Department of Psychology, College of Staten Island,City University of New York, New York, USA
* Corresponding author Email: firstname.lastname@example.org
Joint attention, or coordinated attention between social partners to share interest in entities, objects or events, is a core difficulty in autism. Reduced joint attention in infancy is predictive of an autism diagnosis while variations in joint attention among people on the spectrumpredict development across a range of domains. Joint attention comprises two types of behaviours, initiation of joint attention and response to joint attention, which may exhibit orthogonal but related patterns of development and associations with other domains. The aims of this review are 1) to compare two theoretical accounts of the origins of joint attention impairments in autism and 2) to examine evidence that response to and initiation of joint attention are both core difficulties across the lifespan in autism.
Joint attention may arise from understanding of others’ intentions (the social-cognitive theory of joint attention) or develop with increasing representational skills (the parallel and distributed processing model of joint attention). The lack of clear evidence that joint attention impairments arise from face-to-face difficulties, coupled with associations between joint attention and developmental level, provide more support for the parallel and distributed processing than the social-cognitive theory of joint attention.
Theorists speculate that initiation of joint attention is more of a core difficulty in autism than response to joint attention, partially because it may be more consistently impaired across the lifespan. However, response to joint attention may also be impaired across development when assessed with appropriate measures. Indeed, there is limited evidence that initiation of joint attention is more of a core deficit than response to joint attention in autism.
This review of the literature suggests that both response to joint attention and initiation of joint attention are central to development in autism spectrum disorder. The lack of strong evidence that dyadic atypicalities precede triadic ones is less consistent with the social-cognitive model than the parallel and distributed processing model of joint attention. Future research should examine potential bidirectional associations between response to and initiation of joint attention, non-social attention,reward sensitivity, social, linguistic and cognitive development.
Joint attention, or coordinated attention between social partners to share interest in entities,objects or events, begins to emerge at around 3 months of age and develops gradually until around 18 months of age in typical development[1,2]. Joint attention is apparent across a range of species. Though it is deeply rooted in ontogeny and phylogeny, it is not always observed, particularly among children with autism[4,5]. Indeed, reduced joint attention in infancy is an early predictor of autism spectrum disorder (ASD)[6,7,8,9]. Joint attention is a core deficit in autism–it distinguishes children with ASD from typically developing children and those with other disabilities[4,5].
Joint attention is a pivotal skill that novices can use to acquire information from others–it is related to subsequent development across a range of domains for typically developing individuals and those with ASD[5,10]. Individual differences in joint attention among people on the spectrum are predictive of adaptive skills, symptoms, social functioning, linguistic skills and cognitive development (see Figure 1)[5,11]. However, little research has been carried out to assess both joint attention and developmental outcomes at multiple time points. Such research will be key for understanding the directionality of influences between joint attentionand social-cognitive development.
Joint attention is related to the subsequent development of other social-cognitive skills in autism and typical development.
Longitudinal associations between early joint attention and subsequent development are often considered as evidence for the social-cognitive theory of joint attention wherein the development of joint attention from simpler social behaviours (such as face-to-face engagement) reflects an emerging understanding of others as intentional agents that in turn scaffolds subsequent symbolic development (see Figure 2). Alternatively, rather than arising from and being defined by an understanding of others’ mental states, joint attention may not initially reflect social understanding but may lead to social knowledge (see Figure 3)[1,13,14]. According to the parallel and distributed processing (PDP) model of joint attention, joint attention arises from an increasing ability to integrate information about oneself, another and the conjunction of the self and other in relation to an external object (triadic relations). The key distinction between the two models is the relative importance of understanding another person’s mind versus the importance of practice representing triadic relationships.
In this depiction of the social-cognitive theory of joint attention, the child turns to attend to the object because he realises that the adult has communicative intent.
In this depiction of the parallel and distributed processing model of joint attention, the child practices representing triadic relations by engaging in joint attention.
