Amidst the classics of interaction design on my bookshelf, I’ve kept a handful of books from my time in the psychology department. One of them, Children’s Thinking, by Martha W. Alibali, was a remnant of one of my favorite courses at Notre Dame: a course on Cognitive Development, taught by Nicole McNeil, the badass who runs the Cognition, Learning, and Development Lab. Oftentimes, if my creative juices started to coagulate on the Watson team, I’d turn to this book. It wasn’t a 1:1 comparison between machine learning and human cognition, but in a tech scene where everyone is learning, a good analogy helps. Wouldn’t it be nice, I would say, if the way a machine acquired language did mirror the way young children, who are remarkable learners, do? Then we’d have an entire body of work to draw from? Especially when it came to preventing unwanted outcomes with deep consequences? Nope–that approach was too slow, to start from the basics. People preferred to start from solutions, drawing from communication theory instead (or no theory at all), even in design.
So, how delighted was I when I ran into this article in the Technology Review: A plan to advance AI by exploring the minds of children. It always makes me happy when disciplines collaborate in earnest. The original vision for AI may have been born from within a technical community, but not necessarily a scientific one, and in many ways, the last 50 years have failed to fully leverage a bunch of good juicy basic research in psychology.
I hope that digging into cognitive development will also shift the conversation away from an obsession with neuroscience and cognitive science in the tech industry. There’s a squeamishness, or a dismissive attitude, towards other areas in the psychological research community–like developmental psychology–which could offer plenty of value. If something has the word “psychology” in it, most people just think of therapy. Cognitive science, or neuroscience, promises the comfort of certainty. Unfortunately, much of what ends up being discussed, regardless of what word is used to describe it, amounts to pop psychology, anecdotal information, or just simply, opinion.