The evidence base

Built on research,
not philosophy alone.

The Berek Fano model draws on decades of research across self-determination theory, Montessori principles, project-based learning, and the emerging field of AI and neurodivergent learning.

What the research says

Built on research,
not philosophy alone

We didn't build this model on intuition and then look for evidence. We built it on evidence and designed the environment to deliver on it. These are the research foundations — and the honest caveats where the literature is still developing.

Child-directed learning

Autonomy support produces executive function gains

Students given autonomy show measurable gains in executive function — the skills standard schooling most tries to build through control.

Cross-sectional research across 208 students found that teacher autonomy support and student intrinsic motivation produced significantly higher executive function scores. External regulation styles — compliance-based learning — correlated with lower executive function.
High EF came with autonomy support and intrinsic motivation styles. Low EF was related to external regulation.
Frontiers in Psychology (2015), University of Ulm
>
Self-determination theory

Three conditions produce genuine learning — traditional schools deliver none

Self-determination theory identifies three conditions for genuine learning: autonomy, competence, and relatedness. Traditional schools routinely undermine all three.

SDT identifies autonomy, competence, and relatedness as the three needs whose satisfaction produces intrinsic motivation, healthy development, and wellbeing. Problem-based, student-directed learning operationalises all three. Compliance-based instruction operationalises none.
SDT validated as an effective framework for supporting autistic individuals specifically — across motivation, wellbeing, and engagement.
Deci & Ryan (2000); SDT and ASD, Humboldt Digital Commons
>
Special interests

Interest-based learning activates brain regions that nothing else reaches

When children study their special interests, brain regions associated with deep learning activate that nothing else reaches. Interest isn't a distraction — it's the mechanism.

MIT neuroscientist John Gabrieli's brain-scanning research found that stories about an autistic child's special interest activated key language brain regions far more strongly than neutral content — suggesting interest-based instruction is neurologically distinct, not just motivationally preferred.
Strengths-based interventions incorporating special interests reduced anxiety, increased wellbeing, and improved social engagement across all 9 intervention studies reviewed.
Gabrieli Lab, MIT; Special interests scoping review (2018)
>
Predictability & regulation

Predictable environments do regulatory work the child doesn't have to

Predictable routines do regulatory work the child doesn't have to do themselves — freeing cognitive capacity for actual learning.

Structured, predictable routines significantly reduce anxiety in autistic children by minimising sensory and cognitive overwhelm during transitions. Visual work systems in low-distraction spaces increase on-task behaviour. The environment itself is a regulatory tool.
Predictable routines support executive function, increase participation, and improve social engagement at all educational levels.
Frontiers in Psychiatry, Integrative Review (2021); Dr Barry Prizant, Uniquely Human
>
Project-based learning

PBL benefits neurodivergent learners — honest about where the evidence is strong and where it's still building

Project-based learning shows improved social skills, reduced emotional incidents, and better academic outcomes for neurodivergent learners across multiple studies.

STEM project-based learning research shows improved social skills, reduced emotional incidents, and better cooperation in autistic students. Practitioner evidence is strong. Rigorous RCT-level research specifically on PBL for neurodivergent children remains limited — a gap we're honest about, and one that makes our own documentation meaningful.
STEM-PBL addresses rigid thinking and executive function deficits in autistic students through three core mechanisms.
PMC Behavioural Sciences (2025); CHADD Conference Research (2019)
>
Learning principles (Montessori)

Child-led, hands-on, uninterrupted learning outperforms traditional models at the ages we serve

A 2023 review of 32 studies found child-led, hands-on, uninterrupted learning consistently outperforms traditional models — particularly for children who learn differently.

A 2023 systematic review across 32 studies and 132,249 data points found this style of learning produced meaningful advantages over traditional education — particularly in executive function (effect size 0.36), creativity (0.26), and language (0.17). Effects were strongest at elementary age, in private settings.
Our approach draws on Montessori principles among others — as one of the strongest evidence bases for the learning environment we've designed.
Lillard et al., Systematic Review, PMC (2023)
>
The concept that changes the age question
Monotropism: why this model doesn't age out for autistic learners

Monotropism — the tendency to focus attention deeply on a small number of interests — has been posited as the central underlying feature of autism by autistic researcher Dr. Dinah Murray. It is precisely the cognitive style that long work periods, problem-led learning, and uninterrupted deep time are designed to serve.

