dr. R.S. van Bergen (Ruben)

Postdoc - Artificiële intelligentie
Postdoc - Donders Centre for Cognition
Postdoc - Donders Institute for Brain, Cognition and Behaviour

Our brain is a computer: an information processor. And the main information our brain needs to process is: what is in the world around me? How is that world shaped - what is its physical layout, so I can navigate through it? And what are the objects in it, so I can interact with them?

The brain is not like a man-made computer, however, in two ways that I find very interesting. First, while the computer crisply encodes information as 1's and 0's, the information that our brain has to deal with is often more fuzzy, imprecise, and ambiguous. Our senses do not convey perfectly reliable or complete impressions of the external world. Information may be missing (such as when one object visually occludes another), corrupted by external disturbances (for instance, when trying to understand your friend over the background noise of a crowded bar), or even damaged internally through noise in the sensory organs and neurons themselves. This means that the input we receive from our senses is often consistent with multiple interpretations. A central question of my research is how the brain computes with such uncertain information. In contrast to the very definite 1's and 0's used in computers, our working hypothesis is that the brain employs probabilities to encode information, to allow for multiple paths of interpretation in its computations.

Brains are also not like classical computers, in that they are not programmed by an intelligent supervisor. Through evolution and through experience in our own lifetimes, our brains have (had) to learn how to interpret the world, and how to interact with it. Therefore, I believe the most fundamental level at which we can understand the brain, is in the way it learns to do these things. Interestingly, computers are increasingly able to do this as well, through methods known as "machine learning". It is my hope that insights gained from learning in computers may guide us to understand learning in the brain.