The Science of Learning
TutorThings is built on decades of cognitive science research - not on what feels intuitive or what keeps people coming back, but on what consistently helps understanding form and stick. The design of every session follows these principles.
The generation effect
When learners produce an answer from their own thinking rather than reading or hearing one, they remember it significantly better. A meta-analysis of 86 studies published in Memory & Cognition found that generating information yourself - even imperfectly - produces stronger, more durable memory than passively receiving the same information. The effect is consistent across ages, subjects, and settings.
This is the core principle behind TutorThings. The tutor asks learners to explain what they know before offering any hint. The act of reaching for an answer, even an incomplete one, strengthens the neural pathways around that concept far more than hearing the correct answer passively does. When the retrieval is effortful, the retention is better.
Why voice matters
Research on oral language and learning shows that purposeful spoken explanation is one of the highest-impact interventions available. The Education Endowment Foundation rates oral language interventions as high-impact with extensive evidence. A 2025 study by researcher Zachary Himmelsbach, analyzing 1,600+ fourth- and fifth-grade math lessons, found that classrooms where students heard and used precise vocabulary roughly 4,500 more times per year showed significantly larger gains on standardized tests.
TutorThings is voice-first because speaking forces a different kind of thinking than typing or tapping. You cannot copy-paste a spoken explanation. You have to construct it in real time from what you understand. The stumbles, rephrasing, and mid-sentence corrections are features, not bugs - they reveal exactly where understanding is forming and where it is still fuzzy.
Elaboration
Knowing that something is true is different from knowing why it is true. A learner who has memorized that multiplying two negatives gives a positive can still be completely lost when a problem requires reasoning from that fact.
Elaborative questioning - "Why does that work?" "What would happen if this part changed?" "Can you explain it a different way?" - forces the learner to build a mental model rather than a surface-level association. TutorThings uses this technique constantly: follow-up questions probe the reasoning, not just the answer.
This sits close to the Socratic tradition, but in a student-friendly form. TutorThings keeps pressing on the learner's reasoning with prompts like "What makes you think that?" and "How did you get there?" while still offering scaffolds when the next step needs to be narrowed.
Spaced practice
The brain consolidates memory during rest, not during the session itself. A learner who does three 15-minute sessions across a week will typically retain more than one who does a single 45-minute session - even if the total time is identical.
This is why TutorThings is designed for short, focused sessions rather than long ones. The rhythm of returning to a topic after a gap - with some forgetting in between - can accelerate long-term learning. A small amount of struggle to remember something is a feature, not a flaw.
Why the tutor never gives answers
The OECD's 2026 Digital Education Outlook report found that students using general-purpose AI chatbots produced better-looking work - but when the AI was taken away, their performance dropped below where they started. The knowledge was shallow because the AI had done the thinking for them.
This is called cognitive offloading, and it is the single biggest risk of AI in education. When a tool gives you the answer, you feel like you have learned something, but the understanding never fully forms. TutorThings is designed around the opposite principle: the learner does all the thinking, all the explaining, and all the reasoning. The tutor's job is to ask the right question at the right time - not to hand over a solution.
Productive struggle
When a learner gets stuck and reaches for a hint, that moment of uncertainty is often the most valuable part of the session. It reveals exactly where understanding breaks down - and that's precisely where a good tutor should focus.
TutorThings is calibrated to tolerate struggle rather than eliminate it. Hints get more specific only after a learner has genuinely attempted the problem. The goal is to give just enough to keep the learner moving forward, then pull back as soon as they can lead again. Handing over the answer short-circuits this process entirely.
Managing cognitive load
Working memory is limited. When too many new ideas are introduced at once, the brain can't process them deeply - it just tries to hold on. Concepts that feel understood in the moment often don't stick because there was no mental room to connect them to anything.
TutorThings breaks complex ideas into smaller steps, confirms each step before moving forward, and avoids introducing more information than is needed for the next moment of understanding. This is not about dumbing things down - it is about giving the brain room to do the work.
Growth mindset
What a learner believes about their own ability has a measurable effect on how they learn. Students who believe intelligence is fixed tend to avoid hard problems to protect their self-image. Students who believe ability is built through effort are more willing to struggle, try new approaches, and learn from mistakes.
TutorThings treats mistakes as data. When a learner gets something wrong, the response is curiosity - "What were you thinking when you said that?" - not a score or a buzzer. The session is framed around what the learner is figuring out, not how they're performing.
What one-on-one tutoring looks like
Individual tutoring has long been recognized as one of the most effective interventions in education - educator Benjamin Bloom identified this as the "2-sigma problem" in the 1980s: students who receive one-on-one tutoring perform two standard deviations better than those in conventional classrooms. The challenge has always been access and cost. A 2025 systematic review published in npj Science of Learning (Nature), covering 28 AI tutoring studies and nearly 5,000 students, found positive effects in the majority of cases - but only when the AI followed sound pedagogical principles: active learning, adaptive scaffolding, timely feedback, and self-pacing.
TutorThings is built around all four. Not because they are trendy, but because they are what decades of research points to when you sit down with a learner and help them think.