Introduction

Despite the presence of an Autism Spectrum Disorder (ASD) being an exclusionary criterion for a diagnosis of Attention Deficit Hyperactivity Disorder (ADHD) under DSM-IV criteria (American Psychological Association 1994), clinical and research reports of ADHD-related symptoms in children with a diagnosis of ASD are becoming more common (for a review see Reiersen and Todd 2008). Although diagnostic criteria for these disorders are distinct, there are areas of overlap in their cognitive and behavioural features in clinical (Clark et al. 1999; de Bruin et al. 2007; Gadow et al. 2006; Geurts et al. 2004; Leyfer et al. 2006) and population studies (Ronald et al. 2008, 2010). In this report we explore relationships between cognitive processes and both ASD and ADHD-related behaviours in children with diagnoses of ASD.

Theory of Mind deficits have been a key focus of research in ASD (Baron-Cohen et al. 1985) and it is well-established that children with ASD have pronounced deficits in social cognition. Difficulties have also been observed in children with ADHD on Theory of Mind tasks (Buitelaar et al. 1999), pragmatic language (Bishop and Baird 2001; Oram et al. 1999), peer relationships (Hoza et al. 2005) and emotion recognition (Sinzig et al. 2008), though it may be that these difficulties have different origins to those seen in ASD.

Executive dysfunction has also been a focus in ASD and ADHD research. Increasingly, planning and attention shifting deficits are being recognised as key elements of executive dysfunction in ASD while findings of inhibitory control deficits have been less consistent (for a review see Hill 2004). Executive dysfunction, including working memory, inhibition, and planning difficulties, have been implicated in ADHD; however, as in ASD, while each of these has been seen to be deficient in at least a sub-sample of children with ADHD, none has been seen to be necessary or sufficient to explain the disorder (for a meta-analytic review see Willcutt et al. 2005).

Given these overlaps in cognitive dysfunction, one question that arises is how these relate to each other in cases where features of each disorder are present (Caron and Rutter 1991). Research has started to examine the influence of co-morbid behaviours on the cognitive profiles of children with diagnosed disorders. For example, Sinzig et al. (2008) report that facial affect recognition is more impaired in children with ASD and ADHD symptoms than those without co-morbid ADHD symptoms. Similarly, Sinzig et al. (2008) report that, in addition to planning and flexibility deficits associated with ASD, children with co-morbid ADHD symptoms also had inhibitory deficits that resembled those of the ADHD-only group. However, it is not clear from previous research how executive processes interact with social cognition, nor how ASD and ADHD-related behaviours are differentially influenced by cognitive processes.

The research reported here develops previous findings through the examination of social cognition and executive dysfunction in a large group of children with ASD, assessed for both symptoms of ASD and ADHD. First, it measures ADHD-related behavioural problems reported by parents. Second, it seeks to replicate findings that ADHD-related behaviours are associated with inhibitory control deficits, an area of mixed findings in the ASD literature. Finally, it seeks to identify whether social cognitive deficits, as represented by Theory of Mind (ToM) task performance are related only to ASD-related behaviours, or whether they are also related to symptoms of ADHD. Theory of Mind is of particular interest because of relationships between Theory of Mind and executive function both in typical and atypical development (Carlson and Moses 2001; Russell 1997). In summary, this study attempts to increase understanding of the effect of co-morbid ADHD symptoms on cognition and behaviour in ASD.

Method

Participants

Fifty-five children (mean age 10 years 2 months, range 7 years 7 months–13 years 6 months; 48 boys, 7 girls) with ASD took part. The children were recruited through the Educational Psychology Service and through local schools to take part in a larger research program involving 4 h of cognitive tests; those who managed to complete the battery were therefore relatively high-functioning, as their IQ scores indicate (VIQ mean 105, s.d. 19; PIQ mean 94, s.d. 14). All these children had previously received a diagnosis of autism (n = 9), Asperger syndrome (n = 30) or autism spectrum disorder/PDD-NOS (n = 16) from a qualified clinician prior to participation in the study.

Procedure

Parental Report of ASD and ADHD-Related Behaviours

Parental reports of ASD and ADHD-related behaviours were elicited using the 3Di (Skuse et al. 2004). The 3Di interview is similar to the Autism Diagnostic Interview (ADI-R; Lord et al. 1994) with which it correlates highly (Skuse et al. 2004); it emulates the ADI algorithms in its measures of social and communication impairments and repetitive behaviours. In addition the 3Di provides information about other developmental and psychiatric disorders including ADHD, in accordance with ICD-10 (World Health Organization 1992).

