Guess before you look it up

What happens in your brain

The guess sends you searching. To guess, you have to reach for anything you know that might be related. In one study, guessing helped people remember word pairs that were related, like tide and beach, and did nothing for pairs that weren't. Evidence: decent My reading is that the guess works by pulling up related knowledge the answer can attach to. That step is a reading of the results, not something measured. Evidence: worth trying Another set of experiments found a gain with unfamiliar words, where a guess had little to draw on, so it isn't the whole story. Evidence: decent

You want to know if you were right. In a scan study, 19 students guessed at 40 trivia questions and rated how curious they were. The more curious they were while reading a question, the more active the caudate, part of the striatum, a deep area tied to wanting. Evidence: decent When the answer showed after a wrong guess, curiosity went with more activity in areas tied to memory. Evidence: decent In a follow-up without the scanner, 12 people came back one to two weeks later, and higher curiosity went with better recall of the answers they'd got wrong. Evidence: decent Scans track blood flow, so this shows correlation. It doesn't show that dopamine causes the gain. Evidence: decent Everyone in that study guessed, so it compares curiosity, not guessing against not guessing; that a guess helps partly by making you curious is a link I'm drawing, not a finding. Evidence: worth trying For more on curiosity and memory, see ask someone to explain their thing.

The answer lands on a gap. When the answer isn't what you guessed, there's a gap between what you expected and what you got. Researchers call it a prediction error. One review argues that the surprise at this gap is part of why predicting first helps memory. It makes that case for predicting, and says it's what sets predicting apart from plain guessing. That's theory, not settled. Evidence: worth trying

You underrate it. In three experiments with new word definitions and foreign words, people predicted they'd remember the items they had guessed at worse than the ones they'd simply read. They remembered them better. Evidence: decent

The proof

• In six experiments, students guessed at made-up trivia questions or weakly linked word pairs. Guessing and then seeing the answer beat just reading the answer, on a later test, even though the guesses were almost always wrong. Lab studies, mostly single sessions. Evidence: decent
• In five experiments, students about to read an essay on vision took a short test on its key ideas first. They got most answers wrong. Counting only the ones they missed, they still did better on a later test than students given that time to read more. One essay on one topic. Evidence: decent
• In five experiments with 1,573 online volunteers and students reading factual texts, tested 5 minutes or 48 hours later, a quiz before reading beat the same quiz after reading. In the first experiment, scores were about 11% higher on the items they'd guessed at and 7% higher on items they hadn't. Evidence: decent
• A meta-analysis, which pools many studies, with its methods fixed before it started, found a moderate gain on the guessed item. The effect size, a standard measure of how big a difference is, was 0.54 across 97 comparisons. On material that wasn't guessed at it was 0.04 across 91 comparisons: essentially zero. Evidence: decent
• Guessing helped with related word pairs, not with unrelated ones. Evidence: decent
• In four experiments with unfamiliar words and foreign vocabulary, wrong guesses followed by the answer beat reading it, on a later test. In the three that asked, people predicted the opposite. Evidence: decent

How to do it

• Guess before you look: Google, the manual, the colleague about to tell you. Wrong guesses still helped in the studies. Evidence: decent In two experiments with students reading texts, a guess kept in your head worked as well as a typed one. Evidence: decent The other studies had people type or write their guess. Evidence: decent
• Make it a real attempt. With word pairs, guessing helped only when the guess could draw on something related. Evidence: decent Another set of experiments found a gain even with unfamiliar words, so this isn't settled. Evidence: decent A coin flip gives the answer nothing to hook onto. Evidence: worth trying
• Check straight away. In the studies here the answer came soon after, in the same sitting. Evidence: decent With word pairs, even a short delay before the answer removed the gain. Evidence: decent A guess you never check risks the wrong answer sticking. Evidence: worth trying
• Before a manual or a report, guess the answers to the few things you need from it. A test on key ideas before reading beat extra reading time. Evidence: decent

Watch out

Check the real answer straight away. The studies here all gave it in the same sitting, and with word pairs a short delay removed the gain. Evidence: decent And never act on a guess where a mistake costs something: medicine doses, electrics, gas. Look those up first.

It helps the thing you guessed, not the whole chapter. Evidence: decent One large set of experiments found a smaller gain on items nobody guessed at; pooled across studies, that comes out at about zero, so don't count on it. Evidence: decent And before a text, one recent study found that reading a list of what it would cover worked as well as guessing. Evidence: decent

You'll hear this sold as "failing is the best way to learn". The studies show something narrower: a wrong guess followed straight away by the right answer helps that answer stick. They don't show that failing in general teaches you more, and it isn't a general brain boost. Evidence: decent

The evidence is students and online volunteers, with trivia, word pairs and short texts, mostly in single sessions. Using it at work is an extrapolation. Evidence: decent The brain part is one small scan study that shows correlation, and a review that's theory. Evidence: worth trying