Designing systems that help people learn

I have been thinking for a long time about how to help people learn. This is a noble goal, pursued for selfish reasons. My academic status depends on my ability to learn fast and then produce new ideas and prototypes. For me, simply being ignorant of a hot topic or algorithm in computer science, like… recurrent neural networks, can be anxiety provoking.

If I widen my gaze beyond my own inner world, it’s easy to imagine, or remember, the multitude of folks who want to learn new facts, concepts, and skills efficiently. Investigative reporters must teach themselves about the domain they’re assigned to. Dissatisfied employees teach themselves new skills in order to move up into more interesting or better-paying jobs. My advisor used Duolingo to learn more French before going to a conference in Paris, anticipating, I imagine, the pleasure of getting around in a new language. Kids can be insatiably curious. Learning can be its own reward.

I now have access to thousands of online resources to help me acquire expertise, including videos, books, and question-and-answer sites like StackOverflow. I can Google it all because it’s indexed for nearly instantaneous retrieval. And yet, unless I’ve got a deadline and some particular algorithm or bit of information is critical for it, I get distracted. I run into jargon I don’t understand, or skip to a different video on the same topic as soon as I get bored with the first. (Videos in MOOCs have a high drop-out rate after just a few minutes. I’m not surprised.) It can be tough to find resources at the right level of depth — not too simple, not too detailed. For folks with math anxiety, the anticipation of math activates their brains’ pain centers. The information is never-ending; there are millions of search results. There’s often no defined notion of “done” when my brain can release dopamine, its version of Good job! Even when learning is not painful for us, it is effortful.

Our relationships with our teachers and our active membership in learning communities help us overcome the activation energy necessary to get past the discomfort that is part of and inseparable from learning. That’s why I’m specializing in human-computer interaction. The question that I keep turning over in my mind is… how can the computer side of this interaction help me get to more ‘aha’ moments?

System and interface design is a piece of the solution. I’m not sure how large a piece it is, but I’m heartened by the results of my colleage Carrie Cai’s research. Her WaitChatter system helps people learn vocabulary words in a new language without any modification to their existing habits, because it takes advantage of the times during the day when they’re just simply… waiting. This kind of design doesn’t expect us to keep our New Year’s Resolution of studying Spanish for X minutes every day.

How can computers present me with information that helps me learn about my query, while taking into account what we already know about human psychology? Without paralyzing me with too many choices? Or asking me to keep more than 4 chunks of information in my working memory? Or expecting me to set aside a semester instead of a few days, or a few minutes? I hope to see more research, and contribute some of my own, that helps answers these questions.