Engineering Learning - Building context relevant structures, not shipping software for the average learner
Engineering Learning - Building context relevant structures, not shipping software for the average learner
An interview with Bror Saxberg on the need for a different approach to improving learning
It’s become common to say that education hasn’t changed.
This claim is often backed by pictures of old and new classrooms that “look the same”. The problem with this shallow comparison is that it assumes that to fundamentally change and reform education we need to focus to a large extent on the physical structure of the learning environment. In reality, the most important factors in the learning process are the humans within the building and what their minds are experiencing. We need to focus on the engineering of learning, rather than that of the buildings.
Dr. Bror Saxberg has played a pioneering role in the nascent field of learning engineering. From being the VP of Learning Science at the Chan Zuckerberg Initiative to being trained as a medical doctor and electircal engineer, Dr. Saxberg has a unique vantage point on what learning engineering can offer.
I recently had the chance to sit down with him and walk through why he's bet his career on nurturing the field, trends in the learning sciences, and unsolved questions that keep him up at night.
Below is my attempt to highlight the most insightful parts of the discussion.1 I hope you enjoy it as much as I did.
What is learning engineering, and how can it help us improve learning?
Learning engineering is the use of learning, motivation, and development science to help organisations increase their odds of learning and training success at scale. The goal is not perfection but to increase the odds of getting better outcomes on first launch, and then to increase the odds of improvement with each iteration.
It’s important to underscore that learning is highly contextual - a facet that is often overlooked in education reform efforts. What works with one group of students, from one set of cultures, may not work at all for a different set of students. We’ve made a bit of a mistake in drawing too close an analogy between learning engineering and software engineering. We’ve assumed that, as with much of software development, you can make a single great product offering, and then “scale the heck out of it” with minimal tweaking. That has been amazingly successful for many software tools (Excel, Word, Google tools, etc.). However, there are other kinds of engineering that do not function this way, and still work to great effect.
Civil engineering is one example. You don’t copy and paste one bridge from an area to a different area. You spend a lot of time studying and understanding the context - effectively optimising for the unique problems you need to solve in this new context.
However, under this paradigm, when you scale you don’t have to rebuild the component parts from scratch. You use the principles you’ve learned and the context to create specifications for components most of which you can order and not make anew.
It would be amazing if we could do the same for education: get better at measuring and understanding context, and then have catalogues of interventions to find the best fit for our specific context. Then we continue to monitor to see if we are getting the expected results, and iterate as needed for subgroups of students who are not doing as well as we hoped.
What are some key trends from the “learning sciences” over the past decade that are improving how we learn?
First, the increasing awareness to think of the whole learner is a very important development. This means not just academic capacities, but also things like mental and physical well-being, identity issues, social and emotional skills, and more. You cannot just apply research piecemeal - you have to think about how what you are designing will support the whole learner (and the whole teacher) in their context.
Second, in a similar vein the notion of inclusive design is quite powerful. Traditionally, the way learning science often gets propagated was to start with research done wherever convenient, then create and test a prototype, then scale to a few schools, and ultimately try to take it to a larger scale after this. Under this model, you almost always see that effectiveness drops as you scale.
The wrong conclusion here is that nothing works at scale. The right conclusion is to think about design more the way that IDEO or the Stanford d.school folks think about it. You need to start with the ending context in mind, and bring in the stakeholders from there before you start designing. This way, you ensure that right from the beginning we are gated by “what is the real setting/context?” . Rightfully, this is a more time consuming process because the context is made of people and it takes time to build trust with them.
What do people often get wrong about the role of technology in advancing learning objectives?
We’ve known since at least 1983, due to the pioneering work of Richard Clark that once you control for the actual learning experience, technology alone does nothing to improve learning. What matters is what the mind experiences, not the technology used to deliver that experience. Technology can take good, or bad, learning experiences (unending lectures!), and make them more affordable, reliable, available, and data-rich.
In order to truly leverage technology in a helpful way we have to get the order right. We should start by looking at the learning problem faced by the human mind, and then find out what will be a better experience for that mind to resolve that learning problem based on learning and motivation science. Only then can we ask how technology can take parts of this solution, and make them more affordable, reliable, available, etc.
Finally, we have to be careful to think holistically about the learning environment, and solutions. We run the risk of assuming that solving part of a problem is enough, or that our solution exists in isolation for students and teachers. We often forget that teachers and students experience all of the solutions in their learning environments at once - having multiple, incoherent methods or innovations, or even data from those interventions, can often be counterproductive and overwhelming.
In closing, what are the most pressing educational problems that you are paying attention to now?
Yikes - so many things to worry about! I’d venture four key challenges.
First, despite the growing body of evidence on the importance of understanding what real expert minds decide and do, we are still not properly investing in that area. If we really want to design training targeting high-value expert performance, we need to invest time and effort in understanding what top performers do. Companies and organisations need to invest more time on techniques like cognitive task analysis, to unpack what these valuable professionals decide and do.
Second, I’m interested in seeing how we move away from building education systems for the “average child” - a non-existant charachter. This is related to work by folks like Todd Rose, showing that the concept of “average” has been unhelpful for product design. When the American Air Force early in its history built planes for the “average pilot”, it literally led to higher fatalities. We are killing our children’s dreams because we are still trying to build learning environments for the “average child” as opposed to more flexible and customizable environments.
Third, I believe there’s still a missing piece around the question of designing for motivation. It’s a uniquely different educational problem when students don’t start, persist, or put in mental effort. It’s not about if you are instructionally designing well. It's a separate kind of problem. We need to do more - across the ecosystem - to help recognize and design for motivation challenges, context by context.
Finally, the puzzle of professional development for teachers is one that is still very far from being completed. Many programs teachers experience are not designed according to the best available evidence about learning either, and are not motivating. Any good new program should focus on what teachers need to know to fulfil their new roles with the materials they have and the students they have, and should think longitudinally about how we give them regular feedback and support to keep improving, context by context.
People often don’t appreciate that teaching is very much a time-bound performance art - many things are happening at once in a classroom, and becoming good at applying principles of learning, development, and motivation in such a setting requires significant coaching and training over time, just like any performance task. It’s a craft and needs to be scaffolded as such.
I hope that you enjoyed this interview with Dr. Bror Saxberg - any and all thoughts welcome. You can find me on Twitter.
I also want thank the 30 people that have subscribed to the newsletter since the latest issue. If you enjoyed this piece, please do share it with your network and don’t forget to subscribe.
Bror’s responses are lightly edited. All credit goes to Bror, and any misgivings are solely my own.