Neurotypical children develop increasing capacity to multitask, task-switch, and divide attention as they age. Traditional schooling accelerates and rewards this shift. Autistic learners don't follow that trajectory in the same way. The monotropic strength deepens with age — interests become more sophisticated, domains of genuine expertise emerge. This means our model doesn't age out — it becomes more aligned with how these students actually think as they grow, not less.

Flow states

74% of autistic respondents prefer long focused work; over two-thirds report this helps them stay calm and focused at school

Deep expertise

Intense focus enables mastery in specific areas — indispensable in science, maths, art, technology, and philosophy

Sensory regulation

Flow states support sensory regulation and mental health — a documented benefit of monotropic engagement for autistic individuals

The honest gap

Autistic learners are pulled from deep-focus environments just as their strengths deepen. Berek Fano exists to close that gap.

AI and neurodivergent learning

What the research says about AI
and learners like ours.

AI and neurodivergent learning

Neurodivergent learners interact more openly with AI than with adults in classroom settings

Neurodivergent learners ask more questions and recover faster from mistakes when working with AI than with adults in evaluative settings.

A qualitative study of gifted students and students with learning disabilities found participants expressed apprehension about judgment and social anxiety in classroom settings — but not when interacting with AI. Some autistic students showed a preference for AI interaction due to its predictability and absence of social pressure, reporting greater willingness to ask questions, make mistakes, and try again.
AI can provide consistent, patient instruction without social pressure — particularly valuable for students who struggle with face-to-face interaction or feel uncomfortable asking questions in social situations.
El Naggar et al. (2024); Sağdıç et al. (2024); BMC Psychology (2024)
>
Voice and working memory

Speech-to-text removes the barrier between thinking and expression for ADHD learners

Voice-to-text separates thought generation from transcription — freeing working memory for thinking. The work produced reflects what the student actually knows.

Evidence shows youth with ADHD consistently generate less organised written work and fewer words than peers — not due to lower capability, but because written expression requires simultaneous integration of multiple demands governed by working memory, an area of significant weakness in ADHD. Speech-to-text isolates thought generation from transcription, freeing working memory for the actual thinking. The result is work that reflects what the student actually knows.
TTS and STT tools free up working memory — essential for writing — by allowing focus on the task itself rather than the mechanics of recording it.
OT4ADHD Clinical Review (2022); ADDitude Magazine; Molitor (2016)
>
Why AI belongs in this school — what the research actually says
AI and neurodivergent learners: three evidence-backed roles

The research on AI and neurodivergent learners has moved well beyond "might help" into specific, documented mechanisms. A systematic review of 84 studies from 2018 to 2024 confirms AI applications are expanding rapidly across ASD, ADHD, and dyslexia — with measurable outcomes in written expression, task planning, self-efficacy, and engagement.

Three roles are consistently supported. AI as patient tutor — providing consistent instruction without social pressure, particularly valuable for students who experience judgment anxiety in classroom settings. AI as access tool — voice transcription and text-to-speech that free working memory for thinking rather than transcription. AI as inquiry extension — removing the ceiling on depth for learners whose curiosity outruns what any single adult can support.

The research also raises an honest caution: AI interaction that substitutes for human connection rather than complementing it can reduce motivation for social engagement. This is precisely why Berek Fano keeps AI at a shared station with deliberate checkout protocols — and why the guide relationship is framed as irreplaceable. The tools augment the human. They do not replace it.

No-judgment interaction

Neurodivergent students ask more questions, make more attempts, and recover from mistakes faster when interacting with AI than with adults in evaluative settings

Working memory relief

Voice-to-text separates thought generation from transcription — the most significant barrier for ADHD writers. What gets captured reflects actual thinking, not working memory capacity

Twice-exceptional fit

AI can simultaneously deliver higher-order content and lower-barrier access — the combination twice-exceptional learners need and that no single human can consistently provide across ten students

The honest caveat

Research is still developing, particularly for K–5 age groups. Berek Fano's AI integration is designed to complement human relationship, not replace it — and is reviewed as the evidence base grows

Seen enough?

If the research makes sense to you, the next step is a conversation about your child.

Enroll your child →