Abstract

A recent study conducted by a research team in Rehabilitation Science at the Shanghai Institute of Artificial Intelligence for Education at East China Normal University found that autistic children showed the same amount of social attention to faces and toys during social play as typically developing children. During play, whether the child has autism or not, their gaze was mostly focused on the toys (60%–80%) and only for a small amount of time did they look at adults’ faces (6%–14%) (Figure 1).

The area of interest of social gaze across typically developing, autistic, and developmental delay children.
Autism is a neurodevelopmental disorder that affects social communication (American Psychiatric Association, 2013). One key characteristic in clinical diagnosis is that children rarely look at others’ faces, which has become a significant behavioral marker to identify autism. In fact, typically developing children also seldom make direct eye contact during everyday play interactions. Does this mean that our understanding of autism symptoms might be biased?
The idea that children with autism avoid eye contact has been around for a long time. In 1943, Leo Kanner first described autistic children as those who “never looked at any person's face” (Kanner, 1943). This observation was interpreted as a lack of interest in social interaction and an inability to share attention with others.
Joint attention is an important skill for children's social, communication, and language development, considered as a fundamental skill for language and communication (Mundy & Newell, 2007). The Early Social Communication Scales (ESCS) or the Autism Diagnostic Observation Schedule (ADOS) are often used in the clinical setting to measure children's joint attention skills. These tests involve presenting stimuli (often toys) to children and observing their gaze patterns to assess their ability to follow and initiate joint attention. However, these tests are usually conducted in standardized lab environments and may not fully reflect children's performance in natural settings.
The research team in Rehabilitation Science at the Shanghai Institute of Artificial Intelligence for Education built an experimental environment friendly to both children and artificial intelligence technology, which addressed issues such as “overly artificial experimental environments leading to unnatural children's behavior” and “the complexity of natural behavior making it difficult for AI to recognize.” The team established a non-intrusive, multimodal behavior observation laboratory (Figure 2) and designed two observation scenarios that were tailored to the core symptoms of autism: “Intelligent Observation of Social Communication in Children With Autism” and “Parent-Child Free Play.” The aim was to collect key behavioral data such as gaze, movements, vocalizations, and facial expressions during natural social interactions. The study aimed to deconstruct the core symptoms of autistic children using artificial intelligence and identify behavior indicators for autism identification. Written informed consent was obtained from the caregivers of all participating children.

The multimodal behavior observation laboratory.
Meanwhile, a study published in Current Biology reported similar findings (Yurkovic-Harding et al., 2021, 2022). Yurkovic-Harding et al. (2021) used head-mounted eye trackers to record gaze patterns of autistic and typically developing children during natural interactions with their parents. The study found that both autistic and typically developing children rarely looked at their parents’ faces during play, with only 1% of the time looking at their parents’ faces. Interestingly, the study found that both groups of children spent a significant amount of time looking at their parents’ hands. This finding suggests that children may use communication cues other than eye contact to engage in joint attention.
The results of the two studies have important implications for understanding joint attention in children with autism. Firstly, the hypothesis that autistic children avoid eye contact may have been exaggerated. While autistic children may look at their parents’ faces less often than typically developing children, this difference is not as pronounced as previously thought. Secondly, the findings suggest that children may use other communication cues to participate in joint attention. For example, parents’ hand gestures may be more important communication cues than their gaze in the toy play.
These findings have significant implications for clinical practice as well. Many interventions for autistic children focus on improving joint attention and encouraging eye contact. However, the results of this study suggest that these interventions may not be suitable for all autistic children. Instead, interventions focusing on other joint attention cues, such as parents’ hand positions (communicative gestures), may be more effective.
The new findings by the research team at East China Normal University and Yurkovic-Harding et al. (2022) reshape our understanding of joint attention in children with autism. It indicates that our understanding of the core symptoms of autism needs continuous updating, especially in the new era of artificial intelligence. Classical research paradigms need to be re-examined. We need to continue exploring and researching to understand children's natural behaviors in social communication in a natural environment, thereby providing appropriate and effective support for autistic children.
Takeaway message
Recent findings from the Shanghai Institute of Artificial Intelligence for Education reveal that autistic children exhibit social attention to adult's face and toys similar to typically developing peers during play, with a primary focus on toys over faces. This result challenges the longstanding belief that autistic children inherently avoid eye contact, emphasizing the need to reconsider assumptions about their social behaviors. Both autistic and typically developing children were found to spend minimal time looking at the adult's face during toy play, echoing similar findings published in Current Biology. These insights suggest that interventions prioritizing eye contact alone may not address the diverse needs across the autism spectrum. Instead, focusing on broader communication cues, such as gestures, may be more impactful.
Footnotes
Contributorship
Lu Qu and Qiaoyun Liu contributed to the design and implementation of the research, the analysis of the results, and the writing of the manuscript.
Declaration of conflicting interests
The authors declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.
Ethical statement
The study protocol was approved by the Institutional Review Board of Human Research at the East China Normal University (HR706-2022, HR374-2024). Written informed consent was obtained from the caregivers of all participating children.
Funding
The authors disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This work was supported by the China Postdoctoral Science Foundation (No. 2023M731103), Shanghai Municipal Education Commission Research and Innovation Program (2023SKZD07), and National Key Research and Development Program (2022YFC2705201).
