We sought to quantify the resolution of spatiotemporal representation and to compare performance among two groups of children with neurodevelopment disorders (22q11DS and SCA) and TD controls. Using tasks designed to examine spatial and temporal magnitude estimation, we replicated and extended previous findings reported by Debbané [18] and Simon [1, 17] demonstrating in children with 22q11DS increased sensitivity to the distance effect, whereby participants with 22q11DS required larger differences (i.e., higher detection threshold) to accurately compare magnitude in both visual and auditory modalities. Participants were required to determine difference between sequentially presented spatial, temporal, and pitch stimuli in a series of simple comparison tasks. We interpret these findings as evidence of reduced resolution in the mental representations needed to quantify space and time and heightened vulnerability to attentional crowding. Or in terms of our digital imaging analogy, the resolution of spatial and temporal representations in children with 22q11DS was “grainier” than those of TD children and those with SCA. As expected, these impairments were specific to spatial and temporal domains, as no group differences were observed in pitch detection thresholds. This indicates that group differences in the spatial and temporal tasks are not due to inability to perform comparison tasks in general and supports the hypothesis that cognitive impairment of attention resolution in 22q11DS is preferentially sensitive to the spatiotemporal properties of stimuli and does not extend to impairment of the mental representation of frequency (pitch).
It is important to note that compared with age- and gender-matched TD peers and those with SCA, more children with 22q11DS (between 14 and 34%, depending on the task) than those with SCA or TD children (0–10%) were unable to perform each task. That is, they were unable to perform above chance when the magnitude difference between stimuli was largest. These children may represent a sub-group of children within the larger 22q11DS sample who present with severe impairment in magnitude comparison. However, it is important to note that when examined on a case-by-case basis, the severe impairment sub-group was not comprised of the same participants across the different spatial, temporal, and pitch comparison tasks.
Quantifying the resolution of magnitude representation in 22q11DS and SCA
A significant difference in detection threshold between the TD group and children with 22q11DS and between the SCA group and children with 22q11DS clearly showed that children with 22q11DS needed the differences between the spatial and temporal extent of the standard and target bars to be significantly larger than was required for the children in the TD and SCA groups in order to be able to perform at the same level of accuracy (i.e., 75%). Since our psychophysics-like tasks held performance constant and adjusted detection thresholds, our results allow us to produce the first estimates of the “degree of hypergranularity,” or severity of impairment causing reduction in resolution of spatial and temporal representations that is experienced by our sample of children with 22q11DS. As expected, the SCA group displayed comparable performance to the TD group on tasks of spatial, temporal, and pitch detection. On the spatial resolution task, given the smallest standard/target difference for TD children was 7% and the smallest ratio for children with SCAs was 8%, the 14% ratio for our 22q11DS group suggests that their mental resolutions for spatial extent were 6–7% less complete than was true for the other groups.
Overall, all groups demonstrated poorer temporal magnitude detection threshold compared with the spatial magnitude threshold. On the temporal detection tasks, participants reported higher detection thresholds and more participants were excluded for failing to respond above chance. Given the smallest standard/target ratio for TD children on the temporal duration judgment was between 8% (auditory) and 17% (visual), and the smallest ratio for children with SCAs was between 12% (auditory) and 22% (visual), a standard/target ratio between 23% (auditory) and 31% (visual) for our 22q11DS group suggests that their mental resolutions for spatial extent were between 9 and 15% less complete than was true for the other groups. Further, it is important to note that the 22q11DS group performed comparably with both SCA and TD groups on the pitch detection task (ratio: 22q = 21%; SCA = 18%; TD = 20%).
Though methodological differences prevent direct comparison between studies, our findings are consistent with those reported by Debbané et al. [18] on measures of temporal magnitude discrimination—they also showed that children with 22q11DS required larger difference between stimuli to accurately detect difference in tone duration, but not pitch. Surprisingly, given the similarity in some aspects of task design and presentation, a recent study by Attout et al. [32] did not replicate the findings reported Debbané and colleagues [18] or those of the present study that showed temporal magnitude representation impairment in individuals with 22q11DS. Participant characteristics such as, age range, or differences in methodology and accuracy analysis may account for the difference in findings.