Joint attention is an umbrella term for two broad types of behaviour that develop relatively independently of one another and may rely on different neural substrates: response to joint attention (RJA–gaze or point following; see Figure 4) and initiation of joint attention (IJA–directing others’ attention with eye movements or gestures; see Figure 5). IJA can be further subdivided into gestural indication and gaze alternation. Theorists speculate that IJA may be more of a core difficulty than RJA in autism[13,14]. As will be discussed, recent researchsuggests that both RJA and IJA may be core difficulties in autism.
In this depiction of response to joint attention, the child is not attending to the model (a), exhibits dyadic orienting (b), views the model pointing to the object (c), responds to joint attention cues (d) and thus gains access to a link between a word and its referent (e). While RJA often occurs in contexts unrelated to word learning, it is a powerful tool for learning what others are referring to.
In this depiction of initiation to joint attention, the child is not attending to the model (a), exhibits dyadic orienting (b), points to the object (c), directs the model’s attention (d) and thus gains access to a link between a word and its referent (e). While IJA often occurs in contexts unrelated to word learning, it is a powerful tool for eliciting information about objects.
The studies discussed in this review have been conducted in accordance with the Declaration of Helsinki (1964) and the protocols of these studies have been approved by the relevant ethics committees related to the institution in which they were performed. All human subjects, in these referenced studies, gave informed consent to participate in these studies.
RJA may rely on posterior and temporal attentional networks that develop earlier than the anterior networks believed to subserve IJA[13,14]. While RJA may reflect relatively involuntary orienting in response to social cues, IJA may index more voluntary sharing of experiences. Indeed, RJA was associated with intention understanding among 3-year-olds with ASD. Interestingly, gaze alternation was associated with speed of attention disengagement, indicating contributions of non-social attention to IJA in ASD.
RJA and IJA exhibit dissociable developmental trajectories among typically developing infants;despite within-domaincontinuity, associations between IJA and RJA are infrequently observed. RJA is easier to assess than IJA because it involves presenting a stimuli rather than observing child-initiated behaviours. As discussed above, autism may involve a more fundamental disturbance in IJA than RJA[13,14]. Evidence supporting the idea that IJA is more of acore deficit than RJA in ASD includes: only IJA may be uniquely human (or capable of supporting the complexities of social interaction that are often impaired in ASD), RJA may only be atypical early in development while IJA remains atypical across the lifespan in ASD, IJA may be more closely tied to reward processing than RJA, and IJA may be more predictive of social outcomes than RJA. However, IJA is not uniquely human. As will be discussed, RJA may be impaired across the lifespan in autism, there is limited evidence of stronger links between IJA and reward processing in ASD, and RJA is also associated with social outcomes.
Joint attention is often assessed in a laboratory using structured observational measures designed for children up to around 30 months of developmental level. Ideally, joint attention would be assessed in multiple settings. In one of the most commonly used assessments of joint attention,the Early Social Communication Scales (ESCS), IJA is coded into mutually exclusive categories: gestural indication or gaze alternation. The ESCS distinguishes between RJA proximal (to a book) and RJA distal (to posters). Little research focuses on RJA proximal. The ESCS does not vary the subtlety of cues for RJA distal–children receive overlapping verbal, postural and gestural cues.
Congruent with the finding that children with ASD may have particular difficulty responding to sparser cues, other measures include varying combinations of cues. The Autism Diagnostic Observation Schedule-2 includes opportunities for RJA and IJA. Opportunities for RJA are presented within a hierarchy of increasing prompts. Parent-checklists or a combination of parent-report and direct observation are also used.
Researchers recently developed age-appropriate measures of joint attention for older individuals. Hobson and colleagues used shared looks as a measure of IJA among older children and adolescents with ASD. A measure of RJA appropriate for children and adolescents from 7 to 17 years of age uses six naturalistic prompts.
Retrospective video analyses revealing that decreased dyadic behavioursin the first year of life preceded reduced joint attention in the second year in autism suggest that less attention to people in infancy may be a precursor of reduced joint attention. However, reduced IJA between 6 and 12 months of age in ASD with no dyadic atypicalities has also been documented.
