A lot of research on cognitive load/working memory focuses on math, but of course the same principles apply to all learning. And if you view the typical approach to reading comprehension and writing instruction through the lens of working memory, it becomes clear that we've been making reading and writing much harder than they need to be for kids by routinely asking them to read and write about topics that are unfamiliar.
The standard approach to reading comprehension instruction has students practice comprehension skills and strategies on books or texts that may be on topics they know nothing about . Reading about unfamiliar topics places a heavy burden on working memory.
Writing curricula often include limited information about a topic and then ask students to write about it. It's even harder to write about a topic you know little or nothing about than to read about one!
Having students read and write about topics they're learning about through the core curriculum makes a lot more sense because (for one thing) it modulates the burden on working memory. That's what any effective approach to literacy instruction will do.
Thanks, Natalie! I’m thinking of how my dad, who taught 8th grade social studies, made his students do so much memorizing—Bill of Rights, Gettysburg Address, they actually did quite a bit—he didn’t know about working memory technically, but always swore that the memorizing was key to students’ SS knowledge. It’s good to think about this in terms of writing, too.
Working memory can be trained to efficiently and effectively build verbatim representations of text—memorizing a speech or a part
in a play. Actors, for example, use a variety of personally meaningful cues to build a schema. Rote memory is the least effective and winds up with the least durable schematic representations. In my experience With Oral Language fairs give students opportunities for dramatic performances which activate more of the brain than rote memory
You are confusing short term memory and working memory. They are theoretically very different elements in the most current perspectives on cognitive load.
I think this premise applies just about anywhere. A musician who knows music theory can far more readily grasp a musical score than a novice. The novice may recognize the notes and be able to play them on a keyboard. But they are likely to be limited to precise reproduction of the notes, unable to in any way be interpretive.
Then again, there's physics. I remember a physics teacher telling us that a student of Einstein's once approached Einstein with a complicated problem. Einstein took only a few moments to find the solution. The student was amazed and asked Einstein how he wo easily found the solution. Einstein replied that it 'just seemed right'.
As an employer, I am quite expert at what I do. There are things I am so familiar with I barely think about them. But when I have to instruct an employee, it can be difficult to distill the instructions down to a level they can comprehend.
On the contrary, I can't cook. My wife cooks up meals off the top of her head. Even with careful instruction from her, my results are inferior. (Trust me on this!)
Yes, students should memorize times tables, even in this modern age. Having a realm of knowledge serves beyond just getting the right answer to a specific question.
I’ve been using Brian’s system in my class for over 6 months now. It works. Performance on standardised NFER maths tests jumped 17%. The control group had, as far as possible, the same instruction (minus the fluency intervention) and they remained constant.
Being able to consider a multi step problem, without hunting around the brain for simple maths facts, has had a significant impact on my Year 5 (UK) children.
I'm surprised, but perhaps should not be that most teachers are not well-informed about working memory. There's a "rule" in applied psychology, and human/computer interfaces that a an average person can keep track of or work with 7 things, plus or minus 2. Bell Labs researched this issue at length. That's why telephone numbers are generally 7 digits long, and are written "chunked" as XXX-XXXX, or as (XXX) XXX-XXXX if you're not phoning locally.
Because another useful thing about working memory is that people can learn to increase their working memory's capacity by assigning bits of information into "chunks", and then keeping track of the chunks.
What you see on it that makes it a great post? I would like to know, partly because I perceive serious problems with it as I’ve articulated in several comments. Particularly, I think it is indexed to an old perspective on short term memory in relation to schema construction and executive function, a simplified perspective
I think the post offers important concrete information about the limitations of working memory and how we can bypass limitations by building up students long term memory (knowledge and skills).
Teachers don’t know about these foundations, and so the post helps to address that and talk about that as an issue. I don’t think the author is seeking to write a definitive review of all the nuances of currently competing models.
If you could say it in a few sentences what teachers need to know about the missing parts of the evidence you are suggesting, and how it would improve their teaching (more than what Holly has put forward) - that would be helpful for me to understand better.
The need to pass information from short term memory (very limited in length in quantity) is critical, but the way to accomplish this task is not repetitive action. Data is not transferred directly from short term memory to long term memory. STM sends data to WM (working memory) which is under the regulation of executive function or metacognition). The flaw in this post is conflating short term memory with working memory. The only known strategy to get data from STM to LTM (long term) is rehearsal (a rote of routinized process). If one wanted to memorize a phone number, for example, one would say it over and over again. There is no rhyme or reason for the numbers and therefore it must be learned through repetition. If it is not used over time it will disappear.
