Five research-backed frameworks working together to personalise learning for every neurodivergent child.
Updated February 2026
Universal Design for Learning is a research-backed educational framework developed by CAST (Center for Applied Special Technology). It was originally conceived in the 1990s by David Rose and Anne Meyer, drawing on neuroscience research about how the brain processes information. The core insight is simple but profound: there is no single "average" learner.
Every child's brain processes engagement, representation, and expression differently. Traditional classrooms tend to present content in one way and assess understanding in one way. For neurodivergent children, this rigid approach creates unnecessary barriers. UDL removes those barriers by building flexibility into every aspect of instruction from the start.
CAST's research demonstrates that UDL benefits all students, not just those with identified learning differences. A 2018 CAST meta-analysis found that UDL-based instruction improved academic outcomes across neurotypical and neurodivergent populations, with particularly strong gains for students who had previously struggled in traditional settings.
Key finding: Schools implementing UDL frameworks report a 20-30% reduction in the need for individual accommodations, because the learning environment is already designed to be flexible (CAST, 2018).
Rise Bright implements all three UDL principles throughout the platform. Rather than retrofitting accessibility, every lesson is designed from the ground up with multiple pathways to success.
Children choose topics linked to their interests. Gamification provides variable rewards to maintain motivation. Flexible pacing respects each child's energy and focus levels. Autonomy and choice are built into every session, supporting intrinsic motivation rather than relying solely on external rewards.
Concepts are presented through visual diagrams, interactive animations, text, and audio. Mathematical problems use colour-coded notation and step-by-step breakdowns. Reading tasks offer adjustable font sizes, line spacing, and background colours to reduce visual stress for dyslexic learners.
Students demonstrate understanding through drag-and-drop interactions, typed responses, multiple choice, matching exercises, and drawing. No single response format is required. This removes the barrier of "I know it but can't show it" that many neurodivergent children experience in traditional assessments.
The Concrete-Representational-Abstract (CRA) approach is a three-stage instructional method that builds mathematical understanding from the ground up. It was developed from Jerome Bruner's theory of cognitive development and has decades of research supporting its effectiveness, particularly for students with learning disabilities in mathematics.
Many children with dyscalculia or other mathematical learning differences struggle because traditional maths instruction jumps straight to abstract symbols. When a child sees "7 + 5 = 12" without first understanding what those numbers mean physically, the equation becomes meaningless memorisation rather than genuine understanding. CRA prevents this by ensuring each stage of understanding is solid before progressing.
Students interact with virtual manipulatives: draggable blocks, counters, fraction tiles, and base-ten materials. They physically group, combine, and separate objects to build intuitive understanding of mathematical operations. For example, addition becomes the act of combining two groups of blocks and counting the total.
Once concrete understanding is established, students transition to visual models: number lines, bar models, arrays, and diagrams. These representations bridge the gap between physical objects and abstract symbols. The blocks become drawn circles, then tallies, then structured visual patterns that mirror the mathematics.
Only after demonstrating understanding at both previous stages do students work with mathematical symbols and equations. At this point, "7 + 5 = 12" carries genuine meaning because the child has physically combined 7 and 5 objects, drawn the visual representation, and now writes the symbolic notation with deep comprehension.
Spooner et al. (2019) conducted a comprehensive meta-analysis of CRA interventions and found a Tau-BC effect size of 0.9965, which represents near-perfect improvement across single-case research designs. Dennis et al. (2016) similarly found strong evidence supporting CRA for students with learning disabilities in mathematics, noting that the approach is particularly effective when combined with explicit instruction and systematic feedback.
Rise Bright's AI determines when a child is ready to progress from one CRA stage to the next. If a student demonstrates fluency with virtual manipulatives but struggles at the representational stage, the system provides additional practice with visual models before attempting abstract notation. This prevents the frustration that comes from being pushed into symbolic mathematics before genuine understanding is formed.
Why this matters for dyscalculia: Children with dyscalculia often have difficulty connecting number symbols to quantities. CRA builds this connection explicitly, starting with quantities they can see and touch, then gradually introducing the symbolic shorthand that represents those quantities.
Bayesian Knowledge Tracing (BKT) is a probabilistic model that estimates what a student knows in real-time by analysing their response patterns. Originally developed by Albert Corbett and John Anderson at Carnegie Mellon University in 1994, BKT has become the gold standard for student modelling in intelligent tutoring systems worldwide.
The fundamental problem BKT solves is this: how do you know whether a student truly understands a concept, or whether they guessed correctly? And conversely, how do you know whether a wrong answer means the student does not understand, or whether they simply made a careless error? BKT uses probability to untangle these possibilities.
The model tracks four key parameters for each concept a child is learning:
The probability that the student already knows the concept before encountering it on Rise Bright. This is initially estimated from their baseline assessment and demographic data, then refined as the student interacts with the platform. A child who has been studying Year 3 maths might have a higher P(L₀) for basic addition than for multiplication.
