← Back to SafeSkillVR SafeSkillVRby Frappua!
Learning in public

How we think about learning

We build VR training with an AI mentor inside it. To build it well we have to understand how people learn, so we study, ask questions and show our working. These are the slides Erik Sindonen prepared for a conversation with Professor Kirsti Lonka at the University of Helsinki in autumn 2026. Press Play on any slide to hear it.

SafeSkillVR is a prototype. The slides describe how we want the mentor to behave and what we want to test. They are questions and design intentions, not measured results.

Curiosity can be designed

1 · An old problem

How do you teach so a person thinks for himself, instead of waiting for the answer?

The same answer keeps coming back for 2,500 years: wait for the learner's question, give a part, let him find the rest. Not to scale.

Confucius c. 500 BCE Wait for the question. Show a part. Socrates c. 400 BCE Teach by asking. Rousseau 1762 Let him discover it. Fennoman students 1860s to 1890s Go to the learner. Montessori 1907 The environment teaches. Dewey 1910 Thinking starts with curiosity. Vygotsky 1930s Help only what he can't do alone. Wood, Bruner & Ross 1976 Scaffolding. Loewenstein 1994 The information gap. D'Mello et al. 2014 Confusion can help. Arguel et al. 2019 Optimal confusion. Lonka 2026 Start with questions.

2 · Confucius could wait

One question, two answers

  • "If a man does not ask 'what should I do?', I cannot help him."
  • He showed only a part. The rest the student worked out himself.
  • He taught in conversation. No class, no bell. He had time for each student.
  • Two students, same question: should I act at once?
  • Hot-headed one: "wait, ask your elders first." Timid one: "do it at once."
  • Because the students were different.
in pedagogyscaffoldingA little support, not the whole answer. Wood, Bruner & Ross, 1976.
in pedagogyadaptive supportHelp that changes with the learner, not with the task. Analects 11.22.
act now? act now? wait go, now same question, two answers

3 · Finland, 19th century

The teacher adapted to the learner

  • The learner was far away, in the village.
  • The university spoke Swedish. The village spoke Finnish.
  • So the Fennoman students went to the villages. To teach.
  • Confucius waited for the learner. The Fennomans went to the learner.
Different paths, same point: the teacher adapted to the learner.
university Swedish village Finnish the teacher went to the learner

4 · School

One teacher. Thirty students. Forty-five minutes.

  • You can't wait for each one.
  • You can't go to each one.
  • Thirty students adapt to one teacher. Otherwise there is no time.
The opposite of Confucius and the Fennomans. Not because of the teacher. Because of the numbers.
one 45 min thirty thirty adapt to one no time for anything else

5 · Curiosity

At what moment does curiosity go away?

  • While curiosity is there, he searches. When it's gone, he waits to be told.
  • No single moment. The price of not knowing goes up.
  • At 6: fine. At 16: embarrassing in front of the class. At 45: embarrassing in front of yourself.
  • OECD, Helsinki, 2019 to 2023: curiosity and creativity dropped the most among 15-year-olds.
in pedagogyinformation gap"I don't know this, but I could." The trigger of curiosity. Loewenstein, 1994.
6 ? 16 ? 45 ? free status self-respect price of not knowing the question gets smaller, the price gets bigger

6 · Not alone

The problem is old, and your university talks about it

  • Open University, introduction course.
  • Irene Hein's lecture on Illich: you cannot motivate everyone.
  • With one teacher and thirty, she is right. Two technologies change the numbers: VR and an AI layer.
  • Heikki Pasanen, on the Fennomans: "self-directed learning does not mean the learner is left alone."
Give every teacher what Confucius had: time for each learner.
Irene Hein you cannot motivate everyone Heikki Pasanen never leave the learner alone not alone an old problem, and your university talks about it

7 · Games

A game is one tool. Only one.

  • It raises interest and attention. It can't learn for the person.
  • Diplomacy paper, 2026: the game raised almost all epistemic emotions. Curiosity, and anxiety too.
  • Enjoyment alone gave no result.
in pedagogyepistemic emotionsEmotions about knowing: curiosity, surprise, confusion, frustration, boredom.
curiosity confusion anxiety enjoyment learning: flat every emotion went up. learning did not

8 · The environment

If the environment is ours, curiosity can be designed

  • In VR we build the whole environment. The machine, the gauge, the light, what the mentor does when silent, what the student sees.
  • Not alone: with teachers, with students, with people who know how curiosity works.
  • Curiosity arises between learner, subject and environment.
  • If the environment can be changed completely, curiosity can be designed. We want to test it.
in pedagogyprepared environmentMontessori: the environment is designed so the child learns from it, not from the adult.
mentor waits teacher: they fail here students: scary here researcher: curiosity? built together

9 · The thin line

The mentor no longer interferes, but has not left

  • We see what the student does, where the hand reaches, where he looks.
  • Never looked at the gauge? Attention. Looked for ten seconds and failed? Understanding.
  • Same mistake. Different learners. Different help.
  • The first needs a quiet click. He turns himself, and is sure he noticed it himself.
  • The second needs a question: "what is the pressure right now?"
  • A good master doesn't say "look at the gauge". He walks past and taps it with a wrench.
in pedagogyscaffolding and fadingSupport first, then step back.
VR headset where he looks where the hand reaches what we see 0 the student tries 1 the environment a sound, the needle, steam 2 the mentor's body looks there, walks over 3 a question 4 a hint in words 5 the answer only at the end what we do about it

10 · The opposite direction

Almost all educational AI answers faster

  • Bastani et al., 2025: almost 1,000 high school students with an AI maths tutor. With it, they solved more. Without it, they did worse than those who never had it. A tutor that gave hints, not answers, avoided most of the harm.
  • With us: the student tries, fails, searches. Curiosity stays his.
  • The mentor is there. Always there. Mistakes are not punished. Nobody is left alone.
  • In a classroom things are explained once. In our environment you can try a hundred times. Classmates don't see you fail.
  • One number to watch: how many questions the student asks. First session, tenth session. Curiosity comes back fast. Asking comes back slowly.
Too early: the mentor kills the question.
Too late: the student gives up.
The line: support, then fade.
AI tutor ? answer answer answer no AI after AI did worse a fast answer kills the question

11 · The open question

Did he learn by himself? Or did we trick him?

  • The gauge clicks. He turns his head by himself.
  • He thinks he did it himself. Now he believes he can do it, and his confidence grows.
  • But we helped him, and he didn't know it.
  • So is that good for learning? Or should the learner know when we help him?
in pedagogyself-efficacyThe learner's belief "I can do this". Bandura.
click I noticed it myself the mentor clicked it learned by himself? or tricked?

12 · Where we are

We can build the environment. Not its pedagogy.

  • That has to be built with someone who knows.
  • Where would you start? What would you read first?
Give every teacher what Confucius had: time for each learner.
Confucius: an answer for each learner today AI mentor every learner has one give every teacher what Confucius had
arrow keys work

Sources

Lines attributed to lecturers are our notes from their lectures in the University of Helsinki Open University course Johdatus kasvatustieteisiin, not verbatim quotations. The narration is a synthetic voice reading Erik's text. If we have misread a source, tell us at info@frappua.win and we will correct it.