Tag Archives: metacognition

Episode 24 – An EduBlether with Patrice Bain (Powerful Teaching – Unleash the Science of Learning)⤴


In this episode of EduBlether, we welcome Patrice Bain, co-author of the book ‘Powerful Teaching – Unleash the science of learning’. We discuss knowledge, critical thinking, assessment, curriculum and lots more.

It was a fascinating discussion with some great practical ideas for how to begin using these research-informed strategies or ‘power tools’ in your classroom.


Listen: https://soundcloud.com/edublether/edublether-episode-24-an-edublether-with-patrice-bain

Episode 21 – Cognitive Science and Retrieval Practice⤴


In this Episode we have a discussion about retrieval practice and developments in cognitive science. We look at the implications for teaching and learning, as well as what this means for the role of Education in a wider debate. We have many tangents in this episode discussing hegemony, knowledge, politics and much more. I hope you enjoy as much as we did.

Listen: https://soundcloud.com/edublether/episode-21-cognitive-science-and-retrieval-practice

Misconceptions about learning⤴

from @ Memory & Education Blog - Jonathan Firth

Memory does not work like a video camera.  Image by beegaia  from Pixabay.

Memory does not work like a video camera. Image by beegaia from Pixabay.

Perhaps surprisingly, what people believe about learning and memory is often very different from the scientific consensus.

For example, in a large-scale survey of members of the public, Simons and Chabris (2011) found that over 80% of participants believed that amnesia sufferers forget their own name. This is actually not the case – the memory loss tends to affect recent events rather than their personal identity or childhood memories. In the same study, 63% of members of the public agreed with the idea that memory works like a video camera, while 48% agreed that once you have experienced an event and formed a memory of it, that memory does not change. None of these ideas are supported by mainstream psychological science; a linked study of psychology researchers found 0% endorsement in every case.

Memory seems to be fundamentally counterintuitive, and there are many other myths and misconceptions out there. Guilmette and Paglia (2004) found that 41% of respondents agreed with the idea that a second blow to the head can help a person remember things that were forgotten as a result of a first blow to the head!. Other studies have found widespread support for myths about learning, even among highly educated people.

A couple of years ago I conducted a study which presented a set of statements about learning and memory to teachers; some statements fit with contemporary learning science and others did not. Here I share some of the flawed/wrong statements used, together with my comments (provided as feedback to participants) about why these might best be viewed as misconceptions:

- To improve the effectiveness of learning you have to increase the time spent studying.

This may seem obvious, but is not actually true. Experiments into memory nearly always keep study time constant, but can still demonstrate impressive improvements in memory for information depending on the way that the learning takes place. The idea that attainment directly results from time and effort is therefore over-simplistic. Imagine walking up the down escalator – it would require a lot of time and effort, but wouldn't get you very far. Study advice given to students should therefore be more sophisticated than just "work hard".

- Learners are in the best position to judge what and how they should study.

In fact, learners make many mistakes when regulating their own learning. Learners tend to select easier to-be-learned items, and stop studying when they perceive that they are no longer learning; this approach may be ineffective as it and leads to their avoiding studying harder but important material (Metcalfe & Kornell, 2005). When it comes to specific study techniques, Hartwig and Dunlosky (2012) found that most learners use techniques that are out of step with the strategies which are supported by research, but that those who do use such strategies gain better grades. As Kornell and Bjork (2007) put it, "the task of becoming a metacognitively sophisticated learner is far from simple; it requires going against certain intuitions and standard practices, having a reasonably accurate mental model of how learning works, and not being misled by short-term performance" (p.223).

- The majority of information taught during a class will still be retained by learners 2-3 weeks later.

Forgetting of lesson material is actually very rapid if nothing is done to prevent it. The classic forgetting curve suggests forgetting of approximately 80% within this timeframe. It should be considered that the level of forgetting varies widely depending on the type of material, the learner, and various other factors, but nevertheless, teachers would do well to assume that much of what is taught will be forgotten if nothing is done to prevent this from happening.

- The best way to learn something is to go over it repeatedly within the same hour.

Although this may be a useful thing to do in some circumstances, it is not the most efficient way to learn something because performance over the short term is a poor indicator of long-term learning (Soderstrom & Bjork, 2015), and in addition, this type of activity fails to take advantage of the reliable benefits of spacing out learning over multiple study sessions – a single study session would be followed by a lot of forgetting (see above), which spaced out review activities could minimise. This benefit is known as the ‘spacing effect’.

- It’s always best to simplify things for learners in some way, because making something easier helps it to be processed into long-term memory.

No, this is not necessarily the case. In fact, many factors which slow down learning or make it harder are actually beneficial can improve learning – these are what Bjork (1994) refers to as desirable difficulties (see video clip). It’s true that it is possible to overload learners by presenting too much at once – human working memory can only process a certain amount of information at a time. But that doesn’t mean that we should always make things as easy as possible for students.

- Multiple re-readings are more useful for learning than doing lots of tests.

