Dr. Mihaly Banyai

My primary interest is the way algorithms that people use to maximise their reward shape the learned representation of environmental statistics. In other words, what properties of the environment do we learn depending on what decisions we have to make based on them. I build probabilistic models of perception and action to predict behaviour in experiments probing human statistical learning, and neural representations measurable with imaging methods. In particular, I'm interested in the effect of computational resource constraints on the learning of task-dependent stimulus representations. Previously I worked on modelling neural representations in the visual cortex. List of publications can be accessed here.