-P Sensation Seeking and extraversion), and lack of perseverance (UPPS-P Lack of Perseverance).

Table 1.

Comparison of Genetic Structural Equation Models of Impulsivity Facets

Model χ2 df p CFI AIC Comparison with Model 1
Δχ2 df p
Model 1: five factors 12.52 9 .185 .959 50.52
Model 2: four factors (UPPS-P Positive and Negative Urgency collapsed) 34.96 12 < .001 .807 66.96 22.44 3 < .001
Model 3: four factors (BIS total score, UPPS-P Lack of Perseverance, and UPPS-P Lack of premeditation collapsed) 18.13 11 .079 .939 52.13 5.61 2 .061
Model 4: single factor 63.14 14 < .001 .634 91.14 50.62 5 < .001
Open in a new tab

Note: The best-fitting model is displayed in boldface (and shown in Fig. 3). CFI = comparative fit index, AIC = Akaike information criterion; UPPS-P = UPPS-P Impulsive Behavior Scale; BIS = Barratt Impulsiveness Scale.

Fig. 3.

Fig. 3.

Open in a new tab

Best-fitting model of the genetic-factor structure of impulsivity facets. Individual impulsive personality traits (IPTs) are represented by squares, latent factors are represented by ovals, single-headed arrows indicate factor loadings, and double-headed arrows indicate correlations. All individual IPTs are based on summary statistics from genome-wide association studies. Factor loadings on factors with only two indicators were equated to identify the factor. Factors with only one indicator had factor loadings fixed to 1.0, and the residual variance (R) for that indicator was fixed to 0. Significant correlations are indicated by boldface and black arrows, and significant factor loadings are indicated by boldface (based on 95% confidence intervals). Confidence intervals are shown in Table S2 in the Supplemental Material; confidence intervals were nearly identical to those displayed in Table 2 after adding other constructs to the model. All values reflect fully standardized parameter estimates. UPPS-P = UPPS-P Impulsive Behavior Scale; NU = Negative Urgency; PU = Positive Urgency; Premed = Lack of Premeditation; BIS = Barratt Impulsiveness Scale; SS = Sensation Seeking; Persev = Lack of Perseverance.

We compared Model 1 with models in which the UPPS-P Positive Urgency and Negative Urgency subscales were collapsed into a single factor (Model 2) and in which the BIS total score, UPPS-P Lack of Perseverance, and UPPS-P Lack of Premeditation were collapsed into a single factor (Model 3). Model 2 did not fit the data as well as Model 1, Δχ2(3) = 22.44, p < .001. Although the fit of Model 3 was similar to that of Model 1, Δχ2(2) = 5.61, p = .061, all other model-fit statistics were less favorable (e.g., lower CFI value, higher AIC value; see Table 1). Additionally, parallel analyses that did not include extraversion indicated that Model 3 fit the data less well than Model 1 when all fit statistics were considered, including χ2 (see Tables S3 and S4 in the Supplemental Material). Thus, we also rejected Model 3. Finally, Model 4, which was a single-factor model, fitted the data very poorly (see Table 1, Model 4).

Associations between impulsivity and delay discounting

We first included a separate delay-discounting factor (i.e., with only one indicator) and allowed it to correlate with all other factors in the model. This model, displayed in Table 2, showed acceptable fit, χ2(29) = 154.10, p < .001, CFI = .957, AIC = 228.10. As expected, the delay-discounting factor was positively genetically correlated with all other IPT factors in the model, although some were not statistically significant (see Table 2). Also as anticipated, the delay-discounting factor was most strongly genetically correlated with the lack-of-premeditation factor, r = .47, 95% CI = [.03, .91]. However, when we attempted to incorporate delay discounting as a third indicator of the lack-of-premeditation factor, the model fitted significantly worse, Δχ2(5) = 11.71, p = .039; overall model fit: χ2(34) = 165.81, p < .001, CFI = .955, AIC = 229.81. Thus, delay discounting does not seem to be subsumed as an indicator of lack of premeditation.

Table 2.

