Flying horses, green Martians, and square circles. How our brain can combine meaning.

Monday 17 June 2024, 10:30 am
PhD candidate
R.P. Sommers
prof. dr. P. Hagoort
dr. M.S. Nieuwland

“Did you see that big man with the little dog?” We find it quite normal to combine words to describe things around us. But actually, this is a particularly clever trick of our brains. It requires a flexibility and creativity that our artificial language models – no matter how clever – have apparently not yet developed. Even more extraordinary is the fact that we can combine an increasing amount of information across sentence boundaries. Think for example of everything you find out about Harry Potter when reading the book series. To do this, we need to be able to refer back to a person in an earlier sentence.
In my PhD thesis, I investigated this phenomenon using EEG and MEG, two methods for measuring subjects' brain activity when reading. Among other things, I found that the frontoparietal network becomes activated when reading new reference words (compared to reference words that refer back to previously mentioned individuals). The frontoparietal network is often associated with the working memory, which allows us to flexibly manipulate information. I also used insights from neurobiology to program a computational language model that can build complex reference objects by grouping words together.

Rowan Sommers was born in Uden, the Netherlands, in 1989. He went to school at the Bernrode Gymnasium in Heeswijk-Dinther. He obtained his Bachelor's degrees in Psychology and Philosophy from Radboud University, and his Master's degrees in Cognitive Neuroscience (cum laude) and Philosophy (cum laude), also from Radboud University. He conducted his PhD research at the Max Planck Institute for Psycholinguistics and the Donders Institute for Brain, Cognition and Behaviour. He currently works as a postdoctoral researcher at the Institute for Cognitive Sciences in Osnabrück.