Working Memory and Intraindividual Variability in Response Time Mediate Fluid Intelligence Deficits Associated With ADHD Symptomology

Abstract

Objective: To determine if decreased fluid intelligence was associated with ADHD, and was mediated by deficits in working memory and intraindividual variability in motor responding. Method: The present study tested 142 young adults from the general population on a range of working memory, response time, and fluid intelligence tasks, and an ADHD self-report symptoms questionnaire. Results: Total and hyperactive ADHD symptoms correlated significantly and negatively with fluid intelligence, but this association was fully mediated by both working memory and intraindividual variability in response time. However, inattentive symptoms were not associated with fluid intelligence. Conclusion: These results have important implications for clinicians using speeded psychometric tests as part of their assessment battery, working memory interventions for ADHD patients that focus on performance improvement without controlling for response consistency, and also demonstrate potential differences in the neuropsychological profiles of ADHD subtypes.

Keywords

ADHD fluid intelligence working memory intraindividual variability

ADHD is a neuropsychological disorder characterized by hyperactivity, inattention, and impulsiveness (Barkley, 1997), and is observed in approximately 5.9% to 7.1% of children (Willcutt, 2012). The condition is most commonly diagnosed in early childhood, and has been found to persist into adulthood in 30% to 50% of cases (Klein & Mannuzza, 1991; Weiss & Hechtman, 1993). Individuals suffering from ADHD commonly suffer cognitive deficits, in domains including executive functioning, processing speed, and intelligence (Seidman, 2006), which can, in turn, negatively affect long-term outcomes such as self-esteem, academic performance, and income (Ingram, Hechtman, & Morgenstern, 1999). Recent research has found that an increase in severity of ADHD symptoms was associated with decreased fluid intelligence, and that working memory fully mediated this relationship, but processing speed had no effect (Brydges, Ozolnieks, & Roberts, 2017). However, ADHD-related deficits in processing speed may be dependent upon ADHD subtype (Nigg et al., 2005), as research has found that larger intraindividual variability in response time (the temporal consistency of response times made by an individual) is robust across ADHD subtypes (Kofler et al., 2013), and accounts for a significant proportion of variance of intelligence even when the mean response time (RT) is taken into account (Jensen, 1992). As such, it may be the case that Brydges et al. (2017) examined an inappropriate construct of speed of processing—perhaps variability in the consistency of responding, rather than the speed at which the average response is made, mediates ADHD-related deficits in fluid intelligence.

Previous research has found that ADHD is associated with deficits in a range of higher-order cognitive processes (Theiling & Petermann, 2016), including fluid intelligence (Dige & Wik, 2005), working memory (Schweitzer, Hanford, & Medoff, 2006), and processing speed (Boonstra, Oosterlaan, Sergeant, & Buitelaar, 2005). Fluid intelligence (gF) is the ability to solve novel problems, independently of any previously acquired knowledge (Cattell, 1963). Working memory is the ability to store, manipulate, and update incoming information or stimuli (Baddeley & Hitch, 1974). Processing speed refers to the speed at which individuals can perform cognitive tasks (Sheppard & Vernon, 2008; Vernon, 1983). A large body of research has found that these three constructs are associated with each other (e.g., Ackerman, Beier, & Boyle, 2005; Kail & Salthouse, 1994; Kyllonen & Christal, 1990; Wilhelm & Oberauer, 2006), and that working memory and processing speed are both excellent predictors of gF (e.g., Redick, Unsworth, Kelly, & Engle, 2012).

Given the associations between gF, working memory, and processing speed, and that ADHD appears to be associated with deficits in these three domains, Brydges et al. (2017) tested the hypothesis that ADHD-related deficits in gF are an indirect effect of impaired working memory and/or processing speed. They found that a latent variable of working memory fully mediated the association between self-reported ADHD symptom severity and the Cattell Culture Fair Intelligence Test (a commonly used measure of gF) in 142 young adults, albeit a non-diagnosed sample. However, they also found that a processing speed latent variable did not affect the relationship between ADHD symptom severity and gF. It was suggested that this was due to the different subtypes of ADHD having opposite effects on processing speed. Specifically, Nigg et al. (2005) found that in comparison to a typical control group, inattentive-type ADHD patients had significantly longer RTs on a processing speed task, whereas hyperactive-type ADHD patients had significantly shorter RTs (though it should be noted that neither subtype was significantly associated with processing speed).

Intraindividual variability (IIV) in RT has been studied in clinical samples for more than 90 years (Head, 1926), and has been found to be significantly increased in individuals diagnosed with ADHD relative to nonclinical groups (see Kofler et al., 2013 for a meta-analysis of 319 studies). Seeing as increased IIV is associated with mind wandering and lapses in attention (e.g., West, Murphy, Armilio, Craik, & Stuss, 2002)—a commonly reported observational characteristic of ADHD patients—it stands to reason that there may be an association between IIV and ADHD symptom severity. There are several neurological explanations that could explain this effect. First, brain regions associated with IIV are also commonly implicated in ADHD, including much of the lateral and medial prefrontal cortices, such as the inferior frontal gyrus, superior frontal gyrus, and dorsal anterior cortex (Bellgrove, Hester, & Garavan, 2004; MacDonald, Nyberg, & Backman, 2006). Second, increased IIV in ADHD is theorized to be caused by dysfunctional dopaminergic and noradrenergic neurotransmission, which has been observed in several imaging studies of patients diagnosed with ADHD. This is supported by current pharmacotherapy approaches in treating ADHD, which involves the use of psychostimulants such as methylphenidate and dexamphetamine that target the dopaminergic and noradrenergic systems (Faraone & Buitelaar, 2010). In addition, twin studies have found a genetic link between hyperactivity and IIV (Kuntsi, Oosterlaan, & Stevenson, 2001; Kuntsi & Stevenson, 2001), and adolescents carrying two copies of a high-risk allele for ADHD (a 10-repeat allele of the dopamine transporter gene) displayed increased IIV in an attention task (Bellgrove, Hawi, Kirley, Gill, & Robertson, 2005).