Genetic Correlations Among Impulsivity, Internalizing Psychopathology, and Delay Discounting

Latent factor and indicator Factor loading Genetic correlations
1 2 3 4 5 6
1. Negative urgency —
 UPPS-P Negative Urgency 1
2. Positive urgency .70
[.20, 1.0]		 —
 UPPS-P Positive Urgency 1
3. Lack of premeditation .26
[−.15, .67]		 .63
[.18, 1.0]		 —
 UPPS-P Lack of Premeditation .87
 BIS Total Score .87
4. Sensation seeking −.23
[−.60, .14]		 .24
[−.11, .59]		 .44
[.02, .85]		 —
 UPPS-P Sensation Seeking .70
 Extraversion .70
5. Lack of perseverance 1 −.03
[−.42, .35]		 −.06
[−.45, .33]		 .41
[−.01, .84]		 −.03
[−.31, .26]		 —
 UPPS-P Lack of Perseverance
6. Internalizing psychopathology .55
[.43, .67]		 .38
[.25, .51]		 .25
[.08, .42]		 −.43
[−.59, −.27]		 −.10
[−.22, .03]		 —
 Major depressive disorder .79
 Neuroticism .87
 Subjective well-being .78
7. Delay discounting 1 .20
[−.19, .58]		 .19
[−.22, .59]		 .47
[.03, .91]		 .15
[−.18, .47]		 −.16
[−.57, .25]		 .23
[.10, .35]
Open in a new tab

Note: The table displays the factor loadings from individual measures to latent factors and the genetic correlations (and corresponding 95% confidence intervals) between latent factors. Factor loadings of 1 were fixed to identify these dummy latent factors. Factor loadings for both measures used to assess the lack-of-premeditation and sensation-seeking factors were equated to identify the latent factor. See Figure S1 in the Supplemental Material for a visual depiction of this model. UPPS-P = UPPS-P Impulsive Behavior Scale; BIS = Barratt Impulsiveness Scale.

Associations between impulsivity facets and internalizing psychopathology

As shown in Table 2, and as expected, the internalizing-psychopathology factor was most strongly positively genetically correlated with negative urgency, r = .55, 95% CI = [.43, .67], and positive urgency, r = .38, 95% CI = [.25, .51]. Internalizing psychopathology’s genetic correlation with negative urgency was significantly stronger than its genetic correlation with positive urgency, Δχ2(1) = 10.93, p < .001, but its genetic correlation with positive urgency was not significantly stronger than its genetic correlation with lack of premeditation, r = .25, 95% CI = [.08, .24], Δχ2(1) = 1.25, p = .264. The sensation-seeking factor was negatively genetically correlated with the internalizing-psychopathology factor, r = −.43, 95% CI = [−.27, −.59], whereas the genetic association with lack of perseverance was nonsignificant, r = −.10, 95% CI = [−.22, .03]. The internalizing-psychopathology factor was also associated with delay discounting, r = .23, 95% CI = [.10, .35].

Not surprisingly, this full model confirmed the results of our initial hypothesis regarding the five-factor structure of IPTs. Namely, the UPPS-P Positive Urgency and Negative Urgency subscales could not be collapsed into a single factor, Δχ2(5) = 38.14, p < .001, nor could UPPS-P Lack of Perseverance be collapsed into the lack-of-premeditation factor. Although the model fit was similar, Δχ2(3) = 0.90, p = .825, the CFI and the standardized root mean square residual were lower, and the factor loading for UPPS-P Lack of Perseverance was nonsignificant and in the unexpected direction (−.08).

Discussion

Numerous studies have examined the phenotypic relationship of IPTs; however, ours is the first to use genomic data to address this question. Our results were consistent with the UPPS-P model, suggesting that positive urgency, negative urgency, lack of premeditation, sensation seeking, and lack of perseverance capture distinct genetic IPTs. Delay discounting appeared to represent a unique genetic construct that was genetically correlated only modestly with some of the factors in our models. Of note, the genetic factors identified here do not represent individual genes but rather the contributions of many hundreds or thousands of genetic polymorphisms. This study is also the first to model the genetic structure between IPTs and internalizing psychopathology from unrelated individuals, extending twin research (Gustavson et al., 2019). Internalizing psychopathology was most strongly positively genetically associated with negative and positive urgency, whereas internalizing psychopathology showed a negative genetic correlation with sensation seeking and no genetic correlation with lack of premeditation.

