Language models and brains align due to more than next-word prediction and word-level information
This work addresses a fundamental question in cognitive neuroscience about the mechanisms behind brain-model alignment, but it is incremental as it builds on prior findings without introducing a new paradigm.
The study investigated whether next-word prediction is necessary for aligning language models with brain recordings, finding that improvements in alignment are due to more than just next-word prediction and word-level information.
Pretrained language models have been shown to significantly predict brain recordings of people comprehending language. Recent work suggests that the prediction of the next word is a key mechanism that contributes to this alignment. What is not yet understood is whether prediction of the next word is necessary for this observed alignment or simply sufficient, and whether there are other shared mechanisms or information that are similarly important. In this work, we take a step towards understanding the reasons for brain alignment via two simple perturbations in popular pretrained language models. These perturbations help us design contrasts that can control for different types of information. By contrasting the brain alignment of these differently perturbed models, we show that improvements in alignment with brain recordings are due to more than improvements in next-word prediction and word-level information.