The Best Way to Help Children Remember Things? Not “Memorable Experiences”
The difference between “Episodic Memory” and “Semantic Memory,” and what it means for teaching and learning
Clare Sealy joined Education Next Editor-in-chief Marty West to discuss this article on the EdNext Podcast.
When we look back on our own school days, our strongest memories are probably a mix of big occasions—field trips, plays, and sports days alongside more personal events tinged with strong emotion. Things that happened that were really funny or sad, or that made us feel excited, interested, exhilarated, or angry. We don’t tend to remember vividly, if at all, actually learning the substance of math or English or design technology. We might remember amusing anecdotes from lessons gone awry, or still bristle at past injustices — “but I wasn’t talking” — or have a vague impression of sitting in the science lab, with fleeting snippets of memories of this or that experiment. All of which leads to us making the entirely reasonable hypothesis that if we want students to remember what we teach them, then we need to make our lessons more like the spectacular one-off special events, or, at the very least, involve something specially selected because it’s exciting and possibly unusual. Memorable events, in this view, should form the template for creating memorable lessons.
As reasonable as this seems, this is a myth. It is a myth because human memory works in two different ways, both equally valid but one of which is much better at enabling us to transfer what we have learnt to new contexts. This transfer is an essential prerequisite for creativity and critical thinking.
The two forms of memory are known as episodic and semantic memory. Episodic memory is the memory of the ‘episodes’ of our life—our autobiographical memory. This takes no effort on our part, it simply happens. We don’t have to try consciously to remember what happened yesterday. Those memories just happen automatically. But there is a downside. Episodic memory is “easy come, easy go.” If you try to remember what you had for lunch yesterday, you will probably remember. If you try to remember what you had for lunch a year ago today—unless that happened to be some very significant date and some particularly noteworthy lunch—you will have no idea.
Semantic memory, on the other hand, involves much harder work. We have to expend effort to create semantic memories. This is the kind of memory we use when we consciously study something because we want to remember it. Unlike episodic memory, it does not just happen. The upside, however, is that the effort involved results in a long lasting memory.
Have you ever been in a course where you have really enjoyed listening to the speaker, found the subject matter interesting and the presenter amusing and engaging. Yet when you try to explain to someone the next day what the course was about, all that is really left is a vague impression of your emotions during the day, tinged with the odd snippet of content? You know the course was really good yet can’t really explain what it was actually about beyond the most general of assertions. That’s because at that point your memories are mainly episodic and are already fading. This is particularly likely to have happened if you just listened to the speaker rather than making some notes, and if you didn’t have to do activities during the day that made you think hard about the content. But even if you did, unless you reread those notes some time later, or read the PowerPoint, or the blogs they mentioned, or plan a staff meeting to tell others about what you have learnt, unless you expend some effort in revisiting the message, however inspiring you found the message at the time, your memory of actual specifics will quickly fade away—leaving you with at best fond memories of an enjoyable and interesting day.
Episodic memory is highly contextual—memories come bundled together with the sensory experiences and emotions we experienced at the time. So when we recall our course, we remember the vicious air conditioning, the awesome lunch, the snazzy stationery. Annoyingly, we may remember these things more strongly that the things we really want to remember. Teachers probably have had this experience with classes. When teachers ask students to remember what they were learning the previous day, the students remember all sorts of things: that you used Post-it Notes, that Mollie was late, that you spilt your coffee, that Liam made a hilarious joke. The actual lesson content? Memories of that are much weaker.
These emotional and sensory cues are triggered when we try and retrieve an episodic memory. The problem is that sometimes students remember the contextual tags but not the actual learning. Episodic memory is so tied up with context it is no good for remembering things once that context is no longer present. This means that it has serious limitations in terms of its usefulness as the main strategy for educating children, since whatever is remembered is so bound up with the context in which it was taught. This does not make for flexible, transferable learning that can be brought to bear in different contexts and circumstances. Yet it is this transferability that is the essential prerequisite for creativity and critical thinking.
Fortunately, we also have semantic memory. Semantic memory does not have the limitations of episodic memory. Semantic memories are context free. Semantic memories have been liberated from the emotional and spatial/temporal context in which they were first acquired. Once a concept has been stored in the semantic memory, it is more flexible and transferable between different contexts. Semantic memory is central, therefore. to long-term learning, learning that can be put to use in novel contexts to solve unexpected problems. Semantic memory is what we use when we are problem-solving or being creative. Those involve applying something learnt in one context to another, novel context. Episodic memories, by contrast, aren’t flexible and don’t easily transfer, because they are anchored in specifics.
