Relation between executive functioning, sensory processing, and motor performance in children with autism

Abstract

Purpose

The neurological model of autism proposes that higher-order processing disturbances underpin the condition’s behavioral features, although emerging evidence attributes these executive functioning issues to lower-order processing disturbances influenced by sensory and motor development. This raises an important question concerning the directionality and development trajectories of neurological disturbances in autism. Hence, this study sought to elucidate the overlapping relations among executive dysfunctions, sensory processing atypicalities, and motor performance disruptions in children with autism.

Methods

Data were collected from 119 children with autism and their parents/guardians, who were recruited from Bahrain, Saudi Arabia, and the United Arab Emirates. The participants’ executive functioning, sensory processing, and motor performance were assessed using standardized computerized neuropsychological tests and parent rating scales. Two models were developed to examine whether the downstream effects of sensory processing disturbances and motor performance delays predict/contribute to the cognitive disruptions observed in the children.

Results

The structural equation modeling results revealed there to be significant structural pathways leading from the latent sensory–motor domains to the latent executive functions, which held true for both laboratory and real-world functioning, indicating that sensory–motor issues contribute to more severe disturbances in executive functions. Notably, the model including the motor variable (measured using the BOT-2) was the best predictor of altered executive functioning in everyday and laboratory settings.

Conclusions

The findings of this study indicate the potential of multifaceted and clinically integrated training programs that target both sensory and motor abilities in children with autism to improve their executive functioning. An in-depth understanding of the relations among these parameters may suggest new therapeutic approaches for these children.

Supplementary Information

The online version contains supplementary material available at 10.1186/s12887-025-05756-9.

Introduction

Autism is a complex neurodevelopmental condition thought to affect various brain systems [1, 2], with structural abnormalities potentially playing a crucial role in its neurobiology. While there is currently no consensus regarding autism’s etiology and/or the involved brain substrates, numerous theories consider the conceptual link between the brain and behavior in autism [3–5]. Among these theories, executive functioning theory [6] suggests that the characteristic behaviors associated with autism stem from executive functioning alterations caused by the frontal lobe [7]. In fact, as the frontal lobe, particularly the prefrontal cortex, plays an important role in executive functioning [8], the frontal deficit model of autism has been proposed to reflect the changes observed in these higher-order processes [7]. However, emerging evidence indicates that such issues can be attributed to a primary disturbance in lower-order processing, which may be related to the brain stem or cerebellar control [9], including sensory and motor differences [10]. In this regard, the systems approach to development indicates that an individual’s primary abilities constitute both the foundation and the formative building blocks for their later functioning [11]. Consequently, what might appear to be a minor delay in one developmental aspect, such as motor or sensory functioning, could have a domino effect, impacting other developmental areas [12]. This is likely due to the foundational nature and influence of developmental aspects of motor and sensory functioning relative to the emergence of executive functioning [13]. Hence, this theory raises important questions regarding the directionality and development of neurological disturbances.

While the exact mechanisms that underlie the executive functioning differences seen in autism are not fully understood, they may reflect alterations in interactions across large-scale brain networks [14, 15]. Yet it remains necessary to determine whether impaired executive functioning can be explained by a change in the prefrontal cortex (a reduction in top-down modulation) or the secondary effects of atypical processes in other domains (bottom-up influences). As the global executive functioning difficulties seen in autism may reflect a reduction in critical top-down modulation, potentially due to the frontal lobe [15], a lack of top-down guidance may hinder general executive functioning, sensory discrimination, and motor performance [16]. However, despite the support for the frontal deficit model of autism, a growing body of evidence indicates that sensory–motor issues (as behavioral indices of low-level brain alterations) may lead to various downstream impacts on the higher-order cognitive functions [17, 18]. Sensory–motor symptoms may reflect non-localized disturbances to the connections among the subcortical and cortical regions [19], indicating that sensory–motor processes and the associated neural structures may play a significant role in higher-order cognitive functioning.

Developmental approaches such as the cascading model of autism and the developmental cascades perspective suggest that even minor disruptions to early-emerging behaviors—particularly in domains that are not traditionally linked—can influence subsequent developmental trajectories across functional areas [20, 21]. Furthermore, previous studies found sensory–motor signs to both represent the first clinical symptoms observed in autism [22–25] and serve as good clinical indicators of brain and nervous system integrity [26, 27]. The brain regions associated with more basic functions, including sensory–motor processes, mature before areas involved in top-down behavioral control [28]. Thus, changes in early-maturing regions compound developmental disruptions in subsequently developing areas, including the frontal and limbic structures [13], leading researchers to consider the role(s) that sensory responsivity and motor coordination alterations play in autism. Do disturbances in lower-level functions (sensory processing and motor performance) impact higher-level functions (executive functioning)? To what extent do lower-level functions contribute to higher-level functions?

