--- title: "How people learn: 12 principles Ducky follows | DuckyHelper" url: "https://duckyhelper.com/research/principles/" description: "Twelve learning-science principles behind DuckyHelper, from one-on-one tutoring to spacing, each with the research and the tutor behaviour it drives." updated: "2026-10-09" --- Principles # How people learn Decades of research agree on a short list of things that help people learn: a tutor who guides, worked examples, pictures with words, quizzing yourself, spacing and mixing practice, explaining in your own words and quick feedback. Here are twelve, each with the papers behind it and what DuckyHelper's tutor does because of it. [All films](https://duckyhelper.com/films/#model) ## Twelve principles ### Teach one on one Bloom's 2 sigma problem in plain words: what one-on-one tutoring did in the studies, what later reviews found, and how DuckyHelper tutors. ### Guide, don't hand over answers A 2025 PNAS study found that unguarded GPT-4 help raised practice grades but lowered exam grades. What it means, and how DuckyHelper guides instead. ### Just enough help What scaffolding means in learning science (Wood, Bruner and Ross; Vygotsky), why help should fade, and how DuckyHelper's hints and tutorials do it. ### Show a worked example first The worked example effect: why studying solved problems helps beginners more than solving alone, when it stops helping, and how DuckyHelper shows steps. ### Draw it as you explain Multimedia learning in plain words: why words with pictures beat words alone, why watching a drawing appear helps, and how DuckyHelper draws as it talks. ### Quiz, don't re-read The testing effect: why pulling an idea out of memory beats re-reading it, with the real numbers from Roediger and Karpicke, and how DuckyHelper quizzes you. ### Space it out The spacing effect and the forgetting curve: why reviewing after a gap beats cramming, what 317 experiments found, and how DuckyHelper schedules review. ### Mix it up Interleaved practice in plain words: why mixing problem types helps learners tell them apart, what Rohrer and Taylor found, and what DuckyHelper does today. ### Explain it back in your own words The self-explanation effect: why students who explain ideas to themselves learn more (Chi et al.), the ICAP framework, and how DuckyHelper asks you to think. ### Feedback right away, on the step What research says about feedback: why it is powerful, why over a third of feedback studies found it hurt, and how DuckyHelper gives feedback on the step. ### Keep the learner doing, not watching Active learning in plain words: the 225-study PNAS meta-analysis, why active classes feel like less learning, and how DuckyHelper keeps you doing. ### What the evidence says about AI tutors Randomized trials of AI tutors so far: Harvard physics, Ghana, Nigeria, Tutor CoPilot, and the PNAS study where AI help hurt. What they show, and do not. ## The evidence in one line each Each principle, what the research found, and what DuckyHelper does because of it | Principle | What the research found | What DuckyHelper does | | --- | --- | --- | | [Teach one on one](https://duckyhelper.com/research/principles/teach-one-on-one/) | Tutored students scored about two standard deviations above a regular class (Bloom, 1984). | A live voice tutor for one learner. | | [Guide, don't hand over answers](https://duckyhelper.com/research/principles/guide-dont-hand-over-answers/) | Unguarded GPT-4 help cut exam grades 17%; safeguards largely fixed it (Bastani et al., 2025). | Try first, hints before answers. | | [Just enough help](https://duckyhelper.com/research/principles/scaffolding/) | Tutors handle what is beyond the learner and leave the rest to them (Wood, Bruner and Ross, 1976). | The hint ladder, tutorials that wait for you. | | [Show a worked example first](https://duckyhelper.com/research/principles/worked-examples/) | Worked examples beat problem solving for beginners (Sweller and Cooper, 1985). | Worked steps, one move each. | | [Draw it as you explain](https://duckyhelper.com/research/principles/draw-it-as-you-explain/) | Words with pictures beat words alone (Mayer and Moreno, 2003). | Drawing on your screen and on the blackboard. | | [Quiz, don't re-read](https://duckyhelper.com/research/principles/retrieval-practice/) | A practice test beat re-reading a week later, 56% to 42% (Roediger and Karpicke, 2006). | Check questions and quizzes. | | [Space it out](https://duckyhelper.com/research/principles/spaced-practice/) | Spacing helps across 317 experiments (Cepeda et al., 2006). | Flashcards on a spaced schedule. | | [Mix it up](https://duckyhelper.com/research/principles/interleaving/) | Mixed math practice did far better a week later (Rohrer and Taylor, 2007). | Reviews that mix your topics. | | [Explain it back](https://duckyhelper.com/research/principles/explain-it-back/) | Prompting students to self-explain improved learning (Chi et al., 1994). | Asks what you think, checks after each step. | | [Feedback right away, on the step](https://duckyhelper.com/research/principles/immediate-feedback/) | Feedback about the task helps; about the self, it can hurt (Kluger and DeNisi, 1996). | Not quite and why, the first wrong line. | | [Keep the learner doing](https://duckyhelper.com/research/principles/active-learning/) | Active learning raised exam performance across 225 studies (Freeman et al., 2014). | Simulations, graphs, tutorials you do. | | [What the evidence says about AI tutors](https://duckyhelper.com/research/principles/ai-tutoring-evidence/) | Built-to-teach AI tutors showed gains in early trials; plain answers hurt. | A tutor, not an answer machine. | Every number belongs to the study cited beside it. DuckyHelper has not been tested in a study like these. What these pages do not claim These studies tested other people, other schools and other tools. DuckyHelper has not been tested in a study like these, so none of these results are DuckyHelper's results. They are the reasons behind how it teaches. ## Frequently asked questions ### What are the most effective ways to study? A large review by Dunlosky et al. (2013) rated practice testing and spaced practice as the most useful of ten common techniques. Re-reading and highlighting rated low. ### Which principle matters most for an AI tutor? Not handing over answers. Bastani et al. (2025) found a plain GPT-4 chat hurt exam results, while a version built to give hints mostly did not. ### Does DuckyHelper follow all twelve principles? It is built around all twelve, and each page says what the tutor does today. Where a principle is only partly covered, such as interleaving or explaining it back, the page says so plainly. ## Keep reading ### The learning-science library: 100+ papers 100+ real learning-science papers behind DuckyHelper, sorted by principle, each with a citation, a link to the publisher and a one-line summary. ### A model built to teach DuckyHelper's tutor is trained on 1,000,000+ tutoring examples and inspired by 100+ learning-science papers: a tutor, not an answer machine. ### Papers and technical notes DuckyHelper's white papers: how the tutor is designed from learning science, drawing on the learner's screen, and hints before answers. PDF and HTML. ### Watch: A model built to teach, the film Watch A model built to teach, the one minute film on the learning science behind DuckyHelper: real papers, seven principles and the real app, narrated. ## Ready to learn DuckyHelper is ready on your Mac. Learn with Eddy on its blackboard, or with Ducky right on your screen. [Download for Mac](https://duckyhelper.com/download/) [See what it does](https://duckyhelper.com/features/) The Mac app needs Apple silicon and macOS 14 Sonoma or later. There is no Chromebook, Windows or web version right now.