The probability that the student learns the concept on each attempt. This parameter captures how quickly a child picks up new ideas and is personalised over time. Some children learn quickly with visual representations; others need more repetitions. The AI adapts to each child's natural learning pace.
The probability that the student answers correctly despite not actually knowing the concept. For multiple-choice questions with four options, P(G) might start at 0.25. For open-ended questions, it is much lower. This parameter prevents the system from overestimating knowledge based on lucky guesses.
The probability that the student answers incorrectly despite actually knowing the concept. This accounts for careless errors, misreading questions, or attention lapses. For children with ADHD, slip probability may be higher, and Rise Bright's model adjusts accordingly rather than penalising these natural attention fluctuations.
After each response, BKT updates the probability of mastery using Bayes' theorem. When a student answers correctly, P(mastery) increases. When they answer incorrectly, it decreases. The magnitude of each update depends on the guess and slip parameters, so a correct answer on a difficult open-ended question produces a larger increase than a correct answer on a multiple-choice question.
Zone of Proximal Development: Vygotsky's theory states that optimal learning occurs when tasks are challenging enough to promote growth but not so difficult that they cause frustration. BKT enables Rise Bright to keep every child in this zone by selecting questions where P(mastery) is between 0.4 and 0.8 — challenging but achievable.
This approach is used by some of the world's leading adaptive learning platforms, including Carnegie Learning and Khan Academy. Rise Bright adapts the model specifically for neurodivergent learners by incorporating additional parameters for attention state, time-of-day effects, and session fatigue patterns that are particularly relevant for children with ADHD and autism.
In 1885, German psychologist Hermann Ebbinghaus published groundbreaking research on memory that introduced the "forgetting curve." He demonstrated that newly learned information is rapidly forgotten without review, with approximately 70% of new knowledge lost within 24 hours. However, he also discovered something remarkable: reviewing information at strategically spaced intervals dramatically improves long-term retention.
This discovery, known as the spacing effect, has been replicated in over a thousand studies across more than a century of research. It is one of the most robust findings in all of cognitive psychology. Spaced repetition scheduling applies this principle by automatically determining the optimal time to review each concept.
Rise Bright uses the FSRS (Free Spaced Repetition Scheduler) algorithm, a next-generation scheduling system that significantly outperforms the older SM-2 algorithm used by most traditional spaced repetition tools. FSRS was developed through machine learning analysis of hundreds of millions of review data points and models three key properties of each memory:
How inherently challenging a concept is for this particular child. Some children find fractions easy but struggle with place value. FSRS tracks this per-concept difficulty and adjusts review intervals accordingly. Harder concepts are reviewed more frequently until they become automatic.
How firmly a memory is encoded. A concept reviewed five times over three weeks has higher stability than one reviewed once yesterday. FSRS calculates stability to determine exactly when the child will start to forget, scheduling the review just before that tipping point.
The current probability that the child can recall the concept right now. This decreases over time according to each child's personal forgetting curve. When retrievability drops below 90%, FSRS schedules a review. This ensures children review concepts before they forget them, not after.
The SM-2 algorithm, developed in 1987 by Piotr Wozniak, was revolutionary for its time but relies on fixed mathematical formulas that do not adapt well to individual learning patterns. FSRS uses machine learning to model each learner's actual memory behaviour, resulting in 30% fewer reviews needed to maintain the same level of retention.
For neurodivergent learners, this efficiency is particularly valuable. Children with ADHD often have working memory challenges that make retention harder. Children with autism may have uneven knowledge profiles where some topics are deeply mastered while others need frequent reinforcement. FSRS handles these patterns naturally because it models each concept independently for each child.
Practical example: If a child masters "adding fractions with like denominators" on Monday, FSRS might schedule the first review for Wednesday, then the following Monday, then two weeks later, then a month later. Each successful review extends the interval. If the child struggles at any point, intervals shorten until the concept is re-stabilised.
The 3E Cognition Framework structures every Rise Bright learning session into three distinct phases: Engage, Explore, and Express. This framework is designed around how neurodivergent brains naturally process information, moving from attention capture through guided discovery to creative demonstration of understanding.
Traditional instruction often begins with explanation and ends with testing. This format works against neurodivergent learners because it demands sustained attention for passive listening and then judges understanding through a narrow assessment format. The 3E framework inverts this by starting with active engagement, embedding learning within exploration, and allowing flexible expression of knowledge.
Every session begins by capturing attention through curiosity, novelty, and connection to the child's interests. This might be a surprising fact, a visual puzzle, a "what if" question, or a connection to something the child enjoys. The Engage phase activates the brain's dopamine-driven curiosity system, which research shows is essential for attention regulation in ADHD brains. Sessions are never launched with dry instructions or passive reading. Instead, the opening 30-60 seconds are designed to create a "curiosity gap" that the child wants to fill.