Considering that IIV is predictive of intelligence, even when average RT is taken into account (Jensen, 1992), it is possible that (a) there is more than one measurable “speed” construct in the brain, and if so, (b) ADHD may differentially affect these multiple speed domains. The current study was a reanalysis of the data presented by Brydges et al. (2017). Specifically, the processing speed tasks, from which median RTs were extracted, were recalculated, and IIV in RT measures were extracted from each of the tasks. The current study aimed to determine if associations between ADHD symptom severity and fluid intelligence were mediated by working memory and IIV. It was hypothesized that ADHD is associated with deficits in gF, working memory, and IIV. Hence, it was predicted that the severity of ADHD symptoms combined across hyperactivity and inattentiveness sub-scores would negatively predict gF and working memory performance (i.e., lower scores), and would positively predict IIV (i.e., greater variability in RTs). It was also hypothesized that gF, working memory, and IIV are all related constructs. As such, it was predicted that these three constructs would significantly correlate in a confirmatory factor analysis (CFA), and that working memory and IIV would be predictive of gF in a structural equation model (SEM). In addition, it was hypothesized that ADHD-related deficits in gF are only observable due to deficits in working memory and IIV. Hence, it was predicted that the ADHD-gF association would be fully mediated in an SEM by working memory and IIV. Last, the same analyses were conducted on hyperactivity and inattentiveness sub-scores with the same hypotheses that the sub-score-gF associations would be fully mediated by working memory and IIV.

Method

Participants

The sample examined in the current study is the same as that reported in Brydges et al. (2017). Participants were 142 adults aged 18 to 40 years (M = 22.24, SD = 4.20), of which 91 were female. All participants demonstrated competency in English and reported normal or corrected-to-normal hearing and vision. No other inclusion/exclusion criteria were applied. The majority of the participants were undergraduate psychology students at an Australian university, with the remainder recruited from the general population. The sample had a mean fluid intelligence quotient (IQ) of 116.01 (given that the Cattell Culture-Fair Intelligence Test has a standardized mean of 100 and SD of 24, this is within normal range). Participants were offered a choice of 1.5 participation credit hours to partially fulfill course requirements, or entry to a draw to win a $50 gift card.

Materials

All working memory and IIV tasks were from the Psychology Experiment Building Language (PEBL) version 0.13 (Mueller & Piper, 2014). In addition, the ADHD symptoms questionnaire was administered through the PEBL Survey program.

ADHD questionnaire

World Health Organization Adult ADHD Self-Report Scale (ASRS)

The ASRS is a self-administered questionnaire designed to measure current ADHD symptoms in adults (Kessler et al., 2005). It consists of 18 items based on the Diagnostic and Statistical Manual of Mental Disorders (4th ed.; DSM-IV; American Psychiatric Association, 1994) symptoms or criteria for ADHD that are measured on a 5-point scale (0 = never and 4 = very often), with higher scores indicative of more severe symptoms. Nine items each related to the hyperactivity and inattentive subtypes of ADHD. The total score (which could range from 0-72) was used as the indicator for ADHD symptoms, with each subtype having a range of 0 to 36. The ASRS has internal consistency in the range .63 to .72 and test-retest reliability (Pearson correlations) in the range .58 to .77 (Kessler et al., 2007), and also correlates highly with the clinician-administered ADHD Rating Scale (r = .84; Adler et al., 2006).

Reaction-time IIV measures

Simple motor response

A single stimulus (an “X”) would appear on screen, and participants were required to press the X key as quickly as possible. There were four blocks of 50 trials, and to ensure participants were paying attention and reduce the temporal predictability of responding, the inter-stimulus interval varied between 250 and 2,500 ms.

Two-choice motor response

In the two-choice motor response task, a fixation cross would appear in the center of the computer screen for 1,250 ms, before being replaced with the target stimulus (a letter) for 100 ms. The stimulus would then disappear and be replaced with three non-alphabetical characters (e.g., “&#@”), with the target stimulus appearing either to the left or right of center, and a distractor stimulus on the other side. Participants were required to press a button (Z or /) to indicate which side of the screen the target stimulus appeared on. There were 150 trials administered in a single block.

Four-choice motor response

In the four-choice motor response task, a 2 × 2 grid was presented on screen. A cross would appear in one of the four squares, and the participants were required to press a keyboard button (F, V, J, or N) to indicate which square the cross appeared in. There were 50 trials administered in a single block.

Working memory measures

Backward digit span

Participants were required to remember and then recall lists of numbers of increasing length in reverse order. There were two trials of each list length, starting with three digits. The task ended once participants recorded two consecutive incorrect trials. The indicator of working memory was the number of correct trials.

Memory span

Participants were sequentially presented with a series of pictures on a computer screen and had to recall them in order. This task had 16 trials, starting with four pictures to be recalled, and used a staircase method to determine memory span. This final value from the staircase method was the indicator used.

Corsi block task

Participants were presented with nine navy blue boxes on a computer screen, one of which at a time would light up. Participants were required to remember and then recall the order in which the boxes lit up in. There were two trials of each length, starting with two boxes lighting up. The task ended once participants recorded two consecutive incorrect trials. The indicator of working memory was the number of correct trials.

Fluid intelligence measure

Cattell Culture Fair Intelligence Test (Scale 3, Form A)

The Cattell Culture Fair Intelligence Test (CCFIT) is a commonly used, nonverbal measure of gF (Cattell, 1973). The task requires inductive reasoning about perceptual patterns, and consists of four timed subtests (series completion, odd-one-out, matrices, and topology), with items increasing in difficulty within each subtest. The indicator for gF was the raw score of this measure, which is the total number of correct items across all subtests. The CCFIT is an excellent measure of gF with high reliability and convergent validity with other tests of general intelligence (r = .70-.81; Cattell, 1973).

