In the experiment reported in this paper, we found a consistent pattern of results that supports the view that girls with chromosome 22q11.2 deletion syndrome, fragile X syndrome, or 45,X Turner syndrome suffer from both common and diagnosis-specific impairments in visuospatial attention. Despite significant differences in global intellectual functioning between the three groups, girls with 22q11.2DS, FXS, or TS had very similar performance profiles of unimpaired and impaired functioning for the alerting and executive networks. Comparison of the orienting index revealed that only girls with FXS struggle to orient their attention appropriately when presented with valid spatial cues.
Our first aim was to determine whether the observed attentional impairments arose from general intellectual impairment, as designated by IQ, or from specific neurological impairments. We tested the first hypothesis by examining whether between-group differences in behavior and IQ were observed in the same domains. For TD adults, a non-correlation, a significant positive correlation and a significant negative correlation have previously been found between FSIQ and the alerting, orienting and executive indices, respectively [
64]. While we were able to replicate the significant positive relation between the orienting index and IQ in the sample as a whole, given the large range of FSIQ in each group, we do not interpret this to mean that based on the FSIQ and orienting index interaction, an individual’s diagnosis group could be predicted.
Intellectual disability is commonly reported in children with 22q11.2DS or FXS, but not in girls with TS. Therefore, if attentional impairment is due to a global intellectual impairment, as represented by reduced IQ, then girls with TS should perform like TD girls and girls with 22q11.2DS or FXS should perform like each other and less well than TD girls. Any other pattern would imply that attention impairment is not linked to global intellectual function. This alternative approach has been studied previously using a different task for children with 22q11.2DS or TS and the results suggested that despite a significant difference in FSIQ between children with 22q11.2DS and girls with TS or TD children, the NDD groups performed more similarly to each other in numerical cognition than to TD children [
15]. This supported their interpretation that specific cognitive processing impairments are manifest in these NDDs, and that these differences were not due to differences in IQ. We predicted that our results would replicate this pattern. Here we found that each of the three NDD groups had some form of impaired performance on the ANT. For example, girls with 22q11.2DS, FXS, or TS had significant impairment in the executive control of attention, as had been previously reported for boys and girls with 22q11.2DS [
25,
57]. This difference in performance is not attributable to the time needed to generate and implement a motor response or to a difference in error rate. Therefore, our results are demonstrably not consistent with the predictions made by a global intelligence hypothesis since significant differences were found between the FSIQ of the groups studied here, but these differences were not replicated in the behavioral differences of an attention task. This suggests that selective impairments in attention reflect either a common core impairment due to the NDD or diagnosis-specific impairments to attentional subsystems.
If girls with a NDD performed differently from each other, and differently from TD girls, this would suggest that there exist diagnosis-specific impairments. For example, the propensity for perseverative behaviors in boys with FXS [
12,
65] and to a lesser extent in children with 22q11.2DS [
66] may contribute to the higher probability that girls with FXS or 22q11.2DS will make incorrect responses. Overall, we predicted both possibilities to be true – in some measures of attention, some NDDs will perform like each other and unlike controls, but that in other measures of attention, different NDDs will exhibit distinct attentional profiles.
The only attentional index in which we found no group differences, no matter which measure we used, was the alerting index. Existing models of attention refer to the ability to direct and maintain focus on an item or location as alerting or sustained attention, a component of which is vigilance [
67]. While the definitions may overlap, these concepts are not synonymous with each other. There is a paucity of research comparing cognitive function of attention across NDDs and specifically in girls with 22q11.2DS, FXS, or TS. Evidence suggests that children with 22q11.2DS have impaired sustained attention [
66], but unimpaired vigilance [
68] and alerting [
25,
57,
69] compared to TD children. A study of boys with FXS using a vigilance task did not find group differences in comparison to boys matched by mental age [
9]; however, when the comparison group was divided into those with poor or unimpaired attention, the boys with FXS were slower and less accurate than the boys with better attention [
10]. Using a different attention task, adult males with FXS had impaired vigilance [
11]. Girls with TS, when compared to children with a learning disability, did not have an impairment in sustained attention [
57]; however, when compared to age-matched TD girls, impairments in sustained attention were seen for children [
70,
71] and adult women with TS [
72]. By our approximation, the alerting index is most comparable to other vigilance tasks. Our results are consistent with previous results for children with 22q11.2DS or FXS. To the best of our knowledge, the present study is the first to test girls with TS using a vigilance or alerting task. Our results thus predict that girls with TS will perform comparably to TD girls on a vigilance task.
