Learning sciences

I had him on the show before talking about that, which the UK Department for Education named a best practice.

His current book, he publishes one every year, is The Teacher Toolkit Guide to the Science of Learning.

Ross, thanks for coming back on the show.

It's my pleasure, Vicki.

It's lovely to chat with you again.

The science of learning.

Now, in our pre-show conversation, you said you've been looking at MRI scans.

Okay, what are you doing? This sounds so intriguing.

Today we're going to explore what the future of learning looks like in the age of artificial intelligence.

This is the Science of Learning Podcast, and I'm Keith Sawyer, and my guest today is Dr. Yasmin Kafayi.

Dr. Kafayi, welcome to the Science of Learning.

Thank you very much, Keith, for having me.

Well, learning scientists argue that children don't learn passively from listening to lectures.

I mean, we didn't think that computers were machines for children, but how were we proven wrong? I mean, first with video games and now with social media, they are interfacing, I mean, directly with all these technologies.

Oh, absolutely.

Well, this is the Science of Learning podcast, and I am Keith Sawyer.

Was there anything in particular that kind of led you, whether it was you or whether it was the school in general, to make that decision around this is the direction that we want to go in? Was it student learning outcomes? Was it something in particular that led to these changes being implemented?

So the trigger was really from a combination of our developing understanding around the science of learning and alongside our analysis of our data points.

So our assessment data indicated that students had encountered content but couldn't always retrieve it fluently or transfer it independently.

And alongside this, we were learning more and more about the science of learning.

So we just provided an opportunity for some shared understanding of what that meant.

Yeah, so I guess, you know, just trying to make it really clear as to, you know, like what, when you're saying, you know, you started learning about the science of learning, was it like a couple of individual teachers? Was it, you know, the leadership team? Was it like at the system level they started to say like these are the things that you need to be doing? Was it anything in particular? Because I always find it interesting as to like what that kind of, that trigger was that led to these changes being made.

Yeah, it was definitely some individual champion teachers who had gone out and done some research and brought some fantastic dialogue back to the rest of the team.

So, yeah, once it started to create some interesting conversation, we just saw that as an opportunity to do some further learning.

So behind the scenes, what's happening is that it's a bunch of different agents working together to construct the lesson rather than, like, one monolithic chatbot that is interacting with you.

We think that it can get so far, but there are a lot of different theories in learning science that are interesting here.

For example, self-regulated learning, like if you are able to control, like plan even just a little bit, we know that those lessons are much more effective for adults.

We know that when you have a much more active learning process as opposed to just getting a bunch of videos, like when you're able to ask questions back or we're able to be like, "Do you need more explanation here? You know, what do you need?" That active versus passive learning clearly works better.

Different voices in a podcast, we know works better.

And so we're trying to pull these best learned practices from the learning sciences and use that in the way the lesson is created.

Yes.

I mean, I wanna dig into the multi-agent structure that you talked about, but let's just go back over everything you just said because-

What about the actual

learning science behind this? How does neuroplasticity fit in?

Liemert argues that struggle matters, but only when it's guided.

Students need an error signal, worked examples, then less scaffolding over time.