Sentence-level ERP effects as error propagation: A neurocomputational model
Event-related potentials (ERPs) provide a window into how the brain is processing language. Here, we propose a new theory that argues that ERPs such as the N400 and P600 arise as side effects of an error-based learning mechanism that plays an important role in language learning and linguistic adaptation. We instantiated this theory in a connectionist model that can simulate data from three studies on the N400 (amplitude modulation by expectancy, contextual constraint, and sentence position), five studies on the P600 (agreement, tense, word category, subcategorization and garden-path sentences), and a study on the semantic P600 in role reversal anomalies. Furthermore, since there is a tight link between learning and ERPs, we can explain developmental changes in ERPs, adaptation of ERP amplitude to frequency manipulations within an experiment, and the sensitivity of ERPs to word predictability in previous sentences. The model provides a unified account of the sensitivity of ERPs to expectation mismatch, the relative timing of the N400 and P600, the semantic nature of the N400, and the syntactic nature of the P600. The main innovation of this work is to provide a functional reason why ERPs exist, namely ERPs are the error signals that support language learning and adaptation.
Fitz, H. and Chang, F. (2019). Language ERPs reflect learning through prediction error propagation. Cognitive Psychology, 111, 15–52, 10.1016/j.cogpsych.2019.03.002 pdf.
The code for the model is available here.