While altered executive functioning may stem from atypical sensory, motor, and/or attentional processes [10], few studies have explored the bilateral associations between sensory processing and executive functioning in autism [29–32]. Similarly, information on the bilateral associations between motor skills and executive functions in autism is limited [12, 33, 34]. Moreover, no study has considered the three-way relationship between sensory, motor, and executive functioning, meaning that the contributions of sensory–motor issues to altered executive functioning in autism have not been systematically examined. The affected domains are generally perceived as discrete mental states, although the existence of interrelationships among them has been hypothesized in attention deficit hyperactivity disorder (ADHD) [35] and traumatic brain injury [36]. Research has shown the sensory–motor functions to be integrated with the cognitive functions, indicating interrelations among them [27], which could also exist in autism, although differences in clinical samples and methodologies across studies render comparison of the findings difficult.

Given the above, it is vital to develop a comprehensive model of the interrelations among the three constructs and examine the hypothesis that sensory–motor functions play a significant role in executive functioning in autism. Therefore, this study sought to (i) advance understanding of the downstream effects of sensory–motor disturbances on the altered executive functions seen in autism, (ii) assist in the early and accurate diagnosis of autism, and (iii) identify optimal targets for more effective function-based interventions for autism. To accomplish these aims, the key objective of this research was to determine the correlation among executive functioning, sensory processing, and motor performance in autism. Strong interrelationships were hypothesized to exist among these constructs, with atypical sensory and motor characteristics potentially affecting executive functioning. The Behavior Rating Inventory of Executive Function-2 (BRIEF-2) was used to measure children’s everyday executive functioning, while the Cambridge Neuropsychological Test Automated Battery (CANTAB) was used to investigate this variable in a laboratory setting. Additionally, the Child Sensory Profile-2 (CSP-2) and Bruininks-Oseretsky Test of Motor Proficiency, Second Edition (BOT-2) were used to evaluate various dimensions of sensory and motor functioning, respectively. It was postulated that the parent-rated measures of executive functioning would indicate different relations than the performance-based measures.

Method

Participants

Drawn from a larger study concerning the executive functioning [37], sensory processing [38], and motor performance [39] profiles of children with autism when compared with normative data and data concerning typically developing children, the data analyzed in this study were collected from 119 children with autism and their parents/guardians (mean age = 8.75 years, age range = 6–12 years, 79.8% male participants). A larger proportion of male participants was expected due to the higher prevalence of autism in males than females, with males being four times more likely to be diagnosed with the condition than females [40]. In this regard, the female protective effect hypothesis suggests that females require more extreme genetic mutations and greater environmental risk factors to meet the diagnostic threshold for autism when compared with males [41].The participants were recruited from three Gulf states—namely, Bahrain, Saudi Arabia, and the United Arab Emirates (UAE)—to elucidate how the seemingly disparate constructs might be related in children with autism. Two different models were designed to examine the extent to which lower-level functional disruptions undermine higher-level functions in children with autism.

The autistic group comprised children whose autism was categorized as mild. Their autism diagnosis had previously been determined based on the Diagnostic and Statistical Manual of Mental Disorders, Fourth Edition, Text Revision (DSM-IV-TR) criteria and the Childhood Autism Rating Scale (CARS) by a pediatrician or neurologist. The clinical diagnosis and symptom severity assessment of each participant were verified using the Clinician-Rated Severity of Autism Spectrum and Social Communication Disorders (CRSASSC) measure [42] included in the fifth edition of the Diagnostic and Statistical Manual of Mental Disorders(DSM-5).Additionally, the participants’ parents completed the Gilliam Autism Rating Scale, Third Edition (GARS-3) [43] and Michigan Autism Spectrum Questionnaire (MASQ) [44] to confirm the symptom severity level.

The participants’ autism was categorized as mild according to the following criteria: (a) their intelligence score was ≥ 70 (as determined using the Arabic version of the Wechsler Intelligence Scale for Children, Third Edition [WISC-III; Wechsler, 1991]); (b) their GARS-3 score was ≥ 70; (c) their rating was “mild–requiring support” according to the severity scale included in the DSM-5; and (d) their MASQ was ≥ 22. By contrast, the exclusion criterion was the presence of severe comorbidities (e.g., intractable epilepsy, severe self-injury, aggression, blindness, deafness). The absence of comorbidities was verified by the researcher.

Instruments

Executive functioning tests

This study focused on four executive functioning skills—namely, planning, mental flexibility, inhibition, and working memory. These skills have been extensively studied in psychological laboratories in both typically developing children and children with autism of varying ages [45].

