Multiple pathway models of ADHD (see Section 1.1.5.) explain heterogeneity on an ADHD symptom level through heterogeneity in basal psychological endophenoytpes. For instance, the dual pathway model (Sonuga-Barke, 2002) proposed that deficits in either behavioral inhibition or delay aversion could independently cause ADHD. Yet, the dual pathway model was not able to explain more than about 50% of the variance in ADHD (Coghill et al., 2014; Sonuga-Barke et al., 2010). Thus, the model incorporated a third pathway: temporal processing (de Zeeuw, Weusten et al., 2012; Durston et al., 2011; Sonuga- Barke et al., 2010). But even the thriple pathways model could not fully explain heterogeneity in ADHD and sustained attention was proposed to constitute an additional pathway (de
Zeeuw et al., 2012). However, so far no formal theoretical model of ADHD includes a pathway of sustained attention or a sustained attention deficit as a core deficit in ADHD.
The results of the current study clearly emphasize the importance of a sustained attention deficit in ADHD. Associations for sustained attention with symptoms of ADHD and the categorical ADHD diagnosis were stronger than associations with behavioral inhibition or conflict-monitoring. The strength of this association was further underpinned by the presence of a total effect of sustained attention on continuous ADHD symptoms even when indirect effects through risky decision-making were considered. In addition, the current study revealed that sustained attention could be reliable measured by a factor score derived from measures of different neuropsychological tasks. Furthermore, measures of sustained attention can be related to specific neuronal areas (Langner & Eickhoff, 2013; Petersen & Posner, 2012; Posner & Petersen, 1990; Sarter et al., 2001; Weissman et al., 2006) and EEG components (Arns, Connors, & Kraemer, 2013; Barry et al., 2003; Hoedlmoser et al., 2010). Finally, heritability has been established through behavioral genetic studies (Kuntsi et al., 2010; Uebel et al., 2010) as well as molecular genetic studies (Barnes et al., 2011; Bellgrove & Mattingley 2008; Kollins et al., 2008) indicating that sustained attention can be considered an endophenotype for ADHD.
With sustained attention as an endophenotype, current multiple pathway models of ADHD would extend to four factor models, including sustained attention, behavioral inhibition, delay aversion, and temporal processing. However, this brings up the question how many pathways actually exist and whether all of the proposed pathways can be considered as endophenotypes. There is preliminary evidence that both delay aversion (see Section 1.1.2.) and temporal processing (see Section 1.1.3.) might be endophenotypes for ADHD.
On delay aversion tasks, behavioral differences between children and adults with and without ADHD have been demonstrated both with regard to an altered delay-to-reinforcement gradient (Demurie, Roeyers, Baeyens, & Sonuga-Barke, 2012; Hurst, Kepley, McCalla, & Livermore, 2011; Paloyelis et al., 2010; Scheres, Tontsch, & Thoeny, 2013; Scheres, Tontsch, Thoeny, & Kaczkurkin, 2010; Wilson, Mitchell, Musser, Schmitt, & Nigg, 2011; but see Scheres et al., 2006) and the motivation to avoid or escape delay situations (Antrop et al., 2006; Bitsakou, Antrop, Wiersema, & Sonuga-Barke, 2006; Bitsakou et al., 2009; Marco et al., 2009; Sonuga-Barke et al., 2010). The smaller delay-to-reinforcement gradient in children and adults with ADHD was related to functional (Costa Dias et al., 2013; Hoogman et al., 2011, 2013; Rubia et al., 2009; Scheres, Milham, Knutson, & Castellanos, 2007; Ströhle et al., 2008) and structural changes in fronto-basal ganglia circuits (Carmona et al., 2009;
Durston et al., 2011), whereas the motivation to avoid or escape delay situations can be related to functional and structural changes in the amygdala (Lemiere et al., 2012; Sasayama et al., 2010; Wilbertz et al., 2013). Finally, heritability of an altered delay-to-reinforcement gradient has been established through behavioral genetic studies (Anokhin et al., 2011; but see Kuntsi et al., 2006) and molecular genetic studies (Kawamura et al., 2013; Paloyelis et al., 2010; Sonuga-Barke et al., 2011) indicating that delay aversion might be an endophenotype for ADHD.
Likewise, an examination of the criteria for endophenotypes with regard to temporal processing demonstrated behavioral differences between children with and without ADHD on a range of temporal processing tasks, including time estimation and time reproduction tasks (Barklay, Murphy, & Bush, 2001; Bauermeister et al., 2005; Pollack, Kroyzer, Yakir, & Friedler, 2009; Prevatt, Proctor, Baker, Garrett, & Yelland, 2011; Rommelse et al., 2007; Valko et al., 2010; see Noreika et al., 2013 for a review), tapping tasks (Rubia, Noorloos, Smith, Gunning, & Sergeant, 2003; Toplak & Tannock, 2005; Zelaznik et al., 2012), and time-based prospective memory tasks (Altgassen, Kretschmer, & Kliegel, 2012; Kerns & Price, 2001; Zinke et al., 2010). Deficits in temporal processing in children with ADHD have been associated with a reduced amplitude of the contingent negative variation ERP component (Banaschewski et al., 2003; Doehnert, Brandeis, Schneider, Drechsler, & Steinhausen, 2013; McLoughlin et al., 2010; Valko et al., 2009) and are thought reflect functional (Durston et al., 2007, 2011; Mulder et al., 2008; Vloet et al., 2010) and structural changes (Castellanos et al., 2002; Mackie et al., 2007; Montes et al., 2011) in fronto- cerebellar circuits. Finally, there is first evidence for heritability of temporal processing from behavioral genetic studies (Rommelse, Altink, Oosterlaan et al., 2008; Rommelse et al., 2007) and molecular genetic studies (de Zeeuw et al., 2013; Sysoeva et al., 2010; Wiener et al., 2011) indicating that temporal processing might be a potential endophenotype for ADHD.
Besides delay aversion and temporal processing, other cognitive constructs might be considered as pathways or even endophenotypes of ADHD, as for instance response speed or working memory. Reduced response speed (de Zeeuw et al., 2012; Epstein et al., 2011; Nikolas & Nigg, 2013; but see Epstein et al., 2003) and decreased working memory capacity (Alderson, Kasper, Hudec, & Patros, 2013; Martinussen et al., 2005; Marzocchi et al., 2008; Rapport et al., 2008) have been reported for children with ADHD, related to neuronal structures (Rypma et al., 2006; Turken et al., 2008; Wen et al., 2011), and have been shown to be heritable (Kumsta et al., 2010; McClearn et al., 1997; Neubauer, Spinath, Riemann, Angleitner, & Borkenau, 2000; Rommelse, Altink, Oosterlaan et al., 2008; Vernon, 1989).
However, no formal model of ADHD (see Section 1.1.) focuses on or even includes deficits in response speed or working memory.
Thus, future studies will have to investigate whether response speed and working memory might constitute additional pathways for ADHD. Furthermore, future studies need to examine whether the proposed pathways are truly independent or whether it is more likely that the single pathways interact. For instance, both working memory and sustained attention have been implicated in temporal processing, considering that a time interval can only be compared to a reference interval if attention is paid to the starting point, the endpoint, and the duration of the interval (Casini & Macar, 1997; Macar, Grondin, & Casini, 1994; Minvielle- Moncla, Audiffren, Macar, & Vallet, 2008; Pollack et al., 2009) and if a representation of the reference interval can be retrieved from working memory (Khan, Sharma, & Dixit, 2006; Matell & Meck, 2000; see Block, Hancock, & Zakay, 2010 for a meta-analysis). Similarly, neuropsychological models have been proposed relating attention, working memory, and behavioral inhibition (e.g., Knudsen, 2007). Attentional processes are supposed to control perceptional inputs that compete for entry into working memory. In turn, working memory compares the processed input with inherent task goals and organizes top-down control to modulate attentional processes in accordance with these task goals, for instance through the inhibition of irrelevant distractors (Knudsen, 2007).
Therefore, to investigate heterogeneity in ADHD it might be more useful to examine performance profiles over various different tasks tapping on different possible pathways, rather than to analyze single pathways independently. Accordingly, a recent study using a latent class analysis on performance on a reward sensitivity task and a time-manipulated Go/No-Go task reported three groups of children with ADHD that could either be characterized as “quick and accurate”, “having poor cognitive control”, or “being slow and variable in timing” (van Hulst, de Zeeuw, & Durston, 2014). However, these results need to be replicated in larger samples and with larger number of cognitive tasks.
In addition, the current study revealed indirect effects of sustained attention and behavioral inhibition on ADHD symptoms through risky decision-making. These indirect effects are in line with the assumption that the different pathways of sustained attention, or behavioral inhibition are not directly related to the symptoms or syndrome of ADHD. Rather, they are related to more complex behaviors like risky decision-making that in turn are related to the symptom dimensions. These higher-order, more complex behaviors might be influenced by various basal cognitive functions as risky decision-making was influenced by both sustained attention and behavioral inhibition. Considering the heterogeneous results
regarding the associations between sustained attention, behavioral inhibition, other cognitive functions and ADHD (Coghill et al., 2014; Sonuga-Barke et al., 2010; Willcutt et al., 2005), it might be fruitful to reconsider the level of analysis in favor of higher-order cognitive functions instead of solely focusing on the endophenotypes implied in multiple pathway models (Coghill et al., 2014; de Zeeuw et al., 2012; Sonuga-Barke et al., 2010).
Thus, future studies should identify other higher-order psychological functions that might potentially mediate the associations of endophenotypes on ADHD. As deficits in problem solving and planning have been observed in ADHD (Marzocchi et al., 2008; Nigg, Blaskey, Huang-Pollock, & Rappley, 2002) and given the associations of problem solving and planning with behavioral inhibition (Baughman & Cooper, 2007; Miyake et al., 2000), problem solving and planning might be candidates for higher-order cognitive functions. In addition, goal setting and goal pursuit might be cognitive-motivational higher-order candidates, as deficits in goal setting and goal pursuit have been observed in children with ADHD (Gawrilow et al., 2013; Nyman et al., 2010) on the one hand, and have been related to inhibitory processes and the allocation of attention on the other hand (Kruglanski et al., 2002; Locke & Latham, 2002).