They are not. A study by Roediger and Karpicke (2006) compared multiple-re-readings with multiple tests, and found that when a one-week delay was taken into account, testing was much more effective than re-reading. It is likely that beyond the first reading, there is very little to be gained from subsequent repetitions in most cases. Testing, in contrast, promotes active retrieval of information (see Karpicke et al, 2014, for a discussion of theoretical explanations behind the testing effect).

- Good study advice for learners should include telling them to find a place where they are comfortable and to do all their revision there.

Although this is popular advice and can on occasion provide reassurance to learners, memory researchers such as Robert Bjork advocate varying our study locations. Doing so leads to a more diverse set of associations forming between the studied material and incidental cues in the physical surroundings, and these cues can boost recall at a later date (see also Smith & Rothkopf, 1984). Different locations may also help the revision session to stand out as a unique event in episodic long-term memory.

- If a learner guesses and is not correct they may remember the wrong answer, so it’s best to avoid guessing/predictions during lessons.

Contrary to this popular idea, guessing incorrectly has been shown to be harmless by a number of studies (e.g. Kang et al, 2011) and may even be beneficial (Metcalfe, 2017; Richland et al, 2009). Kornell & Vaughn (2016) have provided evidence that failed retrieval followed by feedback is just as beneficial as successful retrieval; if success is not essential, this gives teachers more flexibility and freedom about how to structure retrieval practice tasks. In addition, doing a quiz during a lesson may help to minimise the extent to which we get old items mixed up with newer ones (Szpunar et al, 2008Wissman et al, 2011).

- Ultimately, learners form new memories through frequent repetition.

Spaced out practice testing and elaborative links (where rich, meaningful connections are made) are more important than simple repetition. The 1960s 'multi-store model of memory' (Atkinson & Shiffrin, 1968) claimed that repetition/rehearsal in STM was necessary and sufficient for memorisation, but by the early 1970s this was already seen as oversimplistic – a study by Craik and Watkins (1973) showed that repetition alone does not lead to memorisation, and other studies at around the same time showed that meaningful items are much better remembered. Repetition/practice is clearly going to be better than no repetition, but is very inefficient unless combined with other well-evidenced strategies.

- It makes sense to do a homework task soon after the material is done in class.

On the basis of the spacing effect, it would actually make more sense to delay homework by at least a few days.

- Once learners have got a question wrong and then been corrected, they will be able to predict whether they will get it right in future.

Learners’ predictions of their own performance can be fairly accurate but are subject to flaws too. Learners draw on their memories of past tests to predict their future recall (Finn & Metcalfe, 2008), so if they got something right in the past, they tend to predict future success too. This heuristic is not always accurate, however – see the previous point about short-term performance v's long-term learning. As Agarwal and Bain discuss in their book Powerful Teaching, desirable difficulties can help learners to make more accurate predictions of future performance.


These beliefs about learning are common, and are arguably not in the same category as ideas such as learning styles or the left-brain right brain myth. However, they can be harmful – if teachers’ (or students’) views of learning are flawed, then their planning and decision making is likely to be flawed as well.

My survey presented these and other items, some of which were in line with the evidence (e.g. “Learners benefit from mixing up lots of different types of problems, rather than doing one type of task at a time”). And what did I find? Well, the teachers in my survey generally performed better than the general public in terms of not endorsing debunked ideas such as repression, or memory working like a video camera. But there were some notable classroom-relevant misconceptions that were widely endorsed. In particular, participants tended to endorse massing rather than spacing of learning, and re-reading/restudying rather than retrieval and testing.

Interestingly, there was no link between the accuracy of their answers and the number of years spent teaching, suggesting that misconceptions don’t self-correct through experience.

This is an article based on my research study:

Firth, J. (2018). Teachers’ beliefs about memory: What are the implications for in-service teacher education? Psychology of Education Review, 42(2), 15-22.

For a more evidence-based approach to memory and learning that links to all aspects of classroom practice, check out my co-authored book, Psychology in the Classroom.

Also on this blog: What should students focus on? Evidence-based study habits.

Harnessing the power of the testing effect⤴

from @ robin_macp

The power of the testing effect has wide currency and is identified by Dunlosky et al as the most effective method that pupils can use in order to build long term memory. This sounds wonderful, but there’s a problem. Does telling a pupil to self-test actually lead to them self-testing? And if they do, are they actually doing it in the right way? To become good at this is actually quite difficult so it needs to be modelled in the classroom first. What follows is a process I used this term with my National 5 history class (the Scottish equivalent to GCSE), so I hope it provides a practical case study that others find useful.

Step 1: calibrating the knowledge base

The most difficult thing for pupils to become familiar with at the start of this particular history course is the standard of the knowledge they need to learn (and be able to recall). The content we covered is a unit called ‘Migration and Empire’ that addresses which groups came to Scotland, as well as where Scots emigrated to, in the period from c.1830 to 1939. The first section looks at Irish immigration after the famine of the 1840s, so pupils need to know technical terms like ‘potato blight’, statistical information (the population of Ireland decreased from 8 million to 5 million), facts about where Irish people settled in Scotland (e.g. areas of Glasgow like the Saltmarket) and concepts like ‘strike-breaking’. They need to be able to recall some fairly precise information and the standard is higher than they are used to.

