Temperament refers to individual differences in patterns of emotional and behavioral reactivity and self-regulation that emerge early in life and exhibit relative stability over time and situations. Temperament traits describe individual tendencies, dispositions, and capacities that influence individuals’ adaptation or maladaptation to the environment throughout life [13–15]. As previously mentioned, two broad, higher-order dimensions of temperament, namely NT and PT, are fundamentally affective. NT refers to a tendency for negative emotional and behavioral reactivity, including fear, sadness, and anger, whereas PT refers to a propensity for positive affective experience, including joy, interest, and excitement, as well as reward sensitivity and sociability [13, 14].

The tripartite model asserts that both anxiety and depression are characterized by high levels of NT, whereas depression, but generally not anxiety, is associated also with low levels of PT [7]. Large bodies of cross-sectional and longitudinal studies of adults (e.g., [16]) and youth (e.g., [17–20]) have provided considerable empirical support for these distinct relations. Further, in a more recent study of inpatient youth with comorbid externalizing and anxiety or depression, NT evidenced significant associations with both depression and various anxiety disorders (e.g., general anxiety disorder [GAD], social anxiety disorder, panic disorder), whereas the combination of high NT and low PT was more consistently associated with depression [21]. Similarly, in a prospective study of community youth, Lonigan et al. [20] found that NT was associated with changes in symptoms of both depression and anxiety, whereas PT was more strongly associated with changes in symptoms of depression 7 years later. Collectively, the extant literature strongly supports the tripartite model across clinical and non-clinical samples of adults and youth in the explanation of anxious and depressive symptomatology.

Executive functioning reflects a set of higher order cognitive processes associated largely, although not entirely, with the prefrontal cortex, that control a wide range of neuropsychological and cognitive abilities, including decision making, planning, problem solving, attentional flexibility, inhibitory control, and working memory [22–24]. Most commonly, EF has been examined with regard to externalizing behaviors, with considerable research finding significant associations between executive dysfunction and the development and maintenance of a range of behavioral disorders including antisocial behaviors, substance use, and attention deficit/hyperactive disorder [25, 26].

In recent years, a growing body of neuropsychological research has also found EF deficits to be associated with anxiety and depression. For example, both individual studies [27, 28] and a recent meta-analysis [29] have demonstrated positive associations between adult depression and impairments in various abilities subsumed under the EF umbrella. Many fewer studies have investigated associations between EF and depression among youth. Among those that have, similar conclusions regarding the importance of considering EF in investigations of depression have emerged. For example, youth with (vs. without) depression evidence difficulties with task initiation, attentional deficits, and slower response time when performing various neuropsychological tasks [8], as well as significant deficits in working memory, cognitive inhibition, and processing speed (for a comprehensive review, see [10]). Thus, a converging literature in both adult and youth samples suggests negative associations between EF and depressive symptoms.

With regard to anxiety, as compared to the literature on depression, many fewer studies have examined associations with EF. Among those that have, results are equivocal, perhaps as a result of foci on various specific forms of anxiety-related psychopathology [e.g., panic disorder, obsessive–compulsive disorder (OCD)]. For example, across a range of DSM-IV anxiety disorders, Airaksinen et al. [30] found that whereas adults meeting diagnostic criteria for panic disorder and OCD exhibited significantly lower attentional capabilities than did healthy controls, those diagnosed with GAD and specific phobia did not exhibit EF deficits.

Although similarly limited, studies among youth also collectively suggest negative associations between EF and anxiety. For example, Günther and colleagues [31] found that, compared to healthy controls, youth diagnosed with GAD, separation anxiety disorder, and social phobia exhibited poor verbal working memory and difficulty sustaining attention, two indicators of EF. Further, youth diagnosed with (vs. without) OCD were found to perform more poorly on tests of cognitive flexibility, verbal comprehension, and visuospatial abilities [32]. Thus, although the adult literature is largely equivocal, as noted above, the few studies that have examined associations between EF and anxiety in youth generally converge on negative associations.

Whereas converging research has confirmed the role of temperament and EF in relation to anxiety and depression, relations between EF and affective dimensions of temperament are more equivocal, particularly with regard to NT. PT-related dimensions have been found to be positively associated with verbal fluency [33], verbal working memory, problem-solving [34, 35], and cognitive flexibility, as well as with increased distractibility [36]. However, with regard to NT-related dimensions, the few studies have found consistent associations with EF, with the exception of a positive association with visual spatial memory [37, 38].