Behavior rating inventory of executive function, second edition (BRIEF-2) [46]

The BRIEF-2 was used to measure the parent-rated behavioral correlates of children’s executive functioning in real-world contexts. It is composed of nine clinical scales, three broad indices, and one overall score. For this study, four subscales from the nine clinical scales were used (inhibit, shift, working memory, and plan/organize). Lower scores for the BRIEF-2 are associated with better executive functioning (and vice versa), meaning that higher scores are indicative of a greater number of difficulties concerning the executive functions. For all the BRIEF-2 clinical scales and indices, t-scores ranging from 60 to 64 are considered mildly elevated, while t-scores ranging from 65 to 69 are considered potentially clinically elevated. Moreover, t-scores ≥ 70 are considered clinically elevated.

Cambridge neuropsychological test automated battery (CANTAB) [47]

Four subtests from the CANTAB—namely, the intra-dimensional/extra-dimensional set-shifting (IED)task, spatial working memory (SWM) task, Stockings of Cambridge (SOC) task, and stop-signal (SST) test—were used to assess specific executive functioning domains (i.e., mental flexibility, working memory, spatial planning, and response inhibition). The CANTAB uses non-verbal tasks displayed on a computer screen and requires non-verbal answers to be given on the same screen.

Child sensory profile-2 (CSP-2) [48]

The CSP-2 measures the parent-rated behavioral correlates of different sensory processing patterns. It comprises six sensory system scores, three behavioral scores and four sensory quadrant scores. The four quadrants including seeking, avoiding, sensitivity, and registration are considered in this study due to aligning more directly with Dunn’s model of sensory processing, which categorizes sensory patterns based on the interaction between the child’s neurological thresholds and their self-regulatory behavioral responses. Previous studies used confirmatory factor analysis to investigate the validity of the Sensory Profile 2 (SP2), which demonstrated a good fit with the four-factor model based on Dunn’s sensory processing framework [49, 50].

For the CSP-2, low scores are just as meaningful as high scores, with both being considered indicative of sensory processing differences. There are cut-off scores for each summary raw score, which are based on the bell curve. Summary scores that deviate from the mean indicate the presence of sensory symptoms. These summary scores are expressed as a categorical variable composed of five descriptive categories: (i) “just like the majority of others” or “LM,” (ii) “more than others” or “M1,” (iii) “much more than others” or “MM2,” (iv) “less than others” or “L1,” and (v) “much less than others” or “ML2.” This means that the CSP-2’s classification system interprets the results based on the normal distribution.

Bruininks-oseretsky test of motor proficiency, second edition (BOT-2) [51]

The BOT-2 is an individually administered test of motor proficiency that quantifies motor skills in individuals aged 4–21 years based on the results of goal-directed activities. The BOT-2 results in four composite scores—namely, the fine motor control (FMC), manual coordination (MC), body coordination (BCO), and strength and agility (SA) composites—and one comprehensive measure of overall motor proficiency. This study utilized the short form of the BOT-2, which uses 14 items from the full scale to achieve an accurate representation of all eight subtests and generate sufficiently reliable scores. A score of ≤ 40 is regarded as an abnormally low score.

Procedures

Ethical approval

for the study was obtained from the Human Research Ethics Committee of Queensland University of Technology (QUT). The parents of children who met the eligibility criteria were interviewed so that the parent rating scales (BRIEF-2 and CSP-2) could be completed. The children were tested individually at their respective educational institutions. All the measures were administered according to the standardized procedures outlined in their manuals. The assessments took place in the morning in a quiet, distraction-free room. Some participants with autism were accompanied by a personal assistant during the assessment due to behavioral concerns.

Data analysis

The statistical calculations were performed using IBM SPSS Amos v.24. More specifically, structural equation modeling (SEM) was applied in Amos. Prior to the analysis, the data were screened using IBM SPSS Statistics v.23 with regard to the accuracy of the data entry, missing data, and violation of parametric assumptions. Then, Pearson correlation coefficients were calculated to determine the degrees of the relationships between the constructs being measured (i.e., executive functioning, sensory processing and motor performance). The results concerning the interrelationships between the variables subject to investigation are reported in the in the supplementary materials (see table S2, S3, S4 and S5). Third, the structural model was used to estimate the latent variables representing sensory processing domains, motor performance skills, and executive functions, as well as the linear pathways between these variables. SEM offers the advantage of assessing latent (unobservable) variables. It also permits researchers to confront a theory-driven model and provides for the assessment of the model’s goodness of fit as a whole, which is crucial in relation to determining the robustness of the hypothesized underlying framework [52, 53]. The statistical analysis involved in SEM comprises two steps: validating the measurement model through confirmatory factor analysis (CFA) and fitting the structural model via a path analysis with the latent variables [54].

