Theory of Mind Carried the Difference

What the research found

Pragmatic language is the practical side of communication. It is the part that lets us read a metaphor, catch a hint, or fill in what a speaker means but does not say outright. A child may have a rich vocabulary and correct grammar and still struggle here. Difficulties with pragmatic language are a well-documented feature of autism, and this study set out to ask a harder question: what underlying skills explain them?

The researchers tested 143 children and adolescents aged 8 to 18 in Italy. Seventy-three had a diagnosis of autism; seventy were typically developing. The two groups were matched carefully. They did not differ in age, in IQ, or in receptive grammar. Every child in the autism group had an IQ of 80 or above, so this was a sample without intellectual disability. That matching matters, because it means any difference the study found could not simply be put down to the autistic children being younger or having lower general ability.

On the pragmatic language tasks, the autistic group scored lower across the board. This held for both parts of the assessment: understanding metaphors and drawing inferences from short stories and pictures. The gaps were moderate rather than dramatic. In statistical terms the effect sizes ran from about 0.51 to 0.63, which describes a real and consistent difference between groups, not a chasm. The autistic children were also weaker on the two skills the researchers suspected might sit underneath pragmatic language: theory of mind (the capacity to reason about what other people think, feel and intend) and executive function (the mental control system for planning, holding information in mind, and shifting between tasks).

The central analysis then asked which of those two skills actually accounted for the pragmatic language gap. The answer was clean. When the researchers built a model that controlled for group, age, gender and IQ, and then added executive function, executive function explained almost nothing (it added 0.2 per cent to the variance and was not significant). Theory of mind, added next, explained a further 17 per cent and was a strong predictor. In the final model, only age and theory of mind still mattered. Strikingly, once theory of mind was in the equation, the difference between the autistic and typically developing groups was no longer statistically significant on its own.

A formal test of mediation told the same story. The path running from group membership to pragmatic language through theory of mind was significant (an indirect effect of β = −0.19, p = 0.001). The parallel path through executive function was not (β = −0.035, p = 0.297). Put plainly: the pragmatic language difficulties seen in the autistic group appeared to travel largely through theory of mind, not through executive function. One further detail is worth noting. Executive function correlated with pragmatic language in the typically developing children but not in the autistic children, a hint that the two skills may relate differently depending on the child.

What it means for you and your child

The useful takeaway is a shift in where to look. If a bright autistic child, with strong vocabulary and sound grammar, keeps missing the point of a joke, taking an idiom literally, or losing the thread of what a friend really meant, this study suggests the sticking point is more likely to be social reasoning than a problem of attention or mental organisation. That reframes the difficulty. It is less “my child cannot concentrate” and more “my child finds it hard to work out what the other person is thinking and therefore what they mean.”

The authors draw a modest and sensible clinical implication from this. Support aimed only at pragmatic language, drilling metaphors or conversation rules in isolation, may miss the foundation. Building the capacity to consider other people’s perspectives, beliefs and intentions may do more of the real work. That does not mean theory of mind training is a proven fix. The authors are careful to note that the wider evidence on such training is mixed: some programmes improve social interaction, others improve theory of mind on tests without that carrying over into everyday conversation. This paper identifies a promising target. It does not demonstrate a treatment.

It is also worth being clear about what a single study of this design can and cannot show. This was a snapshot taken at one point in time. The word “mediation” carries a whiff of cause and effect, but a cross-sectional study cannot establish that weaker theory of mind causes weaker pragmatic language. The two travel together, and theory of mind is the better statistical explanation of the two candidates tested, but the direction of the arrow is not proven here. Age, which improved pragmatic language in both groups, was tangled up in the analysis in ways the authors themselves flag as a limitation.

Two other cautions temper how far this should be generalised. The sample was overwhelmingly male, 65 boys to 8 girls in the autistic group, so it says little about how these skills relate in autistic girls, who often present differently. And every participant had an IQ in the typical range. The findings should not be stretched to children with intellectual disability or co-occurring conditions, whom this study deliberately excluded. Within those bounds, the message is a helpful one: for a verbally able autistic child, the road into better communication may run through understanding other minds, and that is worth attending to alongside language itself.