Learning Requires Effort

Part 5 in the series The Brain, the Screen and the School

We have been all the way around. How tech companies design their products , sleep , digitalisation policy , the brain’s basic needs .

Now we come to the core of it. How does the brain actually learn? And what does that mean for how we use screens in school?

The house with the shelves

Sissela Nutley uses an image that sticks: learning is like a house.

The front door is your attention. You can, to some extent, control what you let in. Inside the door is the hallway, and the hallway is your working memory. There you can hold a small amount of information, but only for a short while. About 20 seconds for random information, before it fades away.

The hallway is also narrow. It fills easily with things other than what you intended to think about. You get distracted, and then you have to start over.

If you are actually going to learn something, it has to make it to the storage room, onto a shelf in your long-term memory. And it only gets there through three steps: direct your attention, process it in working memory, and repeat until the knowledge sticks.

That takes effort. Building shelves is slow work.

And here is the point: it is supposed to be slow. The friction is not a flaw in the system. It is the learning itself.

Working memory is decisive

Working memory is so central that Nutley calls it decisive for all learning.

It is strongly linked to logical thinking. You can measure it as early as preschool age and from that predict which students will complete upper secondary school . Working memory capacity predicts academic performance across all subjects, several years later .

And it develops at different rates in different children. Nutley shows data from Nyköping: within a single Year 2 class, the spread can correspond to eight years’ difference in maturity. Some eight-year-olds are at the level of an average twelve-year-old. Others at that of a four-year-old.

Same classroom. Same task. Completely different conditions for receiving it.

This is worth pausing on for those of us who make decisions about schools. When we talk about a lesson that “works for everyone,” we are talking about brains with dramatically different capacities to hold and process information. Class size matters. The more students per teacher, the harder it is to meet each brain where it actually is.

The phone hijacking the hallway

Remember that the hallway is narrow?

There are experiments where participants were asked to do working memory tests with their phone in different locations. Next to them, in their pocket, or in another room.

The result: those with the phone closest had the worst access to their working memory capacity. Not because they were using it. Just because it was there.

Researchers think it costs mental effort to resist. Part of the hallway is occupied with not picking up the phone. And especially when the task is dull and difficult — precisely when we are building new shelves.

That is one of the strongest arguments for phone-free schools, which becomes law on 1 July. Not because phones are bad, but because their mere presence steals from the space where learning happens.

Books beat screens, sometimes

This is where it gets counterintuitive.

Nutley talks about something called the Google Effect. We put less effort into remembering things we know we can google again. Why build a shelf for information that is just one click away?

And there is a corresponding effect for physical books. Our ancient brain senses that information is more valuable and unique when it arrives in physical form. “This is only here on this page — save it.” The internet, on the other hand: “I can fetch that whenever.”

Add that a book is distraction-free, that it engages multiple senses, and that we remember where on the page something appeared. We remember visuospatial information: it was up in the top right corner. That is much harder on a screen where everything scrolls past as a uniform surface.

That does not mean screens are always worse. But it does mean that format matters, and that we should not pretend otherwise.

When does technology help, when does it hinder?

This is the question I think is most important, and the one our education debate skips over far too often.

Nutley is clear: digital tools can be fantastic for certain things. Visualising how blood flows through the heart is far better with an animation than with a black-and-white worksheet. Practising and repeating with a good quiz can reinforce learning through exactly the kind of retrieval that builds shelves.

But learning the names of the parts of the heart? Then pen and paper are probably better.

Different things suit different tools. Anyone who claims everything should be digitalised is as wrong as anyone who wants to ban all technology.

What does not work is passivity. Clicking on answers. Copy and paste. Photographing notes instead of writing them. No shelves get built that way, no matter how polished the app is.

And AI?

In March 2026, a major review came from Stanford: “The Evidence Base on AI in K-12: A 2026 Review” from their SCALE Initiative.

Researchers went through over 800 studies on AI in schools. But only around twenty were of sufficient quality to say anything about actual cause and effect.

The conclusion is worth heeding. AI can improve student performance while the tool is in use, but the gains appear to disappear when the technology is removed . In other words: the student performs better — but have they actually learnt anything?

There was, however, a clear pattern. AI tools with pedagogical guardrails — systems that offer hints and guide reasoning rather than just serving up the answer — show more promising results than general chatbots. And AI as support for the teacher, to save time on preparation, appears to work well.

My interpretation: AI belongs in school, but not as a shortcut past thinking. As a tool that compels effort, not replaces it. And perhaps most of all: in the older age groups, after the foundational skills have been automated.

What we must not lose sight of

Nationally, around 16 per cent of students do not reach upper secondary eligibility. In Örebro municipality’s compulsory schools, the figure is even higher: in 2025, 23 per cent of Year 9 students were not eligible for upper secondary school — nearly one in four.

That is our real problem.

We should absolutely equip students for a world with AI. But a student who cannot read fluently, cannot sustain a line of thought, has not built their foundational shelves — that student has no use for the world’s best AI tools.

As Nutley puts it: we need AI literacy. But we also need literacy.

You can use digital tools well. But we must safeguard the learning processes that matter.

And those words did not come from a technophobe. They came from the students themselves, in Nutley’s surveys.

Perhaps it is time we adults started listening.

In the final part of the series, I look outward: at the landmark court rulings against tech companies, the EU’s new regulations, and the Swedish inquiry on age limits. Because something is finally starting to happen.