Our findings indicate that processes related to forming frequency-based mental representation and detecting difference between auditory pitches is equivalent in 22q11DS to children with SCA and TD peers. These findings show that properties of the representation of spatial and temporal stimuli that are not found in pitch stimuli are specifically impaired in 22q11DS, and that performance on these tasks is not the secondary consequence of task design. Further, we interpret these findings to indicate that children with 22q11DS present with specific impairment in spatiotemporal magnitude estimation that does not extend to pitch detection.
22q11DS cognitive phenotype: evidence for selective spatiotemporal impairment
As mentioned previously, the broad cognitive phenotype of 22q11DS is characterized by impairment in the non-verbal domain [21, 37, 38]. Moreover, previous literature point to impaired representation of space, time, and number in 22q11DS, with difficulties documented on measures of numeracy, visuospatial processing [17, 19], spatial working memory, and executive functioning [17, 21, 45]. The present study extends previous research in this area by explicitly quantifying spatiotemporal representational abilities in 22q11DS compared with TD children and children with SCA.
A methodological strength of the present study is the addition of the pitch detection paradigm (APC) with identical PEST algorithm and design to the spatial magnitude comparison task (AMC). Thus, if task performance was the consequence of general working memory and/or attention impairment, we could reasonably expect similar patterns of performance across tasks. However, as stated above, when compared with age- and gender-matched children with a different NDD (SCA) or the TD group, children with 22q11DS demonstrated comparable ability to generate mental representations of frequency estimation to discriminate between tones. Thus, the cognitive impairment associated with 22q11DS does not impede the ability to represent and compare frequency information.
Spatiotemporal hypergranularity hypothesis
The profile of performance reported here by children with 22q11DS is consistent with what would be expected from a developmental impairment in spatiotemporal mental representation and relatively intact mental representation of pitch. This hypothesis was tested with the spatial magnitude comparison task, temporal duration tasks (visual and auditory versions), and a pitch comparison task. This study advanced the hypothesis of specific spatiotemporal hypergranularity in 22q11DS, first by demonstrating evidence of reduced resolution of spatial and temporal mental representations on a series of simple psychophysics comparison tasks, and secondly by demonstrating uneven magnitude estimation performance across sensory domains, demonstrating relatively intact pitch perception abilities compared with children with SCA and TD controls. Thus, impairment was specific to spatiotemporal task demands and was not accounted for by more general attentional or working memory aspects of the task design.
Limitations
Before we address the potential contribution of intellectual impairment on task performance, there are several limitations associated with the current study. These relate primarily to task characteristics as well as the measurement of IQ. It should be noted that the number of trials in each task was relatively small compared with typical psychophysiology studies. While this was an intentional methodological decision to minimize participant fatigue, replication with additional trials is recommended. However, similarities between our findings and previous studies utilizing similar paradigms (e.g., [18]) coupled with adequate power would suggest that the reduced number of trials did not adversely impact our results. Secondly, rates of excluded children with 22q11DS were unexpectedly high on the temporal duration judgment (TDJ-A/V) tasks (auditory: 22q11DS: 26% (n = 14/53), visual: 22q11DS: 34% (n = 18/53)). While worth noting, this outcome may not necessarily represent a methodological weakness. Results showed, across all groups, the same pattern of poorer performance on the TDJ-A/V tasks. Therefore, while children with 22q11DS demonstrated impaired performance, the threshold of > 50% accuracy for inclusion in the primary analysis indicated that participants included in analysis understood the requirements of the task. Furthermore, it does not appear that task difficulty explains the TDJ-A/V exclusion rates in the 22q11DS group. We report that only one TD participant failed to meet block 1 accuracy thresholds for these tasks compared with n = 4 (8%) of TD participants who failed to meet block 1 threshold on the APC task.
