How well does surprisal explain N400 amplitude under different experimental conditions?

TL;DR

This study assesses how effectively word surprisal, computed via neural networks, predicts N400 amplitude across various experimental conditions, revealing both its strengths and limitations in modeling neural language processing.

Contribution

It demonstrates the applicability of neural network-based surprisal in predicting N400 responses and identifies conditions where it falls short, offering insights into neurocognitive language processing.

Findings

Surprisal predicts N400 amplitude in many cases.

Certain experimental conditions reveal limits of surprisal's predictive power.

Insights into neurocognitive processes underlying language comprehension.

Abstract

We investigate the extent to which word surprisal can be used to predict a neural measure of human language processing difficulty - the N400. To do this, we use recurrent neural networks to calculate the surprisal of stimuli from previously published neurolinguistic studies of the N400. We find that surprisal can predict N400 amplitude in a wide range of cases, and the cases where it cannot do so provide valuable insight into the neurocognitive processes underlying the response.