LEXIC: Lightweight Eye-tracking eXtension via Injected Complexity
Quick Answer
The LEXIC model enhances gaze-only reading comprehension predictions by integrating word-level difficulty signals, achieving AUROC improvements of 1.8 to 2.9 percentage points on the OneStop task.
Quick Take
This study highlights the potential of lightweight conditioning methods to bridge performance gaps in eye-tracking models, particularly for unseen text and readers.
Key Points
- LEXIC-Base builds on EyeBench AhnCNN baseline for improved gaze prediction.
- Two mechanisms, LEXIC-Concat and LEXIC-Res, inject word-level difficulty signals.
- Statistically significant AUROC gains of +1.8 to +2.2 points on Unseen Text.
- LEXIC-Concat achieves +2.9 points improvement for Unseen Reader, p = 0.010.
- Architectural limitations observed in LEXIC-Res for out-of-distribution readers.
Paper Resources
📖 Reader Mode
~2 min readAbstract:On the recent EyeBench benchmark, predicting reading comprehension from eye movements exposes a stark gap: text-aware models using pretrained language models reach 56--63% AUROC, while gaze-only models operate at chance. We ask how far a gaze-only model can be pushed by lightweight, language-model-free conditioning. Building on the EyeBench AhnCNN baseline, LEXIC-Base, we propose two mechanisms to inject three precomputed word-level difficulty signals, GPT-2 surprisal, word frequency, and word length, into the per-fixation input: direct concatenation, LEXIC-Concat, and a residual mechanism, LEXIC-Res, where a small head predicts typical-reader gaze response and the encoder is conditioned on the deviation. On the OneStop reading comprehension task, with K=5 seed-ensemble training across ten folds, both mechanisms produce statistically consistent AUROC gains on Unseen Text, +1.8 to +2.2 percentage points, Wilcoxon p <= 0.065. LEXIC-Concat additionally lifts Unseen Reader by +2.9 percentage points, p = 0.010. We trace an architectural boundary in LEXIC-Res on Unseen Reader, +1.8 percentage points, p = 0.19, to the prediction head being calibrated to training readers, transferring imperfectly to out-of-distribution readers.
| Comments: | Accepted to APCCAS 2026 |
| Subjects: | Computation and Language (cs.CL); Artificial Intelligence (cs.AI); Human-Computer Interaction (cs.HC); Machine Learning (cs.LG) |
| Cite as: | arXiv:2607.08152 [cs.CL] |
| (or arXiv:2607.08152v1 [cs.CL] for this version) | |
| https://doi.org/10.48550/arXiv.2607.08152 arXiv-issued DOI via DataCite (pending registration) |
Submission history
From: Kyeonghun Kim [view email]
[v1]
Thu, 9 Jul 2026 06:46:19 UTC (319 KB)
— Originally published at arxiv.org
Want this in your inbox every morning?
Daily brief at your local 8am — bilingual EN/中文, free.
More from arXiv cs.CL
See more →TriAgent: Divergence-Aware Committees for Cost-Efficient Financial Sentiment Analysis
TriAgent introduces a cost-efficient multi-agent system for financial sentiment analysis, combining VADER, FinBERT, and Qwen2.5. It achieves an F1 score of ~0.87 with significant savings of $9.3M/year at a 10M-user scale compared to GPT-4o-mini, while also detecting hallucinations with an AUC of 0.90.