Dual Alignment Between Language Model Layers and Human Sentence Processing

TL;DR

This study investigates how different layers of large language models align with human sentence processing, especially in syntactically challenging contexts, revealing that earlier layers mimic natural reading, while later layers better model complex syntactic processing.

Contribution

It demonstrates that different layers of LLMs correspond to distinct modes of human sentence processing, especially under syntactic challenge, and introduces probability-update measures for improved modeling.

Findings

Earlier layers align with naturalistic reading behavior.

Later layers better estimate cognitive effort in syntactic ambiguity.

Probability-update measures offer complementary advantages in reading time prediction.

Abstract

A recent study (Kuribayashi et al., 2025) has shown that human sentence processing behavior, typically measured on syntactically unchallenging constructions, can be effectively modeled using surprisal from early layers of large language models (LLMs). This raises the question of whether such advantages of internal layers extend to more syntactically challenging constructions, where surprisal has been reported to underestimate human cognitive effort. In this paper, we begin by exploring internal layers that better estimate human cognitive effort observed in syntactic ambiguity processing in English. Our experiments show that, in contrast to naturalistic reading, later layers better estimate such a cognitive effort, but still underestimate the human data. This dual alignment sheds light on different modes of sentence processing in humans and LMs: naturalistic reading employs a somewhat weak prediction akin to earlier layers of LMs, while syntactically challenging processing requires more fully-contextualized representations, better modeled by later layers of LMs. Motivated by these findings, we also explore several probability-update measures using shallow and deep layers of LMs, showing a complementary advantage to single-layer's surprisal in reading time modeling.