How Many Bytes Can You Take Out Of Brain-To-Text Decoding?

TL;DR

This paper introduces an information-based metric for evaluating brain-to-text decoders, demonstrating significant performance improvements and analyzing their properties, with implications for future decoder development.

Contribution

It proposes a novel information-based evaluation metric and explores methods to enhance brain-to-text decoding performance, including empirical analysis of decoder properties.

Findings

Decoders can be improved by over 40% with proposed methods.

Brain-to-text decoders exhibit Zipfian power law dynamics.

Estimated ideal performance suggests practical decoders are feasible.

Abstract

Brain-computer interfaces have promising medical and scientific applications for aiding speech and studying the brain. In this work, we propose an information-based evaluation metric for brain-to-text decoders. Using this metric, we examine two methods to augment existing state-of-the-art continuous text decoders. We show that these methods, in concert, can improve brain decoding performance by upwards of 40% when compared to a baseline model. We further examine the informatic properties of brain-to-text decoders and show empirically that they have Zipfian power law dynamics. Finally, we provide an estimate for the idealized performance of an fMRI-based text decoder. We compare this idealized model to our current model, and use our information-based metric to quantify the main sources of decoding error. We conclude that a practical brain-to-text decoder is likely possible given further algorithmic improvements.