I think what’s interesting about how you describe prepping students prior knowledge for complementary angles is that it’s more like activating prior knowledge they already had because, as you said, those are standards from previous grade levels. Activating prior knowledge is definitely hugely important when jumping into a new topic with students, but I wonder how it differs from the way they “teach” prior knowledge in this study which seems to be teaching knowledge that the participants didn’t have before? My hunch is that this is a distinction that matters.
Yea I agree with you. Here's an attempt to reconcile those, curious if this lands for you:
We want prior knowledge to be firmly in long-term memory. This means having high storage strength and high retrieval strength. If it's something students have learned before but are rusty on (high storage strength, low retrieval strength), then a quick refresher works. But if it's something students haven't learned before or only learned shallowly before (low storage strength, low retrieval strength) then we need to spend more time building up both storage strength and retrieval strength.
While my students have seen those pieces of prior knowledge for complementary angles before, many come into my class seemingly having forgotten all of it or never learned it in the first place. So what I'm trying to do is support students in both situations, giving repeated exposure and practice over time to build up both storage strength and retrieval strength.
I think that’s a great way to frame it and it informs how we collect data on prior knowledge and then respond to it before we are introducing new content. It makes me think about how detailed the backwards planning must be to execute effectively on this.
Thanks, Dylan. I was also glad to read Carl Hendrick's article about this study. I wrote a note about it (https://substack.com/@laurensbrown/note/c-125212511) but I didn't fully do what you did and apply it to what you teach students. I'm going to give this a whirl and see if I can come up with a similar example about U.S. history.
Before I do, can you explain your second diagram a bit more? If I'm reading it correctly, it's showing TWO things being taught at the same time and NEITHER being connected to something in long term memory, yes? Which violates your proposed principles: "1. Teach one idea at a time and 2. Connect each idea to something in students' long-term memory." My question is about the arrow in the working memory part of the diagram. Is the arrow supposed to show that you are teaching one thing and connecting it with a second thing? And that both things are relatively new?
The idea I was trying to get at is actually something you allude to in your note. If I want students to understand the causes of the Civil War, I want them to connect that knowledge to the antebellum South, the southern economy, the physical geography of the country, and way more. Students need to have all that in their long-term memory. I can't start a lesson on the civil war by being like "oh btw here's the role that cotton and plantations played in southern society" because it won't be in long-term memory. That prior knowledge has to come first and be secure. So my diagram was trying to get across that "failure mode" - trying to backfill prior knowledge right before it's useful, without making it secure in long-term memory. Does that make sense?
This is a great and useful contribution to this debate.
I am also struggling with getting into the details of the study. I think there is a really important distinction to make here... There is prior knowledge that is a key building block for the new knowledge (like a Kris Boulton "atom"), and there is associated domain knowledge that is linked to/ similar to / related to the new knowledge. I am not sure which of these this study is really testing with its design. Both feel like having a solid knowledge of them prior to learning the related new knowledge must surely help the learning of the new knowledge be faster / more effective. The core atoms of prior knowledge are surely more important to this? I can believe the knowing related but not essential knowledge would also have an effect, but not as strongly. Also, the brain is an amazing thing... and in the example they gave of radioactive decay and isotopes, although the specific example given was classified as prior knowledge to the new knowledge, anyone with GCSE physics could do an alpha decay calculation. So the brain may very well be able to infer and assimilate the new knowledge from a different mix of prior knowledge than the study controlled. I think we have to be VERY careful not to say this study proves we dont need to link to prior knowledge for effective teaching!
Here's a hypothesis. I agree that atoms are especially important, and weren't tested here. Maybe the associated domain knowledge helps, but only if it is solidly in long-term memory and there are clear connections made between that knowledge and new knowledge. Atoms help in a pretty obvious way, but broader domain knowledge only helps when some specific conditions are met. What do you think?
I'm having a hard time following exactly what the protocol used in this study actually was, but a brief 50-minute prior knowledge intervention will limit the generalizability.
Similar to papers which report no significant effect of SRS if the study duration (and thus interval, among other factors) is simply too short for the effects to materialise.
I struggle to see how prior knowledge, assuming it is RR (relevant and retrievable), not only wouldn't aid future (domain-specific) learning, but how couldn't it?
I have a similar reaction that the value of prior knowledge is common sense, but I also worry about trusting my intuition too much here. That's the point of research, to challenge my intuitions about teaching. I'd love to read more studies along these lines.
If 2 students are learning about topic Y that builds on top of topic X, I'm not sure how the student who needs to learn both could possibly learn Y better than the student who is *only* learning Y (assuming all else equal, and that their X knowledge is RR).
This is actually a really well-observed scenario in language learning: the process of learning a third language is less difficult, all else equal, to learning the second.
I think the difference in language learning is that there's a broad and well-connected set of knowledge to draw on. I think one of my takeaways from the study is that without a broad network of knowledge it's less likely prior knowledge will help.
Yes, an insufficiently-low amount of (RR) prior knowledge will certainly help less - but this is just a direct function of less knowledge in-and-of-itself, no?
I.e., if it takes an increase of 2 arbitrary units of 'knowledge' to produce a positive later-learning effect, and a paper finds that an increase of 1 arbitrary unit of knowledge beforehand didn't help, this is not necessarily evidence of a lack of an effect of prior knowledge, this is evidence that 1 arbitrary unit was not enough to produce an effect.
I wonder if the above has produced erroneous conclusions in the literature, which is what I think may be happening here.
