Background
Autism spectrum disorder (ASD) is a neurodevelopmental disorder characterized by deficits in social communication and presence of repetitive and restricted behaviors [
1]. Although there is a general consensus that ASD originates early in life from predominately biological and genetic origins, these behavioral symptoms do not emerge until the second year of life, and a clinical diagnosis is often not received before the third birthday [
2]. To help detect atypicalities that arise prior to diagnosable behaviors, researchers have turned towards studying younger siblings of children with ASD [
3,
4]. These siblings are considered high risk for developing ASD (HRA) with approximately 20% later receiving a diagnosis of ASD [
5,
6] compared to only 1.5% in the general population [
7]. Past research has found that many HRA infants who later received an ASD diagnosis display behavioral symptoms within the first 2 years of life (see [
3,
8,
9] for reviews). However, subtle atypicalities, including social impairments and delays in verbal and gestural communication, have also been found as early as 12 months in HRA infants who do not go on to develop ASD [
10‐
13]. These subtle atypicalities, or endophenotypes, are heritable biomarkers seen in both affected and unaffected family members [
14,
15]. When studying the early development of ASD, it can be difficult to separate endophenotypes from early biomarkers of risk for ASD. Early biomarkers of risk for ASD are specific to the HRA infants who later receive a diagnosis of ASD and are not seen in the unaffected HRA infants. Despite the difficulties, researchers are starting to identify early behavioral differences in HRA infants who later received a diagnosis of ASD, including differences in looking behaviors towards faces and social scenes [
16‐
18].
While identifying early behavioral atypicalities is valuable, use of functional brain measures, such as event-related potentials (ERPs) and electroencephalogram (EEG), might offer more insight as these measures uncover infants’ perception and processing abilities without being restricted to infants’ behavioral capabilities. Indeed, several possible neural ASD risk markers and endophenotypes have been proposed, primarily within the social domain. For instance, in HRA infants, atypical EEG patterns and ERPs have been found in response to direct eye gaze [
19], faces and objects [
20,
21], and familiar and unfamiliar faces [
22,
23]. Additionally, initial evidence from functional near-infrared spectroscopy show that HRA infants exhibit atypical neural processing to dynamic social stimuli [
24,
25]. Finally, atypical trajectories of EEG power [
26,
27] and neural connectivity [
28,
29] have also been found in HRA infants. Most research looking into neural development of HRA infants looks at the group as a whole, investigating all infants at familial risk. However, there is some promising evidence in identifying neural biomarkers of risk for ASD including distinct patterns in neural connectivity [
28,
29] and lateralization in response to faces [
20] in the subset of HRA infants who go on to develop ASD.
Past research has mostly focused on neural response to social stimuli, but fewer studies have looked into those related to language processing. Language ability, while highly heterogeneous and no longer a core symptom of ASD as outlined by the most recent DSM-5 [
1], is often impaired in individuals with ASD with many displaying milestone delays and impairments in their vocabulary development [
30‐
32]. Even children and toddlers with ASD and normal language abilities show subtle atypicalities in their language processing [
33‐
35]. Language is a highly lateralized function with left hemispheric dominance emerging as early as a few days after birth in the majority of typically developing individuals [
36,
37]. Conversely, in toddlers, children, and adults with ASD, numerous studies have demonstrated reduced or reversed hemispheric lateralization of both structure and function for neural regions associated with language (see [
38,
39] for reviews). In addition, in an earlier study from our research program, we found atypical lateralization patterns of ERPs in response to speech sounds in HRA infants as young as 12 months of age [
40]. These HRA infants exhibited reversed lateralization compared to low-risk control (LRC) infants in a later negative-going slow wave (LSW) observed over the anterior regions from 300 to 700 ms after the presentation of speech sounds. However, it is unknown whether the reversed lateralization response in the HRA infants was being driven by those who later received a diagnosis of ASD. The current study expands on Seery and colleagues’ [
40] sample to examine differences in lateralization patterns in response to speech between infants who go on to develop ASD and infants who do not determine if early atypical lateralization patterns are an endophenotype or a biomarker of risk for ASD.