As noted earlier, only a single study to date has examined the role of child temperament and EF on child anxiety: parent-reported shifting, a single dimension subsumed under executive functioning, mediated the effect of parent-reported fear temperament, a NT-related dimension, on anxiety among 7-to-10-year-old children [39]. To date, research has yet to consider task-based neuropsychological assessments in such investigations. Given the relative lack of research concerning associations among neuropsychological indicators of EF and affective dimensions of temperament, consideration of the extant literature that has examined associations between temperamental traits (i.e., NT and PT), and effortful control (EC), which overlaps with EF both conceptually and empirically [40–42], may help to explicate a pattern of associations.

Effortful control refers to the ability to inhibit a dominant, prepotent response to perform a subdominant, less salient response and to detect errors [43] and, as noted above, is thought to share common features, such as attentional focus and inhibitory control, with broad, higher order cognitive processes of EF [44, 45]. The tripartite model underscores the importance of considering the way in which associations between temperament dimensions and anxious and depressive symptomatology may be moderated by a third variable [7, 46–48]. Consistent with this view, the extant literature that investigated two-way interactions between affective dimensions of temperament and EC has found that EC moderates the effect of both NT and PT in relation to anxiety and depression (e.g., [49–53]), indicating the potential moderating role of EF in the association between temperament and anxious and depressive symptomatology.

Taken together, the extant literature highlights the importance of investigating the joint and interactive contribution of the core underlying processes of EF and affective dimensions of temperament in service of advancing our understanding of etiological mechanisms associated with anxiety and depression. Knowledge of how EF moderates the interplay between NT and PT, both with established links to symptoms of anxiety and depression, will contribute to a better understanding of risk factors associated with anxiety and depression.

In the present study, we report the results of the first investigation to date of triangular relations among task-based neuropsychological indicators of EF, two affective dimensions of temperament, and anxiety and depressive symptoms within a community sample of adolescent males. Given the relative lack of task-based measures of EF used in previous research, a clear strength of the current study is our use of a widely used task-based neuropsychological assessment tapping a range of EF subdimensions. Consistent with the tripartite model [7], it was expected that NT would be positively associated with both anxiety and depressive symptoms, whereas PT would be negatively correlated with depression. We further expected that PT would moderate the effect of NT on depressive symptoms. That is, it was expected that the individuals with both higher levels of NT and lower levels of PT would show increased depressive symptoms above and beyond the additive effects of the temperament dimensions. Additionally, consistent with the literature reviewed, it was hypothesized that anxiety and depressive symptoms would be negatively associated with EF.

In a more exploratory set of analyses, we investigated the interactive relations of EF and affective dimensions of temperament with anxiety and depressive symptoms. Given the dearth of research examining three-way interactions among EF, NT, and PT in the explanation of anxiety and depression, our a priori hypotheses were tentative. Nonetheless, drawing from literature examining two-way interactions between EC and NT or PT in association with anxiety and depression (e.g., [49–53]), we hypothesized that PT and EF together would explain associations between NT and anxiety and depressive symptoms above and beyond their additive effects.

Participants were a community sample of 174male youths and their mothers who participated in the Iowa Youth Development Project (I-YDP), a larger study of developmental factors associated with social behaviors in male adolescents. Three youth did not complete all of the measures, resulting in a final sample of 171 for youth-reported data. Participants were predominantly White adolescent males aged 11–16 years (M_age = 13.6 ± 1.35; 87.9 % White) and their mothers (M_age = 44.2 years; 93.1 % White). The families were relatively high in socioeconomic status in terms of education and income, with 34.1 % exceeding an annual combined household income of $100,000. Further, most mothers were married to their son’s biological fathers (81.0 %), had achieved college or post-graduate education (71.9 %), and were employed full-time (93.7 %).

The I-YDP used multiple recruiting methods to obtain a sample of Midwestern male youth; participants were recruited from a child participant database maintained at the University of Iowa Psychology department, as well as through fliers distributed in the community, advertisements placed in newsletters, and on-line advertisements in the affiliated university hospital. To ensure a typically developing sample, exclusion criteria were: intellectual disability, autism spectrum disorder, reading disorder, a history of being held back a grade, neurological disorders, past head injury requiring hospitalization, and life-threatening medical illness, all assessed by maternal report. Participants provided informed consent/assent before beginning the study procedures and were compensated monetarily for their time. The University’s Institutional Review Board approved all study protocols and materials.

