Active recall is easy to describe and easy to do a weaker version of

This page is about the mechanics: what actually counts as an attempt, how to write a cue that has exactly one answer, when the answer should arrive, and the one cost of the technique that almost nobody mentions. The tool below turns your notes into cued prompts so you can see the difference immediately.

Recognising is not remembering, and it feels identical

Seeing the answer and thinking yes, of course is recognition memory, and it is far easier than recall because the item itself is doing the retrieving for you. Every exam that asks you to produce something is asking for the harder one. This matters because your sense of how well you know the material is measured by researchers as a judgment of learning, and judgments made immediately after studying are reliably too high, a core finding of metamemory research. The practical consequence is uncomfortable. The technique that feels most productive is usually the one that is teaching you least, and the only way out is to make the attempt before you look. Studying smarter, if the phrase is to mean anything, means noticing that difference and acting on it rather than working longer.

What counts as an attempt

An attempt has three parts and most people skip the third. You see a cue and no answer. You produce something, aloud or written, even if it is wrong or partial. And you stay with it past the point of discomfort, which is where the useful work happens. Ten seconds is a reasonable floor: quitting after two is a glance, not an attempt. That struggling state has its own literature, since being unable to produce a word you clearly know, tip of the tongue, is a partial retrieval rather than a failure of storage. Writing the attempt down rather than thinking it is worth the extra seconds, because a thought can be vague and a sentence cannot, and vagueness is exactly what lets a half memory pass as a whole one.

A cue with one answer, and no more

Most bad cards fail the same way: the front admits several correct answers, so you produce one, see a different one on the back, and cannot tell whether you were wrong. Front sides such as the heart or chapter four are prompts to recite, not cues. Compare that with which chamber of the heart drives the systemic circuit, which has exactly one answer and can be scored honestly. The reason this matters more than it sounds is the encoding specificity principle: what you can retrieve depends on the match between the cue at test and the cue at learning, so a cue you invent while revising should resemble the question you will actually be asked. Two related results are worth knowing. Context dependent memory means practising in one fixed setting can leave part of the memory attached to that setting, and transfer appropriate processing means practising the operation the exam demands beats practising a more convenient one.

Being wrong first is the productive part

A failed attempt followed by the answer beats reading the answer straight, which is the counterintuitive core of the whole technique and the reason it is grouped under desirable difficulties: conditions that make learning slower and worse in the session and better weeks later. Practically this means you should not narrow a cue until you can answer it comfortably, because the discomfort is doing the work. It also means the answer should arrive after the attempt, never alongside it, so a card whose back is visible while you read the front is not a card. Delaying feedback by a few seconds is enough; delaying it by a day mostly just wastes the attempt.

Free recall, the version with no cue at all

Close everything, take a blank page, and write down everything you can produce about the topic in ten minutes. This is free recall, and it is the most demanding form of the technique because nothing prompts you. It also does something cards cannot: it tests your map of the topic rather than a list of facts, so the gaps it exposes are structural, whole sections you did not know were missing. The reliable procedure is to write first, then open the source and mark what you missed in a different colour, then make cards only from the marks. Doing this once at the start of a revision week is usually worth more than an extra hundred cards, and explaining the same material to somebody, learning by teaching, is the spoken version of the same test.

The cost of the technique, stated honestly

Retrieving one item can make related items you did not retrieve temporarily harder to reach, a measured effect called retrieval induced forgetting. In practice it means drilling a favourite subset of a topic can suppress its neighbours for a while, and the fix is coverage rather than abandoning the technique: test across the whole topic rather than the third you enjoy. This is a real limitation and it belongs on a page that recommends the method. The broader evidence still points one way. Dunlosky and colleagues rated practice testing high utility while summarising and highlighting were rated low, and the Roediger and Karpicke experiments found the advantage over rereading widened as the delay to the final test grew.

Having the cues written for you

Writing good cues by hand is the slow part, and it is why most people stop. Upload a PDF, a photograph of a page or a camera shot and WeSolve+ reads the file itself and writes the cues, then keeps the ones you get wrong in their own list so the next session opens with them rather than with what you already know. Cards return on a schedule instead of when you happen to think of them, which handles the timing side that this page has deliberately left alone. It runs at wesolveapp.com in any browser and as an App Store app on iPhone and iPad. Free to start, with two quizzes and two uploads a day, two recaps a week and two duels a day. There is no separate Android app. Numbers on pricing, and the timing on spaced repetition.

Last updated: 2026-08-15