Having taught the content, the first factual test consisted of 30 questions divided into 6 sections, half of which were multiple choice. The class were given some instruction in metacognition and advised on how to self-test, but this was little more than an introduction. Until they’ve actually tried putting theory into practice, it won’t mean much. The first test was therefore tough and the average mark was just 16.2 out of 30. Pupils marked it immediately (Dylan Wiliam’s advice is that the best person to mark a test is the one who just sat it) and I took in the scripts to do some (swift) number crunching to give whole class feedback. The final section of the test on Irish occupations scored the lowest and needed some reteaching to correct misconceptions.

Step 2: applying metacognition

The second section of the course looks at other immigration groups to Scotland – Lithuanians, Jews and Italians. This time we could prepare for the test using a scaffold of types of knowledge, which we identified by looking back at the last test (the recipients were definitely doing more work than the donor at this stage). They had to go over previous questions and answers to think about knowledge types and we set up this framework:

  1. Key terms and concepts
  2. Events and processes
  3. Key individuals and groups
  4. Statistics/facts

This allowed pupils to categorise information using knowledge organisers I made, but they populated. The second test used the same structure as the first, and the average mark rose to 22.6 out of 30. Number crunching afterwards showed no particular section was weak, so no reteaching was needed.

Step 3: teaching to self-test

The third part of the course is on reasons why Scots emigrated and has some tricky knowledge on the Highland Clearances, poverty in the Lowlands, and incentives like government-sponsored emigration schemes. The knowledge for this is the hardest to break down and organise logically.

This time, once we’d covered the content we did a lesson on how to construct a test. Homework beforehand was for each pupil to come up with 10 questions and answers on the material we’d covered. In class, they worked in pairs on a combined list of 10 questions based on comparing their efforts. They then doubled up with another pair and refined this further. We then came together as a whole class and, with a scribe typing things up on a computer and projected on the screen, we took the best questions and answers and wrote a 30 question test.

This exercise allowed us to stop and think about what knowledge we need to recall, and how best to frame a question that would test our ability to recall that. This led to lively discussion about the wording of specific questions, as the main problem pupils have with history self-testing is they ask open questions that are more like essay titles. For example, instead of asking ‘when was the depression in the fishing industry?’ (answer: 1884-1894) they ask ‘why was there a depression in the fishing industry?’ It’s a valid question, but it’s for full written practice, not a low stakes test. When we had written up 30 questions, we worked on turning half of them into multiple choice (which is trickier than they realise – they need 3 credible wrong answers for every question). I then took the document we drafted and edited/formatted it, with only one question needing a rewrite by me before pupils took the test in the next lesson.

So what was the outcome? The class average was 27.2 out of 30 (so over 90%), and the pupil with the lowest score in round one achieved 100%. You may well say ‘of course they did so well, they knew the questions in advance.’ Yes. That’s the whole point. The assessment here wasn’t so much about answering factual questions but how to make them. In giving feedback I focused on the fact that some pupils had contributed more questions than others to the test, so some still have to practise how to make a good test. However, we’ll keep working on this so it becomes easy – and, hopefully, transferable to other subjects.

Step 4: the forgetting curve and self-regulation

I observed a lesson once where a Year 11 pupil couldn’t remember a piece of information from the year before, and when the teacher followed this up it turned out they couldn’t remember even having learned the entire topic. This shows the half-life of knowledge, so the forgetting curve needs to be defeated with spaced practice.

Forgetting Curve

So what I did with each of these tests was turn them into a Kahoot. We allowed enough time (at least one week) after each test before doing it again as a Kahoot, and this showed that some information types were harder to retain than others (no prize for guessing the chief culprit – statistics).  Over time we’ll build up a bank of these which pupils can use for self-testing and will return to them periodically in lessons. When we’re midway through the Transatlantic Slave Trade next term we’ll still spend 10 minutes doing an old Kahoot on Irish immigration to Scotland. This means I can have a weekly test with much less effort, so my jealousy of maths teachers who seem to test with ease might finally subside…

Making effective use of the forgetting curve works wonders for long term memory, but it also shows pupils how easy it is to self-test. I estimate that we’ll have about 15 such tests by the end of the course which means they will have at least 450 pieces of precise knowledge on entering the exam room, so the ammunition will be there to provide evidence in all essay questions.

The crucial final ingredient is taking the pressure out of the testing in this process. The last test was both the least stressful and the most successful. The message here is that to get pupils to want to self-regulate by testing themselves you need to reduce the stakes first. Emphasise that a test isn’t about data collection (though I obviously did that in this example, as a diagnostic), or reporting. The crucial message is that tests are about practice. The first mark you get is far less important than the last mark you get. The more you self-test, the better your recall, and the more you build confidence. Once they experience this process so it feels real rather than abstract, they will be far more likely to do what we want them to do: take up the challenge of learning for themselves.