Whenever a practice becomes mandated there seems to be a tendency for it to lethally mutate. When I first started writing about retrieval practice (or the testing effect as we used to call it) many people were surprised by the finding that attempting to dredge something up from memory was a more effective way to learn it than simply restudying it. Today, this has become something new teachers are routinely told as part of their initial training and has been accepted as incontestable.
The result is that teachers are told that lessons must contain retrieval practice and schools often specify when and how such practice must be enacted. The depressing but wholly unsurprising consequence is that a great deal of what is referred to as retrieval practice is a either time wasted or, in the worst cases, actively unhelpful to students. Over time, this has hardened into various strange orthodoxies. One of the most common is that lessons must begin with five questions, drawn from different topics, deliberately disconnected from the lesson about to be taught.
In his post, The Anatomy of a Perfect ‘Do Now,’ Dario Emanuele offers a clear and carefully reasoned version of this approach. He opens up his planning and shares a five-question framework for selecting Do Now content. Each slot has a defined purpose. The first two questions are time-lagged recalls designed to build confidence and gently raise challenge. The third checks homework integrity. The fourth targets a high-leverage weakness identified in recent assessment. The fifth stretches the strongest students through novelty or synthesis. Crucially, these questions are intentionally disconnected from the content and objective of the lesson to follow. Prerequisite knowledge, the argument goes, is better checked later using mini whiteboards, allowing the Do Now to focus on long-term retention without derailing lesson timing. It is systematic, defensible, and clearly grounded in experience.
The problem of forgetting
Anyone interested in the flexibility and durability of students’ knowledge needs a strategy for dealing with forgetting. We have unambiguous evidence that memory decays at broadly predictable rates, and that this decay can be slowed.1 Spacing, the deliberate revisiting of knowledge after increasing intervals of time, and interleaving, the practice of mixing different topics or problem types rather than teaching them in blocks, allow us to resist the natural drift towards forgetting. Used well, they help stabilise knowledge, keep it accessible, and reduce the steady cognitive entropy that would otherwise undo much of what we teach.2
The problem is the assumption that the Do Now is the best place to pursue disconnected interleaving. The opening minutes of a lesson are doing a very particular kind of cognitive work, and using them to retrieve knowledge that will not be used shortly afterwards is - in my view - a missed opportunity. Why would you deliberately start a lesson by asking questions disconnected from the lesson you’re about to teach? Whatever the benefits of interleaving might be, the Do Now is possibly not the best place to try to harness them.
Instead, recall starters should seek to activate prior knowledge that will be utilised in the lesson to come in order to capitalise on the effects of retrieval. Retrieval is not a generic good. It works because it increases the accessibility of knowledge at the point it is needed. When retrieved knowledge is immediately applied, it strengthens performance in the moment as well as memory over time. When it is not, we risk generating effort without instructional return.
How does retrieval practice work?
Too often, teachers are told retrieval practice is a Good Thing, without having the underlying theory explained. It used to be believed that when we stopped using information it gradually disappeared from our memories over time. In 1992, Bob and Elizabeth Bjork suggested a New Theory of Disuse: information didn’t disappear, we just couldn’t remember it. This might sound like splitting hairs – if you can’t remember something what’s the difference? – but, when you think about it, the difference is pretty important: not being able to recall something that’s ‘in there somewhere’ is very different to not being able to recall something that is not there. If it’s in there somewhere, they may be a way to access it.3
The Bjork’s posited that each item in long-term memory has both a storage strength and a retrieval strength. Storage strength is our ability to recall something right now; retrieval strength is our ability to recall something elsewhere and later.
This led to the counterintuitive finding that attempting to improve retrieval strength in the here and now can actually reduce storage strength elsewhere and later, whereas practice that attempts to improve storage strength also results in better retrieval strength in the longer term. Many still find this confusing.
Essentially, teaching that focuses on students ‘demonstrating progress’ over the course of a lesson is much less effective than teaching that focusses on regularly trying to recall previously taught content. This explains why students can produce a slick performance at the end of a lesson by recalling what they have just been taught only to have forgotten it all next lesson. The struggle involved in dredging something up from long-term memory strengthens our ability to fluently recall that information again in the future.
This is what Bob Bjork called ‘desirable difficulty’. A desirable difficulty is a task that requires effort without tipping into failure. It slows performance in the short term, but strengthens retention and transfer over time. Crucially, it is not a single hard task but a sequence of carefully calibrated challenges, supported by feedback, that gradually raise the level of demand. The emphasis in the phrase matters. The difficulty must be desirable. That means it is pitched so success is possible, even if effortful. While harder tasks often feel less fluent and can give the impression that learning is slower, the long-term gains are typically greater than those produced by easier work. Once difficulty becomes overwhelming, however, it stops being desirable and starts to work against learning. So, for a difficulty to be desirable it must result in success.4
When retrieval practice goes wrong
This is where a lot of so-called retrieval practice goes wrong: students are asked a question which they fail to answer. In lesson after lesson, students write down ‘do now’ questions from the board, wait for teachers to reveal the answers and then write down the answer. There is no struggle and no success, just a meaningless time wasting exercise.
When questions are too hard, students guess or disengage. When success rates are low, retrieval weakens confidence and entrenches error. When content is disconnected from subsequent learning, retrieval becomes a ritual rather than a lever. A common mistake is to treat difficulty as the goal, rather than successful remembering. What strengthens memory is not struggle for its own sake, but effortful success.
What do students learn from this? All too often, what they learn is a) that recall tasks are a piece of mindless box ticking admin at the start of lessons that they simply have to suffer through or, worse, b) that, right at the start of a lesson, they have – yet again – failed. These failures compound, often resulting in the belief that students are ‘rubbish’ at the subject. Needless to say, this is no one’s interest.
That is why I argue that recall starters should aim for near-universal success. Not because challenge is unimportant, but because failure at the start of a lesson is corrosive. If students cannot retrieve what they have previously been taught, that is not an inconvenience to be worked around. It should be the focus of the lesson. The purpose of retrieval at this point is diagnostic and preparatory, not selective.
My advice for effective recall starters is therefore simple. Questions should conform to three simple rules:
They should be answered correctly by 100% of students, with prompts given until retrieval is successful
They should be useful, and ideally immediately used.
They should then be repeated regularly so that fluency builds and confidence stabilises.