Memorizing the multiplication facts can be accomplished by routine repetition, but the structure built in memory is piecemeal, not schematic. Schemas are knowledge structures which require understanding of relationships among the data, relationships which are grasped when related data (like multiplication facts), are understood in working memory. Working memory sends organized,schematized information to long term memory. The result is deep understanding of the relationships among among the data.
Learners who understand how multiplication works have a strategy to apply on their own if they fail to call up, say, 7x6. If rote memory fails, they can reorganize the data and work with it in working memory. They could, for example, add 3 sevens, get 21, and double 21. This kind of thinking is “mathematical thinking.” Thinking differs from rote recall, and frequent opportunity to solve real world problems using mathematical thinking drawing a schema representing the relationship between addition and multiplication gives learners access not only to a way to figure out a product, but more generally a path to mathematical thinking.
Traditionally, basic math has been taught as a rote system of facts and algorithms. As a result, our national achievement is much worse than in Japan or Germany where it is taught as problem solving. As students use their mathematical thinking in meaningful situations to solve problems, preferably collaboratively, they “memorize” facts as a byproduct. Setting out to memorize facts and testing to see if data “stuck” not only diminishes opportunities for schema building but constructs a habit of mind which is contrary to mathematical thinking.
I just had a phone conversation with a math education expert at Vanderbilt who told me that the approach espoused in this post is the very problem we have with elementary math ed, the explanation for huge numbers of students so underprepared that they don’t have the cognitive resources to profit from advanced math. They have spent years memorizing non-understood bits of data and one-way algorithms that they can’t conceive of any other way.
My point is this: Though what learning science knows about schema building is complex, teachers must understand complexity in order to teach mathematical thinking. Conflating short term memory with working memory and assuming teaching by rote repetition is the best or only way to reduce intrinsic cognitive load, which is where deep learning lives, is the opposite of what is currently known about pedagogy in mathematical thinking. It’s feeding students fish rather than teaching them to fish. Teachers must understand this point on a deep level.
If the goal is to produce kids who can do multiplication problems and quickly get the right answer, drill them all summer. If the goal is to produce kids who can parse the elements of a problem and reduce
intrinsic cognitive load independently, a different approach is needed. Beginning in first grade or earlier, for example, teaching set theory through manipulation of physical objects and visual diagram prepares them for addition and subtraction and helps them build prerequisite schemas for learning multiplication and division. With this understanding they can comprehend fractions and ratios instead of applying rote algorithms without understanding.
I’m wondering which research you are referring to as a framework for your conclusions. Traditional approaches sort of hyperventilate about the limitations of short term memory. Baddely in 1974 developed the notion of “working memory” as a buffer zone between short term memory and working memory. Research in dyslexia using Baddeley’s model studied ways to measure short term memory to determine whether its limitations inhibit working memory. Your focus is on teaching to short term memory with is well-documented time and quantity restraints which may be improved through rote memory—a different issue from teaching to improve working memory where information from long term memory is activated and drawn into working memory where the products of short term memory (immediate input) is anchored to more durable and familiar information. Teaching to rote memory is one way of improving working memory that it may
stock longterm memory but it does not focus on working memory.
John Sweller, who has a long career of research on instruction geared to working memory Dord not focus specifically on rote memory instruction in his cognitive load theory. Instead, Sweller's cognitive load theory emphasizes managing the intrinsic, extraneous, and germane cognitive load on students during learning tasks. The goal is to reduce extraneous cognitive load (through instructional design) and manage intrinsic load to allow students to engage in germane processing that leads to schema development and learning.
Some strategies aligned with cognitive load theory that are mentioned include using worked examples to reduce cognitive load when solving complex problems, avoiding split attention effects by integrating related information sources, chunking information into meaningful segments, making connections to preexisting background knowledge and real-world scenarios, and using principles of explicit instruction
The overall focus is on designing instruction to manage cognitive load and facilitate schema development, not on rote memorization approaches. Schema development is an intrinsic activity that can’t be micromanaged. Teaching cold memory is intuitively appealing, but it doesn’t aid schema construction much.
Sweller's cognitive load theory does not emphasize or advocate for rote memory instruction, but rather focuses on managing cognitive load to support meaningful learning and schema development. Sweller's views are more aligned with strategies to support deeper understanding rather than simple memorization. Teachers are positioned to reduce germane cognitive load (instructional design) but learners must manage intrinsic load. The issue is more complex than presented here.
This makes me want to spend the whole summer drilling my kid on her multiplication tables.
