Introduction

The recently published DSM-5 [1] has imposed major changes on the classification of child and adolescent psychiatric disorders, including the classification of attention deficit/hyperactivity disorder (ADHD) and autism spectrum disorders (ASD) as “neurodevelopmental disorders”, alongside other specific disorders.

This new classificatory convention was based on converging clinical [2, 3], familial [4], and genetic [5, 6] evidence that has suggested not only phenomenological, but also aetiological, overlap between the two disorders [7]. Indeed, while ADHD is one of the most frequent psychiatric conditions in childhood and adolescence (5.29 % worldwide prevalence of ADHD according to [8] and a comparatively infrequent 1–2.65 % prevalence of ASD [9–11]), there is substantial comorbidity of ASD and ADHD symptoms, varying between 30 and 80 % of cases in clinical assessments [7]. The inconsistent rates of comorbidity are probably due to different thresholds in establishing diagnostic criteria for an additional ADHD diagnosis, since symptoms of (social) inattention, fidgety behaviour, and irritability can also be intrinsic to the autistic phenotype. However, studies using factor analysis or modelling the classification of subjects according to typical questions used to detect symptomatology have previously revealed a good differentiation between the two clusters of symptoms [12, 13].

Although the clinical, familial, and genetic evidence is strongly suggestive of an aetiological overlap, examination of specific functions that may be aetiologically relevant reveals a more complicated pattern. Firstly, increased intra-subject variability (ISV) of performance, measured typically as an increase in the intra-subject standard deviation of reaction times seems to be a core abnormality of ADHD [14, 15] and a promising endophenotype of the disorder [16]. However, there is early evidence that increased ISV is not associated with ASD unless patients also suffer from ADHD [17]. Secondly, although executive dysfunctions appear to be present in both ASD and in ADHD [7], the validity of specific and differential profiles of deficits for each disorder is still a matter of debate [18–20]. Thirdly, deficits in Theory of Mind (e.g. emotion recognition, understanding of others’ minds) and weak central coherence are discussed as being specific to ASD, but have rarely been studied in ADHD [7]. Moreover, methodological problems due to the complexity of the ToM concept, a lack of a consensual definition of its basic components [21], and the potential influence of executive abilities on the development of ToM in young individuals with ASD [22–24] raise doubt about its validity as a specific marker of ASD.

Among the putative mechanisms for the development of social cognition, imitation abilities in particular have been intensively studied in ASD and found to be impaired [25] or at least delayed in their development [26]. The nature of the imitation deficits and their contribution to early social skills, however, is still controversial, mostly due to the wide range of methods and conceptual approaches that have been employed to study imitation abilities in autistic persons, hampering direct comparison between studies [27, 28]. Whatever the reason for imitation problems in ASD might be, the ability of “copying behaviour” implies a genuine motor connotation, that is reproducing any type of “action”, and has been mainly explored in tasks exploring imitation of action on objects, gestural skills or facial expressions [25, 27]. This implicit link between imitation and motor/gestural abilities is remarkable from several points of view. Firstly, the mirror neuron system (MNS) theory postulates a motor substrate for the development of social behaviour including imitative behaviour as well as understanding of others’ minds, which could be impaired in ASD [29]. While the role of mirror neurons in imitation is controversial, there is substantial empirical evidence that parieto-frontal mirror neurons circuits offer a plausible, though not unique, neural substrate for action and intention understanding, which is the basis of imitative behaviour [30, 31]. Secondly, Mostofsky et al. [32] claimed that deficits in imitation are secondary to a general dyspraxia in ASD that is, in turn, due to abnormal sensory–motor integration. Finally, Gowen [33] underlined the importance of action kinematics in correctly imitating goal-less/meaningless actions and postulated a deficit in ASD in perceiving and reproducing kinematics in observed actions. Cook et al. [34] found that individuals with autism even show atypical kinematic profiles when simply asked to move their arm back and forth.

