TRACE-ing fixations in the Visual World Paradigm: Extending linking hypotheses and addressing individual differences...

Area of Science:

• Cognitive Science
• Psycholinguistics
• Computational Linguistics

Background:

• The visual world paradigm (VWP) is a key tool for studying language processing using eye-tracking.
• Existing cognitive models, like TRACE, can simulate VWP data, but interpretation of fixation patterns remains ambiguous.
• Decades of VWP research highlight the need for clearer links between data and cognitive theory.

Purpose of the Study:

• To review and critique alternative linking hypotheses for VWP eye-tracking data and cognitive models.
• To present a novel fixation model based on probabilistic sampling for simulating individual trials.
• To clarify misunderstandings and address critiques regarding linking models to VWP data.

Main Methods:

• Review of existing linking hypotheses and computational models (e.g., TRACE model).
• Development and presentation of a simple fixation model using probabilistic sampling from lexical activation.
• Simulation of individual trials to analyze fixation behavior.
• Analogies drawn from simpler paradigms like lexical decision tasks.

Main Results:

• A probabilistic sampling model can simulate VWP data at the individual trial level.
• Averaging trials from the sampling model approximates underlying activation patterns, similar to other models.
• The primary utility of trial-level simulation lies in explaining individual differences, not just central tendencies.

Conclusions:

• Linking eye-tracking data from the VWP to cognitive theories requires careful consideration of model assumptions.
• Probabilistic sampling models offer a viable approach for simulating trial-level VWP data and exploring individual variability.
• Further work is needed to resolve ambiguities in VWP data interpretation and strengthen the link to cognitive processes.