Language learning rates in autistic children decline exponentially after age two

A new study published in the Journal of Autism and Developmental Disorders suggests that the ability to comprehend complex language relies on a developmental window that may close significantly earlier for autistic children than for their neurotypical peers. The findings suggest that while autistic children initially acquire the cognitive skills necessary for language comprehension at a typical rate, their learning pace tends to slow down exponentially starting as early as two years of age.

Complex language comprehension involves more than just understanding individual words. It requires a cognitive function known as Prefrontal Synthesis. This is the ability to mentally combine separate objects or concepts into a novel image or scene. This skill allows a person to understand the difference between “the dog bit the boy” and “the boy bit the dog.”

Without Prefrontal Synthesis, individuals may struggle to understand sentences that rely on word order, spatial prepositions, or recursive structures. This specific deficit is common among individuals with autism spectrum disorder. Estimates suggest that between 30 and 40 percent of individuals with autism experience significant challenges with this type of mental synthesis.

“A long-standing question in autism research concerns why some individuals never fully acquire syntactic comprehension, a limitation that often leads to lifelong difficulties with independent living and employment,” explained study author Andrey Vyshedskiy, a lecturer at Boston, University and author of The Evolution of Language: How the Brain Evolved Syntactic Language from Early Mammals to Homo sapiens.

“Two competing explanations have been proposed. One suggests a persistent, lifelong barrier that consistently slows learning at every age. The other proposes that there is a critical period during which the brain is especially capable of acquiring the neurocognitive mechanisms needed for syntactic comprehension, and that this critical period may be shorter in autism.”

“Despite decades of discussion, there had been little large-scale longitudinal evidence capable of distinguishing between these two possibilities. Most prior studies were small, cross-sectional, or focused on older children, thereby missing the earliest developmental window.”

“Our motivation was to use a very large, real-world longitudinal dataset to directly test which developmental trajectory better fits observed learning patterns,” Vyshedskiy continued. “Understanding this distinction has important practical implications, because it speaks directly to when intervention should occur and how urgently early support may be needed.”

The study utilized data collected through a language therapy application available on major app stores. Parents used the app to track their child’s progress over time. The final sample consisted of 15,183 autistic individuals and 138 neurotypical individuals. The participants ranged in age from 2 to 22 years. Inclusion in the study required that caregivers completed at least three separate evaluations over a period of six months or longer.

The primary measure used was the Mental Synthesis Evaluation Checklist. This is a parent-reported assessment designed to evaluate a child’s Prefrontal Synthesis ability. The checklist includes 20 items that ask about the child’s ability to understand stories, engage in pretend play, and comprehend spatial prepositions.

The researchers used these repeated assessments to calculate a “learning rate” for each participant. This rate represented the change in their score over time. By looking at how this rate changed as the children grew older, the team could model the developmental trajectory for both groups.

The analysis revealed distinct patterns between the autistic and neurotypical groups. At two years of age, both groups exhibited similar learning rates. Autistic children showed a learning rate of approximately 5.9 points per year, while neurotypical children showed a rate of 6.1 points per year.

This finding indicates that at the earliest age measured, autistic children were acquiring these cognitive skills at a pace comparable to typically developing children. However, the trajectories diverged significantly as age increased. For the neurotypical group, the learning rate remained high and relatively constant until about seven years of age.

In contrast, the learning rate for the autistic group began to decline much earlier. The data showed an exponential decrease in the rate of learning that started shortly after age two. As the children grew older, their annual gains in Prefrontal Synthesis became progressively smaller.

“The main takeaway is simple but important: timing matters enormously for syntactic comprehension development in autism,” Vyshedskiy told PsyPost. “We found that autistic children initially learn key comprehension-related skills at roughly the same rate as neurotypical children. However, for many autistic children, this learning rate begins to decline much earlier—sometimes as early as 1 to 3 years of age. After this critical-slowing-down-age, gains become progressively harder.”

The researchers further broke down the autistic cohort by severity levels. The researchers found that the timing of the decline correlated strongly with the severity of the autism diagnosis. The point at which the learning rate began to drop—referred to as the critical inflection point—occurred earliest in those with the most severe symptoms.

For children with severe autism, the learning rate began to slow down significantly around 1.4 years of age. For those with moderate autism, this inflection point occurred around 2.0 years of age. Children with mild autism maintained their initial learning rate for longer, with the decline beginning around 3.2 years of age.

“The effects are substantial, not subtle,” Vyshedskiy explained. “The difference between developmental trajectories is large enough to explain why some individuals eventually acquire syntactic comprehension abilities while others do not.”

“In our modeling, neurotypical children maintain a high syntactic learning rate until at least age 7. Autistic children show an earlier decline, with the timing depending on severity. Children with more severe autism exhibit this learning rate decline earliest (that is, they have the shortest critical-slowing-down-age).”

