Restricted and repetitive behaviors (RRBs) are central to the phenotype and diagnosis of Autism Spectrum Disorder (ASD) (DSM-5 1), RRBs are considered «positive» symptoms of autism as they include behaviors that the person does more often or differently, instead of an absence of behaviors, like those described in the social-communication domain (2). These symptoms include stereotyped and repetitive motor movements (e.g., finger mannerisms, arm movements or hand flapping), stereotyped speech (e.g., echolalia), insistence on sameness (e.g., rigid adherence to certain routines or rituals), intense interests in specific areas such as dinosaurs or public transport maps, and special skills (e.g., calendar calculation or memory for routes) (1, 3, 4). RRBs are observed in autistic children of all intelligence level but not at every age and not in the same proportion (2).
Despite their higher prevalence in autism, RRBs are also observed in populations diagnosed with other conditions such as intellectual disability, Tourette’s syndrome, obsessive-compulsive disorder (5–7) and in typically developing children (i.e. 5, 6, 8). These behaviors are perceived differently in developmental conditions and in typical development. For example, early on in infancy, stereotyped motor behaviors are thought to be intrinsic to the development of complex functional motor behaviors in typically developing children (9, 10). These stereotypies also serve a communication function since they are more frequently observed during child-caregiver interactions and diminish as verbal abilities augment in typically developing children (10–12). Furthermore, for typically developing children, these repetitive behaviors are a way to adapt to one’s environment and by repeating, classifying, and ordering, the brain is able to organize information from the environment (13). Attempts to understand the role of these behaviors in typically developing children could be informative for clarifying the role of RRBs in autism or in other conditions. Indeed, the role of stereotypies in the development of motor skills or the possibility that at least some repetitive behaviors serve adaptive purposes in typical development are promising avenues to explore when trying to understand the role some of these behaviors might have in the development of individuals with autism.
In order to understand the role of RRBs in autism, research to-date has taken three approaches. One approach is to study the domain of RRBs symptoms as a whole, combining all individual behaviors and investigating whether presenting more or less symptoms in that domain is positively or negatively related to other domains of functioning, such as intelligence, language abilities, or adaptive functioning; and how the whole domain changes with age. The second approach is to group individual RRBs into categories, usually derived from factor analysis, and the third is to study each RRB individually.
When studied as a whole, RRBs increase throughout preschool years but then diminish later in childhood (14, 15). This is especially the case for autistic individuals with higher cognitive and language abilities (16). Moreover, the presence of more RRBs is generally linked to lower adaptive functioning (17), language abilities (18) and intelligence (19).
A different pattern of results emerges when RRBs are investigated using factor analyses. The number of factors usually identified depends on the specific RRB measure and the age range of the sample. For example, using the Autism Diagnostic Interview-Revised (ADI-R, 20), when participants older than 3 years of age are included, a third factor, circumscribed interests (CI), emerges in addition to the Repetitive and Sensory Motor (RSM) and Insistence on Sameness (IS) factors (e.g. 21). Using the Repetitive Behavior Scale – Revised (RBS-R, 22), some researchers have identified a three-factor model (23), while others have found five- (21, 23) or six-factor models (24).
The links between the categories of RRBs derived from factor analysis and other developmental domains are somewhat inconsistent (see: 16 for a review). For example, the score on the RSM factor derived from the ADI-R was shown to be negatively correlated to adaptive functioning (3), while none of the factors derived from the RBS-R were shown to correlate to the developmental level (23). The IS factor is often not correlated to IQ (21, 25), but the corresponding factor derived from the RBS-R is negatively correlated to NVIQ (17). Taken together, studying RRBs as a whole, or even as factors has not revealed a consistent pattern of results across studies. A more fine-grained exploration of the RRB domain is warranted in order to clarify the nature of the links between developmental domains and the different behaviors included in the RRB domain. To-date, few studies have looked at individual RRBs. Bishop, Richler (2) explored the association between each of the 15 RRB items from the ADI-R, age, and NVIQ in a large cross-sectional study of 830 autistic children ranging in age from 3 to 11 years. Although the association between NVIQ and the prevalence of specific RRBs was generally negative (i.e. the behavior was more prevalent in groups of children with lower NVIQ), two specific RRBs were more prevalent in children with higher NVIQ, namely Compulsions/rituals and Circumscribed interests. These positive and negative associations with NVIQ tended to become stronger with age. As for the severity of each behavior, it was either not associated with NVIQ or was related in the same fashion as was prevalence (2).
This approach illustrates that taking a more global or factor analytic approach to RRBs could lead to specific behaviors being combined despite their differential associations with intelligence, adaptive behaviors, language level, or age (2, 19). These findings also suggest that the presence of some individual behaviors may be more strongly linked to developmental level and intelligence, while others might be more specific to the autism phenotype (2, 26). The presence of a specific RRB early in development might not have the same implications as the presence of this same behavior later on in development. Indeed, the presence of a behavior at a young age may not necessarily be linked to NVIQ at that particular time in development. However, if that same behavior is still present as the child gets older, it can become an indicator of a lower IQ or level of functioning (2).
The above findings suggest that RRBs include several distinct features that likely have different underlying etiologies and manifest differently over development, thus warranting investigation of individual behaviors. However, to date, cross sectional studies have provided only snapshots of development across rather than within individuals over time. The current study, builds on the previous study from Bishop, Richler (2), by longitudinally examining specific RRBs included in the ADI-R in relation to age and NVIQ in an inception cohort of children followed from age of diagnosis to around 11 years of age.