Theory of Mind Battery (ToM)

The children performed 12 tests of ToM previously established as discriminating between autistic and typically developing samples of children. It has been argued that the development of theory of mind in young typically developing children passes through a series of stages, which can be measured using scaled theory of mind tasks (Wellman and Liu 2004). Tasks were presented in a random order with accompanying props. Seven of the tasks were those used by Wellman and Liu (2004: diverse desires, diverse beliefs, knowledge access, contents false belief, explicit false belief, belief-emotion, real–apparent emotion) and the five additional tasks were expected to be more challenging for the older children involved here (Ice-cream van, Baron-Cohen 1989; Penny hiding, Baron-Cohen 1992; Sally-Ann, Baron-Cohen et al. 1985; Interpretational false belief, Luckett et al. 2002; Birthday puppy, Sullivan et al. 1994). These were scored out of a total of 25 with higher scores reflecting better performance.Footnote 1

Inhibitory Control (IC)

A children’s version of the Hayling Sentence Completion Test (Shallice et al. 2002) was used here as the measure of inhibitory control. This test establishes a prepotent response by making use of over-learnt knowledge of sentence endings. Sentences with the last word missing were read to the child. For example, ‘If you have cold hands you put on…’. In one condition the child must give an appropriate word to finish the sentence, e.g. ‘gloves’, whilst in the second condition the child must produce a word that is unrelated to the sentence, to the missing word or to a previous answer. This task was further adapted (Shallice, unpublished) to make it even more challenging by presenting the conditions in a block of mixed trials rather than presenting separate blocks of each condition. Participants are required to switch between conditions, arguably increasing the inhibitory load as they are required to inhibit responding according to the previously used rule. Whilst switching and inhibition are distinguishable, it has been argued that switching involves an element of inhibitory control (Miyake et al. 2000). The two conditions were alternated, with the experimenter indicating trial type with a hand signal. For the 10 ‘appropriate’ trials, the number of inappropriate responses given was recorded. For 10 ‘unrelated’ trials, three penalty points were awarded when the correct response was given or one point when the response was semantically related to the sentence or the missing word. Higher scores on this measure therefore represent less efficient inhibitory control processes.

Results

There were no correlations between age and either cognitive measures or parental reports of symptomatology (all p > .2).

Parental Report of ASD- and ADHD-Related Behaviours

While some children did not meet criteria on all three ASD domains of the 3Di at the time of this study, no child was excluded as variation in presentation was essential for the study design. ASD-related behavioural data were unavailable for one child; ADHD-related behavioural data were available for this child and were included in ADHD-related analyses only, as exclusion of this data did not influence the pattern of results. See Table 1 for descriptive data for both ASD and ADHD parental reports of behaviours.

Table 1 Descriptive data for parental reports of ASD and ADHD related behaviours, as measured by the 3Di and for Theory of Mind battery and Hayling task
Full size table

87.3% (48/55) of parents reported above threshold levels of behaviour on at least one ADHD component. Figure 1 shows the percentage of children that met criteria for inattention, impulsivity and hyperactivity.

Fig. 1
figure 1

ADHD-related behaviours: % of ASD sample meeting clinical criteria for each component

Full size image

Correlations between ASD- and ADHD-related behaviours revealed significant positive associations (see Table 2) between ASD-communication and all ADHD components, and between ADHD-impulsivity and all ASD components. A further significant correlation was found between ADHD-hyperactivity and ASD-social interaction.

Table 2 Correlations between ASD- and ADHD-related behaviours
Full size table

Relationship Between Cognitive Measures and Parental Report of Behaviours

ToM and IC were significantly correlated (r = −.53, p = .001). The range of scores for the ADHD-related behaviours was limited (see Table 1) and so, in order to examine the relationships between the cognitive measures of ToM and IC, participants were divided about the mean into groups of high and low presentation of each ASD- and ADHD-related behaviour, where high represents increased impairment as measured by a higher number of symptoms reported by parents. T-tests were chosen to analyse the difference in cognitive processes between these groups (Table 3); in this brief report, we have retained the .05 significance level despite multiple comparisons as relationships between parent report and cognitive measures are typically hard to detect (e.g Pellicano et al. 2006).

Table 3 Mean (s.d.) of cognitive measures of ToM and IC for groups of participants presenting as high or low on ASD- and ADHD-related behaviours
Full size table

A significant difference between the high and low ASD-social interaction impairment groups was found on both the IC (t(52) = 2.10, p < .05, d = .55) and ToM (t(52) = 2.63, p = .01, d = .68) measures, with children in the high ASD-social impairment group performing more poorly than those in the low ASD-social impairment group. A significant difference between the high and low ADHD-impulsivity groups was also found on the IC task (t(53) = 2.06, p < .05, d = .55), with children in the high ADHD-impulsivity group performing more poorly than those in the low ADHD-impulsivity group.