“Practically speaking, this means that the same amount of instruction delivered at ages 2 and 4 can have very different effects,” Vyshedskiy said. “Early intervention is not just helpful—it may be developmentally decisive.”

“Parents are often dismissive of their toddler’s delay in language comprehension, assuming the child will ‘catch up’ on his/her own. This assumption can be risky. A lack of syntactic comprehension can often be corrected at age two through additional syntactic conversations, reading fairy tales aloud, and engaging in imaginative play. The same interventions, when introduced at age four, may no longer support the acquisition of syntactic comprehension.”

A potential misinterpretation of these findings is that learning becomes impossible after these ages. The study identifies a slowing of the rate of acquisition, not necessarily a total cessation of learning. However, it implies that the effort required to make gains increases substantially after the critical period begins to close.

As with all research, there are some limitations to consider. The data relies on parent reports rather than direct clinician observation. While the assessment tool has demonstrated high internal reliability, parent reports can be subject to bias or wishful thinking.

Further investigation is also needed to explore the specific neural mechanisms involved. Understanding exactly why the critical period might close earlier could open new avenues for medical or therapeutic interventions. Researchers might also investigate whether specific types of intensive therapy can extend this window.

The implications of these findings are significant for clinical practice. Currently, many children are not diagnosed with autism until nearly four years of age. By this time, according to this data, the learning rate for Prefrontal Synthesis may have already begun to decline significantly.

This suggests that early intervention is a matter of urgency. Waiting for a formal diagnosis before beginning language therapy may result in missing the most fertile window for developing syntactic comprehension.

“Many readers are surprised by the age-dependent nature of syntactic comprehension acquisition, although this should not be unexpected,” Vyshedskiy told PsyPost. “Consider pronunciation as an example: unless a person is raised in a French-speaking environment, acquiring native-like French pronunciation after approximately five years of age is exceedingly difficult. Early exposure and practice are essential for accent-free speech.”

“Similarly, maturation of the networks underlying syntactic comprehension requires sustained engagement in syntactic dialogue, imaginative play, and storytelling. As a result, the critical-period constraints governing the development of syntactic comprehension are likely as strong as those governing accent learning. In most children, the most sensitive period for acquiring syntactic comprehension occurs before age five, although its timing varies across individuals, much like height, weight, and other physiological traits.”

“Autistic children are often disadvantaged on two fronts: first, by reduced spontaneous engagement in conversations, joint attention, and imaginative play; and second, by an abbreviated critical period for syntactic comprehension acquisition,” Vyshedskiy said. “Together, these factors constrain the natural maturation of syntactic networks and help explain why autistic children often derive the greatest benefit from early, intensive language intervention.”

Overall, this research offers a detailed look at the developmental timing of a crucial cognitive skill. It shifts the focus from simply looking at what skills are missing to understanding when the opportunity to learn them is most active. This temporal perspective could help refine how and when therapies are delivered to autistic children.

“About 20 years ago, I posed a critical question: given that many autistic children show limited interest in conversations, imaginative play, and story listening, could a structured set of syntactic exercises effectively complement naturalistic syntactic exposure?” Vyshedskiy added. “Furthermore, if such an exercise set could be developed, would it be possible to deliver this form of brain training during the peak sensitive period for syntactic comprehension development, between the ages of two and four years?”

“To enable early access to therapy, we have gamified language therapy and made it easy for parents to administer by packaging all syntactic exercises in one app: Mental Imagery Therapy for Autism (MITA). In a 3-year observational clinical study of 6,454 children with ASD, children who engaged with the MITA syntactic comprehension intervention showed 2.2-fold greater language improvement than children matched by initial evaluations. This difference was statistically significant (p<0.0001).”

“The study results have been published in the journal Healthcare. Based on this study results, the Food and Drug Administration (FDA) has granted the MITA language therapy intervention Breakthrough Device designation status,” Vyshedskiy said.

“One important point is that this work was only possible because thousands of families contributed longitudinal data over many years. The large number of participants—over 15,000— allowed us to identify developmental patterns that are invisible in traditional small studies.”

“More broadly, the study reframes language development not just as vocabulary learning, but as the maturation of syntactic comprehension. Understanding how and when this ability develops may help bridge neuroscience, linguistics, and clinical practice in a more unified way.”

The study, “Age-Dependent Process Governs Executive Function Disability in Autistic Children,” was authored by Andrey Vyshedskiy, Allegra Marsiglio, Sahil Batham, Alessandro Tagliavia, Rohan Venkatesh, Anel Tarakbay, Sagar Mundhia, Samarth Urs, Edward Khokhlovich, and Eugene Pinsky.