A diffusion decision model analysis of evidence variability in the lexical decision task.

Area of Science:

• Psycholinguistics
• Cognitive Psychology
• Computational Neuroscience

Background:

• The lexical-decision task is a cornerstone paradigm in psycholinguistics.
• Existing frameworks like Signal-Detection Theory and the Diffusion Decision Model (DDM) conceptualize lexical decisions based on accumulating evidence.
• The Retrieving Effectively from Memory model of Lexical-Decision (REM-LD) offers a detailed account of this task.

Purpose of the Study:

• To empirically test the predictions of the REM-LD model regarding evidence variability in the lexical-decision task.
• Specifically, to investigate whether word-likeness evidence is more variable for words than non-words, and for higher frequency words than lower frequency words.
• To analyze existing lexical-decision data using the DDM to assess these REM-LD predictions.

Main Methods:

• Analysis of five distinct lexical-decision datasets.
• Application of the Diffusion Decision Model (DDM) to quantify decision processes.
• Examination of drift-rate variability across different stimulus conditions (words vs. non-words, word frequency).

Main Results:

• Significant changes in drift-rate variability were observed across word frequency and non-word conditions in all analyzed datasets.
• The empirical results largely supported REM-LD's predictions concerning the relative ordering of evidence variability.
• Evidence variability patterns align with REM-LD's hypotheses about word and non-word processing.

Conclusions:

• The findings provide strong support for the REM-LD model's account of the lexical-decision task.
• The study validates key assumptions about evidence accumulation and variability in word recognition.
• This research contributes to a deeper understanding of the cognitive mechanisms underlying lexical access and decision-making.