Minimally speaking autistic preschoolers imitate less often and less accurately than both speaking autistic children and typically developing peers, and they spend less time watching the very action they are being shown, according to a new eye-tracking study of 102 children published in the Journal of Autism and Developmental Disorders. The research, led by Gioia Cavalli of the A.J. Drexel Autism Institute with colleagues at the UC Davis MIND Institute and Drexel University, also found that the emotional tone of the demonstrator matters for everyone: when an adult performed an action with playful, positive affect, children across all three groups looked longer at the demonstrator’s face and copied the action more accurately than when the same demonstrator performed it with a neutral expression.
The study addressed a persistent gap in the imitation literature. Imitation is associated with adaptive functioning and spoken language development in both typical and autistic development, yet comparatively little is known about imitation skills and the processes underlying them in minimally speaking autistic preschoolers, a subgroup defined here as children with an Expressive Communication age equivalent of 14 months or below. By stratifying the sample by spoken language level, the team aimed to clarify why findings on imitation in autism have been so heterogeneous.
The participants were 102 preschool-aged children with an average age of about 32 months, divided into three groups: 44 minimally speaking autistic children, 43 speaking autistic children, and 15 typically developing children. Autism diagnoses were verified with the Autism Diagnostic Observation Schedule, Second Edition, administered by research-reliable clinicians for most autistic participants, with the remainder assessed by service providers within the previous six months. Minimally speaking status was defined using established benchmarks based on the Vineland Adaptive Behavior Scales, Third Edition, or, for six children, the Mullen Scales of Early Learning. The minimally speaking group was similar in chronological age to the typically developing group but younger than the speaking autistic group, and it scored lower on adaptive and cognitive measures.
The experimental task was deliberately designed to minimize language demands. Each child sat roughly 60 centimeters from an eye-tracking monitor and watched short videos in which an unfamiliar adult demonstrator performed one of two object-oriented actions: scribbling with a marker or making five dots on paper with a dot marker. Children passively viewed the demonstrations and were never required to respond to verbal instructions. Each action was shown twice by the same demonstrator under two conditions: a playful condition in which he displayed positive affect, and a neutral condition in which his facial expression remained neutral. Playful and neutral videos alternated in an order counterbalanced across participants, each preceded by a fixation cross and interspersed with child-friendly filler videos.
After each video, the child received the same materials used by the demonstrator and was simply encouraged to play freely, with no specific imitation prompt. This allowed the researchers to examine spontaneous imitation rather than instructed copying. Eye-tracking data were collected with a Tobii Pro Spectrum system, letting children move freely without head-mounted gear. Gaze was extracted from two predefined areas of interest, the demonstrator’s face and the area containing the action and relevant objects, and summarized as total fixation duration, the cumulative time gaze remained within each area. Children’s subsequent behavior was video-recorded and coded offline for imitation frequency, whether any imitative response occurred, and imitation accuracy, scored from 0 to 3 for how closely the child’s behavior matched the demonstration. Coding was performed by an observer blind to group membership and hypotheses, with high inter-rater agreement on 20 percent of videos.
The attention results showed a clear affective modulation. Across groups, children looked longer at the demonstrator’s face in the playful condition and longer at the action in the neutral condition, with no group differences in face attention. For action attention, however, group mattered: minimally speaking autistic children showed significantly lower total fixation duration on the demonstrator’s action than speaking autistic children and typically developing peers, whose attention patterns did not differ from each other. Age was included as a covariate in this analysis because attention to the action was modestly associated with age. In other words, the minimally speaking group’s reduced imitation was accompanied by reduced attention to the task-relevant information itself, not by a global difference in looking at the demonstrator.
Imitation performance followed the same pattern. In the neutral condition, 40 percent of minimally speaking autistic children imitated, compared with 69 percent of speaking autistic children and 75 percent of typically developing children. In the playful condition the figures were 46, 71, and 77 percent respectively. Imitation accuracy was higher in the playful than the neutral condition across all groups, and minimally speaking children showed lower accuracy than both comparison groups, while speaking autistic children performed comparably to typically developing peers. Notably, the effect of the playful condition on accuracy was consistent across groups, with no significant condition-by-group interaction, suggesting that affective cues influence imitation similarly regardless of diagnostic status or language level.
The study also linked attention and imitation directly. Partial correlation analyses controlling for group found that, in the playful condition, greater visual attention to the demonstrator’s action was associated with higher imitation accuracy, independent of diagnosis. No such association emerged in the neutral condition. The authors interpret this as evidence that affective modulation may enhance the functional coupling between attention and imitation: a smiling, playful demonstrator may not only draw children’s gaze but also strengthen the connection between watching an action and reproducing it. This was somewhat unexpected given prior suggestions of reduced sensitivity to social cues in autism, though it aligns with earlier work indicating that emotional signals such as smiles modulate imitation in autistic children.
The finding that speaking autistic children matched typically developing peers on both attention and imitation, despite lower cognitive and adaptive scores, points to relatively preserved imitative abilities in that subgroup and underscores the importance of stratifying by spoken language when studying imitation in autism. The authors caution, however, that the task placed limited demands on broader cognitive and adaptive abilities, and that imitation differences in autistic children are more likely to emerge as the motor complexity of the modeled actions increases. They also note that cognitive functioning was not treated as a statistical covariate, because the team considered broader developmental differences intrinsic to the minimally speaking profile rather than an extraneous confound; consequently, the study cannot determine whether the observed differences are specifically tied to minimally speaking status or to broader developmental differences characterizing this subgroup.
Several further limitations shape how far these results generalize. The typically developing group was small, which may limit statistical power, and the sample was predominantly male. The typically developing children were not systematically screened for conditions such as ADHD, anxiety, or depression, and no autism-specific screening was conducted for that group. Eye-tracking captures visual attention but not the full range of cognitive processes behind imitation, and the task tested only object-directed actions in a controlled setting with a video-recorded, unfamiliar adult demonstrator viewed from a perspective different from the child’s own, which may not reflect imitation in response to a live, familiar social partner. Even so, the authors suggest a practical implication: interventions targeting imitation in minimally speaking autistic children may benefit from explicitly supporting attention to action-relevant information while using affectively engaging contexts to support learning.
Subject of Research: Spontaneous imitation and visual attention in minimally speaking autistic preschoolers under playful and neutral demonstration conditions
Article Title: Spontaneous Imitation and Affective Modulation in Minimally Speaking Autistic Preschoolers
Article References: Cavalli, G., Paolizzi, E., Talbott, M., Stahmer, A. C., Lee, N. R., Orsili, L., Robins, D. L., & Vivanti, G. (2026). Spontaneous Imitation and Affective Modulation in Minimally Speaking Autistic Preschoolers. Journal of Autism and Developmental Disorders. https://doi.org/10.1007/s10803-026-07586-9
Image Credits: AI Generated
DOI: 10.1007/s10803-026-07586-9
Keywords: autism, imitation, eye-tracking, preschoolers, minimally speaking, visual attention, affective modulation, spoken language, child development, social cognition, Spontaneous, Affective
News Source: Ophelia Keating. (October 29, 2026). Playful Demonstrators Boost Imitation in Minimally Speaking Autistic Preschoolers. Scienmag.