Although our results supported a multifactor solution of IPTs, it is debatable whether all factors truly represented facets of a single construct of impulsivity. For example, genetic influences on sensation seeking and extraversion loaded highly on a single genetic factor and demonstrated mostly weak or negative genetic correlations with other IPTs. Combined with phenotypic evidence showing that sensation seeking is weakly correlated with other IPTs (MacKillop et al., 2016; Sharma, Markon, & Clark, 2014) and results from our earlier twin work (Gustavson et al., 2019), this pattern suggests that viewing sensation seeking as a component of impulsivity may be an example of the jingle fallacy (Block, 1995), in which different constructs are referred to with the same label. Sensation seeking and impulsivity may instead represent independent processes, or dual systems, with sensation seeking capturing bottom-up reward processing and impulsivity capturing the inability to exert cognitive control (i.e., top-down processing; Shulman, Harden, Chein, & Steinberg, 2016; Steinberg et al., 2008). This result is unsurprising given that the original UPPS study leveraged the Five-Factor Model of personality to create the UPPS model (Whiteside & Lynam, 2001) and concluded that the sensation-seeking factor corresponded to extraversion. Furthermore, the factor capturing UPPS-P Lack of Perseverance was uncorrelated with all other IPTs, suggesting that it not only is genetically distinct from the lack-of-premeditation factor but also may have minimal genetic overlap with other IPTs. This is consistent with weak phenotypic associations between lack of perseverance and other IPTs in some studies (Whiteside & Lynam, 2003) but not all of them (Cyders & Smith, 2007; MacKillop et al., 2016). In summary, although our genetic analysis supports the current UPPS-P framework that the five IPTs represent five separable constructs (Lynam et al., 2006), it may be more useful to restrict the term “impulsivity” to facets that share some phenotypic and genetic similarities.

On the other hand, the three BIS subscales were so completely genetically correlated that we analyzed them as a single score (as an indicator of the lack-of-premeditation factor). Thus, one framework may be overly inclusive (UPPS-P), and the other may ignore some important aspects of impulsivity (e.g., the BIS does not assess positive or negative urgency).

This work is relevant to our understanding of the role of delay discounting in impulsivity. The theoretical similarity between delay discounting and IPTs has been highlighted across many studies (Moreira & Barbosa, 2019). However, other researchers have argued that low correlations between IPTs and delay-discounting measures suggest that these two constructs tap distinct biological mechanisms (Murphy & Mackillop, 2012; Reynolds et al., 2006). Our finding that delay discounting was positively, but only weakly to moderately, genetically correlated with other IPTs supports the latter possibility. Delay discounting was most strongly associated with lack of premeditation, consistent with the idea that both constructs relate to the valuation and consideration of future events. Despite some similarities, delay discounting could not be collapsed on the same latent genetic factor as the other indicators of lack of premeditation, suggesting that delay discounting may be an incremental source of information to include when studying impulsivity.

Finally, our findings confirmed the hypothesis that positive and negative urgency, which are related to emotional control, are more strongly genetically correlated to internalizing psychopathology compared with other IPTs (Carver & Johnson, 2018; Johnson et al., 2013). Moreover, internalizing psychopathology was more strongly positively genetically correlated with negative urgency than positive urgency, consistent with a previous phenotypic meta-analysis (Berg et al., 2015). The positive genetic correlation between lack of premeditation and internalizing psychopathology we observed here was larger than the phenotypic associations with depression and anxiety symptoms found by Berg et al. (2015), suggesting that lack of premeditation may be associated with internalizing psychopathology primarily through genetic influences. In contrast, sensation seeking was negatively genetically correlated with internalizing psychopathology, providing further support for its genetic distinction.