However, while a variety of motor problems (e.g. gross-motor delay, hypotonia, reduced ankle mobility with odd postures and ambulation [“toe-walking”], stereotyped movements, mannerisms, motor apraxia and clumsiness) are associated with ASD [35], not all appear to be specific to this disorder. In fact, difficulties in motor planning and control or problems with repetitive, rapid limb movements, for instance, are also commonly found in ADHD [36–38]. Atypical motor learning patterns and deficits in perceptual-motor integration, by contrast, appear to be specific to ASD, as they have not been found in patients with ADHD in two recent studies [37, 39].

Given the nosological implications of phenomenological, and potentially aetiological, overlap between disorders that have been grouped together as “neurodevelopmental” in DSM-5, and recognising the importance of both motor and imitation abilities for the understanding of ASD, the aims of the present study were as follows. Firstly, and foremost, we wanted to determine the presence and extent of both motor and imitation deficits in patients with ADHD in direct comparison to autistic patients and typically developing (TD) children and adolescents. Secondly, we aimed to determine the potential impact of comorbid ADHD on motor and imitation abilities of ASD patients, which has so far only rarely been studied [40, 41]. Finally, we aimed to compare the interrelationships between motor and imitation abilities as well as potential differences herein in patients with ASD or ADHD.

To this end, motor performance was systematically tested by means of a comprehensive and well-established battery, the Zurich Neuromotor Assessment (ZNA; [42]), in three age-matched and IQ-parallelised groups of TD, ASD, and ADHD children. Imitation abilities, furthermore, were assessed with test batteries of face [43] and ideomotor apraxia [25, 44]. Both test batteries had been employed already in previous studies with ASD subjects [26, 45, 46].

Method

Participants

All 66 participants were aged 6–13 years, male, had IQ > 80, and had neither epilepsy nor any other neurological diseases. Forty-six children were recruited from our in- and out-patient populations with a confirmed diagnosis of ASD or ADHD (see Table 1). All 22 patients with ASD met the criteria for an “autism” or “autism spectrum” diagnosis according to the Autism Diagnostic Observation Schedule diagnostic algorithm (ADOS-G) [47] and were above the cut-off in at least two of the three autism domains of the Autism Diagnostic Interview-Revised (ADI-R) [48]. For both patient groups the diagnostic interview K-SADS-PL [49] was applied to confirm ADHD diagnosis and to screen for ADHD comorbidity within the ASD group. Accordingly, 11 ASD children also fulfilled DSM-IV criteria for ADHD (ASD+: 9.3 ± 2.5 years, IQ = 106.45 ± 16.81 SD) while 11 ASD patients had no ADHD comorbidity (ASD−: 10.8 ± 2.1 years, IQ = 107.01 ± 13.33 SD). Children taking stimulant medication (in almost all cases long-acting methylphenidate) were free from medication for more than 24 h prior to neuromotor and imitation testing, which has been shown to be sufficient time for observing on/off MPH effects on motor functions [50]. N = 20 participants with typical development were recruited from local schools. They had no history of developmental, psychiatric or neurological disorders. The children samples with ASD and typical development have already been analysed and published as part of a larger subjects sample in a previous paper [26].

Table 1 Sample characteristics
Full size table

For cognitive assessment of all participants, WISC-III or -IV were routinely used [51]. If IQ had last been assessed more than 1.5 years prior to a subject’s testing, an updated IQ-score was obtained using the shorter Raven’s Standard Progressive Matrices [52]. Handedness was assessed with the Edinburgh Handedness Inventory [53]. In the total subject sample, two of the 22 ASD, none of the children with ADHD, and two of the 20 TD children, respectively, were left-handed.

Assessment of motor performance

The ZNA is a neuromotor assessment battery in German that provides highly standardised separate measures of timed motor performance and quality of movement, two components of motor competence that show differential developmental courses and substantial inter-individual variation [54, 55]. Furthermore, the ZNA differentiates between so-called “pure” motor tasks (basic simple repetitive, alternating foot and hand movements as well as repetitive and sequential finger movements, and diadochokinesis) and adaptive motor performance (the pegboard task testing coordination of fine-motor skills and tasks of gross-motor coordination and balance such as the dynamic and the static balance task, which are complex sensorimotor tasks requiring an adjustment of motor functions to task demand) [42].