The model analyses were conducted in Amos on the covariance matrices via maximum likelihood estimation and following a two-step procedure. In the first step, a separate CFA was run for every construct to ensure that the latent variables were well measured by the respective indicators. In the second step, a series of structural equation models were fitted that systematically examined the contributions of the lower-level brain functions (i.e., sensory and motor functions) to the higher-level brain functions (i.e., executive functions). The fitness (or otherwise) of each structural equation model was evaluated based on the following indices: chi-square (χ2, the corresponding p-value for the model), root mean square error of approximation (RMSEA), and comparative fit index (CFI). A non-significant χ2 value (p-value > 0.01) is considered evidence of the model fitting the data [55]. However, the model’s χ2 is sensitive to the sample size and the complexity of the model [56–58]. It has been recommended that other goodness-of-fit indices are used alongside the χ2 [59], with the RMSEA being most commonly used [55, 59, 60]. RMSEA values between 0.05 and 0.08 are considered indicative of an acceptable fit, while values > 0.10 are interpreted as indicating a poor fit [61]. The CFI is an alternative to the χ2, and CFI values between 0.90 and 0.95 are considered indicative of an adequate fit. The Bayesian information criterion (BIC) and Akaike information criterion (AIC) were also used for the comparisons with non-nested models. The smaller the AIC and BIC values, the better the model fit.

Results

The major outcomes of the statistical modeling of the study data, which was performed to sensory processing and motor performance issues seen in children with autism contribute to more severe disturbances in executive functions, are reported in this section. The statistical analysis of the SEM involved two distinct steps. First, a CFA was employed to examine the goodness of fit in relation to the latent variables in the measurement model. Second, a path analysis was performed to examine the direct pathways between the latent variables Table 1.

Table 1.

The study models

Recent research has demonstrated that sensory and motor processes may actually be more complex than originally believed (Levit-Binnun et al., 2013). The likely broader relevance of sensory and motor skills to the overall integrity of the brain led the current research project to consider the hypothesis that strong interrelationships would emerge between executive functioning, sensory processing and motor performance, while atypical sensory processing and motor performance may exert downstream effects on the executive functioning of children with ASD. In order to examine the relationships between these variables, two models were constructed and tested (Figs. 1 and 2). When constructing the models, it was hypothesized that:

Model (1): The latent variable, namely sensory processing (the sum of the seeking, avoiding, sensitivity and registration items), would have a direct effect on executive functioning in everyday settings (as measured using the plan/organize, working memory, shift and inhibit subscales of the BRIEF-2), as well as on executive functioning in a laboratory setting (as measured using the plan, working memory, shift and inhibit subscales of the CANTAB). This is depicted in Fig. 1.

Model (2): The latent variable, namely motor proficiency (as a composite measure of fine motor control, manual coordination, body coordination, strength and agility), would have a direct effect on executive functioning in everyday settings (as measured using the plan/organize, working memory, shift and inhibit subscales of the BRIEF-2), as well as on executive functioning in a laboratory setting (as measured using the plan, working memory, shift and inhibit subscales of the CANTAB). This is depicted in Fig. 2.

Confirmatory factor analysis

A series of CFAs were run to test the goodness of fit of the measurement model. They included the following latent variables: sensory processing, motor proficiency, BRIEF-2, and CANTAB. For the independent constructs, the fit statistics of the measurement models for sensory processing and motor proficiency indicated that both models fit the data well. As shown in Table 2, the χ² value was non-significant and the RMSEA value was < 0.05 for both models. Furthermore, the CFI value was 0.99 for the sensory construct and 1.00 for the motor construct. Additionally, the indicators loaded strongly on their respective factors, with the factor loadings ranging between 0.70 and 0.80 for the sensory latent variable and between 0.67 and 0.86 for the motor latent variable (see Figures S1 and S2).

Table 2.

Final model fit indices for all four measurement models

	χ²	RMSEA	CFI
Sensory model	0.282	0.048	0.99
Motor model	0.512	0.000	1.00
BRIEF-2 model	0.228	0.064	0.99
CANTAB model	0.178	0.080	0.99

Fig. 1.

Diagrammatic representation of the proposed model of the relations among the sensory processing patterns and the executive functioning abilities in children with ASD

Fig. 2.

Diagrammatic representation of the proposed model of the relations among the motor proficiency skills and the executive functioning abilities in children with ASD

Regarding the dependent constructs, the goodness-of-fit statistics concerning the measurement model for executive functioning, as measured using the BRIEF-2, demonstrated that the model fit the data quite well. The results of the fit indices were χ² = 0.228, CFI = 0.99, and RMSEA = 0.064. Moreover, the latent real-world executive functioning was significantly positively reflected by all four modeled indicators, with the BRIEF-2 working memory exhibiting the strongest loading (see Figure S3). However, the CFA showed that the measurement model for executive functioning, as measured using the CANTAB, was not an optimal fit for the data. The model’s overall χ² was significant, the RMSEA was 0.168, and the CFI was 0.97. Thus, only the RMSEA statistic was under the recommended value, meaning that the model required a slight modification to fit the data well. To improve the model’s fit, the errors between the “between errors” variable and the “stop-signal reaction time (SSRT)” variable were correlated, as shown in Figure S4, which resulted in a better fit across all the indices. The factor loadings, as shown on the arrows from the factor to the items in Figures S4, ranged from 0.75 to 0.90.

