Which simple exponential gives R(S) = 90%?

R(t) = 0.9^(t/S), for S > 0. This is an exponential illustration, not the FSRS-6 forgetting curve.

How is relearning savings calculated?

For comparable learning times T₀ and T₁, savings = 100 × (T₀ − T₁)/T₀, with T₀ > 0. It is not a recall probability. T₀ is the initial learning time and T₁ the relearning time to the same criterion.

How should you handle a repeatedly forgotten card?

Check understanding, ambiguous wording and missing context; simplify, split or suspend the card when appropriate.

How do Anki's Again and Hard ratings differ?

Again means recall failed. Hard means the answer was recalled correctly but with difficulty.

Is an FSRS forgetting curve a universal biological law?

No. It is a predictive model fitted to review data; its accuracy depends on the data and learning conditions.

What is retrievability R in FSRS?

The model's estimated probability of recalling a particular card now, given the learner's review history.

What does the spacing effect compare?

Learning spread across separate occasions with learning concentrated together. Benefits depend on the task and test delay.

What usually happens when desired retention is increased?

Intervals shorten and review workload increases. Choose a target that fits your goals and available time.

Do paper cards become impossible above a fixed deck size?

No. Digital tools automate scheduling and tracking, but there is no universal card-count cutoff for paper systems.

What does FSRS difficulty D describe?

A model estimate of how difficult it is to strengthen the card's memory through review, not its current recall probability.

Is a leech always a card missed four times?

No. Software uses configurable lapse thresholds; the threshold is a scheduling rule, not a law of memory.

What does FSRS do after a lapse?

It updates its memory estimates using the rating and prior state; a failed review does not literally erase a memory.

Are FSRS states direct measurements of brain processes?

No. They are estimates from behavior, not measurements of synapses, brain regions or molecular mechanisms.

How do equal and expanding spacing differ?

Equal spacing uses constant gaps; expanding spacing lengthens successive gaps. Neither schedule is universally best.

What intervals does original SM-2 use after successful repetitions?

Initially 1 day, then 6 days; later intervals multiply the previous interval by EF and round up to whole days.

How did Ebbinghaus study forgetting?

He learned lists and measured how much effort was saved when relearning them after different delays.

Does daily review volume always equal ten times new cards?

No. It depends on learning history, difficulty, retention settings, lapses and the number of new cards.

What forgetting curve does FSRS-6 use?

R(t) = (1 + k·t/S)^(−d), where d = w₂₀ > 0 and k = 0.9^(−1/d) − 1. Thus R(S) = 0.9. Here t is time since the last review, in the same units as S.

Why use short learning steps for new cards?

They provide early practice and feedback. Same-day success does not replace checking recall after a longer delay.

Why is R(t) = e^(−t/S) inconsistent with FSRS's definition of S?

At t = S it gives e^(−1) ≈ 36.8%, not 90%. The parameter has a different meaning in that exponential.

Does a long review interval imply permanent memory?

No. Knowledge can still be forgotten or become outdated; check recall and application when needed.

What does a Leitner box represent?

A group of cards with a particular review frequency, rather than a precise measurement of memory.

How do the interstudy interval and retention interval differ?

The interstudy interval separates study sessions; the retention interval separates the last study session from the final test.

What is stability S in FSRS?

The modeled interval for recall probability to fall to 90%. This definition does not change with the chosen retention target.

If SM-2's EF reaches 1.3, is it stuck there?

No. The minimum is 1.3, but a grade of 5 increases EF by 0.1; a grade of 4 leaves it unchanged.

What does desired retention mean in an FSRS scheduler?

The target predicted recall probability when a card is scheduled for review, not an exam pass mark.

Why is there no universal optimal study gap?

The useful gap depends on the material, learner and delay until the final test; longer gaps are not always better.

What is EF in original SM-2?

An interval multiplier called the easiness factor. It starts at 2.5 and has a lower bound of 1.3.

What is a practical response to a review backlog?

Reduce new material temporarily, prioritize important content and plan a sustainable amount of review time.

Why grade recall attempts consistently?

Scheduling estimates use the review history. Mislabeling failed recall as success gives the model misleading evidence.

Must a Leitner system use five boxes and doubling intervals?

No. Box counts and schedules vary; specify the rules and apply them consistently.

What happens after a wrong answer in a Leitner system?

The card moves to a more frequent review group according to the chosen rules, often the first box.

Spaced Repetition, Ebbinghaus Forgetting Curve, Leitner Boxes & FSRS Scheduling

Master Spaced Repetition, Ebbinghaus Forgetting Curve, Leitner Boxes & FSRS Scheduling with 32 active-recall flashcards covering core definitions, equations, mechanisms, and high-yield problem solving.

Level: basic
Languages: en → en
Cards: 32
Markdown: Spaced Repetition, Ebbinghaus Forgetting Curve, Leitner Boxes & FSRS Scheduling (.md)

Flashcards (32)

#	Front (Term / Question)	Back (Definition / Answer)
1	Which simple exponential gives R(S) = 90%?	R(t) = 0.9^(t/S), for S > 0. This is an exponential illustration, not the FSRS-6 forgetting curve.
2	How is relearning savings calculated?	For comparable learning times T₀ and T₁, savings = 100 × (T₀ − T₁)/T₀, with T₀ > 0. It is not a recall probability. T₀ is the initial learning time and T₁ the relearning time to the same criterion.
3	How should you handle a repeatedly forgotten card?	Check understanding, ambiguous wording and missing context; simplify, split or suspend the card when appropriate.
4	How do Anki's Again and Hard ratings differ?	Again means recall failed. Hard means the answer was recalled correctly but with difficulty.
5	Is an FSRS forgetting curve a universal biological law?	No. It is a predictive model fitted to review data; its accuracy depends on the data and learning conditions.
6	What is retrievability R in FSRS?	The model's estimated probability of recalling a particular card now, given the learner's review history.
7	What does the spacing effect compare?	Learning spread across separate occasions with learning concentrated together. Benefits depend on the task and test delay.
8	What usually happens when desired retention is increased?	Intervals shorten and review workload increases. Choose a target that fits your goals and available time.
9	Do paper cards become impossible above a fixed deck size?	No. Digital tools automate scheduling and tracking, but there is no universal card-count cutoff for paper systems.
10	What does FSRS difficulty D describe?	A model estimate of how difficult it is to strengthen the card's memory through review, not its current recall probability.
11	Is a leech always a card missed four times?	No. Software uses configurable lapse thresholds; the threshold is a scheduling rule, not a law of memory.
12	What does FSRS do after a lapse?	It updates its memory estimates using the rating and prior state; a failed review does not literally erase a memory.
13	Are FSRS states direct measurements of brain processes?	No. They are estimates from behavior, not measurements of synapses, brain regions or molecular mechanisms.
14	How do equal and expanding spacing differ?	Equal spacing uses constant gaps; expanding spacing lengthens successive gaps. Neither schedule is universally best.
15	What intervals does original SM-2 use after successful repetitions?	Initially 1 day, then 6 days; later intervals multiply the previous interval by EF and round up to whole days.
16	How did Ebbinghaus study forgetting?	He learned lists and measured how much effort was saved when relearning them after different delays.
17	Does daily review volume always equal ten times new cards?	No. It depends on learning history, difficulty, retention settings, lapses and the number of new cards.
18	What forgetting curve does FSRS-6 use?	R(t) = (1 + k·t/S)^(−d), where d = w₂₀ > 0 and k = 0.9^(−1/d) − 1. Thus R(S) = 0.9. Here t is time since the last review, in the same units as S.
19	Why use short learning steps for new cards?	They provide early practice and feedback. Same-day success does not replace checking recall after a longer delay.
20	Why is R(t) = e^(−t/S) inconsistent with FSRS's definition of S?	At t = S it gives e^(−1) ≈ 36.8%, not 90%. The parameter has a different meaning in that exponential.
21	Does a long review interval imply permanent memory?	No. Knowledge can still be forgotten or become outdated; check recall and application when needed.
22	What does a Leitner box represent?	A group of cards with a particular review frequency, rather than a precise measurement of memory.
23	How do the interstudy interval and retention interval differ?	The interstudy interval separates study sessions; the retention interval separates the last study session from the final test.
24	What is stability S in FSRS?	The modeled interval for recall probability to fall to 90%. This definition does not change with the chosen retention target.
25	If SM-2's EF reaches 1.3, is it stuck there?	No. The minimum is 1.3, but a grade of 5 increases EF by 0.1; a grade of 4 leaves it unchanged.
26	What does desired retention mean in an FSRS scheduler?	The target predicted recall probability when a card is scheduled for review, not an exam pass mark.
27	Why is there no universal optimal study gap?	The useful gap depends on the material, learner and delay until the final test; longer gaps are not always better.
28	What is EF in original SM-2?	An interval multiplier called the easiness factor. It starts at 2.5 and has a lower bound of 1.3.
29	What is a practical response to a review backlog?	Reduce new material temporarily, prioritize important content and plan a sustainable amount of review time.
30	Why grade recall attempts consistently?	Scheduling estimates use the review history. Mislabeling failed recall as success gives the model misleading evidence.
31	Must a Leitner system use five boxes and doubling intervals?	No. Box counts and schedules vary; specify the rules and apply them consistently.
32	What happens after a wrong answer in a Leitner system?	The card moves to a more frequent review group according to the chosen rules, often the first box.