Motor performance was videotaped throughout the assessment. After verbal instructions, along with a brief demonstration of the task through the examiner, children were allowed to practice movements, in order to ensure that they had understood the task, and were instructed to perform the movement as fast as possible. A stopwatch measured the time required for the sequence of movements in each task to be completed according to the manual instructions (timed performance). Quality of movements was provided by counting the degree and frequency of unwanted or superfluous associated movements occurring during all tasks (grouped together as associated movements) and degree of arm deviation during pronation and supination of the hand in the diadochokinesis.

ZNA additionally defines “block components” which aggregate statistically highly redundant variables (e.g. movements of the left and the right limb are pooled together). Block components have been defined to improve intra- and inter-rater reliability as well as test–retest reliability (in excess of 0.9 and 0.7 for block components of timed performance or associated movements, respectively) and consequently do not need further reduction [26, 56].

The following six ZNA block components were used as dependent variables for statistical analyses: four measures of timed motor performance, including (1) pure motor performance (foot, hand, and finger movements); (2) adaptive performance in the pegboard; (3) adaptive performance of dynamic balance and (4) static balance; and the two measures of quality of movement, including (5) diadochokinesis and (6) associated movements (see Table 2).

Table 2 Standardised z scores for the block components of the Zurich Neuromotor Assessment test
Full size table

For all measures, the ZNA provides normative data as z scores for different age groups and gender.

Assessment of imitation abilities

Imitation abilities were assessed in two test batteries explicitly asking participants to imitate a series of oral–facial movements in the (a) German version of the Face Apraxia Tasks (FAT) [43] and socially meaningless hand, finger, and combined finger/hand gestures that were novel, and not object-directed in the (b) Test Battery of Ideomotor Apraxia (TBIA) [44]. During each imitation task the examiner stood in front of the subject and demonstrated the facial expression, the finger and/or hand postures that was to be imitated.

Ad (a) The FAT includes items representing actions of the upper or lower face that are in general familiar but not embedded in a social context in the test situation. Additional verbal instruction is allowed only after direct imitation of the facial expression has been unsuccessful. The two subscales consist of 9 for the upper and 29 items for the lower facial movements. Performance in each item is assessed on a 4-point rating scale (imitation after one attempt (4) or after two attempts (3) and movement execution on first verbal request (2), on second verbal request (1), or no movement execution at all (0)) and thus yield maximum scores of 36 or 116, respectively.

Ad (b) In the TBIA, subjects have to imitate either different configuration of the fingers (finger gestures) or different positions of the hand in relation to the head, with or without additional configuration of the fingers (combined hand–finger gestures and hand gestures alone). All gestures are non-familiar and unrelated to social or tool-use contexts. Equal numbers of trials are performed with the left and right hand. Each gesture imitation is assessed on a 3-point rating scale (no imitation (0), correct imitation with two attempts (1) or correct imitation with one attempt (2)). Each task group includes 14 items and thus yields a maximum score of 28 points. Further details of all test batteries used are given in Biscaldi et al. [26].

Statistical analysis

Statistical analysis comprised two steps. Firstly, for all dependent measures of neuromotor assessment and imitation abilities, a three-way multivariate analysis of variance (MANOVA) was carried out with the independent factor DIAGNOSIS (TD versus ASD versus ADHD). In order to control for possible associations of all motor and imitation variables with general mental abilities, an additional multivariate analysis of covariance (MANCOVA) was computed with IQ as covariate. Post hoc Gabriel’s tests for unequal sample sizes were done to test for significant between-group differences. After splitting the ASD group in ASD+ and ASD− patients, separate ANOVAs compared performance for all motor and imitation variables in these two subgroups. Secondly, since in a previous paper imitation variables were highly correlated within task groups [26], we explored the dimensionality of the imitation variables and possible group differences herein by carrying out Pearson’s correlations (r) and Principal Component Analyses (PCA) for each clinical group separately. Further, after controlling for age effects by normalisation (normative data for motor variables) or residualisation (imitation variables) (see [26]), we determined the interdependencies between the resulting imitation factor (PCA) on the one hand and motor variables on the other, once again using Pearson’s correlations (r) separately for the three groups. For all statistical analyses, a significance level of α = 0.05 was adopted.