Based on the above-mentioned results, the models exhibited an acceptable fit to the data. Therefore, a path analysis was performed to investigate the direct relationships between the predictor variables and dependent variables.

Latent variable modeling

As the independent latent construct of sensory processing was found to be highly correlated with the other independent latent motor proficiency constructs (r = 0.70; see Fig. 3), it proved difficult to account for the sensory and motor aspects in a single model. Such strong correlation between the two predictors impacted both the structural analysis and the subsequent conclusions. If Fig. 3 is considered again, the absence of significant effects from the sensory processing aspects toward the executive functioning domains, as measured using the BRIEF-2, appears unlikely, given that the correlation between the sensory and motor aspects is high. Dismissing the phenomenon of multicollinearity led to unreliable results when testing the measurement models, thereby jeopardizing the development of our theory. Thus, two models were constructed to reduce the confounding results, each involving only one independent latent variable. The two models were then compared using the AIC and BIC indices to determine which offered a better fit for the data (i.e., to determine which model contributed the most to the interpretation of the executive functioning difficulties).

Fig. 3.

Structural model of the relations among the sensory processing patterns, motor performance skills, and executive functioning abilities in children with autism

The first model specified that the seeking, avoiding, sensitivity, and registration scores were reflective of the general latent sensory domains (X1), indicating that they would have a direct linear effect. More specifically, the inhibit, shift, working memory, and plan/organize BRIEF-2 subscales were reflective of a latent variable representing executive functioning in the real world (Y1), while the mental flexibility “i.e., total errors adjusted”, working memory “i.e., between errors”, spatial planning “i.e., problems solved in the minimum number of moves”, and response inhibition “i.e., stop-signal reaction time” CANTAB subscales were reflective of a latent variable representing executive functioning in a laboratory setting (Y2).

The first model’s fit statistics (χ2 = 0.00, CFI = 0.92, RMSEA = 0.09) were inadequate, meaning that the modeled pathways were insufficient to reproduce the covariance matrix. Therefore, modifications to the model were considered to improve its fit. Model indicators were used as appropriate. The largest suggested estimation was considered and only accepted if the items shared the same category and originated from the same factor. After these modifications, the fit indices increased (see Table 3). Based on the recommendations of the fit indices, it can be stated that the model provided a good overall fit.

Table 3.

Final model fit indices for the structural equation models

	χ2	CFI	RMSEA	AIC	BIC
Model 1	0.000***	0.93	0.08	148.68	223.71
Model 2	0.004***	0.95	0.07	135.62	207.88

As shown in Fig. 4, the regression coefficient for the sensory domains on real-world executive functioning was β = 0.64 (p < 0.001), while the regression coefficient for the sensory domains on executive functioning in a laboratory setting was β = 0.72 (p < 0.001). This first model thus accounted for 41% of the variance in executive functioning when using the BRIEF-2 and 52% of the variance in executive functioning when using the CANTAB.

Fig. 4.

Structural model of the relations among the sensory processing patterns and executive functioning abilities in children with autism

The second model specified that the fine motor control, manual coordination, body coordination, strength, and agility scores were reflective of the general latent motor proficiency skills (X2) that would have a direct linear effect on executive functioning when measured using the BRIEF-2 (Y1) or CANTAB (Y2).

The fit statistics concerning the second model indicated that it fit the data relatively well based on the traditional cut-off standards (see Table 3). The results showed that motor performance disturbances may influence the latent factors of the BRIEF-2 and CANTAB subscales, with β weights of − 0.74 and − 0.68, respectively. Hence, the model accounted for 54% of the variance in executive functioning when using the BRIEF-2 and 47% of the variance in executive functioning when using the CANTAB.

Based on the AIC and BIC, the best model had the smallest AIC and BIC values, which indicated the stronger predictive power of motor performance disruptions regarding executive dysfunctions than of sensory processing issues regarding executive functioning disturbances.

In sum, both sensory and motor difficulties were found to be significant predictors of executive dysfunctions. However, the motor variable was a stronger predictor (β = 0.74, p < 0.001; β = 0.68, p < 0.001) than the sensory variable (β = 0 0.64, p < 0.001; β = 0.72, p < 0.001) in terms of the subjective and objective measures, respectively, of executive functioning.

It is important to note that theory prescribes the direction of causal influence (i.e., a bottom-up causal relationship). Thus, the models developed in this study were intended to validate this theoretical direction. The SEM approach involves correlation, meaning that it cannot prove or disprove the direction of influence; however, SEM results can support or oppose a given theory. The estimated models were specifically designed to test the plausibility of the discussed theory. Therefore, the directions of the arrows were pre-determined deductively and the model estimates were known to be capable of disproving the theoretical relationships. The models were not intended to provide evidence of causal effects. Indeed, such causal effects are determined by theory, not by statistical analysis technique.

