Scrolling is the purest form of passive exposure anyone has built. Things go past; you recognise some of them; nothing is ever asked of you. An hour later the feed has shown you hundreds of things, and you'd struggle to recall five.
A question works the other way round. Nothing moves until you've pulled something back out of your own head. Memory research has a name for what that does — the testing effect — and one experiment in particular explains why we built an app blocker around it.
The experiment that made testing look good
In 2006, Henry Roediger and Jeffrey Karpicke published a study in Psychological Science with a simple design. Students had a short prose passage to learn. One group spent four study periods reading it. Another read it in one period and spent the other three writing down everything they could remember — without ever being told what they'd got right.
Five minutes later, the rereaders did better: they recalled 83% of the passage, against 71% for the group that had been testing themselves. A week later the result had flipped. The testing group recalled 61%; the rereaders, 40%. And the testing group had read the passage far less — 3.4 times on average, against 14.2 times for the rereaders.
Familiar isn't the same as remembered
The same study found something stranger. The students rated how well they expected to remember the passage in a week, and the rereaders were the most confident of all three groups. They went on to remember the least.
That's the trap a feed sets. Seeing something again makes it feel familiar, and familiar feels like knowing. Recognising a face, a meme, a fact you half-heard last week — all of it is smooth and easy, and none of it is the same as being able to produce the thing when someone asks.
Aren't quiz questions just recognition?
Fair objection. Mr. Grummel's questions are multiple choice — four options, one right — so the answer is sitting on the screen. The students in the 2006 study were writing from memory, which is harder.
The research is kinder to multiple choice than its reputation. In 2012, Jeri Little, Elizabeth Bjork, Robert Bjork and Genna Angello reported, also in Psychological Science, that properly built multiple-choice tests can trigger the same productive retrieval — provided the wrong options are plausible enough that you have to work out why they're wrong. That condition matters: a question with one sensible answer and three silly ones is a tap, not a test. It's one more reason how we write questions gets as much care as the classroom around them.
What that means at the door of a blocked app
Every time you reach for a blocked app, Mr. Grummel turns the reach into a small retrieval exercise: questions from a curriculum you picked, and three right in a row to get through. A feed would have shown you three more things to recognise. He asks you to come up with three answers instead.
When he asks the same thing again is a separate job — the scheduler's, which we wrote about in spacing, streaks and the forgetting curve. This note is only about the act of being asked.
What we're still unsure about
Both studies were run in controlled settings, with students who had sat down to learn. Someone answering questions to get into a blocked app is in a hurry and mostly wants the questions to be over. We don't know yet how much of the testing effect survives that kind of impatience, and we're not going to claim numbers we haven't measured. What the research does say is which way it points: being asked left more behind than being shown.
Mr. Grummel is out now on iPhone, with Android in closed testing. If you'd like to be asked first, get the app.
Make your next unlock teach you something
Mr. Grummel locks the apps you lose time to. Three questions in a row get you back in, for as long as you choose.
7 days free, then $4.99 a month. iPhone only for now.
On Android? Join the beta.