Mental Imagery Therapy for Autism (MITA) - an Early Intervention Computerized Language Training Program for Children With Autism

Among primates, humans have exceptionally long and widely distributed fronto-temporal and fronto-parietal tracts (such as the arcuate and superior longitudinal fasciculi), yet the maturation of these pathways depends on early experience. During childhood, these tracts are fine-tuned (largely via myelination and synaptogenesis) through engagement in syntactically rich conversations, imaginative pretend play, and narrative storytelling 5-9. Both biological factors-such as genetically shortened sensitive period for pathways development-and social factors-including limited conversational engagement, low syntactic complexity in daily interactions, lack of storytelling, and excessive passive screen time-can substantially reduce the functional development of these connections and profoundly alter a child's developmental trajectory.

Autistic individuals frequently face a compounding disadvantage from both biological and social factors during early development: a potentially abbreviated critical period, combined with reduced spontaneous engagement in reciprocal conversation, joint attention, and pretend play. These factors limit the experiential scaffolding that typically drives the maturation of fronto-parieto-temporal white-matter tracts.

In a subset of autistic adults these tracts remain structurally and functionally underdeveloped. This compromises the executive function of Prefrontal Synthesis (PFS): the deliberate, effortful co-activation and dynamic binding of distributed visuospatial representations in working memory in order to construct an integrated, manipulable "scene" or situation model. When PFS is impaired, individuals struggle to intentionally generate a unified visuospatial scene that simultaneously contains an agent (subject), an action, and a patient (object) in their correct spatial-temporal relations. Because most syntactic constructions (e.g., active vs. passive voice, relative clauses, prepositional phrases) evolved to efficiently encode precisely these visuospatial-relational configurations, PFS deficits can render sentence comprehension fragmentary or reliant on non-syntactic strategies (e.g., rote memorization or simple word-order heuristics). The autism community refers to this phenomenon as the lack of multi-cue responsivity, or stimulus overselectivity, or tunnel vision (since individuals focus on a single word rather than integrating multiple words within a sentence).

In the absence of robust compensatory mechanisms (which some autistic individuals do develop), this core deficit in constructing and manipulating internal situation models contributes to persistent difficulties with complex language comprehension, abstract reasoning, planning, and flexible adaptive behavior-often resulting in lifelong challenges with independent living for those at the more severely affected end of the spectrum.

About 20 years ago, investigators posed a critical question: if many autistic children show limited interest in conversations, imaginative play, and story listening, could structured, non-conversational exercises be developed that nonetheless train the same fronto-parieto-temporal networks underlying syntactic processing and mental scene construction? Moreover, could such brain training be delivered during the peak sensitive period for language and socio-cognitive development (ages 2-4 years)? To address this challenge, Mental Imagery Therapy for Autism (MITA), a tablet-based intervention was built around highly stylized, brightly colored animated characters that are frequently preferred by autistic children. MITA consists of adaptive, game-like exercises designed to systematically train PFS both within and outside the verbal domain.

Verbal exercises begin with basic receptive vocabulary and systematically progress to increasingly complex syntactic and relational structures, including noun-adjective coordination, spatial prepositions, agent-action-patient relations, and recursive embedding 30. For example, a child may be asked to "find the small red ball" or "put the cup behind the table." Vocabulary is deliberately constrained; the goal is not lexical expansion but rather the repeated practice of dynamically binding multiple features and objects into novel configurations using PFS.

Non-verbal exercises provide analogous PFS training through purely visual puzzles that require mental integration of dissociated parts into a unified whole, without any spoken instruction 31. For example, a child can be presented with two separate images of a train and a window pattern, and a choice of complete trains. The task is to find the correct complete train and to place it into the empty square. Difficulty scales by progressively increasing the number of features that must be simultaneously attended to and integrated (up to eight at the highest levels). Prior work has shown that many children who cannot yet follow explicit verbal instructions can nevertheless succeed on visually isomorphic non-verbal versions of the same integrative task 30.

Since its release in 2015, and without paid advertising, MITA has consistently ranked among the top language-therapy applications across major app stores (Apple App Store, Google Play Store, and Amazon App Store), accumulating more than 3 million downloads worldwide and indicating sustained parental interest. The app includes quarterly in-app parent-report assessments, yielding the largest longitudinal naturalistic dataset on the development of autistic children.

An observational analysis of this real-world longitudinal cohort (N = 6,454) found that children who engaged with MITA showed 2.2-fold greater gains on a composite parent-reported PFS outcome measure compared to age-, gender-, and severity-matched controls who did not engage with the app (p < 0.0001) 33.

These promising naturalistic (citizen-science) findings prompted the design and execution of a prospective randomized controlled trial (RCT) to establish causal efficacy under more rigorously controlled conditions. To minimize environmental and therapeutic confounders, a study was designed in which participants could be randomized and matched within the same clinical and educational setting. After contacting more than 300 specialized early-intervention centers across North America, South America, and Europe, only one institution-Somare Therapeutic and Educational Clinic in Foz do Iguaçu, Brazil- had both access to an appropriate pediatric population and the willingness to conduct an independent, investigator-initiated RCT at no cost to the research team. The clinic serves a homogeneous population of preschool-aged children with autism spectrum disorder (ASD) and provided an ideal setting for tight experimental control of concomitant interventions and daily routines.

Preschool children (aged 2-5 years) attending Somare Therapeutic and Educational Clinic with a confirmed diagnosis of ASD will be enrolled in the RCT after informed parental consent. Eligible participants will be pair-matched by age and CARS-2 total score, then randomized 1:1 to either an intervention group that received standard clinic-based early intervention plus daily MITA training for a minimum of 12 months, or active-control group, that received standard clinic-based early intervention plus an equivalent duration of computerized executive-function exercises (non-MITA). Both groups will continue to receive the clinic's usual intensive behavioral and speech-language interventions. An assessor blinded to group assignment will evaluate all participants every six months for 2.5 years using the Childhood Autism Rating Scale (CARS-2) as the outcome measure.