Autistic Children Imagine Movement More Slowly, Not Less Well

What the hand task shows

Shown a picture of a hand at an odd angle and asked whether it is a left hand or a right one, most children answer by imagining their own hand turning until it matches. The timing gives the strategy away, because answers arrive later for angles a real wrist could not comfortably reach. Autistic children produce the same pattern, which indicates that they too use their own body as the reference rather than the picture as an abstract shape. Their answers simply take longer to come.

Motor imagery is the term for imagining a movement without making it. Researchers cannot observe it directly, so they infer it from that timing pattern, called the biomechanical effect, which is taken as the signature of an imagined body rather than an imagined object.

Emma Gowen and Eve Edmonds, with Ellen Poliakoff at the University of Manchester, assembled that evidence in a systematic review published in February 2024 in Frontiers in Integrative Neuroscience, searching six databases for work appearing before September 2023, screening 220 papers, reading 30 in full and retaining sixteen studies across fourteen articles. Nothing in the review is pooled; each figure it reports comes from a single small study.

Chen and colleagues, in 2018, found no difference in accuracy between autistic and non-autistic children and adolescents and a clear biomechanical effect in both groups, with slower reaction times in the autistic group. An apparently larger effect in that group disappeared once overall speed was accounted for. The same participants showed no group difference on a rotation task using objects instead of hands, which the review reads as evidence that the slowing concerned the hand rather than mental rotation or motor coordination in general.

Conson and colleagues reported the same combination in 2016, slower responses alongside a biomechanical effect in both groups, and the review reads the two studies together as showing that autistic individuals can perform motor imagery, though the process is more effortful, inefficient or, in the authors’ word, deeper.

Other findings point elsewhere. An earlier study from Conson in 2013 found no slowing and no significant biomechanical effect in the autistic group, whose timing pattern suggested rotation of the image as an object, and performance on a letter rotation task was equivalent across groups. Johansson and colleagues, testing the same children at seven, eight and nine, found the biomechanical effect in the autistic group at the first and third ages but not the second, a gap the review attributes to the difficulty of detecting the effect in so few participants. The hand rotation studies converge on a strategy that is present and effortful; the rest of the literature does not converge at all.

What is not yet known

The studies are very small. In the hand rotation experiment reported by Xie and colleagues in 2022, eight autistic children were compared with eight non-autistic children, and that paper was rated for exclusion on quality assessment and kept in the review anyway. Most included studies enrolled mainly boys, which matters here because females are more likely to take an embodied approach to tasks of this kind. All the hand rotation evidence comes from children and adolescents, and the review is explicit that it cannot be generalised to autistic adults.

The review raises the possibility that in one study the autistic participants were behaving as expected and the non-autistic participants were using a shallower form of imagery, in which case the timing difference records a difference of strategy. Where children were given explicit instructions to imagine a movement, no one asked them afterwards how they had gone about it, so a child who found the task hard cannot be told apart from one who did not attempt it.

Beyond the hand task the picture loosens further. Gowen and colleagues found no difference between autistic and non-autistic adults on a questionnaire measuring either the visual or the physical sensation dimension of imagined movement, and David and colleagues found no group differences in speed or accuracy on an adult perspective taking task. Of the body rotation studies, two found processing differences and a third found none. The review’s own conclusion is that evidence for every task type other than hand rotation was mixed and inconclusive.

Motor imagery training is under investigation in other clinical populations, where patients practise a movement mentally in order to improve its execution. A parent who learns that autistic children imagine movement more slowly might reasonably wonder whether their own child should be doing that practice. The review does not support that step. It offers no link between motor imagery and any measure of coordination in daily life, and no evidence that training the one improves the other, and what the slowness represents remains an open question in the authors’ own account. If a therapy is offered to your child on the strength of this literature, it is being offered ahead of the evidence.