Individual Differences in Cue Weighting in Sentence Comprehension: An Evaluation Using Approximate Bayesian...

Area of Science:

• Cognitive Science
• Psycholinguistics
• Computational Linguistics

Background:

• Cue-based retrieval theories propose content-addressable search for syntactic dependencies.
• Recent models suggest cue-weighting, where one cue dominates, to explain average sentence processing behavior.

Purpose of the Study:

• To investigate systematic individual-level variations in cue weighting during sentence processing.
• To examine the relationship between reading speed and cue weighting strategies.

Main Methods:

• Utilized the Lewis and Vasishth cue-based retrieval model.
• Estimated individual-level parameters for reading speed and cue weighting across 13 published datasets.
• Employed hierarchical approximate Bayesian computation (ABC) for parameter estimation.

Main Results:

• Found significant individual differences in cue weighting; not all participants weight cues uniformly.
• Identified that only faster readers tend to exhibit the predicted higher weighting for structural cues.
• Demonstrated a potential association between reading proficiency (approximated by reading speed) and cue weighting.

Conclusions:

• Individual variation in cue weighting is a key factor in sentence processing, challenging uniform cue-weighting models.
• Reading speed may be a significant predictor of how individuals weight retrieval cues for syntactic dependencies.
• The study validates a method for exploring individual differences within complex computational models of sentence processing.