# A Framework for Decoding Event-Related Potentials from Text

## TL;DR

This paper introduces a new framework combining convolutional decoders and language models to predict ERPs during reading, revealing that hybrid models outperform individual approaches in reconstructing neural responses.

## Contribution

The paper presents a novel framework for decoding ERPs from text using combined neural and language models, advancing understanding of neural language processing.

## Key findings

- Hybrid models outperform individual models in ERP reconstruction
- Contextual embeddings underperform surprisal-based models alone
- Combining models yields superior decoding accuracy

## Abstract

We propose a novel framework for modeling event-related potentials (ERPs) collected during reading that couples pre-trained convolutional decoders with a language model. Using this framework, we compare the abilities of a variety of existing and novel sentence processing models to reconstruct ERPs. We find that modern contextual word embeddings underperform surprisal-based models but that, combined, the two outperform either on its own.