Our findings should be interpreted in the context of the following limitations. First, we demonstrated that certain IPT factors could not be collapsed together on the same factor without poorer model fit, but some factors had only one indicator and others had only two indicators with equated factor loadings. The latter can contribute to poor model fit to the extent that one measure is a better index of the true latent factor than the other (e.g., UPPS Sensation Seeking vs. extraversion). Second, IPT measures were based on self-reports and may have different factor structures than laboratory tasks assessing similar constructs (MacKillop et al., 2016; Mallard et al., 2019). However, we anticipated that IPT and task measures would be genetically distinct, given their low phenotypic and genetic correspondence (Duckworth & Kern, 2011; Friedman et al., 2019; Sharma et al., 2014). Third, the GWAS data used here reflect ascertainment strategies that may have biased our results. For example, the cohorts were generally older, and participants had higher socioeconomic statuses than the general population and may have had below-average levels of impulsivity (Sanchez-Roige et al., 2019). In addition, the current findings cannot be used to draw inferences about variation among individuals of non-European ancestry, which reflects the underrepresentation of non-Europeans in the field of human genetics. Finally, although the study was based on GWASs of approximately 20,000 to 400,000 individuals, many genetic correlations had wide confidence intervals. As GWAS sample sizes continue to rapidly increase, more precise estimates of associations (and model testing) will be possible in future studies.

Conclusion

Impulsivity is increasingly recognized as a phenotypically heterogeneous construct (Niv, Tuvblad, Raine, Wang, & Baker, 2012), and our genomic-SEM analyses provide novel genetic evidence to support this view. The current data support the idea that IPTs tap overlapping but distinct genetic influences. Although sensation seeking and lack of perseverance are considered impulsivity-related traits within the UPPS-P framework, our data suggest that they are genetically distinct from the other IPTs, consistent with earlier phenotypic observations. Delay discounting also appears to be a distinct genetic factor. Our findings also support the hypothesis that although internalizing psychopathology is positively associated with all impulsivity facets except sensation seeking, this genetic association is most pronounced for IPTs related to the control over negative emotions (Carver & Johnson, 2018). This work demonstrates that large-scale GWAS results can be used to evaluate theoretical models of impulsivity and psychology more broadly.

Supplemental Material

Gustavson_OpenPracticesDisclosure_rev – Supplemental material for The Latent Genetic Structure of Impulsivity and Its Relation to Internalizing Psychopathology
Click here for additional data file. (591.7KB, pdf)

Supplemental material, Gustavson_OpenPracticesDisclosure_rev for The Latent Genetic Structure of Impulsivity and Its Relation to Internalizing Psychopathology by Daniel E. Gustavson, Naomi P. Friedman, Pierre Fontanillas, Sarah L. Elson, Abraham A. Palmer and Sandra Sanchez-Roige in Psychological Science

Gustavson_Supplemental_Material – Supplemental material for The Latent Genetic Structure of Impulsivity and Its Relation to Internalizing Psychopathology
Click here for additional data file. (261.7KB, pdf)

Supplemental material, Gustavson_Supplemental_Material for The Latent Genetic Structure of Impulsivity and Its Relation to Internalizing Psychopathology by Daniel E. Gustavson, Naomi P. Friedman, Pierre Fontanillas, Sarah L. Elson, Abraham A. Palmer and Sandra Sanchez-Roige in Psychological Science

Acknowledgments

We thank the research participants and employees of 23andMe for making this work possible. The 23andMe Research Team consists of Michelle Agee, Babak Alipanahi, Adam Auton, Robert K. Bell, Katarzyna Bryc, Sarah L. Elson, Pierre Fontanillas, Nicholas A. Furlotte, David A. Hinds, Naomi Iwata, Jennifer C. McCreight, Karen E. Huber, Aaron Kleinman, Nadia K. Litterman, Joanna L. Mountain, Elizabeth S. Noblin, Carrie A. M. Northover, Steven J. Pitts, J. Fah Sathirapongsasuti, Olga V. Sazonova, Anjali J. Shastri, Janie F. Shelton, Suyash Shringarpure, Chao Tian, Vladimir Vacic, Catherine Weldon, Keng-Han Lin, Yunxuan Jiang, Kimberly McManus, David Poznik, Ethan Jewett, Xin Wang, and Barry Hicks.

