Thesis defense Micha Heilbron (Donders series 555)

28 June 2022

Promotors: Prof. dr. F.P. de Lange, Prof. dr. P. Hagoort

Getting ahead: Prediction as a window into language, and language as a window into the predictive brain

The ability to transform linguistic signals into meaning is one of the most astonishing feats of the human brain. A recent theory describes the brain as a prediction machine, comparing incoming signals with internal predictions. The aim of this thesis is to evaluate this theory, using language as a testbed. The thesis asks two questions: first, when does language processing invoke prediction (i.e. under which conditions). And second, what is being predicted (i.e. at which level are predictions taking place)? The questions are addressed in five studies, combining computational modelling with fMRI, EEG, MEG and eye tracking, under diverse conditions – from participants viewing single letters to participants reading an entire novel. Overall, the results suggest that language processing constantly involves prediction, and that these unfold at many levels of abstraction: from predictions about the exact shape of a letter in a word, to predictions about the sound, meaning or grammar of the next word in a sentence. Together, the thesis supports the theory of the predictive brain, and suggests that the brain’s ability to understand or ‘follow’ any piece of language relies on its ability to get ahead and predict it.