The pleasure and privilege of doing cognitive science
I am writing this on my third day as assistant professor at the Experimental Psychology Group at the University of Groningen. And this article is my way to say: hi! I think the best way to introduce myself is to tell you why I think these are exciting times to be a cognitive scientist.
Because that’s how I see myself: as an old-school cognitive scientist, or cognitive psychologist, if you prefer; as someone who is interested in how the mind works. To me, cognitive science is about the computations that our mind performs, and not primarily about the details of how and where neurons process information—that’s the domain of neuroscience. Granted, neuroscience and cognitive science are related, as is perhaps best illustrated by research on visual perception: Neuroscientists have discovered a lot about how neurons process visual input; for example, they have discovered that there are neurons that respond selectively to particular orientations (i.e. they respond to a line that is tilted by 45° but not—or less so—to a line that is tilted by 90°). Cognitive scientists have used these discoveries to build cognitive theories of vision; for example, most theories assume that visual input is processed hierarchically, and that orientation processing is one step in this hierarchy. But these cognitive theories are more than just summaries of neuroscientific discoveries: rather, they are attempts to distill, from these discoveries, general principles of how the mind works. And neurons in the brain just happen to be the hardware that runs the mind. But different hardware—such as a computer—could, presumably, run the same mind.
But wait … did I call this brand of cognitive science ‘old school’? I shouldn’t have! Because cognitive science is now more relevant than ever. Take so-called deep neural networks: artificial neural networks that—among many other impressive feats—correctly recognize images about 97% of the time; in other words, they are almost as good as humans at classifying images. How can they do this? Well, these networks are massive, interconnected collections of artificial neurons, which are inspired by neuroscience. But they are not an attempt to copy the details of the brain; rather, they copy general principles of learning and connectivity that seem to govern the mind. And they copy these principles very successfully—cognitive science at its finest!
Many computer scientists would probably balk when I say that deep networks are an achievement of cognitive science—surely they are pure computer science! And so, probably, would many neuroscientists—surely neuroscience is to thank for that we now understand (small parts of) the brain so well that we can simulate it! But isn’t that exactly what makes research exciting? That nowadays computer science, neuroscience, and cognitive science are working together productively toward a better understanding of the human mind.
This is why I feel that this is an exciting time to do cognitive science. Real progress is being made. Even for someone like me who hasn’t been in the field for that long, it’s clear that things are now progressing faster than they have in many years. I remember that when I started my PhD in 2008, I felt that vision science (my subfield of cognitive science) was, to put it bluntly, a bit like stamp collecting: there were many experiments that led to many findings; but I felt that these findings were often isolated, not integrated into theory, and therefore didn’t deepen our understanding of the mind. But now I feel that this is slowly changing for the better. To you give you an example: a little over a year ago I saw a lecture by Laurent Itti, who works at the frontier between computational neuroscience and cognitive science. Itti’s models of visual attention2 are world famous; I’ve used them in my own work as well. In his lecture, Itti showed how he and his team had used insights from cognitive science to build artificial intelligences that could do all kinds of crazy human-like things, such as flawlessly detecting cyclists in a video, or navigating through an environment based on vision alone (look ma, no GPS!).
So how do I hope to contribute to this progress? My own research focuses on vision, eye movements, and pupil size. I find this fascinating, and I’m proud of the small discoveries that I occasionally make—and about which I may tell you more here on Mindwise in the future. But of course my own contribution to science is tiny. Sometime ago I read an interview with someone—I forgot who, but it was a well-known scientist—who said that only 0.1% of all scientists ever contribute anything substantial, and that the remaining 99.9% are mostly there to create an environment for this 0.1% to work in; and he thought it unlikely that he would ever be among that 0.1%. And so do I, but that’s not terribly important—only someone with even bolder illusions of grandeur than mine would be bothered by the idea of belonging to the bottom 99.9%. What is important is that society gives us, scientists, the opportunity to play a tiny role in better understanding the world. And that’s a pleasure and a privilege.