I think what’s interesting about how you describe prepping students prior knowledge for complementary angles is that it’s more like activating prior knowledge they already had because, as you said, those are standards from previous grade levels. Activating prior knowledge is definitely hugely important when jumping into a new topic with students, but I wonder how it differs from the way they “teach” prior knowledge in this study which seems to be teaching knowledge that the participants didn’t have before? My hunch is that this is a distinction that matters.
Yea I agree with you. Here's an attempt to reconcile those, curious if this lands for you:
We want prior knowledge to be firmly in long-term memory. This means having high storage strength and high retrieval strength. If it's something students have learned before but are rusty on (high storage strength, low retrieval strength), then a quick refresher works. But if it's something students haven't learned before or only learned shallowly before (low storage strength, low retrieval strength) then we need to spend more time building up both storage strength and retrieval strength.
While my students have seen those pieces of prior knowledge for complementary angles before, many come into my class seemingly having forgotten all of it or never learned it in the first place. So what I'm trying to do is support students in both situations, giving repeated exposure and practice over time to build up both storage strength and retrieval strength.
I think that’s a great way to frame it and it informs how we collect data on prior knowledge and then respond to it before we are introducing new content. It makes me think about how detailed the backwards planning must be to execute effectively on this.
Thanks, Dylan. I was also glad to read Carl Hendrick's article about this study. I wrote a note about it (https://substack.com/@laurensbrown/note/c-125212511) but I didn't fully do what you did and apply it to what you teach students. I'm going to give this a whirl and see if I can come up with a similar example about U.S. history.
Before I do, can you explain your second diagram a bit more? If I'm reading it correctly, it's showing TWO things being taught at the same time and NEITHER being connected to something in long term memory, yes? Which violates your proposed principles: "1. Teach one idea at a time and 2. Connect each idea to something in students' long-term memory." My question is about the arrow in the working memory part of the diagram. Is the arrow supposed to show that you are teaching one thing and connecting it with a second thing? And that both things are relatively new?
The idea I was trying to get at is actually something you allude to in your note. If I want students to understand the causes of the Civil War, I want them to connect that knowledge to the antebellum South, the southern economy, the physical geography of the country, and way more. Students need to have all that in their long-term memory. I can't start a lesson on the civil war by being like "oh btw here's the role that cotton and plantations played in southern society" because it won't be in long-term memory. That prior knowledge has to come first and be secure. So my diagram was trying to get across that "failure mode" - trying to backfill prior knowledge right before it's useful, without making it secure in long-term memory. Does that make sense?
Yes, makes perfect sense. So I had the diagram right?
Yes!
This is a great and useful contribution to this debate.
I am also struggling with getting into the details of the study. I think there is a really important distinction to make here... There is prior knowledge that is a key building block for the new knowledge (like a Kris Boulton "atom"), and there is associated domain knowledge that is linked to/ similar to / related to the new knowledge. I am not sure which of these this study is really testing with its design. Both feel like having a solid knowledge of them prior to learning the related new knowledge must surely help the learning of the new knowledge be faster / more effective. The core atoms of prior knowledge are surely more important to this? I can believe the knowing related but not essential knowledge would also have an effect, but not as strongly. Also, the brain is an amazing thing... and in the example they gave of radioactive decay and isotopes, although the specific example given was classified as prior knowledge to the new knowledge, anyone with GCSE physics could do an alpha decay calculation. So the brain may very well be able to infer and assimilate the new knowledge from a different mix of prior knowledge than the study controlled. I think we have to be VERY careful not to say this study proves we dont need to link to prior knowledge for effective teaching!
Here's a hypothesis. I agree that atoms are especially important, and weren't tested here. Maybe the associated domain knowledge helps, but only if it is solidly in long-term memory and there are clear connections made between that knowledge and new knowledge. Atoms help in a pretty obvious way, but broader domain knowledge only helps when some specific conditions are met. What do you think?
I'm having a hard time following exactly what the protocol used in this study actually was, but a brief 50-minute prior knowledge intervention will limit the generalizability.
Similar to papers which report no significant effect of SRS if the study duration (and thus interval, among other factors) is simply too short for the effects to materialise.
I struggle to see how prior knowledge, assuming it is RR (relevant and retrievable), not only wouldn't aid future (domain-specific) learning, but how couldn't it?
I have a similar reaction that the value of prior knowledge is common sense, but I also worry about trusting my intuition too much here. That's the point of research, to challenge my intuitions about teaching. I'd love to read more studies along these lines.
Agreed! I'm just not sure how it couldn't.
If 2 students are learning about topic Y that builds on top of topic X, I'm not sure how the student who needs to learn both could possibly learn Y better than the student who is *only* learning Y (assuming all else equal, and that their X knowledge is RR).
This is actually a really well-observed scenario in language learning: the process of learning a third language is less difficult, all else equal, to learning the second.
I think the difference in language learning is that there's a broad and well-connected set of knowledge to draw on. I think one of my takeaways from the study is that without a broad network of knowledge it's less likely prior knowledge will help.
Yes, an insufficiently-low amount of (RR) prior knowledge will certainly help less - but this is just a direct function of less knowledge in-and-of-itself, no?
I.e., if it takes an increase of 2 arbitrary units of 'knowledge' to produce a positive later-learning effect, and a paper finds that an increase of 1 arbitrary unit of knowledge beforehand didn't help, this is not necessarily evidence of a lack of an effect of prior knowledge, this is evidence that 1 arbitrary unit was not enough to produce an effect.
I wonder if the above has produced erroneous conclusions in the literature, which is what I think may be happening here.