How to Space Retrieval So Everyone Remembers More

Our brains are forgetting machines.

While everything we ever learn appears to be stored in the brain, remembering is effortful, uses up calories, and so is something we biologically try to avoid.

So, the brain only optimises memories it thinks will be useful.

A good measure of whether it is useful, is whether the memory is revisited.

Strong Neural Pathways

A strong memory means your brain has developed a strong neural pathway. If the memory is recent, it will be easy to recall, so no strong neural pathway will be built.

The further away it is, the longer ago you last retrieved it, the stronger that neural pathway will be.

Until it isn’t.

If the gap is too long, you will forget it. The neural pathway will have shrunk. And you’ll have to start over again, near the beginning.

Spacing: Choosing the Right Gaps

The rule is that you want to test the memory when it can have a 90% chance of remembering, but not 89% and not 91%. This is an ideal.

Put it another way, if we have to retrieve 10 pieces of information we have studied, the gap means that we would remember 9 of them.

For the first retrieval, the first gap might be anything from 1 hour to 3 days, depending on the learner and the difficulty of the information.

After that, the gaps increase approximately by doubling.

But the gap should never be more than 20% of the time between learning it and the exam.

SuperMemo Research on Memory (Improving on the Ebbinghaus Forgetting Curve)

Anki, Quizlet, Duolingo – all memory apps, apply versions of the SuperMemo algorithm which spaces questions at exactly the right time for each individual.

This is a gold standard we will never reach. But it is a standard we can use in our curriculum design for:

1.       In class knowledge retrieval

2.       Online knowledge retrieval

3.       Homework design and planning

4.       Assessment design and planning

5.       How we analyse the results of our assessments in order to improve our curriculum

6.       How we motivate students with a high success rate

Key Stage 4 Example

It is day one of year 10. There are 20 months to the exam – call it 600 days. 20% of that is 60 days.

1.       You learn something on day 1. You retrieve it on day 3.

2.       You double the gap to retrieve it again on day 6.

3.       You double again to retrieve it 12 days later, on day 18.

4.       You double it again to retrieve it 24 days later, on day 42.

5.       You double it again to retrieve it 48 days later, on day 90.

6.       You double it again to retrieve it 96 days later, on day 186.

7.       You double it again to retrieve it 192 days later, on day 378.

8.       You can’t double it, as that takes us beyond our 600 days. So you test 60 days before the exam, day 540.

9.       You must also factor in what students learned in year 7, 8 and 9. Some of that may well be knowledge needed for year 11 exams. Still more is knowledge that you would like students to have for life. So, that too will need to be retrieved, probably at gaps between 6 months and 12 months, depending on how long ago it was learned.

(Source: How We Learn, by Benedict Carey)

Implications for Mock Exams

Yes, the optimum time to have your mocks is 60 days before the exam – two months before, as the gap can’t be greater than 20%. March, or April for May and June exams.

For knowledge learned in year 11, the optimum time is the same. September to April is 8 months, roughly 242 days. So new learning can’t have a gap greater than 20% of this, 48 days before the exam. So, again, for an exam on the 10^th June, you need a mock in the last week of March.

A massive advantage of this is that students would only need to revise what they don’t know. Everything else will be in long term memory – at least from the point at which it will need to be retrieved – in May and June.

Implications for Start of Year 11

There are 46 weeks leading up to the summer holidays.

The 6 week break means that a lot of new knowledge, learned in the last 6 weeks will mostly be forgotten, as the gap of the summer holidays has been too great. This knowledge will all need to be tested again in the first week of year 11, and retrieved at smaller intervals.

Each letter represents the essential knowledge learned in that week.

Then we can plot when it needs to be retrieved, beginning with the next week, then doubling to 2 weeks, then doubling to 4, and so on.

The blank columns, however, will be partially filled with the knowledge from Key Stage 3 which you will be testing at intervals between 6 and 12 months.

How do we know if the spacing is working?

The crucial piece of information is that this spacing predicts getting 90% right.

There will be variability based on:

• What the student already knows

• How difficult the new information is

• Factors about the student’s brain which we do not know

• Whether they paid attention at the time the knowledge was taught

This is why all learning apps try to tailor to the individual. We can’t do this in schools with whole-class teaching.

But we can test the impact of our spacing by looking at percentage scores. If no student scores 90%, the gap is too big.

Of course, if our spaced retrieval includes homework, we don’t know how many students have simply been lazy. Or indeed, if the retrieval is in class, in class, we don’t know how many students have simply not bothered to think hard and retrieve.

So, we have to make a best guess of how many students need to get 90% or more. If not enough students score 90%, then our intervals are too long for this class, and we will need to alter them.

Implications for Selecting Knowledge

It is unlikely that your subject has more than 4 lessons per week. But you can see that the curriculum soon builds up to having more than 4 areas of knowledge to test each week.

So, you will be able to set many of these retrievals as homework and include in assessment. But there will be many lessons where you have to test more than one week’s essential knowledge.

This means you need to be ruthless in how you select what students must know. That’s where you have to be very careful in deciding how much information makes it onto your knowledge organiser.

And that is another post.

The takeaway from this post is that you simply can’t leave it up to the individual teacher to plan the best retrievals for their class.

It is too complex for that. Every teacher will get it wrong.

But, as a department deciding together, every teacher will have a brilliant first place to start.