Learning the Forgetting Curve

WIRED magazine had a great article by Gary Wolf a while back, that I can’t seem to get out of my head :
Want to Remember Everything You’ve Ever Learned?  Surrender to this Algorithm

The article chronicles the hermetic exploits of one Piotr Wozniak, creator of SuperMemo. SuperMemo is a computer program that uses concepts from the field of experimental psychology and memory research (in particular, the Forgetting Curve) to optimize retention of learned information.  Here is a graph from the WIRED article, that illustrates how this works:


HOW MANY TIMES MUST I REMIND YOU-- oh, wait, four times.  Okay.

HOW MANY TIMES MUST I REMIND YOU-- oh, wait, four times. Okay.

What This Graph Describes: We start forgetting new information almost as soon as we learn it.  For the information we want to retain, there is an optimum time to review information: not too soon, not too late, but just around the time where the information is just about to become difficult to retrieve. If we reinforce our memory at those key moments  (and it varies by person and situation, although there are some rough averages: Day 2, 10, 30 & 60) the Forgetting Curve can be substantially altered. This idea of having strategically placed intervals of time between reviews is sometimes called the “Spacing Effect.”

Why This Graph Matters: As most of us know from our college days, we can get pretty good at cramming before a test and regurgitating a whole bunch of information, only to forget it completely within just a few days. That form of education is great if you want to be an efficient test-taker, but it’s not true learning. We may know that continuous learning is important, and we may even be passionate about continuous learning, but in reality, our time is limited and we are being constantly bombarded with new, competing information. It’s unrealistic to assume we can review everything we learn, all the time. Understanding the dynamics of the Forgetting Curve is very useful for learners and instructors alike… and it’s especially useful for those who design Learning & Development plans.  (People in marketing can apply this stuff too, to optimize their targeted ad spend and avoid unnecessary carpet-bombing, but we’ll leave that aside in this discussion.) 

The original source for the Forgetting Curve is Hermann Ebbinghaus, a German philosopher who in 1885 conducted a series of experiments on memory. The concept of the Forgetting Curve was further developed and explained by experimental psychologists throughout the twentieth century, and has received a lot of attention in the last few decades.

Popular author Tony Buzan has a variation of the Forgetting Curve that illustrates the idea that subjects will not actually be able to immediately recall ALL of the new information they’ve been presented with, but rather, can achieve 100% recall if they are given a short break to properly “absorb” or “process” the information. Here’s what his diagram looks like:


Tony Buzan's take on the Forgetting Curve

Tony Buzan's take on the Forgetting Curve

Buzan does not provide a source or citation for his variant on the Forgetting Curve (and others have tried to find his source, but to no avail). He might be simply making a semantic distinction between the moment when something is initially TAUGHT, versus the moment when initially LEARNED.  However, the Forgetting Curves that Ebbinghaus and others have drawn always start at 100%, because these scientists set their “Time Zero” at the successful completion of a test proving that the new material was, in fact, learned. So I think Buzan was picking up on something else, namely, the role of sub-conscious processing of information.  Buzan is asserting that memory recall actually IMPROVES during the period immediately following initial learning. I think this is supported by some of the classic research on Forgetting Curves (Jenkins & Dallenback, 1924) which shows that sleeping helps with reinforcement, and not just the ability to learn in the first place.


Those of us who do teaching and training may have an intuitive grasp of these ideas of sub-conscious processing and the Spacing Effect.  For example, as instructors we often find it helpful to review a section of material with our participants AFTER a ten minute break, as opposed to immediately BEFORE the break.  


The following additional variation on the Forgetting Curve and the Spacing Effect comes from the University of Waterloo (some other good links here:  http://www.ravenelbridge.net/~cfs/forgetting/). This diagram illustrates how the time required for each review will decrease with each interval, i.e., a learner may require 10 minutes of review the first time, 5 minutes of review the second time, and so forth. This graph was designed to reflect the reality of university life, where student are being firehosed with information every day. Waiting a week before doing the first review would mean the student now has to do a much longer review — versus a quick 10 minute review the next day.


So now that we’ve locked this little meme into the menagerie, let’s tweak it some more:


I think we can use this Forgetting Curve idea not just to reinforce ideas we’ve already learned (ie, as a follow-up planning tool), but we can also use the curve as a kind of planning guide IN ADVANCE of learning new material, before it’s even presented. 