Footnotes

ORCID iDs: Daniel E. Gustavson Inline graphic https://orcid.org/0000-0002-1470-4928

Naomi P. Friedman Inline graphic https://orcid.org/0000-0002-4901-808X

Supplemental Material: Additional supporting information can be found at http://journals.sagepub.com/doi/suppl/10.1177/0956797620938160

Transparency

Action Editor: Brent W. Roberts

Editor: D. Stephen Lindsay

Author Contributions

D. E. Gustavson and S. Sanchez-Roige developed the study concept and study design. D. E. Gustavson and S. Sanchez-Roige created correlation matrices from available genome-wide-association-study summary statistics and conducted all analyses. All of the authors provided critical revisions and approved the final version of the manuscript for submission.

Declaration of Conflicting Interests: The author(s) declared that there were no conflicts of interest with respect to the authorship or the publication of this article.

Funding: D. E. Gustavson was supported by National Institutes of Health (NIH) Grants DC016977 and HD098859. N. P. Friedman was supported by NIH Grants MH063207, DA046064, DA046413, and DA042742. S. Sanchez-Roige was supported by an NARSAD Young Investigator Award from the Brain & Behavior Research Foundation (Grant No. 27676). S. Sanchez-Roige and A. A. Palmer were supported by funds from the California Tobacco-Related Disease Research Program (TRDRP; Grant Nos. 28IR-0070 and T29KT0526).

Open Practices: The R data files containing the genomic-SEM matrices for all analyses in this study, along with R analysis scripts, have been made publicly available via the Open Science Framework at https://osf.io/5x3ft/. The design and analysis plans for this study were not preregistered. The complete Open Practices Disclosure for this article can be found at http://journals.sagepub.com/doi/suppl/10.1177/0956797620938160. This article has received the badges for Open Data and Open Materials. More information about the Open Practices badges can be found at http://www.psychologicalscience.org/publications/badges.

Inline graphic Inline graphic

References

  1. Barratt E. S. (1993). Impulsivity: Integrating cognitive, behavioral, biological, and environmental data. In McCown W. G., Johnson J. L., Shure M. B. (Eds.), The impulsive client: Theory, research, and treatment (pp. 39–56). Washington, DC: American Psychological Association. [Google Scholar]
  2. Berg J. M., Latzman R. D., Bliwise N. G., Lilienfeld S. O. (2015). Parsing the heterogeneity of impulsivity: A meta-analytic review of the behavioral implications of the UPPS for psychopathology. Psychological Assessment, 27, 1129–1146. doi: 10.1037/pas0000111 [DOI] [PubMed] [Google Scholar]
  3. Block J. (1995). A contrarian view of the five-factor approach to personality description. Psychological Bulletin, 117, 187–215. doi: 10.1037/0033-2909.117.2.187 [DOI] [PubMed] [Google Scholar]
  4. Bulik-Sullivan B. K., Loh P.-R., Finucane H. K., Ripke S., Yang J., Schizophrenia Working Group of the Psychiatric Genomics Consortium, . . . Neale B. M. (2015). LD Score regression distinguishes confounding from polygenicity in genome-wide association studies. Nature Genetics, 47, 291–295. doi: 10.1038/ng.3211 [DOI] [PMC free article] [PubMed] [Google Scholar]
  5. Carver C. S., Johnson S. L. (2018). Impulsive reactivity to emotion and vulnerability to psychopathology. American Psychologist, 73, 1067–1078. doi: 10.1037/amp0000387 [DOI] [PMC free article] [PubMed] [Google Scholar]
  6. Costa P. T., McCrae R. R. (1992). Normal personality assessment in clinical practice: The NEO Personality Inventory. Psychological Assessment, 4, 5–13. doi: 10.1037/1040-3590.4.1.5 [DOI] [Google Scholar]
  7. Cyders M. A., Littlefield A. K., Coffey S., Karyadi K. A. (2014). Examination of a short English version of the UPPS-P Impulsive Behavior Scale. Addictive Behaviors, 39, 1372–1376. doi: 10.1016/j.addbeh.2014.02.013 [DOI] [PMC free article] [PubMed] [Google Scholar]
  8. Cyders M. A., Smith G. T. (2007). Mood-based rash action and its components: Positive and negative urgency. Personality and Individual Differences, 43, 839–850. doi: 10.1016/j.paid.2007.02.008 [DOI] [Google Scholar]
  9. Dalley J. W., Everitt B. J., Robbins T. W. (2011). Impulsivity, compulsivity, and top-down cognitive control. Neuron, 69, 680–694. doi: 10.1016/j.neuron.2011.01.020 [DOI] [PubMed] [Google Scholar]
  10. Duckworth A. L., Kern M. L. (2011). A meta-analysis of the convergent validity of self-control measures. Journal of Research in Personality, 45, 259–268. doi: 10.1016/j.jrp.2011.02.004 [DOI] [PMC free article] [PubMed] [Google Scholar]
  11. Evenden J. L. (1999). Varieties of impulsivity. Psychopharmacology, 146, 348–361. doi: 10.1007/PL00005481 [DOI] [PubMed] [Google Scholar]
  12. Eysenck H. J. (1993). The nature of impulsivity. In McCown W. G., Johnson J. L., Shure M. B. (Eds.), The impulsive client: Theory, research, and treatment (pp. 57–69). Washington, DC: American Psychological Association. doi: 10.1037/10500-004 [DOI] [Google Scholar]
  13. Eysenck H. J., Eysenck M. W. (1985). Personality and individual differences: A natural science perspective. New York, NY: Plenum Press. [Google Scholar]
  14. Friedman N. P., Hatoum A. S., Gustavson D. E., Corley R. P., Hewitt J. K., Young S. E. (2020). Executive functions and impulsivity are genetically distinct and independently predict psychopathology: Results from two adult twin studies. Clinical Psychological Science, 8, 519–538. [DOI] [PMC free article] [PubMed] [Google Scholar]
  15. Grotzinger A. D., Rhemtulla M., de Vlaming R., Ritchie S. J., Mallard T. T., Hill W. D., . . . Tucker-Drob E. M. (2019). Genomic structural equation modelling provides insights into the multivariate genetic architecture of complex traits. Nature Human Behaviour, 3, 513–525. doi: 10.1038/s41562-019-0566-x [DOI] [PMC free article] [PubMed] [Google Scholar]
  16. Gustavson D. E., Franz C. E., Kremen W. S., Carver C. S., Corley R. P., Hewitt J. K., Friedman N. P. (2019). Common genetic influences on impulsivity facets are related to goal management, psychopathology, and personality. Journal of Research in Personality, 79, 161–175. doi: 10.1016/j.jrp.2019.03.009 [DOI] [PMC free article] [PubMed] [Google Scholar]