Discussion

This study sought to determine the interrelations among lower- and higher-level brain functions to ascertain whether the sensory processing and motor performance alterations seen in children with autism aggravate the altered executive functioning typically observed in this population. It has been suggested that the executive functioning disturbances in autism can be explained by competing “low-level” symptoms—namely, a reduction in the efficiency with which children with autism process sensory information or a decrease in the performance efficiency of children with autism regarding gross and fine motor skills. Both these explanations reflect a bottom-up process where the efficiency of sensory processing and motor performance in autism decrease. Thus, the executive functioning processes that depend on sensory–motor perception and performance are also diminished. Prior research found that any disturbance in a sensory or motor ability is almost pathognomonic of dysfunction due to its impact on academic and cognitive function [35]. Hence, sensory–motor disturbances represent domains where further assessment and intervention should occur. Additionally, recognition of these domains’ importance could lead to improved diagnosis and treatment of autism.

To the best of the researcher’s knowledge, no previous study has investigated the interrelations among the above-mentioned constructs in autism. Hence, this study is the first to specifically consider the correlation between sensory–motor and executive functions in autism, revealing a novel set of relations among the lower- and higher-order cognitive processes. Moreover, this study measured executive functioning using multiple clinical assessment tools, involving both direct measures and parent ratings. The identified relations were found to vary according to the type of assessment used, suggesting the usefulness of combining subjective and objective measures.

The SEM analysis revealed there to be significant structural pathways leading from the latent sensory domains and motor skills to the latent executive functions. This held true for laboratory-based and real-world functioning, indicating that rather than occurring in isolation, the executive functioning disturbances observed in autism may be related to sensory–motor issues. Therefore, this study took as its starting point the idea that the alterations in executive functioning in autism are not only a product of a primary alteration in the functions themselves but may also represent a primary disturbance to a lower-level process on which the executive functions depend. If so, interventions targeted at the processing of sensory information and performance of motor functions may have a cascading effect, leading to improved executive functioning in children with autism.

Notably, the model including the motor variable, as measured using the BOT-2, proved to be the best predictor of altered executive functioning in both real-world and laboratory settings. The results also indicated that, as hypothesized, this relation varied based on the type of assessment used. The pathway from the latent motor skills to the BRIEF-2 was stronger than that to the CANTAB, which is unsurprising given that performance of laboratory-based tasks often captures different executive functioning characteristics than revealed by real-world observations. Similarly, the model including the sensory processing variable, as measured using the CSP-2, was the strongest predictor of altered executive functions. This model revealed sensory processing to have a direct influence, albeit to a lesser extent than the model including the motor variable. The sensory model also indicated the pathway from the latent sensory domains to the latent executive functions in a laboratory setting to be stronger than that to the executive functions in a real-world setting, suggesting that unique information can be gleaned from each type of assessment.

According to the model including motor proficiency (Fig. 5), the motor tasks had higher structural coefficients with the executive functioning domains. This result accords with the pilot study by Hilton et al. (2014), which identified strong negative correlations between certain executive functions (measured using the BRIEF-2) and the motor scores for the BOT-2 in children with autism, indicating a relation between the constructs. While they did not directly explore the association between executive functioning and motor performance, Hilton et al. (2014) assessed the effectiveness of an intervention to improve the development of executive functioning and motor performance in school-aged children with autism [33]. The present results also accord with those of Schurink et al. (2012), who investigated the relation between motor performance and the planning-related executive functions during a Tower of London (TOL) test in children with pervasive developmental disorders not otherwise specified (PDD-NOS). Their findings revealed a significant inverse relation between the manual dexterity and balance subtests and the TOL scores, suggesting that motor performance disturbances may be related to executive functioning, and therefore, that reduced motor skills performance is related to altered executive functioning (and vice versa) [34]. However, both studies had certain limitations—namely, Hilton et al. (2014) examined a small sample, while Schurink et al. (2012) investigated one aspect of executive functioning. Furthermore, while cognitive–motor relations have been reported in autism, neither study extensively investigated motor performance or executive functioning. Still, the present results replicate and extend the findings of previous studies, revealing a relation between motor performance and executive functioning based on a large sample of children with autism and various aspects of executive functioning.

Fig. 5.

Structural model of the relations among the motor performance skills and executive functioning abilities in children with autism

The results regarding the connection between executive functioning and motor performance also reflect the findings of correlational studies, particularly studies conducted in a nonclinical population and other clinical populations [62–68]. Moreover, this connection was evident in research suggesting that interventions designed to improve motor skills in autism may also improve aspects of executive functioning [33, 69]. However, it was proposed that enhanced motor performance is not sufficient to improve executive functioning, meaning related programs must include cognitive challenges requiring the use of various executive functions [70–72].

