Across the Levels of Analysis: Explaining Predictive Processing in Humans Requires More Than Machine-Estimated Probabilities
TLDR
This paper critiques the over-reliance on LMs for explaining human language prediction and proposes integrating psycholinguistic models.
Key contributions
- Critiques two claims: LMs are central to language processing and essential for psycholinguistics.
- Argues human predictive processing needs more than machine-estimated probabilities.
- Proposes combining LLM strengths with psycholinguistic models for future research.
- Frames the critique using Marr's levels of analysis for a deeper understanding.
Why it matters
This paper challenges the prevailing view that large language models alone can fully explain human language prediction. It advocates for a more nuanced approach, integrating psycholinguistic theories with LLM capabilities. This shift is crucial for advancing our understanding of human cognition.
Original Abstract
Under the lens of Marr's levels of analysis, we critique and extend two claims about language models (LMs) and language processing: first, that predicting upcoming linguistic information based on context is central to language processing, and second, that many advances in psycholinguistics would be impossible without large language models (LLMs). We further outline future directions that combine the strengths of LLMs with psycholinguistic models.
📬 Weekly AI Paper Digest
Get the top 10 AI/ML arXiv papers from the week — summarized, scored, and delivered to your inbox every Monday.