Procedure

Testing took place in laboratory settings in a single session that lasted approximately 1.5 hr. The order of task administration was fixed for all participants, with the constraint that no two tasks that were supposed to tap into the same construct occurred consecutively. The order was CCFIT, Two-Choice Reaction Time, Backward Digit Span, Simple Reaction Time, Corsi, Four-Choice Reaction Time, Memory Span, ASRS.

Computation of IIV

For the three IIV tasks, extremely fast or slow RTs were removed as they may be a source of measurement error. A lower bound of 150 ms was set for all three tasks based on previous research (Hultsch, MacDonald, & Dixon, 2002), and upper bounds involved computing the task mean and SD, and removing any trials that fell more than 3 SDs from the mean. Missing values were then replaced with the mean RT, as this decreases within-subject variation and represents a conservative approach to examining IIV.

To calculate IIV, the statistical procedures proposed by Hultsch, Strauss, Hunter, and MacDonald (2008) and Bielak, Hultsch, Strauss, MacDonald, and Hunter (2010) were used to calculate the across-trial intraindividual SD. This method is statistically superior to other measures of IIV, such as raw RT SD and the intraindividual coefficient of variation (ICV), because both systematic within-subject and systematic between-subject variability are controlled for. For instance, IIV is well-known to increase with age (Bielak, Cherbuin, Bunce, & Anstey, 2014), and decrease as a result of practice effects (Hultsch, MacDonald, Hunter, Levy-Bencheton, & Strauss, 2000). As Hultsch et al. (2000) state, IIV is defined as the within-subject variability that is independent of relatively durable changes over time (e.g., practice effects). As systematic within-subject variation (e.g., practice effects) is not of interest in the current study, any systematic effects associated with trials must be removed prior to statistical analysis. In addition, if there are group differences in mean levels of performance, it is possible that group differences in IIV are an artefact of average differences in performance. Therefore, between-subjects effects must also be statistically removed prior to any analyses. As such, potential confounds (e.g., age, practice effects) and their higher-order interactions were partialled out using split-point regression. The SD of the residuals extracted from this regression of the processing speed tasks for each participant was used as the measure of IIV for the three tasks.

Transformations and Outlier Analysis

The raw distributions of the 10 measures all had a satisfactory level of normality, so no transformations were conducted. As CFA and SEM are very sensitive to outliers, univariate and multivariate outlier analyses were conducted on the nine dependent variables. Specifically, a test score was considered a univariate outlier if it was greater than 3 SDs from the between-subjects variable mean, and was replaced with a value that was 3 SDs from the mean. This affected no more than 2.1% of the observations for each task. No multivariate outliers were identified when using a Cook’s D value of >1 (Cook & Weisberg, 1982), or had a larger Mahalanobis distance than the critical value. Little’s (1988) missing completely at random (MCAR) test was nonsignificant, χ²(79) = 100.20; p = .054, indicating that missing data were missing completely at random. These scores were estimated using the expectation maximization method implemented in SPSS version 24 (IBM).

Statistical Analysis

Based on guidelines summarized by Schweizer (2010), the CFA/SEM models were evaluated to be well-fitting according to the following criteria: comparative fit index (CFI) ≥ .95; root mean square error of approximation (RMSEA) < .06; and the standardized root mean square residual (SRMR) < .08, although values of .90 (for the CFI) and .10 (for the RMSEA and SRMR) are acceptable (Blunch, 2008). For thoroughness, the implied model chi-square statistics were reported. All models were tested in Amos 24 (Arbuckle, 2016) via maximum likelihood estimation. Significance of correlation and path coefficients was determined by conducting χ² difference tests when removing an individual regression parameter. If the difference in model fit was significant, it indicated that the regression path makes a significant contribution to model fit. This method is more reliable than using test statistics that are based upon comparing standard errors of parameters (Gonzalez & Griffin, 2001).

Results

Descriptive Statistics

Descriptive statistics of the eight measures are presented in Table 1, and the correlations between measures are presented in Table 2.

Table 1.

Descriptive Statistics of the ASRS and Cognitive Measures Used in the Analyses (N = 142).

Measure	M (SD)	Range
ADHD symptoms
ASRS Total^a (/72)	31.70 (11.39)	4-65
ASRS Inattentive^a (/36)	16.96 (6.29)	3-35
ASRS Hyperactive^a (/36)	14.51 (6.31)	0-32
Intraindividual variability
Simple Motor Response^b	61.94 (20.79)	32.61-153.18
Two-Choice Motor Response^b	120.56 (46.90)	40.40-271.55
Four-Choice Motor Response^b	122.61 (83.47)	27.41-450.03
Working memory
Backward Digit Span^c (/16)	6.51 (2.71)	1-14
Corsi^c (/16)	9.02 (1.64)	5-14
Memory Span^d (/9)	4.83 (1.03)	3-8
Fluid intelligence
CCFIT^e (/46)	27.23 (4.48)	14-39
Median Reaction Time
Simple Motor Response^b	318.04 (26.22)	256.0-389.0
Two-Choice Motor Response^b	505.81 (71.31)	390.0-725.5
Four-Choice Motor Response^b	475.13 (80.18)	319.0-740.0

Note. ASRS = ADHD Self-Report Scale; CCFIT = Cattell Culture Fair Intelligence Test.

Number of points.

ms.

Total trials correct.

Span.

Total items correct.

Table 2.

Correlations Between the ASRS and Cognitive Measures Used in the Analyses (N = 142).

	1	2	3	4	5	6	7	8	9	10
ASRS Total	—
ASRS Inattentive	.89**	—
ASRS Hyperactive	.90**	.60**	—
Simple Motor Response	.16	.13	.19*	—
Two-Choice Motor Response	.13	.18*	.05	.41**	—
Four-Choice Motor Response	.12	.07	.15	.23**	.22*	—
Backward Digit Span	−.23**	−.27**	−.15	.08	−.16	−.17*	—
Corsi	−.11	−.08	−.13	−.22**	−.26**	−.15	.12	—
Memory Span	−.05	−.03	−.05	−.08	−.07	−.03	.26**	.09	—
CCFIT	−.17*	−.11	−.20*	−.34**	−.28**	−.23**	.31**	.20*	.24**	—

Note. ASRS = ADHD Self-Report Scale; CCFIT = Cattell Culture Fair Intelligence Test.

p < .05. **p < .01.

Associations Between ADHD Symptoms and Cognitive Functioning

A six-construct CFA model was created, with correlations between the two latent variables (IIV and working memory) and four manifest variables (the three ASRS scores and CCFIT) all free to vary, presented in Figure 1. It was found that the ASRS total and hyperactivity scores significantly correlated with the CCFIT (based on the significant worsening of model fit when the paths were removed; Total Δχ² = 4.18, Δdf = 1, p = .041; Hyperactivity Δχ² = 5.93, Δdf = 1, p = .015), but the ASRS inattentiveness score did not (Δχ² = 1.71, Δdf = 1, p = .19). The working memory and IIV factors also significantly correlated with CCFIT (working memory Δχ² = 14.87, Δdf = 1, p < .001; IIV Δχ² = 20.70, Δdf = 1, p < .001), although the working memory and IIV factors did not significantly correlate with each other (Δχ² = 0.66, Δdf = 1, p = .42). After removing nonsignificant correlations from the CFA, the final model reported acceptable fit indices, χ²(27) = 43.74, p = .022, CFI = .982, RMSEA = .066, SRMR = .085.

Figure 1.

The estimated six-construct model.

SEM and Mediation Analyses

Two separate SEMs were created, with ASRS Total score and ASRS Hyperactivity score as the predictor in each model (no SEM and mediation analysis were conducted with ASRS Inattentiveness as a predictor as it was not significantly correlated with CCFIT in the CFA). Each model had the relevant ASRS variable predicting CCFIT, with this association being mediated by working memory and IIV (which were not correlated with each other, given the nonsignificant correlation in the CFA model).

For the ASRS total score, initial model fit statistics were mixed, χ²(19) = 27.80, p = .047, CFI = .896, RMSEA = .067, SRMR = .078. It was hypothesized that associations between ASRS and deficits in gF are due to impaired working memory and IIV. As such, it was predicted that the ASRS-CCFIT relationship would be fully mediated by working memory and processing speed. Removing the ASRS-CCFIT path from the model did not significantly worsen fit (Δχ² = 1.01, Δdf = 1, p = .31), therefore, the relationship was no longer significant and was fully mediated.

To determine whether working memory and/or IIV were mediators, further testing was conducted. Removal of the ASRS-working memory (Δχ² = 6.16, Δdf = 1, p = .013) and working memory-CCFIT (Δχ² = 16.49, Δdf = 1, p < .001) paths both significantly worsened model fit, indicating that these paths are both significant, and that working memory mediates the ASRS-CCFIT relationship. The same procedure was then conducted with the IIV factor. Removal of both the ASRS-IIV (Δχ² = 4.28, Δdf = 1, p = .039) and the IIV-CCFIT (Δχ² = 19.51, Δdf = 1, p < .001) paths significantly worsened model fit, indicating that these paths are both significant, that the ASRS is associated with IIV, that IIV is predictive of CCFIT, and that both working memory and IIV fully mediate the ASRS-CCFIT relationship. The final model reported acceptable fit indices, χ²(18) = 28.81, p = .051, CFI = .90, RMSEA = .065, SRMR = .080, and is presented in Figure 2. In addition, the model yielded an R² = .45. Thus, 45% of the variance in the CCFIT was accounted for by the model that included working memory and IIV as predictors of the CCFIT.

Figure 2.

The final mediation model.

This process was then repeated for the ASRS Hyperactivity variable, and similar results were observed. ASRS Hyperactivity did not significantly predict CCFIT (Δχ² = 0.04, Δdf = 1, p = .84), and this association was fully mediated by both working memory (WM-CCFIT: Δχ² = 20.90, Δdf = 1, p < .001; though ASRS-WM was not significant, Δχ² = 3.54, Δdf = 1, p = .060) and IIV (ASRS-IIV: Δχ² = 4.66, Δdf = 1, p = .031; IIV-CCFIT: Δχ² = 19.56, Δdf = 1, p < .001). The final model fit statistics were low, χ²(19) = 33.48, p = .021, CFI = .86, RMSEA = .074, SRMR = .090.

Discussion

The aim of this study was to determine if associations between ADHD symptomology and fluid intelligence were mediated by working memory and RT IIV. It was hypothesized that there would be ADHD-related deficits in gF, working memory, and IIV. It was also hypothesized that gF, working memory, and IIV are all related to each other. Third, it was hypothesized that ADHD-related deficits in gF are only observable due to deficits in working memory and IIV. On the whole, these hypotheses were generally supported.

The first hypothesis of the current study was that ADHD-related deficits would be observed in the three cognitive domains of interest: gF, working memory, and IIV. Previous research has shown behavioral deficits in ADHD patient groups in each of these three constructs (Dige & Wik, 2005; Kofler et al., 2013; Schweitzer et al., 2006), and the results of the current study were consistent with these findings. Given that neuroimaging research has found that ADHD is associated with significant structural and functional differences across the frontoparietal network (which is implicated in successful higher-order cognition; Bush, Valera, & Seidman, 2005; Niendam et al., 2012; Seidman et al., 2006), it follows that these constructs may be impacted as a result.

The second hypothesis stated that gF, working memory, and IIV would all be related to one another. While gF was associated with both working memory and IIV, supporting previous research that has examined the relationship between the two constructs (e.g., Ackerman et al., 2005; Jensen, 1992), in our study, the correlation between working memory and IIV was not significant. Brydges et al. (2017) found a significant association between working memory and processing speed factors when processing speed was measured with median RTs. However, Jensen (1992) suggested that “traditional” processing speed (i.e., average RT) and RT variability, may be two different constructs, as these measures shared significant proportions of variance but were both significantly associated with general intelligence. It may well be the case that traditional processing speed (i.e., how fast a response is on average) is not significantly affected by ADHD or that processing speed is differentially affected by different ADHD subtypes (Nigg et al., 2005). In addition, research has shown that in complex, cognitively demanding tasks (e.g., novel rule-learning paradigms) and/or tasks that require prolonged sustained attention (e.g., the sustained attention to response [SART] task or Attentional Network Test [ANT]), IIV (or lapses in attention), and gF are highly correlated, but only for infrequent events (e.g., rarely presented trials), with little or no relationship observed between gF and IIV for the most commonly presented trial type (Roberts, Jones, Davis, Ly, & Anderson, 2014). Future research should attempt to examine the separability of average RT and RT consistency in a clinical sample, and how each are modulated by task demands to better understand and examine their effect on ADHD subtypes.

The third hypothesis was that ADHD-related deficits in gF would only be observable due to deficits in working memory and IIV. That is, the ADHD-gF association would be fully mediated by working memory and IIV. This hypothesis was supported. As previously stated, previous research has suggested close associations between working memory, IIV, and gF from both behavioral (e.g., Ackerman et al., 2005; Jensen, 1992) and neuroimaging (MacDonald et al., 2006; Niendam et al., 2012) perspectives. Hence, the results of the current study imply that decreased gF could possibly be an indirect result of ADHD, although we acknowledge the study design does not imply causality. That is, ADHD is associated with impaired working memory and increased IIV in RT, which in turn leads to decreased gF.

These results show that both working memory and IIV are important predictors of ADHD symptomology. In 1926, Henry Head stated that “an inconsistent response is one of the most striking consequences of lesions to the cerebral cortex” (p. 145). Although our sample was a nonclinical sample of university students—and the results of the study should be qualified as such—there are potential clinical implications, as simply examining the mean or median RT of a cognitive task may potentially give a misleading picture for diagnosis, assessment, and treatment of ADHD, although this needs to be tested in a clinical sample. As Brydges et al. (2017) stated, it is possible that low gF could behaviorally manifest in a similar manner to ADHD. That is, it is possible that gF is not directly affected by ADHD symptomology, but multiple predictors of gF—including working memory and IIV, based on the results of the current study—are impaired. Difficulties in clinical diagnosis of ADHD may be somewhat due to the significant conceptual overlap and behavioral similarity between low gF and ADHD symptoms.

The results of the current study are congruent with the broader neuropharmacological perspective on ADHD and clinical disorders involving the catecholamines: dopamine and noradrenaline. Both dopamine and noradrenaline are known to play crucial roles in executive attention and motor control (Braver & Cohen, 2000; Paus, 2001), and are both heavily implicated in ADHD and sustained attention (Biederman & Spencer, 1999; del Campo, Chamberlain, Sahakian, & Robbins, 2011; Levy, 1991). More important, medications used in the treatment of ADHD that modulate levels of dopamine and noradrenaline, such as methylphenidate, have been shown to improve IIV in RT in healthy controls (Nandam et al., 2011; Noreika, Falter, & Rubia, 2013). Reduced consistency of responding is also seen in a wide spectrum of conditions associated with these neurotransmitters, such as old age (e.g., Bielak & Anstey, 2015; Strauss, Bielak, Bunce, Hunter, & Hultsch, 2007), Parkinson’s disease (Scatton, Javoy-Agid, Rouquier, Dubois, & Agid, 1983), schizophrenia (Davis, Kahn, Ko, & Davidson, 1991), and depression (Dunlop & Nemeroff, 2007). In summary, although our sample was a nonclinical sample of university students, the results are consistent with the broader neuropharmacological framework of catecholamine dysfunction in ADHD.

The results of the current study indicate several potentially fruitful avenues for future research. First, as Brydges et al. (2017) stated, future research could examine the validity of the results, and this mediation model in a group of clinically diagnosed ADHD patients. One limitation of the current study was the use of the general population, which restricts the validity of the findings with regard to clinical groups, especially given that the current sample was mostly female and well-educated (see Seidman, 2006; Theiling & Petermann, 2016, however, who suggest that gender differences in ADHD symptomology are minimal). Alternatively, having a wide range of symptoms scores allows for greater variation than a binomial distribution of diagnosed with ADHD versus not diagnosed with ADHD. Given that self-report symptom scores have been found to correlate highly with clinicians’ diagnoses (Adler et al., 2006; Silverstein et al., 2018), future research should consider using symptom severity in a group of ADHD patients, rather than simply examining whether or not an individual has been diagnosed. In addition, examining this mediation model in a sample of children may be of benefit. Given that the prevalence of ADHD is as high as 7.1% in children (Willcutt, 2012), examining a potential hierarchical model with a view toward developing attention-training programs could be a fruitful area of research.

In conclusion, the current study has shown that both working memory and IIV fully mediate deficits in gF that are associated with ADHD symptoms, albeit in a nonclinical sample. The results of this study highlight the need for clinicians and researchers to examine the consistency of behavior in psychometric assessment of speeded-tasks, especially when treating or investigating complex psychiatric disorders associated with abnormal functioning of the brain’s dopaminergic and noradrenergic systems. Although conducted with healthy adults from the normal population, the results of the current study suggest that ADHD symptomology and its relationship to gF are likely to be due to normal variation between individuals in their ability to store information in working memory and respond consistently.

Footnotes

Declaration of Conflicting Interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The author(s) received no financial support for the research, authorship, and/or publication of this article.

ORCID iD

Christopher R. Brydges

Author Biographies

Christopher R. Brydges is currently a postdoctoral research fellow in the Department of Human Development and Family Studies at Colorado State University. He earned his PhD from the University of Western Australia, in Perth, Australia. His research interests include higher-order cognition and its development, from both behavioral and electrophysiological perspectives.

Krista L Ozolnieks is a registered provisional psychologist following completion of a Master of Professional Psychology at Curtin University, Australia. Currently practicing in rehabilitation settings with a focus on disability.

Gareth Roberts is a postdoctoral research fellow in the School of Psychology at the University of Sydney. He earned his PhD from the University of Western Australia. His current research is focused on the cognitive neuroscience of categorisation, novel task learning, non-verbal reasoning, and theories of prefrontal cortex function.

References

Ackerman

P. L.

Beier

M. E.

Boyle

M. O.

(2005). Working memory and intelligence: The same or different constructs? Psychological Bulletin, 131, 30-60. doi:10.1037/0033-2909.131.1.30

Adler

L. A.

Spencer

Faraone

D. V.

Kessler

R. C.

Howes

M. J.

Biederman

Secnik

(2006). Validity of pilot adult ADHD self-report scale (ASRS) to rate adult ADHD symptoms. Annals of Clinical Psychiatry, 18, 145-148.

American Psychiatric Association. (1994). Diagnostic and statistical manual of mental disorders (4th ed.). Washington, DC: Author.

Arbuckle

J. L.

(2016). IBM SPSS Amos 24 user’s guide. New York, NY: IBM.

Baddeley

A. D.

Hitch

G. J.

(1974). Working memory. In Bower

(Ed.), The psychology of learning and motivation (Vol. 8, pp. 47-89). New York, NY: Academic Press.

Barkley

R. A.

(1997). Behavioral inhibition, sustained attention, and executive functions: Constructing a unifying theory of ADHD. Psychological Bulletin, 121, 65-94. doi:10.1037/0033-2909.121.1.65

Bellgrove

M. A.

Hawi

Kirley

Gill

Robertson

I. H.

(2005). Dissecting the attention deficit hyperactivity disorder (ADHD) phenotype: Sustained attention, response variability and spatial attentional asymmetries in relation to dopamine transporter (DAT1) genotype. Neuropsychologia, 43, 1847-1857. doi:10.1016/j.neuropsychologia.2005.03.011

Bellgrove

M. A.

Hester

Garavan

(2004). The functional neuroanatomy of response variability: Evidence from a response inhibition task. Neuropsychologia, 42, 1910-1916. doi:10.1016/j.neuropsychologia.2004.05.007

Biederman

Spencer

(1999). Attention-deficit/hyperactivity disorder (ADHD) as a noradrenergic disorder. Biological Psychiatry, 46, 1234-1242. doi:10.1016/S0006-3223(99)00192-4

10.

Bielak

A. A. M.

Anstey

K. J.

(2015). Intraindividual variability in attention across the adult lifespan. In Diehl

Hooker

Sliwinski

M. J.

(Eds.), Handbook of intraindividual variability across the lifespan (pp. 160-175). New York, NY: Routledge.

11.

Bielak

A. A. M.

Cherbuin

Bunce

Anstey

K. J.

(2014). Intraindividual variability is a fundamental phenomenon of aging: Evidence from an 8-year longitudinal study across young, middle, and older adulthood. Developmental Psychology, 50, 143-151. doi:10.1037/a0032650

12.

Bielak

A. A. M.

Hultsch

D. F.

Strauss

MacDonald

S. W. S.

Hunter

M. A.

(2010). Intraindividual variability in reaction time predicts cognitive outcomes 5 years later. Neuropsychology, 24, 731-741. doi:10.1037/a0019802

13.

Blunch

N. J.

(2008). Introduction to structural equation modelling using SPSS and AMOS. London, England: SAGE.

14.

Boonstra

A. M.

Oosterlaan

Sergeant

J. A.

Buitelaar

J. K.

(2005). Executive functioning in adult ADHD: A meta-analytic review. Psychological Medicine, 35, 1097-1108. doi:10.1017/S003329170500499X

15.

Braver

T. S.

Cohen

J. D.

(2000). On the control of control: The role of dopamine in regulating prefrontal function and working memory. In Monsell

Driver

(Eds.), Attention and performance XIII: Control of cognitive processes (pp. 713-737). Cambridge, MA: MIT Press.

16.

Brydges

C. R.

Ozolnieks

K. L.

Roberts

(2017). Working memory—Not processing speed—Mediates fluid intelligence deficits associated with attention deficit/hyperactivity disorder symptoms. Journal of Neuropsychology, 11, 362-377. doi:10.1111/jnp.12096

17.

Bush

Valera

E. M.

Seidman

L. J.

(2005). Functional neuroimaging of attention-deficit/hyperactivity disorder: A review and suggested future directions. Biological Psychiatry, 57, 1273-1284. doi:10.1016/j.biopsych.2005.01.034

18.

Cattell

R. B.

(1963). Theory of fluid and crystallized intelligence: A critical experiment. Journal of Educational Psychology, 54, 1-22. doi:10.1037/h0046743

19.

Cattell

R. B.

(1973). Cattell Culture Fair Intelligence Test. Champaign, IL: Institute for Personality and Ability Testing.

20.

Cook

R. D.

Weisberg

(1982). Residuals and influence in regression. New York, NY: Chapman & Hill.

21.

Davis

K. L.

Kahn

R. S.

Davidson

(1991). Dopamine in schizophrenia: A review and reconceptualization. The American Journal of Psychiatry, 148, 1474-1486. doi:10.1176/ajp.148.11.1474

22.

del Campo

Chamberlain

S. R.

Sahakian

B. J.

Robbins

T. W

. (2011). The roles of dopamine and noradrenaline in the pathophysiology and treatment of attention-deficit/hyperactivity disorder. Biological Psychiatry, 69, e145-e157. doi:10.1016/j.biopsych.2011.02.036

23.

Dige

Wik

(2005). Adult attention deficit hyperactivity disorder identified by neuropsychological testing. International Journal of Neuroscience, 115, 169-183. doi:10.1080/00207450590519058

24.

Dunlop

B. W.

Nemeroff

C. B.

(2007). The role of dopamine in the pathophysiology of depression. Archives of General Psychiatry, 64, 327-337. doi:10.1001/archpsyc.64.3.327

25.

Faraone

S. V.

Buitelaar

(2010). Comparing the efficacy of stimulants for ADHD in children and adolescents using meta-analysis. European Child & Adolescent Psychiatry, 19, 353-364. doi:10.1007/s00787-009-0054-3

26.

Gonzalez

Griffin

(2001). Testing parameters in structural equation modeling: Every “one” matters. Psychological Methods, 6, 258-269. doi:10.1037/1082-989X.6.3.258

27.

Head

(1926). Aphasia and kindred disorders of speech. Cambridge, UK: Cambridge University Press.

28.

Hultsch

D. F.

MacDonald

S. W. S.

Dixon

R. A.

(2002). Variability in reaction time performance of younger and older adults. The Journals of Gerontology: Series B, 57, P101-P115. doi:10.1093/geronb/57.2.P101

29.

Hultsch

D. F.

MacDonald

S. W. S.

Hunter

M. A.

Levy-Bencheton

Strauss

(2000). Intraindividual variability in cognitive performance in older adults: Comparison of adults with mild dementia, adults with arthritis, and healthy adults. Neuropsychology, 14, 588-598. doi:10.1037/0894-4105.14.4.588

30.

Hultsch

D. F.

Strauss

Hunter

M. A.

MacDonald

S. W. S.

(2008). Intraindividual variability, cognition, and aging. In Craik

F. I. M.

Salthouse

T. A.

(Eds.), The handbook of aging and cognition (3rd ed., pp. 491-556). New York, NY: Psychology Press.

31.

Ingram

Hechtman

Morgenstern

(1999). Outcome issues in ADHD: Adolescent and adult long-term outcome. Mental Retardation and Developmental Disabilities Research Reviews, 5, 243-250. doi:10.1002/(SICI)1098-2779(1999)5:3<243::AID-MRDD11>3.0.CO;2-D

32.

Jensen

A. R.

(1992). The importance of intraindividual variation in reaction time. Personality and Individual Differences, 13, 869-881. doi:10.1016/0191-8869(92)90004-9

33.

Kail

Salthouse

T. A.

(1994). Processing speed as a mental capacity. Acta Psychologica, 86, 199-225. doi:10.1016/0001-6918(94)90003-5

34.

Kessler

R. C.

Adler

Ames

Demler

Faraone

Hiripi

. . . Walters

E. E.

(2005). The world health organization adult ADHD Self-Report Scale (ASRS): A short screening scale for use in the general population. Psychological Medicine, 35, 245-256. doi:10.1017/S0033291704002892

35.

Kessler

R. C.

Adler

L. A.

Gruber

M. J.

Sarawate

C. A.

Spencer

Van Brunt

D. L.

(2007). Validity of the World Health Organization Adult ADHD Self-Report Scale (ASRS) Screener in a representative sample of health plan members. International journal of methods in psychiatric research, 16, 52-65. doi:10.1002/mpr.208

36.

Klein

Mannuzza

(1991). Long-term outcome of hyperactive children: A review. Journal of the American Academy of Child & Adolescent Psychiatry, 30, 383-387. doi:10.1097/00004583-199105000-00005

37.

Kofler

M. J.

Rapport

M. D.

Sarver

D. E.

Raiker

J. S.

Orban

S. A.

Friedman

L. M.

Kolomeyer

E. G.

(2013). Reaction time variability in ADHD: A meta-analytic review of 319 studies. Clinical Psychology Review, 33, 795-811. doi:10.1016/j.cpr.2013.06.001

38.

Kuntsi

Oosterlaan

Stevenson

(2001). Psychological mechanisms in hyperactivity: I Response inhibition deficit, working memory impairment, delay aversion, or something else? The Journal of Child Psychology and Psychiatry and Allied Disciplines, 42, 199-210. doi:10.1111/1469-7610.00711

39.

Kuntsi

Stevenson

(2001). Psychological mechanisms in hyperactivity: II The role of genetic factors. The Journal of Child Psychology and Psychiatry and Allied Disciplines, 42, 211-219.

40.

Kyllonen

P. C.

Christal

R. E.

(1990). Reasoning ability is (little more than) working-memory capacity? Intelligence, 14, 389-433. doi:10.1016/S0160-2896(05)80012-1

41.

Levy

(1991). The dopamine theory of attention deficit hyperactivity disorder (ADHD). Australian & New Zealand Journal of Psychiatry, 25, 277-283. doi:10.3109/00048679109077746

42.

Little

R. J. A.

(1988). A test of missing completely at random for multivariate data with missing values. Journal of the American Statistical Association, 83, 1198-1202. doi:10.1080/01621459.1988.10478722

43.

MacDonald

S. W.

Nyberg

Backman

(2006). Intra-individual variability in behavior: Links to brain structure, neurotransmission and neuronal activity. Trends in Neurosciences, 29, 474-480. doi:10.1016/j.tins.2006.06.011

44.

Mueller

S. T.

Piper

B. J.

(2014). The Psychology Experiment Building Language (PEBL) and PEBL test battery. Journal of Neuroscience Methods, 222, 250-259. doi:10.1016/j.jneumeth.2013.10.024

45.

Nandam

L. S.

Hester

Wagner

Cummins

T. D.

Garner

Dean

A. J.

. . . Bellgrove

M. A.

(2011). Methylphenidate but not atomoxetine or citalopram modulates inhibitory control and response time variability. Biological Psychiatry, 69, 902-904. doi:10.1016/j.biopsych.2010.11.014

46.

Niendam

T. A.

Laird

A. R.

Ray

K. L.

Dean

Y. M.

Glahn

D. C.

Carter

C. S.

(2012). Meta-analytic evidence for a superordinate cognitive control network subserving diverse executive functions. Cognitive, Affective, & Behavioral Neuroscience, 12, 241-268. doi:10.3758/s13415-011-0083-5

47.

Nigg

J. T.

Stavro

Ettenhofer

Hambrick

D. Z.

Miller

Henderson

J. M.

(2005). Executive functions and ADHD in adults: Evidence for selective effects on ADHD symptom domains. Journal of Abnormal Psychology, 114, 706-717. doi:10.1037/0021-843X.114.3.706

48.

Noreika

Falter

C. M.

Rubia

(2013). Timing deficits in attention-deficit/hyperactivity disorder (ADHD): Evidence from neurocognitive and neuroimaging studies. Neuropsychologia, 51, 235-266. doi:10.1016/j.neuropsychologia.2012.09.036

49.

Paus

(2001). Primate anterior cingulate cortex: Where motor control, drive and cognition interface. Nature Reviews Neuroscience, 2, 417-424. doi:10.1038/35077500

50.

Redick

T. S.

Unsworth

Kelly

A. J.

Engle

R. W.

(2012). Faster, Smarter? Working memory capacity and perceptual speed in relation to fluid intelligence. Journal of Cognitive Psychology, 24, 844-854. doi:10.1080/20445911.2012.704359

51.

Roberts

Jones

T. W.

Davis

E. A.

T. T.

Anderson

(2014). Building tasks from verbal instructions: An EEG study on practice trial exposure and task structure complexity during novel sequences of behavior. Cognitive, Affective, & Behavioral Neuroscience, 14, 1356-1374. doi:10.3758/s13415-014-0276-9

52.

Scatton

Javoy-Agid

Rouquier

Dubois

Agid

(1983). Reduction of cortical dopamine, noradrenaline, serotonin and their metabolites in Parkinson’s disease. Brain Research, 275, 321-328. doi:10.1016/0006-8993(83)90993-9

53.

Schweitzer

J. B.

Hanford

R. B.

Medoff

D. R.

(2006). Working memory deficits in adults with ADHD: Is there evidence for subtype differences? Behavioral and Brain Functions, 2, Article 43. doi:10.1186/1744-9081-2-43

54.

Schweizer

(2010). Some guidelines concerning the modeling of traits and abilities in test construction. European Journal of Psychological Assessment, 26, 1-2. doi:10.1027/1015-5759/a000001

55.

Seidman

L. J.

(2006). Neuropsychological functioning in people with ADHD across the lifespan. Clinical Psychology Review, 26, 466-485. doi:10.1016/j.cpr.2006.01.004

56.

Seidman

L. J.

Valera

E. M.

Makris

Monuteaux

M. C.

Boriel

D. L.

Kelkar

. . . Biederman

(2006). Dorsolateral prefrontal and anterior cingulate cortex volumetric abnormalities in adults with attention-deficit/hyperactivity disorder identified by magnetic resonance imaging. Biological Psychiatry, 60, 1071-1080. doi:10.1016/j.biopsych.2006.04.031

57.

Sheppard

L. D.

Vernon

P. A.

(2008). Intelligence and speed of information-processing: A review of 50 years of research. Personality and Individual Differences, 44, 535-551. doi:10.1016/j.paid.2007.09.015

58.

Silverstein

M. J.

Faraone

S. V.

Alperin

Biederman

Spencer

T. J.

Adler

L. A.

(2018). How informative are self-reported adult attention-deficit/hyperactivity disorder symptoms? An examination of the agreement between the adult Attention-Deficit/Hyperactivity Disorder Self-Report Scale V1.1 and adult Attention-Deficit/Hyperactivity Disorder Investigator Symptom Rating Scale. Journal of Child and Adolescent Psychopharmacology, 28, 339-349. doi:10.1089/cap.2017.0082

59.

Strauss

Bielak

A. A.

Bunce

Hunter

M. A.

Hultsch

D. F.

(2007). Within-person variability in response speed as an indicator of cognitive impairment in older adults. Aging, Neuropsychology, and Cognition, 14, 608-630. doi:10.1080/13825580600932419

60.

Theiling

Petermann

(2016). Neuropsychological profiles on the WAIS-IV of adults with ADHD. Journal of Attention Disorders, 20, 913-924.

61.

Vernon

P. A.

(1983). Speed of information processing and general intelligence. Intelligence, 7, 53-70. doi:10.1016/0160-2896(83)90006-5

62.

Weiss

Hechtman

(1993). Hyperactive children grown up. New York, NY: Guilford Press.

63.

West

Murphy

K. J.

Armilio

M. L.

Craik

F. I.

Stuss

D. T.

(2002). Lapses of intention and performance variability reveal age-related increases in fluctuations of executive control. Brain and Cognition, 49, 402-419. doi:10.1006/brcg.2001.1507

64.

Wilhelm

Oberauer

(2006). Why are reasoning ability and working memory capacity related to mental speed? An investigation of stimulus–response compatibility in choice reaction time tasks. European Journal of Cognitive Psychology, 18, 18-50. doi:10.1080/09541440500215921

65.

Willcutt

E. G.

(2012). The prevalence of DSM-IV attention-deficit/hyperactivity disorder: A meta-analytic review. Neurotherapeutics, 9, 490-499. doi:10.1007/s13311-012-0135-8