Atypical lateralization to language is not specific to ASD. Reduced or reversed language lateralization is also seen more frequently in typically developing children and adults with mixed- and left-handedness [
41‐
43], though few studies have extended this investigation to include individuals with ASD. Using functional neuroimaging, Knaus and colleagues [
44] found that adolescents with ASD who showed more typical left language asymmetry were less likely to be left-handed. However, it is still unclear how this relation manifests in early development of ASD and HRA infants.
In typically developing populations, a right-handed bias may emerge as early as 15 weeks of gestation. Approximately 90% of fetuses show a strong right-handed preference for thumb-sucking [
45], and this preference continues postnatally as infants are more likely to reach and grasp with their right hand [
46]. Despite these early biases, handedness is malleable into early childhood [
47]. Because of this, there is no consensus on when adult-like handedness is acquired. However, some researchers have argued that the direction of handedness is fixed by the age of 3 and then continues to strengthen through late childhood (see [
48] for a review). Despite the malleability in infancy and toddlerhood, there does seem to be a relation between early hand preferences and language abilities [
49‐
51] with increased rates of left- and mixed-handedness in children with language difficulties [
52] and developmental disorders that often have language impairments including dyslexia [
53] and ASD [
54‐
57].
Given the increased prevalence of left- and mixed-handedness in ASD, it is important to consider its development when looking at differences in lateralization patterns of ASD. Atypical lateralization patterns and higher incidences of left- and mixed-handedness might represent a single underlying biomarker of risk for ASD, such as atypical cortical specialization. By exploring the early development of hand preferences, we would better understand the relation between these two domains. The purpose of this study was to examine differences in lateralization to speech sounds in infancy as well as the relation between early lateralization patterns and later hand preferences in toddlerhood. First, we expand on the findings of Seery et al. [
40] by examining differences in lateralization of the ERPs to speech sounds at 12 months in LRC and HRA infants. We were specifically interested in whether infants who later develop ASD differed in the lateralization of their ERP responses compared to the LRC and HRA infants. We hypothesized that infants who received a diagnosis of ASD would show reversed lateralization patterns compared to LRC infants, and HRA infants without ASD would show an intermediate, dampened lateralization pattern. We also measured handedness in these infants at 36 months utilizing a novel observational approach. We hypothesized that three-year-old children with ASD will show more left- and mixed-handedness tendencies compared to their typically developing peers, who will begin to show a stronger right-hand preference. Finally, we investigated the relations between early lateralization patterns to speech, later handedness patterns of behavior, and language and cognitive abilities. We hypothesized that across all groups, infants who show more typical lateralization patterns at 12 months will show stronger right-handedness at 36 months. We also predicted that differences within these patterns would relate to the child’s cognitive abilities, specifically language.
Discussion
In this study, we examined early asymmetry patterns measured through neurophysiological responses to speech sounds and behavioral observations of handedness in children with and without ASD. By including both children with ASD and children who were at high familial risk for ASD, we could separate early ASD risk biomarkers from early ASD endophenotypes by comparing affected and unaffected family members. Analyses revealed group differences in their lateralization of ERPs to speech sounds at 12 months, but no group differences in handedness at 36 months. Handedness at 36 months was related to the children’s cognitive abilities, specifically fine motor skills and expressive language abilities within the LRC group. In children with ASD, we found an association between early lateralization patterns to speech and later handedness.
Hemispheric lateralization
As we predicted, 12-month-olds who later developed ASD had reversed lateralization in the LSW in response to speech sounds compared to their typically developing peers. Moreover, this difference remained even in a subsample of infants without any language delays, suggesting that this finding is not simply a result of potential language difficulties within the ASD population. Similarly, we found no relations between this laterality index and cognitive abilities, including language abilities. Past work has found reversed lateralization in response to language in toddlers who were already exhibiting behavioral symptoms of ASD [
67,
68]. Our current findings extend this by suggesting that atypical lateralization may be present as young as 12 months of age, before the overt symptoms of ASD emerge.
Our findings suggest that reversed lateralization to speech could serve as an early biomarker of risk for ASD, present before behavioral symptoms emerge. Importantly, we did not find reversed lateralization in the HRA group who did not go on to develop ASD, which suggests this is not an endophenotype. However, we cannot conclude that this atypical laterality to speech is specific in identifying children who go on to develop ASD. First, atypical asymmetry to language stimuli has also been found in children and adults with other complex neurodevelopmental disorders including schizophrenia [
69] and specific language impairment [
70,
71] and in two-month-old infants at risk for dyslexia [
72]. Second, although not significantly different from either the LRC or ASD groups, HRA infants who did not go on to develop ASD within our sample showed an intermediate, dampened LSW laterality index with variability including values overlapping with the LRC and ASD range. Reversed lateralization by itself, therefore, might not be sensitive enough to identify which infants will go on to develop ASD from those who are simply at a familial risk for ASD or other neurodevelopmental disorders.