All three NDD groups had impairments in executive attention when both speed and accuracy were taken into account. Unlike many other tests of executive function that require the selection of an appropriate rule, the ANT primarily requires the selection of the appropriate input to determine the correct response [
73], which is made more difficult by the incongruent flankers, as they increase the amount of interference surrounding the informative central arrow. Our findings partially replicate previous studies using the ANT for children with 22q11.DS where the executive but not the alerting or orienting indices were impaired compared to TD control children [
25,
57,
69]. Impairments in other components of executive cognitive function, such as working memory [
16] and cognitive flexibility [
66], have also been consistently found for 22q11.2DS. Our results therefore concur with this trend. Impairments of executive function in FXS are also well established; however, to our knowledge, this is the first report of the use of a flanker paradigm for this population. For children with FXS, tasks that more directly test cognitive control or resistance to distractors have provided clear evidence of attentional impairment [
74], while tasks that require response inhibition or impulse control are commonly impaired for boys with FXS [
9,
75]. Our results using the ANT suggest that impulse control may also be impaired for girls with FXS. The comparison of error rate also agrees with previous findings of poorer inhibitory control with FXS [
12,
75]. In the present study, girls with FXS made more errors (10%) than any other groups (less than 1%). Previous studies of girls with TS that tested the relationship between global intellectual functioning and both attentional processing and executive functions found a positive correlation between PRI and executive function [
39]. Our analysis did not replicate such a correlation. This could be due to the higher mean IQ of the girls with TS in our study. Another possibility is that the demands of the Attention-Executive function domain subtests of the NEPSY may be more demanding than the ANT. Simultaneous recruitment of both executive function and attention in the context of spatial processing appears to be particularly difficult for girls, adolescents and adults with TS [
70,
71]. Whether the same is true for attention tasks that require executive function and spatial processing remains to be determined.
While the alerting and executive indices were similar amongst groups, performance on the orienting index was different. Two mechanisms are thought to direct the orientation of attention: voluntary, endogenous shifts of attention and the involuntary, exogenous capture of attention by salient stimuli [
76], with the latter maturing earlier than the former [
77]. Studies of children with 22q11.2DS have reported impaired endogenous orienting [
24] but typical exogenous orienting [
25,
57,
69,
78]. While the orienting index is an exogenous cueing, the proportion of valid to invalid cues is not the typical 50:50 ratio. Therefore, our results only resemble previous findings of a typical validity effect, an added benefit to their RTs after the presentation of a valid cue, for 22q11.2DS [
25,
57,
69]. As with girls with 22q11.2DS, the girls with TS in our study responded with a typical validity effect. To the best of our knowledge, no prior studies have specifically measured orienting attention in girls with TS. Interestingly, girls with FXS did not demonstrate any validity effect (i.e., the RTs did not differ between validly and invalidly cued trials). In a follow-up analysis, comparison of adjusted RTs to alerting and orienting cue conditions together revealed no significant facilitation by any cue type for girls with FXS. It is possible that the girls were not using the cue information to assist their responses to the eventual target, as has been seen previously in a mixed gender study [
79]. Additionally, in response to an exogenous cueing paradigm, infant boys with FXS performed eye movements as quickly for valid cues as for invalid cues [
9]. This indifference to the validity of a cue was also reported in adolescents with FXS, for both exogenous and endogenous orienting [
79]. This may also provide insight into the aforementioned higher error rates measured for the FXS group in the current study.