Results

Descriptive data for our three groups of participants are shown in Table 1. There were no significant group differences in age (F 2,63 = 0.84, p = 0.43), but we found a significantly higher IQ in the TD as compared with the ADHD group (F 2,63 = 7.62, p = 0.001). In spite of this bias, controlling for IQ made little difference to the results reported below (compare the “MANOVA” and “MANCOVA” columns in Tables 2, 3). To deal with the ambiguity that still surrounds the question of controlling for IQ in studying neurodevelopmental disorders (for a methodological discussion, see Dennis and colleagues [57]), both groups of results, with and without IQ as covariate, are illustrated in the tables.

Table 3 Scores for the two of imitation tasks of facial movements and the three gestural imitation tasks of non-meaningful gestures with MANOVA and MANCOVA results
Full size table

Neuromotor performance (ZNA)

The MANOVA and MANCOVA results for the neuromotor group comparisons are documented in Table 2. The columns “MANOVA” and “Gabriel” together with the MANCOVA results of this table reveal that, with two exceptions, both patients with ASD and ADHD show similarly impaired motor performance when compared to TD children. The first exception is a lack of any group difference for the variable “Pegboard”; the second exception refers to the balance variables. While patients with ASD showed poorer “dynamic balance” than patients with ADHD and controls, patients with ADHD performed inferior than the other two groups in “static balance”. It should be noted that while the finding of inferior “static balance” in the ADHD group was significant in the MANOVA (F 2,63 = 4.07, p = 0.022), it fell to the trend level after controlling for IQ (F 2,62 = 3.02, p = 0.056).

The separate ANOVA with the two ASD subgroups (ASD− and ASD+) revealed no influence of additional ADHD comorbidity on most of the variables (Fs < 2.71). Only in “static balance”, was there a trend towards poorer performance of ASD+ patients compared to ASD− patients (p = 0.059), consistent with the findings above, linking ADHD symptoms and static balance.

Imitation

Table 3 summarises the MANOVA and MANCOVA results for the imitation variables. The variable “imitation of upper facial movements” (very common everyday facial expressions such as “close your eyes” or “close one eye”, “wrinkle your nose”, “frown”, “blink”) was associated with a ceiling effect in all groups, probably because they were very simple and hence easy to be reproduced by the high-functioning autistic children of this study. All other imitation variables yielded robust overall group differences that were exclusively due to poorer imitation abilities in patients with ASD. Conversely, in none of the imitation variables were ADHD patients impaired. The within-group standard deviations and ranges, however, show that the ASD group as a whole performed more heterogeneously than the other two groups across all aspects of imitation behaviour. As mentioned before, controlling for IQ only slightly reduced our test statistics.

In the separate ANOVA, ASD+ patients underperformed ASD− patients only in the imitation of lower facial movements (F 1,20 = 6.43, p = 0.02).

Relationships between ZNA and imitation variables, and intelligence

While correlations between intelligence and imitation abilities were mostly negligible (0 ≤ rs ≤ 0.25), correlations between intelligence and motor abilities covered a range from small to large (0.13 ≤ r ≤ 0.40), according to conventional effect size classifications [58]. Additionally, IQ was a significant predictor of the block-component pure motor performance (F 1,62 = 5.91, p = 0.018). Consequently, including IQ as a covariate had some effect on the group comparisons in motor variables, but no effect on group differences in the imitation variables (Tables 2, 3, MANCOVA column).

Relationships between ZNA and imitation variables

Since imitation variables generally exhibited high inter-correlations (0.539 ≤ rs ≤ 0.784), a PCA run for all three groups yielded a single imitation factor (called “IMI”), explaining 72 % of the variance in imitation abilities, with factor loadings between 0.802 and 0.903. Within groups, this factor was readily replicable in the TD (explained variance: 69 %, bivariate correlations: .412 ≤ rs ≤ 0.746) and ASD groups (66.3 %; 0.464 ≤ rs ≤ 0.800) (see also [26]). The ADHD group, by contrast, showed correlations that varied between r = −0.326 and r = 0.414 and required 2 factors for their reconstruction (factor 1 loaded finger, combined gestures and facial movements with factor loadings: 0.751, 0.856 and 0.553, respectively; factor 2 loaded hand gestures and again facial movements: 0.853 and −0.706, respectively).