Given that reversed lateralization in response to speech might not be specific to children with ASD, a cumulative risk model with additional potential ASD biomarkers of risk might be better in identifying infants who will go on to develop ASD from those who will not [
13,
73]. Other potential neural ASD biomarkers include various indices of reversed lateralization, such as atypicalities in structural lateralization patterns [
74,
75] and reversed functional lateralization in domains outside of language, including face processing [
76]. In fact, within our same sample, we have found atypical hemispheric lateralization in response to faces in HRA infants as young as 12 months. Moreover, preliminary evidence from this study found that infants who later received a diagnosis of ASD showed the most reversed lateralization within the HRA infants [
20]. More work needs to be done to confirm this preliminary finding, but by understanding how these lateralization patterns manifest within the early development of ASD, we could increase the predictive power of early identification and explore the possibility of a larger underlying biomarker such as atypical development of cortical specialization.
Handedness
Utilizing a novel observational approach, we investigated hand preferences in young preschoolers with and without ASD to create a constructed measure of handedness. Final analyses indicated an underlying unidimensional construct as the one-factor model with the following behaviors providing a good fit: hand used to draw, complete a picture-maze, cut with a knife, use a fork/spoon, cut with scissors, use a toothbrush, and throw a ball. Our findings are consistent with past work using the same analytic approach to demonstrate a more precise measure of handedness using the EHI in adults and adolescents [
77,
78]. Interestingly, Dragovic’s [
77] final revision of the EHI is almost identical to the measures included in our construct after he removed two items with large measurement error (i.e., hand used to open box with lid and upper hand when using a broom).
While these seven behaviors provide a measure of handedness in children, our models showed lower factor loadings compared to models of handedness in adults using similar behaviors [
77,
79]. These lower factor loadings might represent the malleability of handedness in 3-year-olds. It is unclear when full, adult-like handedness emerges, but some researchers argue that the direction of handedness (left or right leaning) is fixed by 3 years and then continues to strengthen until late childhood with right-handed children showing stronger and earlier hand preferences than left-handed children [
48]. Early inconsistencies in hand preference at age 3, therefore, could be reflected in the reduced factor loadings. Given these inconsistencies and the idea that adult-like handedness does not emerge until late childhood, using handedness as an early biomarker of ASD may be limited. Future work should follow the development of handedness from early toddlerhood into later childhood. Understanding the development of handedness, specifically emerging consistencies in hand preference, would lead to better understanding of potential differences in handedness across these groups.
Using these behaviors to create a handedness composite score, we found that the handedness measure did not differ across the groups at 36 months as most children showed a tendency to be right-handed or right-leaning. Given that older children with ASD have been found to exhibit less right handedness than their typically developing peers [
55], the lack of a difference in our sample could be interpreted in several ways. First, our ASD sample showed stronger right-handed tendencies than past studies as our proportions of atypical handedness are on the lower end of the estimated ranges. In our sample, 22.7% of children with ASD were classified as mixed-handers while 13.6% were left-handers. Past studies investigating handedness in older children and adults with ASD have estimated higher proportions of incidences for both mixed- (17–47%) and left-handedness (18–57%; [
38]). Our lower levels of left- and mixed-handedness might be a reflection of our sample’s characteristics. All but two of our children with ASD had average or above average cognitive abilities at 36 months as measured by the MSEL. While lower cognitive abilities are not present in all individuals with ASD, it has been estimated that as many as 55% of children with ASD have an intellectual disability [
80]. Perhaps if our sample were more representative of the entire spectrum, we would find more variability in handedness given that individuals with intellectual disability are more likely to show left- and or mixed-handedness [
81]. Second, our LRC children are only slightly right-handed biased. While the majority of LRC children were considered right-handed (73.9%), this proportion is lower than the estimated 90% in the general adult population [
48]. The difference between groups in handedness, therefore, may emerge with age as the LRC children continue to strengthen their hand preference, particularly becoming stronger right-handers, relative to children with ASD.