A lot of research on cognitive load/working memory focuses on math, but of course the same principles apply to all learning. And if you view the typical approach to reading comprehension and writing instruction through the lens of working memory, it becomes clear that we've been making reading and writing much harder than they need to be for kids by routinely asking them to read and write about topics that are unfamiliar.
The standard approach to reading comprehension instruction has students practice comprehension skills and strategies on books or texts that may be on topics they know nothing about . Reading about unfamiliar topics places a heavy burden on working memory.
Writing curricula often include limited information about a topic and then ask students to write about it. It's even harder to write about a topic you know little or nothing about than to read about one!
Having students read and write about topics they're learning about through the core curriculum makes a lot more sense because (for one thing) it modulates the burden on working memory. That's what any effective approach to literacy instruction will do.
Thanks, Natalie! I’m thinking of how my dad, who taught 8th grade social studies, made his students do so much memorizing—Bill of Rights, Gettysburg Address, they actually did quite a bit—he didn’t know about working memory technically, but always swore that the memorizing was key to students’ SS knowledge. It’s good to think about this in terms of writing, too.
Working memory can be trained to efficiently and effectively build verbatim representations of text—memorizing a speech or a part
in a play. Actors, for example, use a variety of personally meaningful cues to build a schema. Rote memory is the least effective and winds up with the least durable schematic representations. In my experience With Oral Language fairs give students opportunities for dramatic performances which activate more of the brain than rote memory
You are confusing short term memory and working memory. They are theoretically very different elements in the most current perspectives on cognitive load.
I think this premise applies just about anywhere. A musician who knows music theory can far more readily grasp a musical score than a novice. The novice may recognize the notes and be able to play them on a keyboard. But they are likely to be limited to precise reproduction of the notes, unable to in any way be interpretive.
Then again, there's physics. I remember a physics teacher telling us that a student of Einstein's once approached Einstein with a complicated problem. Einstein took only a few moments to find the solution. The student was amazed and asked Einstein how he wo easily found the solution. Einstein replied that it 'just seemed right'.
As an employer, I am quite expert at what I do. There are things I am so familiar with I barely think about them. But when I have to instruct an employee, it can be difficult to distill the instructions down to a level they can comprehend.
On the contrary, I can't cook. My wife cooks up meals off the top of her head. Even with careful instruction from her, my results are inferior. (Trust me on this!)
Yes, students should memorize times tables, even in this modern age. Having a realm of knowledge serves beyond just getting the right answer to a specific question.
I’ve been using Brian’s system in my class for over 6 months now. It works. Performance on standardised NFER maths tests jumped 17%. The control group had, as far as possible, the same instruction (minus the fluency intervention) and they remained constant.
Being able to consider a multi step problem, without hunting around the brain for simple maths facts, has had a significant impact on my Year 5 (UK) children.
Thanks for sharing this!
I'm surprised, but perhaps should not be that most teachers are not well-informed about working memory. There's a "rule" in applied psychology, and human/computer interfaces that a an average person can keep track of or work with 7 things, plus or minus 2. Bell Labs researched this issue at length. That's why telephone numbers are generally 7 digits long, and are written "chunked" as XXX-XXXX, or as (XXX) XXX-XXXX if you're not phoning locally.
Because another useful thing about working memory is that people can learn to increase their working memory's capacity by assigning bits of information into "chunks", and then keeping track of the chunks.
Great post yet again, Holly!!
I’d love to read a substantial comment from you, Dr. Swain.
There’s a lot in this thread. What specifically did you want me to comment on?
What you see on it that makes it a great post? I would like to know, partly because I perceive serious problems with it as I’ve articulated in several comments. Particularly, I think it is indexed to an old perspective on short term memory in relation to schema construction and executive function, a simplified perspective
I think the post offers important concrete information about the limitations of working memory and how we can bypass limitations by building up students long term memory (knowledge and skills).
Teachers don’t know about these foundations, and so the post helps to address that and talk about that as an issue. I don’t think the author is seeking to write a definitive review of all the nuances of currently competing models.
If you could say it in a few sentences what teachers need to know about the missing parts of the evidence you are suggesting, and how it would improve their teaching (more than what Holly has put forward) - that would be helpful for me to understand better.
The need to pass information from short term memory (very limited in length in quantity) is critical, but the way to accomplish this task is not repetitive action. Data is not transferred directly from short term memory to long term memory. STM sends data to WM (working memory) which is under the regulation of executive function or metacognition). The flaw in this post is conflating short term memory with working memory. The only known strategy to get data from STM to LTM (long term) is rehearsal (a rote of routinized process). If one wanted to memorize a phone number, for example, one would say it over and over again. There is no rhyme or reason for the numbers and therefore it must be learned through repetition. If it is not used over time it will disappear.