Correlations between the age-residualised IMI factor and the age-normalised ZNA variables, for all three groups separately, revealed large associations only in the ASD group for the block components pure motor performance, peg board, and static balance (0.470 ≤ r ≤ 0.614, ps < 0.05). Correlations corresponded to medium effect sizes in the TD group (r = 0.351, p = 0.13 for the block component associated movements) and negligible-to-medium ones in ADHD children (0.017 ≤ rs ≤ 0.310).

Discussion

In accordance with the recent debate about potential aetiological overlap between ASD and ADHD [7], the present study set out to investigate impairments in motor and imitation abilities in patients with ASD or ADHD that hitherto have been reported primarily for patients with ASD [26, 45, 46]. Our results reveal (1) basic motor dysfunctions in both groups as well as no (significant) differences between the ASD+ and ASD− subgroups; (2) imitation deficits in ASD patients only; and (3) differential patterns of correlations between motor and imitation variables for the ASD (and control) and ADHD groups. These results suggest, that motor dysfunctions may reflect aetiological commonalities between ASD and ADHD, and possibly neurodevelopmental disorders in general. Imitation deficits, by contrast, appear to be specific to ASD and may thus not be part of an overlapping neurodevelopmental aetiology.

Ad (1) The Zurich Neuromotor Assessment revealed on the one hand a similar pattern of deficits in both patient groups, including slower basic movements of fingers/hand, decreased quality of movement (dysdiadochokinesia) with increased involuntary associated movements, but no deficits in fine-motor skills as assessed with the Pegboard test. On the other hand, balance difficulties were once again found in both groups, but were specific for different sets of tasks.

Firstly, we confirm the existence of a similar basic motor deficit both in ASD and ADHD [37, 59] which involves both speed and quality of movements and can be attributed at least partially to problems in the cortico-cerebellar motor networks also involving the basal ganglia and the brainstem [60–62]. Abnormalities in the cortico-brainstem circuitry can be probed with saccade tasks that have recently provided evidence of impaired functioning in ASD patients [63]. The commonality between ADHD and ASD with regard to a general basic motor impairment, reflecting a genuine motor component less dependent on sensory experience and motor learning, is consistent with the putative aetiological links between the two disorders [7], and confirms previous notions that motor deficits could be associated with neurodevelopmental disorders in general [60]. This holds all the more as there were no comorbid autistic disorders in our ADHD sample, as ascertained by our exclusion criteria. Therefore, our results contradict the claim that impaired motor performance in patients with ADHD can be specifically related to autistic traits that may be present but are usually not assessed in the ADHD diagnostic process [41]. Having said that, social responsiveness as a broader ASD-related trait was not directly quantified in our sample. Therefore, we cannot rule out the possibility of a more subtle “sub-clinical” relationship between motor performance and social impairment [40].

Secondly, in spite of these similarities, the findings point to some differences that are worthy of discussion. Whereas the deficit in dynamic balance was more pronounced in those with ASD, static balance problems were found in both groups, but more so in patients with ADHD and in ASD+, compared to ASD− patients, suggesting that impaired static balance is principally related to ADHD symptoms, independent of their co-occurrence with ASD. Although in the present study it narrowly failed to reach significance, the trend is in line with findings previously reported in the literature [64, 65]. Furthermore, while this finding is somewhat sensitive to IQ differences between the samples, the potential link between static balance and ADHD symptoms is also supported by findings showing that postural stability and motor problems in general, like ADHD symptoms, improve with methylphenidate treatment [50, 64, 66]. Yet, further research is necessary to confirm these findings and highlight the mechanisms of this improvement. However, if this finding of balance difficulties in children with ADHD can be replicated, it could be primarily due to their attentional/hyperactivity problems, and this could also be the case when these problems are comorbid with autism. By contrast, the clear impairment in dynamic balance in ASD, independent of comorbidity with ADHD, could point to the aforementioned cortico-cerebellar circuitries via basal ganglia and brainstem, which are responsible for sensorimotor control [60]. Furthermore, this deficit shows a clear persistence in older individuals with ASD [26].