Given that both IC and ToM were related to each other and to ASD-social interaction impairment scores, these variables and both performance and verbal IQ were entered into a stepwise regression analysis as predictors of ASD-social. ToM was found to be the only significant predictor of ASD-social interaction (R2 = .08, F(1,52) = 4.30, p < .05); IC and IQ were unable to predict any further variance.

Discussion

The results clearly show high levels of parent reported ADHD-related behaviours in this sample of children with a diagnosis of ASD. 87% of children in this study crossed threshold for at least one component of ADHD. This is in accord with similarly high estimates from previous studies (e.g. 83% (Frazier et al. 2001); 75% ADHD or DAMP (Sturm et al. 2004) 65% (Holtmann et al. 2005); 60% (Goldstein and Schwebach 2004); 55% (Leyfer et al. 2006); 44.7% (de Bruin et al. 2007).

The data presented are not able to indicate whether the behaviours reported by parents are a reflection of true ADHD symptoms or epiphenomena associated with ASD. In order for an ADHD diagnosis to be given, behaviours must be present in more than one situation; this requires information from an additional informant such as a teacher, which was not available in the current data set. This may explain the high incidence reported here, which may not have been corroborated by a second informant. However, the data do hint that the presence of co-morbid symptoms of ADHD may exacerbate ASD-related behaviours. These findings are congruent with others reporting elevated social interaction symptoms (Holtmann et al. 2007) and greater difficulties in everyday functioning (see also Goldstein and Schwebach 2004) in the presence of co-morbid ADHD symptoms.

While the present results show strong associations between ASD- and ADHD-related behaviours, this may be due to this information being provided by the same informant for each child; associations within the behavioural data should therefore be treated with some caution. However, the relationships between these parental reports of clinically relevant behaviours and aspects of the children’s cognitive profiles is indicative of at least a degree of separation in the aetiology of ASD- and ADHD-related behaviours in these children, as the cognitive tasks provide an important independent source of information.

Performance on the ToM tasks was significantly related to the social interaction component of ASD, supporting previous findings that social cognitive processes may lead to the social difficulties that are characteristic of autism (Happe and Frith 1996). ToM performance was not related to any component of the ADHD-related behaviours, suggesting that although reports of impulsivity and hyperactivity were related to reports of social difficulties, these behaviours have different antecedent cognitive processes.

Inhibitory control was associated with increased impulsivity and difficulties in social interaction. However, inhibitory control was unable to account for any further variance in social interaction independent of theory of mind ability. This indicates that inhibitory control, as measured here, may relate to impulsivity and social interaction abilities in quite different ways, in the latter case, through its relationship with social cognition. One possibility is that the inhibitory control task involved a social component; either through its requirement that the child observe and respond to a social cue (a hand signal) given by the experimenter, through the open-endedness of the task, or through its relatively high load on verbal competency (White et al. 2009). However, the processes tapped by the Hayling task have been seen to be impaired in patients with frontal damage (Burgess and Shallice 1996, 1997) and children with ADHD (Shallice et al. 2002); our finding that performance on this task is related to impulsivity is consistent with this. This suggests an alternative explanation: inhibitory control deficits have a closer association with the ADHD profile than with ASD itself, but that they exacerbate symptoms of ASD through their relationship with social cognition. This is congruent both with the current finding that the effect of inhibitory control on ASD related behaviours was mediated by social cognition and by previous conflicting reports over the status of inhibitory control processing in ASD (Hill 2004).

Somewhat surprisingly, inhibitory control was not related to repetitive behaviours even though repetitive behaviours and impulsivity were correlated in the parental reports. This contrasts with previous research which has reported relationships between inhibitory control and repetitive behaviour (Lopez et al. 2005; although see Bishop and Norbury 2005 for an example of a negative finding). The current data suggest that the impulsivity, reported in these children to be related to inhibitory control, may differ in developmental course to the repetitive behaviours; that is, the cognitive processes that precede these behaviours may differ.

The research presented here replicates previous reports that co-morbid symptoms of ADHD are associated with an increase in ASD-related behaviours. Further, it suggests that inhibitory control difficulties may exacerbate ASD-related behaviours through its impact on social cognition. In contrast the relationship between inhibitory control and ADHD related behaviours appears to be more direct. It is not possible to identify fully the direction of causality of relationships between these cognitive processes and with behaviour in cross-sectional data but these results provide an early indication that there may be some separation between the social and executive cognitive features relating to ASD and ADHD. Future research should seek to identify whether this represents true co-morbidity, through the concurrent cognitive and behavioural assessment of children with ADHD as well as those with ASD, and should seek to track these processes through development in order to fully ascertain how they interact.