This explains the frustration teachers feel at the beginning of each school year when children they have been assured are very competent appear to have absolutely no clue. It is not that their previous teacher was lying or deluded when they said they understood fractions. It was that the previous teacher had not realized that this understanding was not yet secure in semantic memory and was still highly reliant on episodic memory. It was therefore highly dependent on strong contextual cues to be remembered. Move the child to a different classroom, with a different teacher, sitting next to different classmates, and, without the familiar context, the learning simply cannot be recalled. This is bad enough when children move from one class to another but is greatly amplified when children change schools, such as moving from primary to secondary school. Here the context is hugely different, different building, different journey to school, different uniform, much that is familiar gone. No wonder secondary teachers often think primary teachers over-estimate what their former pupils know. With so much familiar context removed, only what has been securely remembered in the semantic memory will be able to be transferred to life in year 7.
Forming semantic memories requires work and practice. Unlike episodic memories, they don’t just happen. If you want to remember something, you need to think about it, not just experience it. The cognitive psychologist Daniel Willingham explains that “memory is the residue of thought.” The more you have thought about something, the more likely it is that you will remember it. So teachers have to make sure that lessons give students the opportunity to think the things we actually want them to remember, rather than some extraneous other thing. We need them to think about the message of the lesson, rather than the medium we use to teach it. This is where “fun” lessons can unintentionally prevent learning happening. If the medium chosen to deliver the lesson is too obtrusive, it is that the students will think about, rather than whatever it is we actually want them to learn. For example, I remember someone suggesting using water pistols to teach children about angles. This is a suggestion not without some merit. In particular, I like the way here angle is being taught vertically as trajectory rather than just horizontally as is more usual with paper, pencil, and protractor. However, maybe the author of this suggestion is a much more skilled teacher than most, but I can’t help thinking that many children would be thinking more about opportunities to accidentally on purpose squirt Hamid than on the nature of acute angles. Even a rather less adventurous activity—a math game for example—runs the risk that children think hard about the rules rather than the actual math. This is not a problem if a teacher decides to sacrifice one lesson to learn the rules of a game that will then be used repeatedly throughout the year. But is highly problematic if teachers are not alert to the possibilities that they might be unintentionally sabotaging learning by making the medium of learning more prominent than the actual message. When teachers plan lessons, we need to be mindful of what children will be thinking about during each part of the lesson, rather than what they will be feeling or doing. Have we planned activities that will ensure children think hard about the right things? If not, don’t be surprised when children remember very little beyond the confines of that specific lesson.
Some kinds of activity do involve hard thinking, but not always about the central things we want children to understand. For example, doing practical science experiments involves lots of thinking about planning, what to do next, monitoring what is happening. In fact, the practical elements require so much mental energy that there isn’t much cognitive bandwidth left over for actually thinking about the concepts the experiment is meant to prove. When Ofsted, the British Office for Standards in Education, investigated science teaching in primary schools, it found that many schools mistakenly attempted to teach the concepts of science almost entirely through practical experiments:
“The misconception here is that ‘working scientifically’ becomes the mechanism for teaching knowledge and concepts. However, approaching the teaching of science in this way leads to a recurring problem that pupils are engaged in these lessons, but it is the experiment that is memorable and not the underlying knowledge intended to be learned. For instance, when inspectors questioned pupils during the research visits, pupils could easily recall the task carried out, but struggled to explain how the processes they were investigating actually worked.”
Which is not to say that children shouldn’t do scientific experiments, but that teachers need to be acutely aware that students are highly unlikely to gain understanding of scientific concepts unless the experiments are conducted after these concepts have already been taught. Once the scientific concepts are secure, children are much more able to really ‘think like scientists’ with the added benefit that the practical activity then consolidates understanding of the previous learning.
The same is true for activities that require children to research information for themselves. If this is done as a way of acquiring information in the first instance, then the cognitive effort of locating the correct information is unlikely to leave space for children to actually remember much about what they have found out. Children will be thinking about where they can find what they need to find out and whether what they are reading is relevant, rather than actually thinking about what it says. If remembering what it says is important – and presumably it was or why waste time finding out about it—then teachers need to bear in mind that extra time, either before or after the research activities, will need to be spent thinking hard about it.
