Our Research

The number of neurons in the human brain is enormous, but the number of meaningful thoughts you could possibly entertain is much larger still. You can understand the sentences "The strongest man in Iceland learned to play bassoon at the Sorbonne" or "The robot with a mustache can't make a free-throw", or “The United States congress is a well-functioning institution”, even though no one has ever said or even thought these particular things.

But if someone asks you to remember the following digits in the order in which they were presented: (5,9, 0, 1, 5, 3, 6, 7, 8), it’s unlikely that the same billions of neurons (and trillions of synapses) will be up to the task.

We draw on neural networks and information-theory to better understand our ability to efficiently make sense of the world in some cases, but not others. We evaluate computational models models using human behavioral and functional magnetic resonance imaging (fMRI) experiments, and are part of the new AI Research Institute for Interaction with AI assistants (ARIA), a multi-university research center based at Brown University.