In other words, given that learning is a continuous process (and a continuous requirement for pretty much everyone these days), it would be interesting to see if one could organize the long term “learning flow” so that ideas and concepts got introduced at just the right time, and in just the right way, so that they optimally cement together with previously learned material, based on the Forgetting Curve. 


This is something that isn’t addressed by  traditional curriculum scheduling techniques (which usually follow a seasonal, budgetary or event-based schedule), nor is it something that’s addressed by ad hoc or self-directed study plans.


Here is my own rendering of the Learning Curve / Spacing Effect diagram:

 Taking the above idea and thinking about it visually, with a simple scenario of having a series of four lessons every 10 days (assuming Day 10 is also the optimum time for a second review) you get something that looks like this:


Let’s take it one step further and introduce a new lesson, after Day 60, to give us a sense of the inflows and outflows of data that our brains are constantly dealing with:


So to review:  This is information, and this is how our brains process information.  Any questions?




About danspira

My blog is at: http://danspira.com. My face in real life appears at a higher resolution, although I do feel pixelated sometimes.

Posted on July 24, 2008, in Information Design, Learning. Bookmark the permalink. 11 Comments.

  1. on a culinary side note,

    eggs are supposed to be cooked using slow heat.

    even if fried.

    but i guess the punch of the message is kinda lost then…

    • Not universally true at all. Julia Child, for example, will tell you to use very high heat to cook an omelet. The whole cooking process should take a minute or so.

  2. in school, we were always told the half-life (the time it takes to
    lose half of something) of everything you learn and don’t use, is 3
    years. (obviously there is some variation between different people)

    (half-life’s are used in order not to get into “Xeno” style messiness
    where the curve never reaches the end… the half way point is a nice
    constant reference point.)

    My personal belief and system – and i’m glad to see it supported with
    what u wrote – is to make multiple correlation with the new material,
    as diverse as possible.

    relate a cooking concept to a business concept, seeing the Krebs cycle
    in a flower (ok im pushing it), etc.. lateral connections.

  3. it is not conceivable to review everything learned on a regular basis.
    However, with lateral connections in place, the probability of coming
    across the concept in our day to day lives (ie, using the information)
    increases… thus raising the half-life.

    if i were a teacher, i would try and connect the new information to as
    many varied examples as possible. As such different situations come up
    in life, they will (to the attentive student) trigger a reminder and
    increase the overall retaining percentage. this is, of course, much
    easier to do in a field like marketting than let’s say, biochemistry.

    speaking of lateral connections, i had read somewhere (and i forget
    where lol) that one definition of intelligence is the ability to
    relate a learned concept to as many seemingly unrelated scenarios as
    possible. or in database terms, the ability to cross-index

    • Nareg — Just doing a little review here (18 months later… the Spacing Effect is, in fact, infinite) and noticed that I didn’t say this before, so I’ll say it now: You’ve done a beautiful job summing up much of what is known as Adult Learning (or sometimes, Accelerated Learning) Principles.

      Learning happens because people are able to connect new information with what they already know. Good instructors facilitate the making of those connections, allowing information to move from working memory (temporary storage) into longer-term memory. Who is a good learner? One who can make those connections on their own… they have, as you said, the ability to cross-index information.

  4. Is there a comuter program out there that would remind you of what needed to be reviewed and when?

    • Yes, there are all kinds of computer programs out there related to the forgetting curve, although I haven’t tested any of them for learning design purposes. “SuperMemo” is the original one. It seems that many of these personal memory management (PMM?) software programs are used by people who are learning foreign languages.

  1. Pingback: How To Learn Better: Space Out and Get Tested, Early and Often « Meme Menagerie

  2. Pingback: 2010 in review (auto-generated navel-contemplation) « Meme Menagerie

  3. Pingback: What are you good at remembering? « Meme Menagerie

  4. Pingback: Wash, Rinse, Repeat | Dan Spira

Leave a Reply -- for humans only, no spambots

Please log in using one of these methods to post your comment:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Google+ photo

You are commenting using your Google+ account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )


Connecting to %s

%d bloggers like this: