AI Search Summary
This video explains the 85% rule for optimal learning: people often learn fastest when a task is challenging enough that they make errors, but not so hard that they fail constantly.
- Main question: How can the 85% rule help you learn or teach skills more effectively?
- Short answer / core takeaway: Aim for practice where you succeed roughly 85% of the time and fail about 15% of the time, then adjust difficulty as you improve.
- Evidence type: Theoretical learning-model research, classroom-learning support, and related human rehabilitation adherence evidence.
- Search topics: 85% rule, optimal learning, error rate, desirable difficulty, study hacks, skill acquisition, learning science.
Common Search Questions
What is the 85% rule for learning?
The 85% rule says learning is often optimized when you get about 85% of attempts right and about 15% wrong. That balance gives enough success to learn patterns while preserving enough error to reveal what needs improvement.
How should students apply the 85% rule?
If practice problems feel easy and you get more than about six out of seven correct, increase the difficulty. If you are failing most attempts, lower the difficulty or build prerequisite skills first.
Is the 85% rule based on a study?
Yes. The video links to “The Eighty Five Percent Rule for optimal learning,” which modeled optimal learning across AI and neuroscience systems and reported an optimal error rate of about 15.87%.
Key Takeaways
- Learning tends to suffer when practice is either too easy or too hard.
- An error rate around 15% can create a useful contrast between successful and unsuccessful strategies.
- The rule can apply to studying, physical skills, games, teaching, and rehab-style practice when the task difficulty is adjustable.
- The exact number should be treated as a practical guide, not a rigid law for every learning situation.
Transcript
The study and the rule
This study found a simple rule that can make you learn or teach almost any skill faster. It can help you study for tests, learn to rock climb, master a video game, and a whole lot more.
It’s called the 85% rule, and we’re going to cover what it is, why it works, and how to apply it.
Welcome back to 30 Studies to Change Your Life: Learning Optimization Week, where we’ll be covering science-backed methods to help you learn faster and retain that information longer.
Two bad difficulty levels from grad school
To explain the rule, I’m going to give you two examples for my grad school days. I was always one of those annoying, overachiever, straight-A students. But I had this one crotchety professor. Let’s call him Binns.
I remember walking out of his first exam freaking out. The class was on transistor theory, but the test had almost nothing to do with that. I later found out that it was more based on his current research project. I got a 37% on that test. I was devastated.
But I spoke to a friend who was super relaxed about it. He told me that his father had also had Professor Binns, and that the class average, even back then, was usually curved up. I remember almost nothing from that whole semester.
That same year, I had a different class on electromagnetism. It was really hard to get into because the teacher, let’s call him Gilderoy, was known to be good but a really lazy test giver. He always used the same questions from previous years and barely even changed the numbers around. Most people got hundreds.
But just like the first professor, a year later, almost nobody remembered anything from that class.
Why the sweet spot matters
This all intuitively makes sense. We learn best when we’re being challenged, but make that challenge too hard, and there’s nothing to hold on to, so we just check out. But there’s a sweet spot, which brings us to the study.
Researchers tested out multiple learning models, both from AI research and neuroscience, trying to see how training difficulty impacts learning speed. They found an optimal error rate to be 15.87%, or roughly 85% accuracy, meaning that you’re trying something, but you’re failing roughly one in every seven attempts.
Applying the rule in practice
This study was theoretical, but it’s been backed up by classroom data and, most recently, a study on computer-guided rehab exercises in humans, where those who found them either too easy or too hard didn’t stick with it as much as those who had a roughly 90% success rate.
So if you’re studying and getting more than six out of seven problems right, raise the difficulty level. But also don’t start out too tough unless that’s part of the actual education process.
Back when I was trying to get good at Beat Saber, I made sure to set it so that I was playing at a difficulty where I could hit roughly 85% of those blocks, rather than spending too much time perfecting those levels where I was already at 95%, or wasting my time with those where I could barely hit any.
I want to hear from you guys. What are the best ways you would apply this rule? Let me know in the comments.
Additional Notes
Caption context
The caption frames the 85% rule as a favorite study hack and warns that people often spend too much time in the wrong parts of the error-rate curve.
Keywords and topics
- Learning optimization
- Study hacks
- Skill acquisition
- Error rate curve
- Desirable difficulty