Solving Problems is an Inefficient Way to Learn How to Solve Problems

I am the Director of Educational Technology at an independent school, which in normal times means I do a lot of coaching and strategizing around technology-enhanced instruction. I chair a department and a committee of pedagogically savvy EdTech coordinators and teachers, and we work on ways to improve the academic program. However, due to some short-term turnover in IT and the finicky nature of hybrid technologies, I’ve recently been spending a lot of my day filling in as IT Support (i.e., “The audio isn’t working, can you please come in and help me troubleshoot the problem!). To be honest, it’s not my favorite part of the day, but it’s what the school needs right now.

I should also mention that this is my first year at this school. In a non-pandemic year, my colleagues in IT would have taught me how the technology works in each of the rooms, which at this particular school means learning the differences between recently added technology and older technology, and memorizing the peculiarities of specialized spaces such as the theatre, the dance studio, and the gym compared to the average classroom. It would be bizarre, ineffective, and massively unsupportive to leave me to figure everything out on my own given that there are experts in the building readily available to teach me. But, because things are crazy in hybrid land, I’ve had no choice but to learn on my own through trial-and-error problem solving.

Learning through trial-and-error problem solving is obviously slow and inefficient. When I go into a classroom to troubleshoot a technology problem, my brain does the same thing everyone’s does automatically; It searches for previously acquired knowledge stored in long-term memory that might be applicable to the problem (e.g., “The audio in Zoom isn’t always set to the room speakers; Switch it to the room speakers”) and then uses that knowledge to solve the problem. Whenever I do not know the specific strategy that will solve the problem, I have to resort to randomly testing things out, such as turning things on and off, unplugging things and plugging them back in, and so on. Sometimes it works, sometimes it doesn’t.

What is fascinating to me – and I’m sure you’ve had a similar experience if you’ve tried to solve a complex problem for which you have no immediate solution in memory – is that it’s entirely possible to solve a technology problem in a classroom and then not be able to solve that exact same problem the next time it comes up in the future. Even though I try to take mental notes of what I do, because my mind is so consumed by all of the options and theories for how to solve the problem, not to mention it sometimes isn’t clear which random test accidentally resolved the problem, I’m often unable to learn anything from the experience! The cognitive demands of unguided, trial-and-error problem solving limits our ability to acquire and generalize solutions to use on subsequent problems.

This rather counterintuitive phenomenon; that solving problems is an inefficient way to learn how to solve problems; is what led John Sweller to conceive of cognitive load theory. In the earliest cognitive load theory experiments, learners were given a starting number and a goal number. The learners were only allowed to use two moves, such as “multiply by 3” and “subtract 27”, as many times as they liked, and the only way to get to the goal number was by alternating between the two moves (Sweller describes this in more detail at about 2:40 of this podcast). What Sweller and colleagues found for math problem solving was the same thing that I experienced in my technology problem solving: Despite demonstrating the ability to correctly land on the right answer for dozens of the same sorts of problems during the experiments, many of the learners didn’t learn the “alternate between the two moves” procedure. The explanation from cognitive load theory is that the learners’ working memories were so overwhelmed by the heavy load of trying to work out the problem that there weren’t many resources left to allocate towards integrating the “alternate between the two moves” procedure into long term memory.

Do we learn from unguided, trial-and-error-style problem solving? Of course we do. It is obvious to those around me that I am getting better everyday at this narrow aspect of my job. But working problems out for yourself is a slow, slow crawl compared to learning from instruction, and even when students somehow manage to succeed at solving a problem in the absence of instruction, their ability to bring that solution to bear the next time it’s required is in no way guaranteed.

-Zach Groshell

---

If you want to read more about the origins of cognitive load theory, check out this article.

If you’re looking for a book on how cognitive load theory applies to teaching, consider Oliver Lovell’s book, Sweller’s Cognitive Load Theory in Action. I’m reading it now and it’s excellent.

More resources:

This, from NSW Centre for Education Statistics and Evaluation, has good materials to create a PD on cognitive load theory for your school.

This for recent developments in cognitive load research.

This is an important article on cognitive load and instructional guidance and this is more of a magazine version of the same thing in the American Educator.

This is for cognitive load and educational technology, but it’s paywalled. See this for ideas of how to get around paywalls.

Here’s an article applying cognitive load theory to teacher training.