Indeed, dyadic atypicalities during live interaction have not been observed in the first year of life in prospective studies of the infant siblings of children with autism. Six-month-old infants with ASD exhibited marginally significantly more attention to a face relative to typically developing infants followed by reduced RJA and gestural indication, but not gaze alternation, at 12 months7. Decreased RJA at 14, 15 and 24 months of age predicted ASD[8,9]. Infant siblings who exhibited severe enough symptoms to be diagnosed with ASD at 14 months showed reduced RJA, gaze alternation and gestures at 14 months. Infants who were diagnosed later with ASD only differed in gaze alternation at 14 months. Gaze alternation also discriminated between children with ASD, intellectually disabled and typically developing children better than other joint attention measures, though gestural indication and RJA were also impaired among the children with ASD. Thus, gaze alternation after infancy may be a more powerful predictor of ASD than RJA or gestural indication.
Despite the absence of documented predictive relations between dyadic interactions and later joint attention, concurrent associations between the two have been observed. Three-to-four-year olds with autism oriented less to social and non-social stimuli (particularly social) and exhibited decreased concurrent IJA and RJA. Reduced social orienting wasconcurrently associated with RJA and IJA in children with ASD[24,25,26]. These studies suggest that dyadic atypicalities may contribute to reduced joint attention. However, predictive relations between social orienting and joint attention have not been documented in ASD and joint attention impairments occur in the absence of dyadic impairments[5,23].
Verbal and cognitive skills may also influence joint attention, particularly RJA. Reduced RJA may not be observed among children with ASD who have a non-verbal mental age above 19 months, a verbal mental age above 47 months or a non-verbal IQ in the normal range[26,27,28]. Leekamand colleagues suggest that the absence of RJA deficits in children with higher verbal skills might be evidence of bidirectional relations between joint attention and language in ASD.
This circumscribed developmental period within which RJA deficits may be apparent in ASD contrasts with the difficulty even high-functioning people on the spectrum have initiating joint attention[13,14]. Indeed, the first module of the Autism Diagnostic Observation Schedule includes both RJA and IJA in its diagnostic criteria while the second module, designed for children with better verbal skills, includes only IJA.
Despite changes in joint attention with development, assertions that only IJA remains a difficulty across the lifespan in autism may be attributable to the fact that most assessments of joint attention were designed for toddlers. A measure of IJA designed for older individuals revealed IJA impairments in older children and adolescents with ASD. Similarly, anage-appropriate measure of RJA disclosed RJA impairments among older children and adolescents with ASD. Thus, both RJA and IJA may remain problematic across the lifespan in ASD when assessed with developmentally appropriate measures.
Not only is joint attention concurrently related to other domains, it is also predictive of development. RJA and IJA between 12 and 18 months were associated with social responsiveness at 3 years for children with ASD. Gaze alternation at 2 years was associated with social symptoms at 4 years among children with ASD. Social-cognitive researchers expect joint attention to be associated with subsequent theory of mind in autism because it has been associated with theory of mind in typical development. However, prospective relations between joint attention and theory of mind in ASD have yet to be established.
IJA in early childhood was associated with later peer engagement. Longitudinal relations between RJA in early childhood and cognitive development have been documented in ASD. RJA in early childhood was associated with adult social functioning, social symptoms and non-verbal communication. Associations between RJA and adult outcomes often appeared to be due to relations between RJA and changes in language and cognitive skills. This is consistent with a view of joint attention as a pivotal skill for learning other skills.
Joint attention may be particularly important for acquiring relevant social cues to map words to objects. Prospective relations between both RJA and IJA and structural language skills are commonly observed among children with ASD. Among children with ASD, RJA at  months was associated with language and cognitive skills between 30 and 36 months of age while gaze alternation at 20 months was associated with vocabulary at 42 months[8,30]. IJA and RJA at 4 years of age were associated with language skills a year later among children with ASD. However, RJA (but not IJA) at approximately 4 years of age was associated with expressive language approximately 8 years later. Predictive associations between gestural indication and syntactic development have also been observed among school-age children with ASD. Future research should assess joint attention and a range of outcomes at multiple points across development to examine potential bidirectional associations. Given strong associations between joint attention and language, it is important to control for linguistic ability when relating joint attention to other domains.
