Alternative terms for spaced repetition
This text is a part of a series of articles about SuperMemo, a pioneer of spaced repetition software since 1987
Terminological confusion
There is a monumental confusion in terminology related to spaced repetition. Back in 1985, in the university library, I could not find any good research in the area memory that would tell me how to space repetitions. Today, with Google and Wikipedia, the job is still hindered by the fact that multiple lines of thought, research, and practice need to coalesce under a single terminological umbrella. This text is intended to clear up a bit of the fog.
Adoption of the term: "spaced repetition"
SuperMemo adopted the term spaced repetition in 1999 after quite a deal of deliberation, word count searches at AltaVista, and several prior attempts at other names (e.g. "repetition spacing"). Due to popular adoption of SuperMemo, and then Anki and Mnemosyne, and then popular on-line applications such as Quizlet, Memrise, and Duolingo, the term "spaced repetition" became pretty well established in development of learning software. Spaced repetition now got its own template at Wikipedia: {{Spaced repetition}}. The template neatly refers to associated terminology such as the spacing effect, testing effect, forgetting curve, etc., it lists all major competitors to SuperMemo, as well as major names from the history of spaced repetition (e.g. Ebbinghaus, Bjork, etc.). The remaining terms at Wikipedia do not get listed in the template, and await merger propositions.
Alternative terminology
In the meantime, other names, such as "spaced retrieval" still retain their own Wikipedia entries, often with more focus on memory research. Scientific community is pretty resistant to the pressure of pop science, however, the fact that the term spaced repetition has been used in literature for quite a while, will hopefully break that resistance at some point to facilitate further integration of terminology, thought, and inspiration.
Some of the terminology listed below overlaps only partly, may form a subset or super-set of the term "spaced repetition", depending on the context. For example, the schedule may not be expanding (e.g. using equal or uniform intervals), the intervals may be measured in unusual units (e.g. intervening items, trials, minutes, etc.), memories might be procedural, or passive, items might be complex (e.g. chapters of a book), etc.
Definition of spaced repetition
Here is the proposed list of conditions necessary to delineate the term "spaced repetition":
- memories learned must be atomic (e.g. chapters of the book do not qualify, nor do videos, nor procedures)
- memories must be subject to active retrieval/recall (i.e. evoke the testing effect)
- intervals must be proven as optimum by measurement using a selected criterion (e.g. forgetting index, stability increase, universal metric, etc.)
- the criterion chosen must evoke the spacing effect (i.e. popular cramming in some most popular systems would not qualify)
By the above criteria, original Leitner system does not qualify. Nor does Pimsleur. Nor wild guesses published all over the web. Algorithm SM-2 itself is too popular to disqualify, however, it is still a heuristic. It does not fully meet the condition of "computing optimum intervals using a criterion". The algorithms used by Duolingo and Quizlet use state-of-the-art machine learning toolset, but the design is too conducive to cramming (which actually is a popular demand of users affected by the system of schooling). FullRecall is a great idea, but the design leaves a bit of doubt, esp. in the area of proving optimality.
The list of terms
Here is the list of alternative names (please write to me if you see inaccurancies):
- spaced repetition, popularized by SuperMemo, Google count 332,000 (May 1, 2018), on the rise in popularity, recommended in all contexts
- repetition spacing or spacing of repetitions, used by SuperMemo in the years 1992-1999, used in scientific literature, including the first ever publication about computational spaced repetition, Google count 13,300 and 19,600, e.g. Hintzman 1973, Melton 1970 (as a superset of distributed practice), in decline
- spaced retrieval, scientific literature (incl. memory therapies), sometimes considered a superset of expanded retrieval, Google count 32,400, in slow decline, spaced retrieval at Wikipedia should be merged as the research part of the spaced repetition article at Wikipedia
- expanded retrieval, or expanding retrieval (Karpicke, with spaces measured in trials), Google count 7,000 (expanding and expanded, each)
- graduated interval, Pimsleur (1967), Google count 40,000 (May 2018)
- spaced practice (Google count 31,000), expanded practice, expanding practice, or distributed practice (Google count 92,000)(e.g. Melton 1970). Distributed practice may be used to mean the same as expanding practice, or just be used in opposition to massed practice where frequent review is opposed by sparse review
- spaced rehearsal (Google count 12,000), expanding rehearsal (Google count 4,600), or expanded rehearsal (Google count 2,700). Used in literature, esp. older literature, e.g. Landauer & Bjork)
- retrieval practice (Google count 99,000) (e.g. Karpicke)
- optimum schedule (literature, e.g. Pavlik, SuperMemo)
- spaced learning (Google count 50,000)(Kelly): focuses on building long-term memories in very short spans of time
- spaced education (Google count 27,000)
- repeated retrieval (Google count 13,700) (literature, superset of spaced retrieval)
- SuperMemo method (1989-1995)(Google count 7,200)
- optimum intervalization, used only by SuperMemo 1990-1992 (absent from Google)
- scheduling repetitions (1992-1995)(present at Google mostly in old SuperMemo documents)
- progressive schedule (my Master's Thesis, 1990)(Google count 11,800, often with a different meaning)
Disclaimer
My brain is not a good evaluator of spaced repetition terminology. After decades of being submerged in the issue it instantly translates all terminology to "spaced repetition", classifies most as boring or irrelevant, and remains primarily focused on the meaning employed in the most up-to-date versions of SuperMemo. Please assist me by mailing your corrections. Knowledge can be a curse too.