The primary limitation of this study relates to several aspects to do with the measurement and treatment of IQ. Participants completed either the 4-subscale WASI-2 or the more comprehensive WISC-IV assessment. Although the comparability of these measures is acceptable, with correlation coefficients of 0.82 (PRI), 0.85 (VCI), and 0.91 (FSIQ) respectively, this may have resulted in slightly higher WASI scores compared with their WISC-IV equivalent. Another limitation to the study relates to IQ and the NDD groups. In addition to characterizing spatial and temporal representation abilities in a group of children with SCA, the inclusion of an SCA group was intended to address several methodological matters common to neurodevelopmental disorders research. Matching samples on IQ runs the risk that IQ-matched group heterogeneity will impede meaningful comparisons, while controlling for features (such as IQ) that are highly correlated with group membership creates similar impediments to interpreting findings. Thus, in the present study, we recruited participants from a cohort with a defined chromosomal duplication with mild intellectual impairment and a cognitive profile “opposite” to 22q11DS. Though statistically different from both the 22q11DS and TD groups—the SCA group had a mean FSIQ within the average range. Thus, our NDD contrast group was not FSIQ matched to our 22q11DS group and did not present with mild intellectual impairment at the group level. Furthermore, our 22q11DS group did not display relative strengths in VIQ and members of our TD control group had FSIQ well above average. Together, these factors limited our interpretation of the data in relation to the effect of IQ on task performance.
For these reasons, we chose to run analysis with and without IQ included as a predictor in the regression model, and performed correlational analysis to examine the patterns of association of general functional ability (IQ) and our experimental measures. It is our contention that IQ is a group-defining characteristic of 22q11DS. Moreover, we consider IQ largely unsuited as a covariate in neurocognitive studies of individuals with NDDs (see [43] for a nuanced discussion of the topic). To summarize, the authors argue that IQ does not meet the requirements of a covariate, and incorporating IQ into analysis of NDDs risks “overcorrected, anomalous, and counterintuitive findings about neurocognitive function” ([43], pp331). However, as the SCA group FSIQ fell within the average range, the contribution of general cognitive ability to task performance remained unclear. When IQ was included as a predictor, we saw several changes to our findings—first, for temporal measures (TDJ-A/V), none of the model predictors remained significant. For spatial representation (AMC), findings shifted to include significant group differences between the SCA and TD groups, as well as a significant contribution of IQ to the model. While for the pitch detection task, we showed a group effect, but that only the SCA group differed from TD children on this measure. IQ significantly contributed to the model, but did not do so differentially for the three groups (unlike its pair AMC). In addition, correlational analysis confirmed these different patterns of association between IQ and the experimental measures among the groups. Given the similarity in task design and demands, if performance was explained—at least in part—by IQ, we would expect the effects of IQ to be reasonably stable between the tasks and groups. The SCA and TD groups both reported FSIQ in the average range and with similar variance in performance, yet we see marked difference in correlation of IQ and experimental measures. The SCA group performance on these measures shows no association with IQ, while the TD group reports fairly consistent associations with some measure of IQ for magnitude and frequency, but not temporal detection. Therefore, based on these combined findings and previous reports on this topic, we contend that the model excluding IQ is the more parsimonious of the two models, and as such, we consider these the basis of our key findings in this study. However, we must acknowledge the complex role that IQ may be playing in our findings. Future studies will help to clarify its involvement through recruitment of NDD contrast groups that are a closer match on IQ.
A final important consideration regards spatiotemporal representation causing downstream difficulties in non-verbal cognition in 22q11DS. We see some support for this hypothesis. Higher PIQ was associated with lower detection threshold for three of the four measures (AMC, APC, TDJ-A). It is difficult to explain why PIQ would be associated with the auditory measure of temporal representation but not the visual. Participants with 22q11DS reported poorest and least variable performance on TDJ-V; however, though it was not associated with PIQ, better performance was significantly positively correlated with FSIQ and VIQ on this measure in the 22q11DS group only.
Methodological strengths of the study relate to the inclusion of age- and gender-matched TD and SCA groups as well as the addition of the pitch detection task that allowed us to explore the specificity of spatiotemporal representation impairment in these groups. Finally, the application of PEST instead of reaction time measures, which are more typical to psychophysical and behavioral paradigms, allowed us to quantify each individual’s detection threshold. Reaction time is notably heterogeneous, often age-dependent, and not normally distributed in neurodevelopmental disorders. Thus, the PEST method allowed us to more directly measure the cognitive feature that we were most interested in (i.e., detection threshold) rather than an indirect manifestation of detection threshold (e.g., reaction time).
Clinical implications
Clinical implications primarily relate to the potential for remediation of the cognitive processes underpinning spatiotemporal representation. Specifically, we will be interested to know whether targeted training can improve spatiotemporal resolution; whether improvements generalize to related cognitive skills; what is the optimal “dose” of cognitive training required to yield improved performance; and importantly, whether training can translate to improve functioning in everyday life.