  17. Howard D. M., Adams M. J., Shirali M., Clarke T.-K., Marioni R. E., Davies G., . . . McIntosh A. M. (2018). Genome-wide association study of depression phenotypes in UK Biobank identifies variants in excitatory synaptic pathways. Nature Communications, 9, Article 1470. doi: 10.1038/s41467-018-03819-3 [DOI] [PMC free article] [PubMed] [Google Scholar]
  18. Hu L. T., Bentler P. M. (1998). Fit indices in covariance structure modeling: Sensitivity to underparameterized model misspecification. Psychological Methods, 3, 424–453. doi: 10.1037//1082-989x.3.4.424 [DOI] [Google Scholar]
  19. Johnson S. L., Carver C. S., Joormann J. (2013). Impulsive responses to emotion as a transdiagnostic vulnerability to internalizing and externalizing symptoms. Journal of Affective Disorders, 150, 872–878. doi: 10.1016/j.jad.2013.05.004 [DOI] [PubMed] [Google Scholar]
  20. Kirby K. N., Petry N. M., Bickel W. K. (1999). Heroin addicts have higher discount rates for delayed rewards than non-drug-using controls. Journal of Experimental Psychology: General, 128, 78–87. doi: 10.1037/0096-3445.128.1.78 [DOI] [PubMed] [Google Scholar]
  21. Littlefield A. K., Stevens A. K., Ellingson J. M., King K. M., Jackson K. M. (2016). Changes in negative urgency, positive urgency, and sensation seeking across adolescence. Personality and Individual Differences, 90, 332–337. doi: 10.1016/j.paid.2015.11.024 [DOI] [PMC free article] [PubMed] [Google Scholar]
  22. Luciano M., Hagenaars S. P., Davies G., Hill W. D., Clarke T. K., Shirali M., . . . Deary I. J. (2018). Association analysis in over 329,000 individuals identifies 116 independent variants influencing neuroticism. Nature Genetics, 50, 6–11. doi: 10.1038/s41588-017-0013-8 [DOI] [PMC free article] [PubMed] [Google Scholar]
  23. Lynam D. R., Smith G. T., Whiteside S. P., Cyders M. A. (2006). The UPPS-P: Assessing five personality pathways to impulsive behavior. West Lafayette, IN: Purdue University. [Google Scholar]
  24. MacKillop J., Weafer J. J. C. G., Oshri A., Palmer A., de Wit H. (2016). The latent structure of impulsivity: Impulsive choice, impulsive action, and impulsive personality traits. Psychopharmacology, 233, 3361–3370. doi: 10.1007/s00213-016-4372-0 [DOI] [PMC free article] [PubMed] [Google Scholar]
  25. Mallard T. T., Linnér R. K., Okbay A., Grotzinger A. D., de Vlaming R., Meddens S. F. W., . . . Harden K. P. (2019). Not just one p: Multivariate GWAS of psychiatric disorders and their cardinal symptoms reveal two dimensions of cross-cutting genetic liabilities. bioRxiv. doi: 10.1101/603134 [DOI] [PMC free article] [PubMed] [Google Scholar]
  26. Moreira D., Barbosa F. (2019). Delay discounting in impulsive behavior: A systematic review. European Psychologist, 24, 312–321. doi: 10.1027/1016-9040/a000360 [DOI] [Google Scholar]
  27. Murphy C., Mackillop J. (2012). Living in the here and now: Interrelationships between impulsivity, mindfulness, and alcohol misuse. Psychopharmacology, 219, 527–536. doi: 10.1007/s00213-011-2573-0 [DOI] [PMC free article] [PubMed] [Google Scholar]
  28. Niv S., Tuvblad C., Raine A., Wang P., Baker L. A. (2012). Heritability and longitudinal stability of impulsivity in adolescence. Behavior Genetics, 42, 378–392. doi: 10.1007/s10519-011-9518-6 [DOI] [PMC free article] [PubMed] [Google Scholar]
  29. Okbay A., Baselmans B. M., De Neve J. E., Turley P., Nivard M. G., Fontana M. A., . . . Cesarini D. (2016). Genetic variants associated with subjective well-being, depressive symptoms, and neuroticism identified through genome-wide analyses. Nature Genetics, 48, 624–633. doi: 10.1038/ng.3552 [DOI] [PMC free article] [PubMed] [Google Scholar]
  30. Patton J. H., Stanford M. S., Barratt E. S. (1995). Factor structure of the Barratt Impulsiveness Scale. Journal of Clinical Psychology, 51, 768–774. [DOI] [PubMed] [Google Scholar]
  31. Reynolds B., Ortengren A., Richards J. B., de Wit H. (2006). Dimensions of impulsive behavior: Personality and behavioral measures. Personality and Individual Differences, 40, 305–315. doi: 10.1016/j.paid.2005.03.024 [DOI] [Google Scholar]