The multiple imputation program in SAS Version 9.1 was used to impute missing items for all participants, as none had more than 10 % of items missing. This approach uses maximum likelihood estimates for missing data and includes a random error component to prevent artificial inflation of item inter correlations.

First, zero-order correlations were calculated to examine the bivariate associations among affective dimensions of temperament (i.e., NT, PT), EF, and anxious and depressive symptomatology. Then, hierarchical multiple regression analyses were performed to examine how temperament and EF jointly and interactively explained anxious and depressive symptomatology. To directly examine the contribution of temperamental dimensions interactively, a NT × PT interaction term was created as were two temperament-by-EF terms: NT × EF and PT × EF. Lastly, a three-way NT × PT × EF product term was calculated. All variables were standardized (using z-scores) before calculating the interaction terms. Given the relatively wide age range of the sample, and known associations between age and anxious and depressive symptomatology, age was included in Step 1 as a covariate in all regression analyses. The remaining variables were entered into the hierarchical regression in the following order: Step 2, NT and PT; Step 3, the NT × PT interaction term; Step 4, EF; Step 5, the NT × EF and PT × EF interaction terms; and Step 6, the three-way NT × PT × EF interaction term. To probe the effect of any significant interactions, simple slopes analyses were conducted.

Of note, ratings from multiple informants consistently evidence low convergent correlations across multiple psychopathological symptoms in youth, particularly with regard to anxious and depressive symptomatology [63, 64]. As context is important in youth assessment, the use of multiple informants, each with their own unique perspective, captures a more comprehensive and accurate picture of youth psychosocial functioning, which may vary across different settings (e.g., home, school; [65]). Indeed, the use of multi-informant ratings is considered essential in evidence-based assessment of youth psychosocial functioning [66] and recent work suggests that discrepant reports from youth and parents may serve as an indicator of poor family functioning (e.g., [67]), highlighting the importance of examining reporter perceptions separately. As such, to examine potential differences by informant, all analyses were run separately for youth- and mother-report.

As shown in Table 1, both youth- and mother-reported A/D and W/D scales were strongly correlated with each other (r = .54 and .67, respectively). Associations of temperament with anxious and depressive symptomatology were largely consistent across informants. Specifically, both the A/D and W/D scales were positively correlated with NT and negatively correlated with PT, with the magnitude of the associations significantly greater for NT than for PT for the A/D scale (Z = 8.05, p < .001 for youths; Z = 2.65, p < .01 for mothers) and vice versa for the W/D scale (i.e., the magnitude of the association greater for PT than NT). However, for the W/D scale, the difference was significant only for mother report (in youths, Z = 0.96, p > .10; in mothers, Z = 2.18, p < .05). In contrast, associations between both anxious and depressive symptomatology and temperament with EF varied by informant: EF was not associated with either youth-reported anxious and depressive symptomatology or temperament, whereas mother-reported A/D and NT scales were both significantly and negatively associated with EF. Moreover, the association with mother report was significantly stronger than with youths’ self-report for the A/D scale (Z = 1.98, p < .05) and marginally stronger for the NT scale (Z = 1.68, p < .10).

To examine the unique associations between affective dimensions of temperament, EF, and anxiety and depression, a series of hierarchical linear regressions were performed predicting each of the psychopathology dimensions separately for youth- and mother-reported temperament and anxious and depressive symptomatology. As shown in Table 2, age was associated only with youth-reported A/D scale (β = .12, t = 2.15, p < .05) and not W/D scale (β = −.03, t = −.59, p > .10) or either scale in mother-reported data (βs < −.02, ts < 1.19, ps > .10). After accounting for age, NT was found in Step 2 to be positively associated with both the youth- and mother-reported A/D (βs = .76, .47, ts = 15.48, 7.21, respectively, ps < .001) and W/D (βs = .49, .30, ts = 7.70, 4.89, respectively, ps < .001) scales. Further, PT was negatively associated with both youth- and mother-reported A/D (βs = −.11, −.22, ts = −2.38, −3.27, respectively, ps < .05 and W/D (βs = −.25, −.49, ts = −3.95, −7.92, respectively, ps < .001) scales.