Memorizing the multiplication facts can be accomplished by routine repetition, but the structure built in memory is piecemeal, not schematic. Schemas are knowledge structures which require understanding of relationships among the data, relationships which are grasped when related data (like multiplication facts), are understood in working memory. Working memory sends organized,schematized information to long term memory. The result is deep understanding of the relationships among among the data.
Learners who understand how multiplication works have a strategy to apply on their own if they fail to call up, say, 7x6. If rote memory fails, they can reorganize the data and work with it in working memory. They could, for example, add 3 sevens, get 21, and double 21. This kind of thinking is “mathematical thinking.” Thinking differs from rote recall, and frequent opportunity to solve real world problems using mathematical thinking drawing a schema representing the relationship between addition and multiplication gives learners access not only to a way to figure out a product, but more generally a path to mathematical thinking.
Traditionally, basic math has been taught as a rote system of facts and algorithms. As a result, our national achievement is much worse than in Japan or Germany where it is taught as problem solving. As students use their mathematical thinking in meaningful situations to solve problems, preferably collaboratively, they “memorize” facts as a byproduct. Setting out to memorize facts and testing to see if data “stuck” not only diminishes opportunities for schema building but constructs a habit of mind which is contrary to mathematical thinking.
I just had a phone conversation with a math education expert at Vanderbilt who told me that the approach espoused in this post is the very problem we have with elementary math ed, the explanation for huge numbers of students so underprepared that they don’t have the cognitive resources to profit from advanced math. They have spent years memorizing non-understood bits of data and one-way algorithms that they can’t conceive of any other way.
My point is this: Though what learning science knows about schema building is complex, teachers must understand complexity in order to teach mathematical thinking. Conflating short term memory with working memory and assuming teaching by rote repetition is the best or only way to reduce intrinsic cognitive load, which is where deep learning lives, is the opposite of what is currently known about pedagogy in mathematical thinking. It’s feeding students fish rather than teaching them to fish. Teachers must understand this point on a deep level.
If the goal is to produce kids who can do multiplication problems and quickly get the right answer, drill them all summer. If the goal is to produce kids who can parse the elements of a problem and reduce
intrinsic cognitive load independently, a different approach is needed. Beginning in first grade or earlier, for example, teaching set theory through manipulation of physical objects and visual diagram prepares them for addition and subtraction and helps them build prerequisite schemas for learning multiplication and division. With this understanding they can comprehend fractions and ratios instead of applying rote algorithms without understanding.
I’m wondering which research you are referring to as a framework for your conclusions. Traditional approaches sort of hyperventilate about the limitations of short term memory. Baddely in 1974 developed the notion of “working memory” as a buffer zone between short term memory and working memory. Research in dyslexia using Baddeley’s model studied ways to measure short term memory to determine whether its limitations inhibit working memory. Your focus is on teaching to short term memory with is well-documented time and quantity restraints which may be improved through rote memory—a different issue from teaching to improve working memory where information from long term memory is activated and drawn into working memory where the products of short term memory (immediate input) is anchored to more durable and familiar information. Teaching to rote memory is one way of improving working memory that it may
stock longterm memory but it does not focus on working memory.
John Sweller, who has a long career of research on instruction geared to working memory Dord not focus specifically on rote memory instruction in his cognitive load theory. Instead, Sweller's cognitive load theory emphasizes managing the intrinsic, extraneous, and germane cognitive load on students during learning tasks. The goal is to reduce extraneous cognitive load (through instructional design) and manage intrinsic load to allow students to engage in germane processing that leads to schema development and learning.
Some strategies aligned with cognitive load theory that are mentioned include using worked examples to reduce cognitive load when solving complex problems, avoiding split attention effects by integrating related information sources, chunking information into meaningful segments, making connections to preexisting background knowledge and real-world scenarios, and using principles of explicit instruction
The overall focus is on designing instruction to manage cognitive load and facilitate schema development, not on rote memorization approaches. Schema development is an intrinsic activity that can’t be micromanaged. Teaching cold memory is intuitively appealing, but it doesn’t aid schema construction much.
Sweller's cognitive load theory does not emphasize or advocate for rote memory instruction, but rather focuses on managing cognitive load to support meaningful learning and schema development. Sweller's views are more aligned with strategies to support deeper understanding rather than simple memorization. Teachers are positioned to reduce germane cognitive load (instructional design) but learners must manage intrinsic load. The issue is more complex than presented here.