In terms of the model including sensory processing, the SEM results indicated that the investigated characteristics played a significant role in predicting executive functioning disturbances in autism. Furthermore, this relation was significantly predictive of altered executive functions during real-world and laboratory tasks in this sample. The results of this study regarding the association between sensory processing and executive functioning alterations in autism are consistent with previous studies documenting similar associations in other clinical populations, including preterm preschoolers [73] and children with Williams syndrome [74]. The results also accord with those of Nesayan (2017), who found a significant relation between sensory processing patterns and working memory in 50 Iranian children with autism [75].

However, the present results supporting an association between executive functioning and sensory processing contrast with those of Boyd et al. (2009), who found no relation between executive functioning and sensory alterations in children with autism. The lack of such a relation might indicate that altered executive functioning is not a shared mechanism underlying sensory processing and repetitive behaviors in cases of high-functioning autism and that neurobiological mechanisms might better explain the link [29]. The difference between the results of Boyd et al. (2009) and those of this study may be attributed to the variability in the methodological approaches, including the selection of measures. Still, a relation between executive functioning and sensory processing has been evidenced in studies examining the influence of sensory integration training on executive functioning. For instance, Faramarzi et al. (2016) investigated the effect of sensory integration training on executive functioning in ADHD, finding that such integration could improve executive functioning [76].

The developed models performed equally well in predicting the presence of altered executive functioning. Both the sensory and motor models yielded identical fit indices and parameter estimates, which could indicate sensory–motor skills to substantially overlap. Such an overlap should prompt detailed consideration of the roles that sensory processing and motor performance issues play in autism. In general, the present findings confirmed a link between sensory and motor functioning while also demonstrating the unique relationship between sensory–motor skills and executive functioning in children autism. These findings lend support to Reitan and Wolfson (2003), who suggested that the close relation between lower-level functions and the brain’s biological status renders their measurement highly valuable in supplementing the neuropsychological interpretation of higher-level disturbances [27]. The findings also reflect the developmental cascades perspective, which proposes that early delays in motor skills in autistic children can lead to subsequent alterations in other areas [20].

Additionally, the present results support the indication from neuropsychology that sensory–motor functions are closely associated with cognitive abilities, especially in terms of executive functioning [35, 36]. Davis et al. (2009) found strong correlations between sensory–motor functioning and higher-order cognitive processes in children with ADHD. They also determined that the measurement of sensory–motor skills could predict 31% of the variance in overall cognitive performance and 65% of the variance in academic achievement, with around 91% shared variance [35]. Mazur-Mosiewicz (2011) examined the relation between sensory–motor skills and higher-order cognitive processes in 136 adults with traumatic brain injury, finding that the measurement of sensory–motor skills could predict 31% of the variance in overall cognitive performance and 23% of the variance in cognitive performance [36].

Additional support for an association between lower- and higher-level brain functions has been derived from studies investigating the relation between neurological soft signs and cognitive disturbances in patients with chronic schizophrenia [77–79]. These studies found that neurological soft signs are strongly correlated with executive functioning in schizophrenia. Yet, while this study provided additional evidence of the interrelations among sensory processes, motor skills, and executive functions, the present results cannot be directly compared with those obtained in different samples investigated using different testing protocols [35, 36, 77–79].

One explanation for the significant relation identified between sensory–motor and executive functioning is that higher-order cognitive functions depend on sensory–motor perception and performance [36, 80]. This hierarchical perspective is supported by research concerning embodied cognition, which found pre-existing sensory–motor functioning to form the basis for the higher-order cognitive processes underlying complex actions [81]. This approach accords with the observation that development of brain structures follows a pattern whereby brain areas responsible for basic functions mature first, followed by areas responsible for more complex functions. Hence, sensory–motor impairments are likely to undermine subsequently developing cognitive functions. Another possible explanation is a non-directorial relationship whereby issues related to cognitive performance are explained by problems associated with the overall integrity of the brain and nervous system, which determine an individual’s performance on cognitive and sensory–motor tests [36]. However, this explanation implies that sensory–motor and cognitive functions do not vary in their sources but constitute different areas of cognitive functions [36].

Overall, this study supports the notion that sensory–motor aspects are related to executive functioning abilities. The implications of this finding concern the potential contributions of multifaceted and clinically integrated training programs targeting sensory–motor abilities to improve executive functioning. Therefore, a better understanding of the relations among these constructs may indicate new therapeutic approaches for autism.

Conclusion

This study provided behavioral evidence of the interrelations among executive functioning alterations, sensory processing atypicalities, and motor performance challenges in children with autism. Furthermore, it proposed two models of the ways in which the former disruptions might contribute to the development and manifestation of the latter. Of course, in terms of supporting the existence of a bottom-up causal relationship, the proposed models are simply two among several potentially valid models. The possibility of a top-down effect (i.e., executive functioning influencing sensory and motor behavior) cannot be ruled out by the findings of this study. Overall, the downstream influences of the observed sensory and motor disturbances may partially overlap and so contribute in parallel to the upstream effects (i.e., the chicken or egg paradox). Nevertheless, the proposed models highlighted an important direction for future work and provided the basis for subsequent empirical and theoretical research endeavors.

