Belief Formation

René Magritte's "Decalcomanie"

 

Most neurobiological studies of learning focus on reward learning (or avoidance learning in the case of potential losses). We have started to dissociate learning about rewards (values) and learning about the chances of the events leading to rewards (probabilities). The latter is really what belief formation is about. In most realistic situations, humans have to learn the chances of events without knowing beforehand whether these events will be rewarding. Classical conditioning theories such as reinforcement learning are not adequate for such situations. And indeed, we have discovered that the brain system engaged in probability learning is quite different from that involved in reward learning. Mathieu d'Acremont is spearheading this research; see http://sites.google.com/site/mathieudacremont/