Indeed, the few studies that have recently investigated motor performance in ASD and ADHD have reported that, while such deficits can be found in both groups, they seem to be more severe and persistent in those with ASD [59, 67–69]. This finding indicates that, although motor problems may be present in both ASD and ADHD and thus constitute, on the face of it, a commonality between the two neurodevelopmental disorders, their causes and correlates may be different. Whereas motor problems of ADHD seem to be at least partly associated with their hyperactive–impulsive symptoms [66], motor deficits in ASD are often explained in the light of specific kinematics and sensorimotor problems [34, 37, 70], as also stated by Trevarthen and Delafield-Butt [71].

Moreover, although the sensorimotor circuitry involving the cerebellum may also be related to the balance problems described in ADHD [72], the deficits in dynamic balance in ASD were more severe than those in ADHD. This could point to difficulties in the autism group with increasing task complexity, challenging integration of sensory input with motor output [26, 60]. Increasing difficulties in motor coordination in ASD due to increasing motor complexity have previously been shown for the pegboard test [73]. In our study, however, patients showed no impairment in the pegboard task, relative to the control sample.

Ad (2) In contrast to the commonalities of motor problems deficits, Dewey and colleagues [59] reported a generalised deficit of instructed or imitated gestures in patients with ASD only. In this study, only ASD patients without comorbid ADHD were included. The results of the present study complement Dewey and colleagues’ findings, showing for the first time that ADHD comorbidity is unrelated to the imitation problems of children with ASD. We also show that ASD-specific deficits involve both familiar (facial movements) and completely novel actions and postures that require subjects to copy the form of a demonstrated action without a goal [25]. The results of Dewey and colleagues’ and the present study, furthermore, together suggest that imitation deficits in ASD are independent of intelligence across a wide range of general mental abilities, ranging from intellectual disability (IQ < 70) and low IQ (<80) in Dewey’s sample to average intelligence in the high-functioning autistic patients of the present study. Hence, imitation impairments are specific to ASD and present across the intellectual spectrum (see also [25] and [74]. Moreover, Vivanti et al. [75] tested spontaneous imitative behaviour in pre-schoolers with ASD, in comparison to TD and children with global developmental delay using a more naturalistic setting. They found specific imitation difficulties in ASD, such as reduced spontaneous imitation, reduced attention to faces, and a tendency ‘to emulate’, that is to pursue the goal of an action without reproducing its form and meant meaning. These results extend the findings of imitation difficulties in ASD to settings challenging in the first place attentional mechanisms focusing on others’ actions and motivational mechanisms to imitate them [71]. Intriguingly, individuals with autism are also impaired in recognising [76] and imitating [77] the “style” or “vitality form” (i.e. the dynamic components of movement, form, timing, intention) of an action, that is “how” an action is performed. Vivanti et al. [75], Rochat et al. [76], and Hobson and Hobson [77] are excellent examples of experiments in laboratory settings that nevertheless try to provide some of the naturalistic context in which spontaneous imitation and recognition or, respectively, imitation of emotional components of motor acts normally occurs. Nevertheless, even these studies are somewhat limited in the ecological validity and thus generalisability to proper naturalistic contexts with their high motivational and emotional saliency.Footnote 1 A potential solution to this problem would be designing field studies addressing the spontaneous behaviours of children with ASD in different environments, and associated emotional states. The obvious gain in ecological validity of such a study, however, would be achieved at the expense of its internal validity and thus clarity.

Ad (3) The reported correlation analyses indicate strong associations between the three block components of pure motor performance, pegboard, and static balance, and the imitation factor only in the ASD group, whereas in the TD group one correlation with medium effect size could be found for the block component of associated movements. By contrast, and rather surprisingly, in patients with ADHD these correlations were patchy and not reducible to a single factor.