It is likely that RJA and IJA are subserved by unique but overlapping mechanisms. Potential mechanisms underlying atypical joint attention in autism include: atypical reflexive gaze following, but see, impaired integration of joint attention and affect, decreased recognition of the referential significance of gaze, decreased social motivation and recognition of the reward value of social interaction[4,36] or atypicalities of non-social attention. Recent evidence suggests that reflexive gaze following may contribute to RJA impairments in ASD while understanding of referential intent is more likely tied to developmental level.
As discussed, there is mixed evidence that decreased social motivation in infancy leads to joint attention impairments. While dyadic orienting has been concurrently related to both RJA and IJA[25,26], deficits in joint attention are often not preceded by dyadic difficulties[7,22]. Indeed, it is possible that dyadic difficulties are concurrently associated with joint attention impairments because joint attention impairments lead to dyadic difficulties rather than the reverse. Social behaviours typically become increasingly object-centred around 1 year of age. Difficulty with triadic representation could reduce opportunities for social learning and lead to dyadic difficulties. The absence of strong evidence for predictive relations between dyadic skills in infancy and joint attention is more consistent with the PDP model than the social-cognitive theory of joint attention.
Consistent with the PDP model, non-social attention may contribute to joint attention in ASD. Both social and non-social orienting is impaired in ASD. A non-social measure of attention, disengagement latency, was associated with IJA. Among typically developing infants, visual attention is related to subsequent IQ and language skills. Future research should assess non-social aspects of attention and joint attention at multiple points across development, as well as measures of social-cognition, in order to determine if there are developmental relations between non-social attention and joint attention, and the relative contributions of each to subsequent development.
Future research should assess reward sensitivity and joint attention from infancy into early childhood in order to determine if reward sensitivity contributes differentially to the development of RJA and IJA. While neuro imaging research in typical development suggests that IJA may be more strongly associated with regions of the brain subserving reward processing, an executive function measure believed to index reward processing was associated with a composite measure of both IJA and RJA among children with ASD.
This review suggests that both RJA and IJA are central to development in ASD. Little evidence that IJA is more of a core deficit was observed: RJA and IJA atypicalities are apparent among older individuals, there is limited evidence that reward processing contributes more to IJA than RJA, and both IJA and RJA predict social development. The lack of strong evidence that dyadic atypicalities precede triadic ones is less consistent with the social-cognitive model than the PDP model of joint attention.
Because joint attention provides a tool for learning from others, change in joint attention across time may be as informative as joint attention at specific points in time. Change implies that something is influencing change. Thus, assessing joint attention and other developmental domains at multiple points across development is crucial for understanding the mechanisms relating change in joint attention to other domains.
Joint attention predicts outcomes across a range of domains in autism and is inexpensive and relatively easy to assess. Given the strong predictive power of a behavioural measure such as joint attention, brain imaging and eye-tracking measures developed to predict outcomes in ASD should be compared with assessments of joint attention in order to demonstrate that they are more effective and thus worth their additional costs, including assessment difficulties commonly associated with eye-tracking or brain imaging such as data loss due to difficulty calibrating or movement.
Throughout this review, I have discussed joint attention as a core difficulty in ASD. It is important to note that atypicalities of joint attention may emerge from strengths. Future research should examine relations between covert attention, motor skills and joint attention in autism. We should also think creatively about ways that people on the spectrum may engage in joint attention that are not captured by current measurement systems.
All authors contributed to the conception, design, and preparation of the manuscript, as well as read and approved the final manuscript.
All authors abide by the Association for Medical Ethics (AME) ethical rules of disclosure.