This problem raises its head yet again in teaching math through problem solving. Some teachers believe that rich tasks such as investigations where children discover relationships for themselves are a much better way of teaching than telling children explicitly how to do things. Discovering is seen as more creative, requiring more imagination, more interesting and therefore much more likely to result in children really understanding the math conceptually, rather than just regurgitating a procedure. However, while rich tasks have their place, they are completely inappropriate for the initial stage of learning, when children are encountering a concept for the first time. If we want children to become independent problem-solvers, we need to teach them carefully and explicitly so that semantic memory can begin to form. Counter intuitive as it may seem, children do not become independent problem solvers by independently solving problems. This is because when children are trying to solve problems before they know the necessary math to do so, they will be expending considerable mental energy tracking what they are meant to be solving against what they have found out so far, so much so that even when they are successful, they will have forgotten what they actually did en route to finally finding the answer!
There is a lot of research that shows that teaching is much more effective when the teacher explicitly explains material in small, carefully thought out steps, giving children lots of opportunity to practise before going onto the next small step. This is particularly true in the early knowledge acquisition phase of learning. For example see Barak Roshenshine’s “Principles of Instruction,” or Coe et al What Makes Great Teaching or Centre for Education and Statistics of New South Wales, Australia, report, What works best: evidence-based practices to help improve NSW performance, all of which reach similar conclusions.
Even if lessons break material down into small steps and involve children thinking hard about the right things, teachers might still have the following all too familiar experience. At the end of a sequence of lessons on a particular topic, a teacher feels reasonably confident that most of the class have learnt how to do whatever it was they have been learning—the grid method, for example. Children may perform reasonably well in an end-of-unit assessment. However, revisit the topic a few weeks later, and not only can children not remember how to do the grid method, several deny all knowledge of having even heard of it before! This situation can leave teachers feeling dispirited and wondering where they have gone wrong.
Unless teachers plan opportunities to revisit concepts again later, some time removed from recent teaching of the concept, it is likely that semantic memory will not yet be strong enough to do the job we need it to do. We need to multiply the opportunities pupils have to think hard about the important things we want them to learn. Are we giving students opportunities to think about concepts in a less highly cued environment? It’s not that hard to do the grid method in the middle of a lesson where the teacher has just modelled how to do the grid method. To build strong semantic memory, students need opportunities to do things when the specific teaching of how to do them is less recent. They also need opportunities to do them when they are less tightly cued. When. For example, they haven’t been told — “to work this out this you will need to multiply, using the grid method.”
In lessons, teachers provide all sorts of clues and prompts that held children give the right answers. This is fine as a first step, but unless teachers also have strategies that enable learners to move beyond performance, little might actually be retained longer term. In a 2015 article in Perspectives on Psychological Science, Nicholas Soderstrom and Robert Bjork explain the difference between performance and learning. Performance is what we can see happening during teaching. Learning, on the other hand, is something invisible that goes on inside children’s heads. We cannot observe learning, we can only infer it. Frustratingly, current performance is a terrible guide to knowing whether or not learning has actually happened or not. Teachers and leaders are at risk at being fooled by current performance and think that change in the long term semantic memory (aka learning) has taken place.
It is not uncommon for people to disagree with this emphasis on building semantic memory. Sometimes people argue “but I don’t remember anything I learnt at school.” When people say this, what they usually mean is “I don’t have strong episodic memories of learning specific things at school.” This is actually a good thing for which they should be grateful. We don’t want to have to remember the context alongside content—that would really impede our ability to think. Much of what we learnt at school we don’t remember learning, and yet we know it. This is because the episodic memory of the actual lesson has long since faded, while the semantic memory formed through thinking hard about the content—albeit perhaps somewhat reluctantly at the time—endures. This is why people know about triangles and oxygen, Anne Boleyn and paragraphs, square numbers and ox bow lakes, color-mixing and Lady Macbeth. Do people remember actually learning about these things? Probably not, yet you know it (or most of it) and though a person may have not thought about ox bow lakes for decades, at the very mention, back the memory comes, effortlessly. That’s the beauty of semantic memory. It isn’t, and doesn’t need to be, tied up with episodic clutter. We don’t need to have fond memories of sitting on the carpet in Reception whilst Mrs. Blackburn told us all about triangles to know about triangles. You have, as Solomon Kingsnorth so gloriously put it “a private tour guide to the universe, living inside your brain” instantly available.
What you know shapes what you see. You or I might look at a mountain and just see…a mountain. A geologist might look at mountains rather differently, as might a keen cyclist. One sees the mountain as we do, yet also sees rock formations, the other sees gear ratios. To have a richly stocked semantic memory is, as Ian Leslie put it in the book Curious, to have access to “an augmented reality, everything [you] see is overlaid with additional layers of meaning and possibility.”