Since sample sizes were rather small, one should be cautious in interpreting the specific pattern of related parameters. Bearing in mind this limitation, the results are interesting and deserve some discussion. Biscaldi and colleagues [26] already argued that in children with ASD motor coordination skills could be related more specifically to imitation abilities whilst typically developing children show a less specific association between delayed motor functions (measured by the number of needless associated movements) and imitation. With regard to this specific association in autism, the reported deficits in (sensory-) motor coordination and imitation abilities could have been linked through action kinematics.

Action kinematics are impaired in ASD [34, 78] and considered fundamental for the development of social responses [33]. They provide qualitative information of the movement, e.g. about its velocity, acceleration, trajectory and displacement, which is essential to understand the purpose of a motor action and therefore contribute to learning and communication skills [33]. A disruption of this function in ASD is corroborated by the aforementioned study by Rochat et al. [76], who found a deficit in recognising the vitality form or style of an action (also called the “how” a person is doing something) in this group.

Action kinematics thus provide a further link between neural motor “substrates” and the more “cognitive” nature of imitative behaviour beyond the theory of a disruption in the neural network sub-serving the organisation and understanding of motor chains [79]. Considering that children with ASD are also impaired in the imitation of meaningful gestures such as skilled motor acts pretending to use an object [59], this link could be seen in one of the postulated parieto-frontal circuits involving the ventral intraparietal area (VIP) that are supposed to be responsible for mirror mechanisms of body-directed motor acts [30]. This putative relationship seems to be developmentally transient though, as the motor deficits may persist at least into young adulthood, whereas imitation deficits disappear and the relationship between motor variables and imitation weakens during adolescence [26]. Interestingly, Rochat et al. [76] found that the deficit in recognising the vitality form of an action in children with ASD remains stable in adolescents. Furthermore, they stated that, in line with the development of ToM functions, deficits in cognitive-driven processes such as action understanding but not the impaired basic sensorimotor mechanisms necessary for recognising and reproducing the form (including the vitality form) of an action may improve with age.

The strong link between motor and imitation abilities that we found in the ASD patients of our study was greatly reduced in those with ADHD, apparently pointing to another difference between the two disorders. Patients with ADHD, however, show excessive short-term fluctuation of performance that has been called “intra-subject variability” (ISV; [15, 16, 80]). While the stability of individual differences in ISV shows that this construct reflects a trait [81], the variance in performance created by ISV overlays inter-individual differences in (motor or imitation) performance in idiosyncratic ways, thus diminishing correlations between tests (or reliabilities of tests). In line with this reasoning, greater differentiation (that is, more factors) has been reported for ADHD patients compared to controls in a previous study employing a much more homogeneous battery of (cognitive) tests than used here [15].

Limitations of the study

Although subjects were accurately selected, diagnosed and well matched for age, we could not avoid some differences in intelligence level due to bias effects in recruiting subjects with typical development. Additionally, testing of imitation abilities was confined to gestures removed from social or meaningful context. In future research it would be useful to directly compare both kinds of imitation with and without objects to explore the relationship between basic motor functions, the ability of reproducing the motor “form” but also the goal of an action. Furthermore, the exploration of similarities and differences between ADHD and ASD in comparison with comorbid forms would benefit from extending the present results to tasks also eliciting spontaneous imitation and imitation of emotional states.

Conclusion

In conclusion, pending replication in larger samples and more complex study designs, the findings of the present study suggest partial overlap in deficits that have so far mainly been postulated for ASD, in patients with ASD and ADHD. While motor dysfunctions appear common to both disorders and have been shown to be developmentally stable in ASD, imitation deficits appear specific to ASD and decrease during the course of childhood and adolescence [26, 69]. However, longitudinal studies will be necessary to confirm the present findings.

Moreover, ADHD seems to be characterised by increased variability and impulsivity (which may, in turn, reduce motor control). ASD, by contrast, may be better characterised by a genuine impairment of complex sensorimotor integration, leading to stable motor deficits and impaired recognition of kinematics and forms of action and hence reduced ability in reproducing (correctly imitating) gestures and facial expressions. The present study has shown how valuable the joint exploration of basic motor functions and more complex imitation abilities can be when addressing the potential aetiological overlap between these two disorders

It is this “mixture” of disorder-specific and -overlapping features that makes the concept of neurodevelopmental disorders so challenging [82].