Related Experiment Videos
Discrete threshold versus continuous strength models of perceptual recognition
1Department of Psychology, New Mexico State University, Las Cruces 88003, USA. kenp@crl.nmsu.edu
Summary
This study compared discrete-threshold and continuous-strength models of word recognition. Continuous strength models better predicted experimental results, suggesting decision criteria are applied to continuous measures rather than discrete thresholds.
Area of Science:
- