Attention-deficit/hyperactivity disorder (ADHD) is a neuropsychiatric disorder characterized by motor restlessness and symptoms of impulsivity and inattention. The prevalence in the child population is estimated to be about 5%, and the disorder frequently persists into adulthood [1–4], with symptoms of inattention rather than impulsivity/hyperactivity as the main persistent symptoms [5, 6].

Neuropsychological studies have related changes in neural networks involving the frontal lobe to impairment on tests of attention, primarily those defined within the concept of executive function (EF) [7, 8]. The importance of EF is emphasized by the fact that impairment in childhood tends to increase into adulthood [9], is associated with severity of ADHD symptoms [10] and overall cognitive and everyday functioning [11]. However, it is well documented that not all individuals with ADHD show impairment on all core tests of EF [12, 13], and multiple pathway models have been developed to link different aspects of EF to neurobiology [14, 15]. Furthermore, not all impairments associated with ADHD can be explained within the concept of EF [16]. Alterations in more basic perceptual processing [17, 18], activation [19] and tempo of information processing [20, 21] are reported to influence everyday functioning of individuals with ADHD. A cognitive model of ADHD should therefore describe and operationalize different levels of information processing, their interactions and neurobiological substrates [15].

Posner and colleagues have presented one such model [22]. Their Attention Network Model defines an alerting or vigilance network, an orientation network and an executive or conflict network. The alerting network maintains a high sensitivity to incoming stimuli, the orienting network is involved in selection of information from sensory input, whereas the control network is involved in resolving conflicts between thoughts, feelings and responses [22]. These networks have been associated with different anatomical locations, neurotransmitter systems and genetic markers [23, 24], and the model has inspired studies of attention deficits in ADHD as well as in other neuropsychiatric disorders (see [25]).

The Attention Network Model has been used to develop test paradigms that have become popular during the last decades. Fan and collaborators developed the Attention Network Test (ANT) [23, 26], and presented updated versions on the Internet. The ANT has recently been used to study cognitive characteristics of individuals with ADHD. A deviant activation pattern in all three networks has been found in fMRI studies of children with ADHD, but impairment was only found on the control network when the test was administrated according to standard procedure outside the scanner [27, 28]. A recent Norwegian population-based study could not confirm impairment on any measure of the attention networks in a group of primary school children with ADHD, but a more detailed analysis showed that the children in general performed less accurate and more variable throughout the task than controls [29]. These characteristics were confirmed in an ANT study of college students (age 18-30 yrs) with a combined type of ADHD [30].

The impact of ADHD symptoms on cognitive function is well documented (e.g., [20, 31]). Recent studies have also shown that symptoms associated with affective disorders are crucial to understand characteristics of cognition in children [32] and adults [8] with ADHD. Murphy and Barkley illustrated a close association between what they referred to as emotional regulation and metacognition [8], and a longitudinal study emphasized the predictive value of such symptoms on future cognitive and everyday functioning [33]. Other studies have shown that symptoms associated with affective disorders are related to impairment of EF and motivation [34–36], and that children with ADHD and affective disorder (i.e., anxiety) are cognitively distinct from and more impaired than individuals in either condition on measures defined within the concept of EF [32, 37]. This may be explained by the heightened arousal characterizing individuals with affective disorders [38], and that this arousal contributes to cognitive impairment through its effect on EF [39]. Due to the high frequency of affective symptoms in adults with ADHD [40], these results motivate further studies using the ANT to investigate characteristics of alerting and control networks.

The aim of the present study was twofold. First, we investigated ANT results in a group of adults with ADHD and a control group from the general population. From earlier studies we expected the ADHD group to show impairment on the ANT measure of the control network, as well as on measures of accuracy, vigilance and variability. Secondly, we investigated ANT results in an ADHD subgroup reporting affective fluctuations. From earlier studies we expected that high arousal and EF impairment in this subgroup would influence their results on the alerting and control networks. Finally, we asked if the results would change when we controlled for ADHD symptoms and intellectual function.

The present study included a subset of adults participating in a Norwegian national study of adults with ADHD.

The present study showed that adults with ADHD did not differ from controls on ANT measures of the three attention networks, but they showed a lower accuracy, a higher intra-individual variability, and lower vigilance across the task. The effect sizes were mainly moderate, but only the accuracy measures retained statistical significance when we controlled for age and intellectual function. An important and novel aspect of the present study was the inclusion of a binary MDQ score to define an ADHD subgroup with affective fluctuations. In this subgroup we found an impact on the alerting and control networks, and on the measure of hit reaction time. The subgroup was significantly more distracted by conflicting stimuli and slower to respond. At the same time their results on the measure of the alerting network suggest that they were more alert. Furthermore, the total ASRS and IQ scores had a major impact on the reaction time and the control network in this ADHD subgroup, while the alerting network was left unaffected.