In addition to main effects of temperament, the interaction of NT and PT significantly contributed to the explanation of both anxiety and depression syndrome scales across informants (all βs > I.09 I, ts > I2.29I, ps < .05). Specifically, the association between NT and anxiety and depressive symptomatology was found to be moderated significantly by PT. To probe the nature of these interactions, as described above, simple slopes analyses were conducted. As shown in Figs. 1 and 2, these analyses indicated that PT moderated the effect of NT on both anxiety and depression syndrome scales. More specifically, at higher levels of NT, a lower PT level contributed more to anxiety and depressive symptoms compared to when NT was lower.

In Step 4, EF was not found to contribute significantly to predicting either youth- or mother-reported A/D or W/D beyond the two temperament dimensions and their interaction (all βs < |.08|, ts < I1.44I, ps > .10). Further, for both youth- and mother-report, none of the temperament by EF interactions emerged as significant contributors to either of anxiety and depression syndrome scales (all βs < |.13|, ts < |1.57|, ps > .05).¹ However, EF did significantly interact with both NT and PT within a three-way interaction in the explanation of youth-reported A/D (β = .13, t = 2.79, p < .01), but not youth-reported W/D nor either of the mother-reported anxiety and depression scales. To probe the nature of the significant three-way interaction for youth-reported A/D, simple slopes analyses were conducted at high and low levels of EF. As shown in Fig. 3, results indicated that at low levels, EF moderated the effect of the two-way NTxPT interaction on youth-reported Anxiety/Depressed. Specifically, when EF was high, the previously observed NT × PT interaction disappeared, whereas when it was low, the previously observed NT × PT interaction was strengthened. That is, when EF is high, NT, but not PT, remains positively associated with A/D. However, if EF is low, not only is NT associated with A/D, but low PT is as well. However, because the effect was found in only one of four cases examined and was not predicted, it needs replication before being interpreted further.

The cross-sectional, correlational nature of the current study does not allow causal or temporal inferences. Additionally, the current sample represented a relatively homogeneous sample comprised of predominantly White youths and their mothers who were moderate to high socioeconomic status, and some exclusion criteria (e.g., youth never held back a grade) may have restricted sample variability, especially in EF. Although this design may result in fewer potential confounding variables, the nature of the sample may also limit the generalizability of the results. For example, our use of an exclusively male sample precludes our ability to consider potential gender differences. Meta-analytic results (i.e., [74]) suggest that female youth evidence higher NT and lower PT than male youth, which may at least partially help to explain higher prevalence of anxious and depressive symptomatology among females than males [1, 2]. However, as compared to males, female youth have also been found to evidence higher levels of EC [74], a construct that conceptually and empirically overlaps with EF [39–41], which may potentially serve to offset the temperamental risk for developing anxious and depressive symptomatology. Nonetheless, it will be important for future research to examine more diverse samples to confirm that results of the current study reflect differences in informants’ temperament, EF, and anxious and depressive symptomatology rather than racial/cultural, gender, or socio-economically based differences. Further, because of our use of a community sample, the inclusion of youth with more clinical levels of anxious and depressive symptomatology was limited, potentially resulting in attenuated associations. As such, future research is needed within clinical populations in which the potential for associations, particularly with regard to EF, is more likely to emerge.

Additionally, our examination of youth- and mother reports separately allowing for examination of potential divergent and unique perspectives represents a significant strength of the current study. Nonetheless, the use of within-informant data may result in observed effects potentially being explained, at least partially, by shared informant variance. Consistent with typical clinical practice, future research would benefit from the inclusion of additional informants (e.g., teachers) as well as other research methods to test whether differential outcomes may emerge with different sources of information in the investigation of youth anxious and depressive symptomatology.

Finally, although our decision to examine a single indicator of EF was based on several considerations, including the number of independent variables we planned to include in our analytical models, EF has been found to consist of a number of sub-components (e.g., [75]). As noted earlier, the use of D–KEFS, which generally represents cognitive self-regulation in affective-neutral conditions, which might have attenuated associations with affective dimensions of temperament (i.e., NT and PT). Future researchers are therefore encouraged to examine potential differential associations between various EF dimensions and anxious and depressive symptomatology in youth using EF measures designed to assess emotion-regulation in affectively based conditions.