If you know about the kind of things listed above—Anne Boleyn and paragraphs and ox bow lakes, for example—you are knowledge privileged. You have been given opportunities to think hard about stuff you didn’t know and therefore have a vast repository of semantic memory on hand, readily available whenever you want it. Yet it is all too easy to overlook this privilege and vastly under-estimate how much we do in fact know and how much our schooling benefited us. Because we don’t remember learning what we know, we don’t remember the effort that went into teaching it. We may assume that the really important stuff we know was learnt once we left school, not realizing that it was the semantic memories acquired at school that created the foundations we were able to build upon so fruitfully later. Had things been otherwise, our ability to think would have been greatly diminished. We might have gone to a school that thought it most important to provide interesting and exciting experiences, and that while that might have been fun, it might well have left us with an impoverished ability to think or be truly creative, especially if our access to powerful knowledge at home was limited. Before we decide to impose our own agendas onto children’s education, we need to check our knowledge privilege before making decisions that will deprive children of their fair share of the rich cultural inheritance our world affords and to which they are entitled.
Episodic memory might appear, at first glance, the more ‘human’ of the two forms of memory, the memory of people, feelings and places that makes us who we are. Semantic memory seems colder, more robotic. Yet it is our amazing ability to store culturally acquired learning in our semantic memory that makes as so successful as a species. The key purpose of education is to build strong semantic memory, to pass on the knowledge built up over centuries to the next generation; how to read and write, how stories work, how to use mathematical reasoning to solve problems, science with its amazing predictive power and the myriad of other concepts, ideas, and practices. That is not to say that building semantic memory is the only purpose of education. We want to help form children who are emotionally literate and morally responsible too. That will involve thinking about the kind of episodic memories we try to build for our children. If we treat our children with kindness and respect, they will have episodic memories of what it was like to be treated kindly and respectfully, which makes it more likely they too will treat others with kindness and respect themselves. Alongside the building of cultural capital in the semantic memory, we should also be building social, moral, and spiritual capital. While building semantic memory will play an important part in this, so too will be developing episodic memory as for these vital dimensions of human experience recalling the emotional context is perhaps even more crucial.
Nor is it to say that there should be no consideration of creating the kind of memorable experiences that trips and plays and so forth afford. Such special events that punctuate the day-to-day routine of school life are the festivals, the “Christmas dinner” of the school year. They are special because they are infrequent and resource-heavy and different. They contrast with the everyday, bread and butter hum drum familiarity of ordinary school life. But the everyday is our core purpose.
Lest I create in turn more myths, let me make some caveats.
Where we educate children who have limited life experience and who do not have rich episodic memories, it will be all the more important for us to provide the kind of rich experiences— trips to the seaside, to the theatre and art galleries, to see first-hand mountains, cities, forests. For many children, such exposure is part and parcel of family life. For others, experience beyond the immediately local is pretty much non-existent. Where this is the case, we owe it to the children to try and provide the kinds of first-hand experience that expands horizons. Very young children, regardless of background, by definition have less rich life experience, so need more memorable experiences as a sort of “welcome to the world” primer pack.
It is also possible to overstate the separateness of episodic and semantic memory. There is a degree of overlap between the two. They are not water-tight compartments. For example, in one experiment, people were asked to list kitchen utensils. First, participants drew on the contents of their semantic memory, but when that ran out they turned to thinking specifically about their own kitchens and remembering what was there. They augmented their recall from semantic memory with context-specific episodic memories.
Strong emotion makes things stick in episodic memory, as does novelty. So doing some sort of less routine, novel or exciting event to round off learning about something might complement semantic memory—a trip at the end of a topic of work for example. Returning to the early example of teaching angle using water pistols, maybe doing this after a series of more traditional lessons might be a good way of reaping the benefits of both forms of memory, as might be doing that science experiment that proves the concept you have been learning about. It is not a simple binary choice between always only doing one or the other. Nor is it the case that episodic memory is in some way “bad” or inferior. It’s just different. The deliberate building of semantic memory is much more likely to result in long lasting, flexible and transferable memory than putting most of your energies into the episodic basket, so should form the bulk of what we spend our time on. But not every moment of every day. Knowing the limitations of both forms of memory can help us make wiser and more productive choices.
Clare Sealy has spent 30 years as a primary school teacher, 22 of them as head of St Matthias Primary School in Tower Hamlets in the East End of London. She has recently left headship to become of head of curriculum and standards for the States of Guernsey. She blogs at www.primarytimery.com. Her particular interests are the application of cognitive science in the classroom, rethinking assessment for learning, and curriculum development. This article is based on a chapter she wrote for The ResearchEd Guide to Education Myths: An evidence-informed guide for teachers (John Catt Educational, 2019, $15, 160 pages.)