Given the original description of the test, it was somewhat surprising that adults with ADHD and controls did not differ on any measure of attention networks. According to findings in previous studies we expected to find impairment of the control network in the ADHD group [27, 28, 50]. Functions such as inhibition and cognitive flexibility are essential to solve cognitive conflicts as they are presented in the task. Dysfunction of these EFs are regarded as core cognitive deficits in the daily life of adults with ADHD [12, 51, 52], and are described as essential to understand their core symptoms [8]. In the present study, EF deficits as assessed by the conflict network were only found in the ADHD subgroup defined with affective fluctuations. The clinical importance of this finding is emphasized by the high frequency of affective symptoms among adults with ADHD [40], and that persistent affective symptoms tend to contribute to impaired function in occupational and academic areas of life [33]. Due to the lower intellectual function and higher ADHD symptom scores in the MDQ+ subgroup, our results may be explained by allocation of the individuals with the most severe ADHD to this subgroup. Although symptoms assessed by MDQ are frequently reported by individuals with ADHD [47], we argue that we used a definition of the term affective fluctuations that indicates an add on to the core ADHD symptoms: the participants should answer "yes" to the symptoms, but also confirm that the symptoms co-occurred and caused moderate to severe impairment in their daily life.

Arousal and alertness are known to be affected in ADHD [19], with a level that was expected to be different from the one shown by the control group. This was only suggested by the results in the ADHD subgroup with affective fluctuations. As anxiety symptoms are frequently found in individuals with ADHD and since anxiety symptoms are associated with high arousal, we speculate that individuals with such symptoms were mainly allocated to this ADHD subgroup. The adjustment to the new setting of ANT may have made them more prone to increased arousal. According to Eysenck, an individual with anxiety and ADHD will also be affected by a top-down influence of the EF dysfunction associated with ADHD [38]. We suggest that this may explain why both the alerting and conflict networks are affected in the ADHD subgroup defined with affective fluctuations. Such a complex interaction between the two networks has been illustrated in studies by Fan and collaborators and is supported by the fact that the alerting and conflict networks share brain networks [53, 54]. MacLeod and collaborators suggested another difference between the two networks that may contribute to explain the results: the executive network is more trait-like and the alerting network more state-like [25]. It is tempting to assume that both trait and state have influenced the results in the ADHD subgroup with affective fluctuations. When we controlled for the ASRS and IQ scores, probably reflecting traits associated with ADHD, only the state-like alerting network was influenced by the affective fluctuations assessed by the binary MDQ score. However, it is important to emphasize that further studies are necessary to obtain firm conclusions about alertness in adults with ADHD and symptoms of affective disorders. The interpretations of the results in the present study should be considered with caution, not at least because the results on the alerting network are combined with slow RT in the MDQ+ subgroup.

The present study showed impairment in the ADHD group on measures of accuracy, variability and vigilance, supporting findings from earlier studies of children [29, 55, 56] and adults [30, 35, 57]. Impaired accuracy and vigilance remained when intellectual function (IQ) was included as a covariate. Earlier studies have demonstrated a more general impact of IQ on ANT measures in children with ADHD (e.g., [29]). The retained group differences in the present study indicates that results on an IQ test are not as important for ANT results in adults as they are in children. However, intellectual function did influence the results when ADHD symptoms were combined with affective symptoms, illustrating the complex relation between core and comorbid symptoms of ADHD, EF and intellectual function.

The present study has limitations related to the diagnosis of ADHD and the definition of affective fluctuations. The adults with ADHD included in the study were evaluated by several clinicians before inclusion in the study, probably yielding a large heterogeneity within the ADHD group. On the other hand, our ADHD sample probably represents adults with an ADHD diagnosis as encountered in a clinical setting. Affective fluctuations were assessed according to a self-report questionnaire, originally designed to measure symptoms of bipolar spectrum disorders. However, it has been shown that most adults with ADHD with a positive MDQ score do not fulfill criteria for this disorder [40], and that a positive MDQ score may be found in a range of psychiatric diagnoses [47]. The conclusions of the present study will therefore not be restricted to a specific comorbidity group. It is also a limitation that the adults with ADHD were only asked not to take medication at the day of testing. Although we thus do not know if the wash-out of the medication was complete, our results are supported by the fact that methylphenidate use was more frequently reported in the MDQ- than the MDQ+ group. Finally, the study did not include biomarkers of the attention networks. Previous studies suggest that data derived from imaging techniques may be more sensitive than behavioral data collected in a standard test condition outside the scanner [27, 28, 50]. To extend the exploration of competition between the neural substrates of the alerting and conflict networks to neuroimaging of the here described subgroups would therefore be an important asset.

The present study suggests that adults with ADHD are less accurate, have a higher level of variability and a lower vigilance than adults without ADHD, and that affective fluctuations make adults with ADHD more alert, but slower and more distracted by conflicting stimuli. Our results indicate that the cognitive heterogeneity among adults with ADHD at least partly is explained by affective symptoms. By these results the present study emphasizes the importance of characterizing and taking these symptoms into account in research and clinical work with adults with ADHD.