Six studies, five dimensions, one direction of travel
Ziyu Yang, at Cambridge’s Faculty of Education, has published a systematic review asking a question the early-childhood literature has largely skipped: not what does the research find about autistic children’s play, but what does it assume before finding anything. The review searched ERIC, PsycINFO and Scopus for empirical studies of solitary, parallel and pretend play in autistic children in early-childhood education, found six eligible studies, and analysed each across five dimensions: the language used, the intervention’s aim, the measurement standard, the theoretical orientation, and how the child is positioned.
The result is uniform in a way that six heterogeneous studies rarely are. Across every dimension, in every study, autistic play is measured against neurotypical standards — the question is always how closely the child’s play resembles their non-autistic peers’. Pretend play appears as an intervention context in which autistic children are positioned as subjects to be moved toward neurotypical social behaviour. Solitary and parallel play appear consistently as behaviours requiring redirection toward “more socially complex” forms. Engagement with neurodiversity-affirming frameworks, across all six studies and all five analytical dimensions: none.
The review’s central claim is the one that earns it a slot in this week’s coverage: deficit assumptions are embedded in the measurement design, not merely in the language. Softening the vocabulary — swapping “impairment” for “difference” in the write-up — changes nothing, because the verdict was cast when the outcome variable was defined. If “improvement” is operationalised as movement away from solitary play, then a child who plays alone has failed by construction, before a single observation is recorded. The instrument encodes the conclusion.
The measure is the verdict — self-regulation absent from every study
Readers will recognise this argument, because the corpus met it twice in a fortnight from the empirical side. The O’Leary scoping review found that ADHD social-media research could only discover harm because harm was all its instruments measured; the van Asselt piece found a coding scheme that filed community knowledge as misinformation by definition. Yang’s review is the same finding at the sandpit: the play literature could only discover deficient play, because the measures defined good play as looking like somebody else’s.
The single most telling absence: self-regulation appeared in none of the six studies’ outcome measures. Not one. This in a literature about early childhood, where the developmental function of play most strongly supported by the general research base — Yang cites it from the mainstream literature, not the affirming one — is precisely the building of self-regulation. The studies measured social complexity, peer engagement, movement toward interactive pretend play. Whether the play was doing its actual developmental job — regulating a nervous system, restoring capacity, enabling the child to stay in the room — was not on any instrument, in any study, at any point.
That absence is not neutral, because for a monotropic child, solitary deep engagement is not a withdrawal from development. It is where the regulation happens — the same architecture this corpus traced through hyperfocus in adults, present at age four in the corner with the trains. A research programme that codes that corner as the problem to be intervened away is measuring its own assumptions and publishing them as findings about children.
What the studies' own data showed and their frameworks couldn't say
The review’s most interesting move is that it doesn’t rest on critique. Yang re-reads the six studies’ own data and finds a pattern the studies generated but their frameworks could not articulate: solitary and parallel play functioned as enabling conditions for participation, not obstacles to it. Children observed in the primary studies used solitary and parallel play as they used breath — as the regulatory base from which social engagement became possible, and to which they returned to recover from it. The corner with the trains was not the opposite of joining in. It was what made joining in affordable.
And the second unacknowledged pattern: where pretend-play outcomes improved, the peer ecology, not the play form, emerged as the primary driver — who the other children were, how the interaction was structured, what the environment permitted. Which lands squarely on the double empathy problem, cited by Yang as one of the review’s guiding frames: the “deficit” the interventions targeted inside the autistic child was substantially a property of the social configuration around them. Change the ecology and the participation changes, without the child’s play ever needing correction.
Put the two patterns together and the six studies, read against their own grain, describe the corpus’s whole argument in miniature: the behaviour coded as the problem was doing regulatory work; the variable that moved outcomes was environmental; and the instruments were pointed at the child throughout. The data knew. The frameworks didn’t have the vocabulary.
Provenance, and why the argument stands anyway
Now the flag, stated as plainly as the argument, because this corpus doesn’t launder sources. This is a single-author systematic review — six studies, narratively synthesised — published in a minor open-access outlet (the International Journal of Contemporary Education and Teaching Research, High-Tech Science Press, with a Zenodo DOI) rather than a major journal, by an early-career researcher. Single-author reviews cannot demonstrate independent screening; a six-study evidence base is small even for a conceptual review; and the venue provides little of the editorial scrutiny a Autism or JADD publication would. Weight it accordingly.
But weight the argument on its own terms too, because its load-bearing components don’t depend on the venue. The five-dimension framework is transparent and checkable. The six primary studies are published, mainstream, and re-readable by anyone. The claim that self-regulation appears in none of their outcome measures is verifiable by inspection. And the theoretical frames doing the work — the double empathy problem, the enabling function of self-directed engagement — are established literature this corpus has covered from a dozen better-credentialed sources. The review’s contribution is not new evidence; it is a lens correctly applied to evidence that was already public. Lenses can be checked without trusting the lens-grinder.
The practical stakes sit in early-years classrooms now. Play-based assessment and intervention are standard in early childhood provision, and the instruments this review examined are the kind that decide which four-year-olds get flagged, redirected, and enrolled in social-play interventions. Yang’s closing recommendations are modest and right: regulatory outcomes on the instruments, autistic-referenced measures rather than peer-resemblance scores, and autistic children’s own perspectives in the research. Until then, the field will keep producing what it has produced for two decades — evidence about distance from a norm, published as evidence about children — and every “improvement” it records will carry the question this piece’s title asks. Playing less like yourself is a strange definition of getting better.
Citations
Yang, Z. (2026) — Exploring the Assumptions Shaping Autistic Children’s Play Research in Early Childhood Education: A Systematic Review — International Journal of Contemporary Education and Teaching Research
Milton, D. E. M. (2012) — On the ontological status of autism: the ‘double empathy problem’ — Disability & Society
Murray, D., Lesser, M. & Lawson, W. (2005) — Attention, monotropism and the diagnostic criteria for autism — Autism
Kapp, S. K., Gillespie-Lynch, K., Sherman, L. E. & Hutman, T. (2013) — Deficit, difference, or both? Autism and neurodiversity — Developmental Psychology
