5 min read

Why Modern Study Tools Are Failing Learners

learning scienceedtechmemory

Most study tools are extremely good at one specific thing, which is helping you remember that you once saw something. They're close to useless at the thing that actually matters, which is helping you use it.

There's an assumption sitting quietly underneath nearly every flashcard app, summariser and revision planner ever shipped: that if you can recall a fact, you've learned it. It's a comfortable assumption to build on, largely because recall is easy to measure and understanding isn't. It's also wrong, and almost every failure below comes out of it.

The recall trap

A student can memorise the definition of osmosis perfectly, word for word, and still not be able to predict what happens to a plant cell dropped into salt water. The fact is sitting right there in their head. What's missing is the connection between the fact and anything you could do with it.

Retrieval that feels effortless is usually retrieval that isn't going to last, and the difficulty involved isn't a bug to be designed away, it's the mechanism doing its job (Bjork and Bjork, 2011).

The summary illusion

Upload an entire textbook to most tools and what comes back is a summary of the entire textbook. It reads well. It's also close to useless to somebody trying to learn the material, and the reason is a bit circular in a way that's easy to miss: a summary is only legible to a reader who already knows the thing being summarised.

Which gives you two cases and neither of them helps. If you understood the material well enough for the summary to make proper sense, you didn't need to study it. If you didn't, the summary has skipped exactly the build-up you were missing.

So the student who reads it hasn't learned the material, they've read a summary and mistaken the warm glow of that-made-sense for understanding (Dunlosky and Metcalfe, 2009).

Nobody ever checks the difference

Here's the failure sitting underneath all the other failures: nothing anywhere in the process forces the student to verify their understanding while there's still time to act on it.

You read, you highlight, you feel ready. And the first occasion anybody tests whether that feeling was accurate is the exam itself, at which point knowing is no longer useful. A tool can generate infinite material for you and still never once require you to demonstrate that you understood a single line of it.

First principle

Nobody is bad at learning. They're using methods that were never designed to produce understanding, methods that test surface recall and call it learning.

Two more cracks

Two smaller failures compound the big ones.

Review scheduled off the wrong signal. Good spaced repetition brings a card back at expanding intervals tuned to how long you actually need to hold it (Ebbinghaus, 1885; Cepeda et al., 2006). Cruder schedulers over-review the things you've already got and under-review the things you nearly lost, so you spend the same hours for worse retention and never notice the difference until the moment it costs you something.

Everything arriving as one undifferentiated blob. One summary, one deck and one quiz for an entire book makes it nearly impossible to work through a specific topic in order, or to concentrate where you're genuinely weak. Real studying is sequential and targeted, and a single blob is neither of those.

And because no individual tool closes the loop, serious students end up running three at once, one for cards, one for document questions, one for explanations, with nothing passing between them. Which leaves the student doing the integration by hand, and integration was the job they needed help with in the first place.

What testing for understanding looks like

ApproachAsksReveals
FlashcardCan you recall the definition?Memory of words
SummaryDoes this read as coherent?Familiarity, not learning
UnderstandingCan you apply it to a new case?Whether you actually get it

A tool that never asks you to produce anything can't tell the difference between a student who understands and one who recognises the page.

the short version

Building the other kind, one that makes you apply an idea, explain it simply, and defend it somewhere unfamiliar, is harder to build and a good deal harder to sit through than another summariser. It's also the only kind that reliably works.

Sources

Read more