One fact per question

A useful card-design judgement is to ask for one short, specific response at a time. Woźniak calls this the minimum information principle, but presents it as practitioner guidance rather than the result of a controlled comparison.¹ There is no universal definition of one atomic fact. The practical test is whether the prompt contains targets that could succeed, fail or need review independently.

Bundling targets creates a scheduling problem. If one card asks for a definition and an example, the harder part can determine how often the whole card returns. Splitting the targets permits separate review.¹ Anki’s guidance likewise recommends separate cards for separately assessable tasks, while its note system can generate and track recognition and production prompts independently.²³

Memory research supports making cues specific, but it does not prove an exact card-size rule. Reviews of cue overload describe how a cue becomes less diagnostic when it points toward several possible targets.⁴ In the original fan-effect experiments, people took longer to verify a learned proposition when its concepts had more learned associations.⁵ That demonstrates a retrieval cost from competing associations, not a maximum answer length for flashcards.

With a deck and twenty minutes a day, inspect cards that produce partly correct answers. If the definition is reliable but the example is not, turn them into separate prompts.¹² If recognition is easy but production fails, test those directions separately. Keep enough context to identify the intended target, but remove incidental material that does not need to be retrieved.⁶

Boundary. This workflow is a reasonable procedure built above the established retrieval-practice foundation. No direct long-term experiment has compared otherwise identical decks containing single-target and bundled cards. Atomising can also change the capability being learned: separate prompts may not preserve fluent recitation or spontaneous recall of an unprompted idea.¹