  32. Rosseel Y. (2012). lavaan: An R package for structural equation modeling. Journal of Statistical Software, 48(2). doi: 10.18637/jss.v048.i02 [DOI] [Google Scholar]
  33. Sanchez-Roige S., Fontanillas P., Elson S. L., Gray J. C., de Wit H., MacKillop J., Palmer A. A. (2019). Genome-wide association studies of impulsive personality traits (BIS-11 and UPPSP) and drug experimentation in up to 22,861 adult research participants identify loci in the CACNA1I and CADM2 genes. The Journal of Neuroscience, 39, 2562–2572. doi: 10.1523/JNEUROSCI.2662-18.2019 [DOI] [PMC free article] [PubMed] [Google Scholar]
  34. Sanchez-Roige S., Fontanillas P., Elson S. L., the 23andMe Research Team, Pandit A., Schmidt E. M., . . . Palmer A. A. (2018). Genome-wide association study of delay discounting in 23,217 adult research participants of European ancestry. Nature Neuroscience, 21, 16–18. doi: 10.1038/s41593-017-0032-x [DOI] [PMC free article] [PubMed] [Google Scholar]
  35. Sharma L., Markon K. E., Clark L. A. (2014). Toward a theory of distinct types of “impulsive” behaviors: A meta-analysis of self-report and behavioral measures. Psychological Bulletin, 140, 374–408. doi: 10.1037/a0034418 [DOI] [PubMed] [Google Scholar]
  36. Shulman E. P., Harden K. P., Chein J. M., Steinberg L. (2016). The development of impulse control and sensation-seeking in adolescence: Independent or interdependent processes? Journal of Research on Adolescence, 26, 37–44. doi: 10.1111/jora.12181 [DOI] [Google Scholar]
  37. Steinberg L., Albert D., Cauffman E., Banich M., Graham S., Woolard J. (2008). Age differences in sensation seeking and impulsivity as indexed by behavior and self-report: Evidence for a dual systems model. Developmental Psychology, 44, 1764–1778. doi: 10.1037/a0012955 [DOI] [PubMed] [Google Scholar]
  38. Tellegen A. (1982). Multidimensional Personality Questionnaire manual. Minneapolis: University of Minnesota Press. [Google Scholar]
  39. van den Berg S. M., de Moor M. H., Verweij K. J., Krueger R. F., Luciano M., Arias Vasquez A., . . . Boomsma D. I. (2016). Meta-analysis of genome-wide association studies for extraversion: Findings from the Genetics of Personality Consortium. Behavior Genetics, 46, 170–182. doi: 10.1007/s10519-015-9735-5 [DOI] [PMC free article] [PubMed] [Google Scholar]
  40. Whiteside S. P., Lynam D. R. (2001). The Five Factor Model and impulsivity: Using a structural model of personality to understand impulsivity. Personality and Individual Differences, 30, 669–689. doi: 10.1016/S0191-8869(00)00064-7 [DOI] [Google Scholar]
  41. Whiteside S. P., Lynam D. R. (2003). Understanding the role of impulsivity and externalizing psychopathology in alcohol abuse: Application of the UPPS impulsive behavior scale. Experimental Clinical Psychopharmacology, 11, 210–217. doi: 10.1037/1064-1297.11.3.210 [DOI] [PubMed] [Google Scholar]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

Gustavson_OpenPracticesDisclosure_rev – Supplemental material for The Latent Genetic Structure of Impulsivity and Its Relation to Internalizing Psychopathology
Click here for additional data file. (591.7KB, pdf)

Supplemental material, Gustavson_OpenPracticesDisclosure_rev for The Latent Genetic Structure of Impulsivity and Its Relation to Internalizing Psychopathology by Daniel E. Gustavson, Naomi P. Friedman, Pierre Fontanillas, Sarah L. Elson, Abraham A. Palmer and Sandra Sanchez-Roige in Psychological Science

Gustavson_Supplemental_Material – Supplemental material for The Latent Genetic Structure of Impulsivity and Its Relation to Internalizing Psychopathology
Click here for additional data file. (261.7KB, pdf)

Supplemental material, Gustavson_Supplemental_Material for The Latent Genetic Structure of Impulsivity and Its Relation to Internalizing Psychopathology by Daniel E. Gustavson, Naomi P. Friedman, Pierre Fontanillas, Sarah L. Elson, Abraham A. Palmer and Sandra Sanchez-Roige in Psychological Science

Articles from Psychological Science are provided here courtesy of SAGE Publications

RESOURCES