However, the findings must considered in light of certain limitations. First, the cross-sectional approach adopted in this study meant that it was not possible to demonstrate causation regarding the relations among the analyzed variables. The kinds of lower-level functional disruptions common in individuals with autism could lead to disturbances in subsequent high-order control processes (or vice versa), or there may be bidirectional relations. Future studies designed to evaluate the long-term functioning of children with executive functioning alterations, sensory processing atypicalities, or motor performance challenges would help to determine the cause and effect between these parameters, and to determine whether the same relations persist or change longitudinally. Interestingly, as preterm birth is a known risk factor for autism [82], future studies should investigate its role in shaping associated neurodevelopmental pathways, particularly the impacts on sensory, motor, and executive functioning and neurobiological mechanisms, elucidating how such disruptions contribute to autism traits. Given the brainstem’s role in the neurodevelopmental hierarchy, its perturbed maturation in preterm infants could also render them susceptible to altered development of higher-order functions, leading to observable behaviors during later development.

The second limitation is the lack of an adequate typical development (or control) group for comparison with the study’s participants. Hence, it would be beneficial for future studies to employ larger sample sizes to allow for direct comparisons via SEM modeling between clinical and non-clinical samples. Another limitation concerns the enrollment of only high-functioning autism cases, which may limit the generalizability of the findings. It is, therefore, important for future studies to explore the relations noted in this study in other samples of individuals with autism through employing appropriate assessment approaches for the target population to determine whether the present results are consistent with those concerning other populations.

Electronic supplementary material

Below is the link to the electronic supplementary material.

Abbreviations

AIC

Akaike Information Criterion

AMOS

Analysis of Moment Structures

ADHD

Attention Deficit Hyperactivity Disorder

BCO

Body Coordination

BIC

Bayesian Information Criterion

BOT-2

Bruininks-Oseretsky Test of Motor Proficiency, Second Edition

BRIEF-2

Behavior Rating Inventory of Executive Function, Second Edition

CANTAB

Cambridge Neuropsychological Test Automated Battery

CARS

Childhood Autism Rating Scale

CFA

Confirmatory Factor Analysis

CFI

Comparative Fit Index

CRSASSC

Clinician-Rated Severity of Autism Spectrum and Social Communication Disorders

CSP-2

Child Sensory Profile, Second Edition

DSM-IV-TR

Diagnostic and Statistical Manual of Mental Disorders, Fourth Edition, Text Revision

DSM-5

Diagnostic and Statistical Manual of Mental Disorders, Fifth Edition

FMC

Fine Motor Control

I/ED

Intra-/Extra-Dimensional

MASQ

Michigan Autism Spectrum Questionnaire

MC

Manual Coordination

Model χ2

Chi-square Statistic

PDD-NOS

Pervasive Developmental Disorders Not Otherwise Specified

RMSEA

Root Mean Square Error of Approximation

SA

Strength and Agility

SEM

Structural Equation Modelling

SOC

Stockings Of Cambridge

SPSS 23

Statistical Package for the Social Sciences, Version 23

SSRT

Stop Signal Reaction Time

SST

Stop Signal Task

SWM

Spatial Working Memory

TOL

Tower Of London

TLI

Tucker-Lewis Index

UAE

United Arab Emirates

WISC-III

Wechsler Intelligence Scale for Children, Third Edition

Rehab H Alsaedi

Ph.D. in autistic disorder, Queensland University of Technology (QUT), Brisbane, Australia.

Author contributions

The author (Rehab H. Alsaedi) confirms the sole responsibility for the conception of the study, data collection, presented results and manuscript preparation.

Funding

This study was partly funded by Taibah University, Saudi Arabia. More specifically, Taibah University funded the study instruments used during the large-scale evaluation project intended to identify neurobehavioral problems in children with autism, which was conducted by the author as part of her Ph.D. research.

Data availability

The data that support the findings of this study are not openly available due to sensitivity concerns; however, the data are available from the corresponding author upon reasonable request.

Declarations

Ethical approval and consent to participate

This study was conducted in accordance with the principles of the Declaration of Helsinki. Ethical approval for all aspects of the research was obtained prior to the study’s commencement from the Human Research Ethics Committee of QUT (approval number: 1500000980). The participants were not involved in any research activity prior to their written informed consent being obtained. The nature of the study population (i.e., children with autism aged 6–12 years) meant that both parent/guardian consent and child assent were required. Each parent/guardian was asked to discuss the study with their child and seek the child’s consent to participate (as per the informed consent form). At the start of the assessment session, the children were also verbally invited to participate in the study.

Consent for publication

Not applicable.

Competing interests

The authors declare no competing interests.