Despite finding no differences on a group level, we did find an association between handedness and cognitive abilities. Specifically, across all participants the better the child was at fine motor skills at 36 months, the stronger their right-handedness. Closer inspection of this association indicated that this was driven by the LRC group. Most of the fine motor tasks require the child to be successful at activities involving one hand (e.g., drawing, placing pennies in a bank with a slot). Since young children, particularly left-handed children, often show inconsistencies in their hand preferences [
48], this might be the result of a practice effect. Perhaps young children with stronger right-handed tendencies use their right-hand more in motor tasks requiring precision thus leading to better performance on these tasks. Additionally, we found that LRC children who were right-handed had stronger expressive language abilities. This is consistent with past work investigating this association in young typically developing populations [
50,
51]. The lack of relation with handedness and cognitive abilities within the HRA and ASD population is surprising. However, perhaps this is, again, a reflection of our sample as the majority of these children had average or above average cognitive abilities. Future work should investigate the developing relation between handedness differences and cognitive abilities over a more representative sample.
Behavioral and neural asymmetry
Across all participants, we did not find an association between speech lateralization at 12 months and handedness at 36 months. Looking within each group, although we found no relation in children without ASD, we did find a moderate association in children with ASD. Specifically, infants with ASD who exhibited a more typical lateralized response to speech at 12 months had stronger right-handed tendencies at 36 months.
On a group level, there is evidence of atypical cortical specialization in individuals with ASD with differences in structural lateralization [
74,
75] and functional lateralization across many domains including face processing [
76] and language [
67,
82]. Moreover, atypical cerebral specialization in ASD is also reflected in the visible behavior of handedness as there are increased incidences of left- and mixed-handedness in children with ASD [
55]. On an individual level, perhaps for the children with ASD in our sample, atypical lateralization to speech at 12 months and atypical handedness (left/mixed) at 36 months is a result of an underlying atypical cerebral specialization. Moreover, the children with ASD with more typical asymmetry patterns might have a protective factor in the early development of cortical lateralization which later manifests into more typical language lateralization and handedness.
Our findings of an association between early lateralization patterns and handedness in a small group of children with ASD as well as the lack of an association in children without ASD highlights many possibilities for future research including investigating different indexes of laterality and hand preferences. Specifically, by employing a longitudinal design and using a variety of neuroimaging methods, future studies could determine when atypicalities in structural hemispheric organization and functional lateralization in multiple domains, including face and language processing, first arise in low- and high-risk samples of children. On a behavioral level, observations of handedness and its consistency should be studied from infancy through childhood, starting with simpler actions in infancy such as grasping and manipulating objects and then moving to more complex adult-like behaviors such as writing. Comparing early lateralization patterns alongside with the growing stability of hand preferences would offer greater insight into how these might interact and lead to cortical specialization in individuals with ASD. Importantly, exploring individual differences within these patterns would potentially help in understanding heterogeneity of behaviors in ASD.
Limitations
There are limitations to the current study that should be considered when interpreting the findings. One is the small sample size of the ASD group. A follow-up investigation with a larger number of infants would advance understanding about the early development of language and handedness within this population. Also, our study focused on a particular group of children who develop ASD: those at familial risk, and so it could be that atypical lateralization has a genetic or environmental underpinning and only runs in families with ASD. Future work should include other high-non-familial risk groups such as premature infants or infants who fail an ASD screener to test the generalizability of these findings. Another limitation is the absence of a reliable handedness measure to validate our novel observational approach for assessing handedness. While we did base our behaviors on an established handedness questionnaire, the EHI, we did not directly validate our measure against the EHI. Finally, we are unable to determine if the reversed lateralization to speech at 12 months is specific to ASD or whether this extends to other disorders. Future work should investigate infants at early risk for other neurodevelopmental disorders.
Acknowledgements
We would like to acknowledge the Infant Sibling Project staff for their hard work in collecting the data. We would also like to thank the research assistants, Erin Ferguson, Rennie Pasquinelli, Priyanka Shah, and Claire Troia, in helping with the handedness codes. Finally, we would like to thank all of the families for their dedication and years of contribution to the Infant Sibling Project.