During the Explore phase, children interact directly with concepts through guided discovery. Rather than being told how something works, they experiment and discover patterns themselves with scaffolded support. The AI provides hints and feedback calibrated to the child's current knowledge level (via BKT). For mathematics, this phase uses the CRA stages. For English, it might involve interactive word building, sentence manipulation, or comprehension activities. The scaffolding is gradually reduced as the child demonstrates understanding, following the principle of productive struggle.
The Express phase allows children to demonstrate understanding in multiple ways, aligned with UDL's third principle. Instead of a single test format, children might drag and drop elements, type responses, select from options, draw representations, or complete matching exercises. This flexibility is critical for neurodivergent learners who may understand a concept deeply but struggle with a particular expression format. A child with dysgraphia, for example, should not be marked as "not understanding" simply because they find writing difficult.
The 3E flow mirrors how neurodivergent brains naturally move through information processing. Research on ADHD, autism, and dyslexia consistently shows that these learners benefit from interest-driven engagement, hands-on exploration, and flexible expression formats. By structuring every session around these three phases, Rise Bright ensures that the learning format itself supports neurodivergent cognition rather than working against it.
Session structure: Each 15-minute Rise Bright session follows the E-E-E pattern: approximately 1-2 minutes of Engage, 8-10 minutes of Explore (with embedded practice and feedback), and 3-4 minutes of Express. Movement breaks are integrated between phases for children who benefit from physical activity to reset attention.
The 3E framework draws on research from multiple domains. The Engage phase builds on Hidi and Renninger's (2006) four-phase model of interest development. The Explore phase applies Kapur's (2008) theory of productive failure, where initial struggle leads to deeper conceptual understanding. The Express phase implements Bloom's taxonomy by allowing children to demonstrate understanding at higher cognitive levels through creation and application, not just recall.
Each framework is powerful on its own, but the real strength of Rise Bright's methodology is how all five frameworks integrate seamlessly within a single learning session. Here is what a typical 15-minute session looks like for a Year 3 student learning multiplication.
The session opens with a visual puzzle: "A farmer has 4 paddocks, and each paddock has 6 sheep. Can you figure out how many sheep the farmer has altogether?" The scenario connects to Australian rural life and creates a curiosity gap. The child's interest system is activated before any formal instruction begins. This takes approximately 60 seconds and sets the tone for active participation rather than passive receiving.
Behind the scenes, Bayesian Knowledge Tracing checks the child's current mastery probability for "multiplication as repeated addition" and "multiplication facts for 4 and 6." If P(mastery) is 0.3 for this concept, the system knows the child is still building understanding and selects questions at the concrete CRA stage. If P(mastery) were 0.7, it might present representational or abstract challenges instead.
Following the CRA approach, the child first sees virtual paddocks with draggable sheep (concrete stage). They physically group 6 sheep into each of 4 paddocks and count the total. Next, the sheep become dots in an array: 4 rows of 6 (representational stage). UDL ensures the child can interact with the array through clicking, dragging, or typing. Finally, they see "4 x 6 = ?" and connect the symbolic notation to what they have built (abstract stage).
The child answers "24." Within one second, they see a celebration animation and earn XP points. BKT updates P(mastery) upward, accounting for the question type and difficulty. If they had answered incorrectly, the system would provide an encouraging hint and step back to the representational stage rather than repeating the same abstract question. The slip and guess parameters ensure the model accurately reflects true understanding.
Now that the child has demonstrated understanding of 4 x 6, FSRS calculates when this specific fact should be reviewed. Based on the child's personal difficulty, stability, and retrievability parameters, it might schedule a review in 2 days. If the child answers correctly again in 2 days, the next review might be in 5 days, then 12 days, then a month. The concept is never "finished" but gradually transitions to long-term memory with decreasing review frequency.
After 3-5 minutes of focused work, the session prompts a brief movement break. The child might be asked to stand up and do 5 star jumps, touch their toes, or walk to the window and back. Research shows that physical activity improves working memory and executive function, particularly for children with ADHD. These breaks are not interruptions to learning; they are a critical component of the methodology that helps maintain focus and processing capacity for the remainder of the session.
The session concludes with the Express phase. The child might be asked to create their own multiplication story problem, match array images to equations, or complete a visual pattern that demonstrates multiplicative thinking. UDL's multiple means of expression ensure the child can show their understanding in whatever format works best for them. This is not a test; it is a celebration of what they have learned in the session.
Throughout the session, the parent dashboard shows which concepts were practised, the child's current mastery levels, time spent in focused learning, and upcoming review schedule. Parents can see exactly where their child is progressing and which areas need more support. This transparency builds trust and helps parents have informed conversations with their child's teacher about learning progress against Australian Curriculum standards.
The integration advantage: No single framework could deliver this experience alone. UDL ensures accessibility, CRA builds mathematical understanding, BKT selects the right difficulty, FSRS optimises long-term retention, and the 3E Framework structures each session for engagement and expression. Together, they create an adaptive learning experience that meets each neurodivergent child exactly where they are.
Our methodology adapts for each child's unique needs:
Join Australian families using research-backed adaptive learning for their children.
Start free trial Learn about us