- Noteworthy News
Building Cognitive Computing Bridges
“Alexa, tell me a joke.”
“Why shouldn’t you arm-wrestle a T. rex? He’s a sore loser, and he will eat you.”
Artificial intelligence is more widespread than ever, but in even the most commercially successful cases such as Amazon’s Alexa assistant, it’s pretty easy to tell you’re not dealing with a real human. Natural intelligence has a flexibility and adaptability that we are just beginning to understand and are far from replicating in machines. Learning more about the computational underpinnings of the brain and mind might help make a better Alexa. But a more enduring value of such research would be illuminating the biological processes that underlie sensation and perception, decision-making, learning, memory, language, and other cognitive processes.
“Computational approaches are central to achieving our vision of reverse engineering the brain,” says department head Jim DiCarlo, the Peter de Florez Professor of Neuroscience. “There is a real need to build formal, testable bridges that integrate across multiple levels of analysis, from molecules to systems to cognition.” DiCarlo’s lab, for example, has developed an iterative approach where models of visual processing and empirical electrophysiological data play off each other, resulting in computational models optimized to replicate—and thereby better understand—brain processes.
Bridging computational and biological approaches is one of BCS’s most distinctive features, made even stronger when the bridges cross to other departments and disciplines at MIT.
Over the past two years, the department has launched the 6-9 Computation and Cognition major in cooperation with the Department of Electrical Engineering and Computer Science; secured a major graduate student training grant from NIH to help train the next generation of leaders in these integrative methods; expanded computational core services within BCS; and formed meaningful connections to the MIT Stephen A. Schwarzman College of Computing (SCC).
The rapid development of the SCC presents exciting opportunities for BCS through both intellectual and physical proximity. The college is creating 25 new faculty positions spanning six strategic areas, that aim to connect computing with other disciplines across MIT. One of those areas, “Computing and Natural Intelligence: Cognition, Perception, and Language,” aims to close the gap between the science and engineering of intelligence, a natural fit for programmatic connections with BCS. Construction of the new SCC building began this summer next door to Building 46, and current plans are for there to be a connection between the two: a physical bridge to complement the strategic one.
Literal or figurative, the connections are vital, says DiCarlo. “Computation is essential to the future of the field of brain and cognitive science; BCS is already a world leader because this is part of our DNA. Building bridges between BCS and other parts of campus help connect us to the best resources and expertise in the world so we can create the meaningful collaborations needed to drive the field forward.”
Image credit: Rachel Donahue