What these three attentional networks have in common is the implementation of selection between competing items or attributes that results in the facilitated processing of what is selected [
41,
80]. A more recent framework divides attention tasks into those that test the selection between competing inputs or selection between competing rules [
73]. Alerting and orienting are clearly examples of processes that place demands on the implementation of selection to use cues effectively, while the executive component, as stated previously, places a higher demand on the implementation of selection, as opposed to the control of selection [
81]. This framework aligns itself with proposed neural systems that subserve the attention networks. The alerting and orienting systems are associated with separate regions of the right frontal and parietal lobes [
43,
82,
83] while the executive system is associated with the anterior cingulate and lateral prefrontal cortex [
84]. In accordance with our results, volumetric reductions have been noted in children with 22q11.2DS [
85] and females with FXS [
86]. Though volumetric changes to regions associated with the executive system have not yet been noted in girls with TS, a recent functional connectivity study found reduced functional connectivity with dorsal frontal regions [
33]. It can be imagined that these distinct anatomical impairments could result in grossly similar behavioral outcomes within the executive system.
As predicted, behavioral impairments in attention subsystem were confirmed for girls with NDDs. Importantly, not every subsystem was negatively affected and some impairments were restricted to a subset of the NDD groups studied. Given possible anatomical and cellular convergence [
6,
8,
45] between children with 22q11.2DS, FXS, or TS and the distributed nature of behavior, in general, and attention, in particular, it is possible that for each disorder different network ‘nodes’ are perturbed. These perturbations will then interact with the unaffected nodes and disrupt the behavior of higher-order cognitive systems in similar ways, much like a mechanical watch will not keep time correctly if any single cog is misaligned.
Our final aim was to determine whether the development of attention subsystems was delayed in any or each NDD relative to the pattern seen in the TD sample. Studies of typical development consistently demonstrate improvements in the ability to perform increasingly difficult cognitive tests of attention with increasing age [
46], generally followed by a period of stabilization [
47,
87]. For example, there is evidence that the ability to maintain alertness matures around the age of ten [
88], while executive control mechanisms continue to develop throughout adolescence and into early adulthood (for a review see [
46]). Using the ANT, the indices of the three attentional subsystems in typical development were found to be stable between the ages of six and ten [
47]. For children with NDDs, it was not clear whether cognitive impairments stem from late but normal maturation of the requisite neurocognitive system or from a developmental trajectory that stabilizes at a similar rate but poorer level of achievement, despite exposure to relevant stimuli and ensuing practice to build the cognitive skill in question. If the former were the case, then we expected to uncover age-related improvements in performance akin to early stages of development found in younger TD girls. A lack of a significant age-effect in an attentional index implies that stabilization of the attentional index was similarly timed in TD girls and girls with a NDD.
In a cross-sectional analysis and consistent with the findings for TD children, we found no linear age-related changes in the alerting and orienting functions for the sample as a whole or for any of the diagnostic groups. This suggests that for girls with 22q11.2DS, FXS, or TS, the attentional indices for alerting and orienting follow the trajectory of typical development and likely stabilize at ages younger than tested here. For the executive control of attention, in the present study there was no indication of age-related changes for girls with FXS, or TS, like the TD girls. In contrast, the results for girls with 22q11.2DS suggest a potential deviation away from the typical developmental trajectory due to a difference in efficiency between the younger and older girls, evidenced by the smaller differences in RTs between congruent and incongruent flanker conditions in the older participants. Interestingly, for the executive index, a previous study of 22q11.2DS that included both boys and girls, found that age was negatively correlated with RT for incongruent flankers. In this study, where we only included girls in the analysis, this age effect for RT did not have statistical significance. This contrasting finding may highlight gender differences in development for children with 22q11.2DS as we only tested girls in the present study. It is possible that the facilitation by congruent distractors or the interference by incongruent distractors arises from neurocognitive factors relating to gender-specific maturational characteristics, such as increased estrogen in pubertal girls, which affects dopaminergic function [
89]. A preliminary analysis comparing the performance of girls grouped by their Tanner stage has suggested this may be the case; however, few girls had reached the later stages at the time of testing. Gender and developmental differences for 22q11.2DS have been understudied, and these results indicate an important direction for future study. Additionally, whether the attentional indices are truly stable in the diagnostic groups during this age range will have to be directly tested through longitudinal studies.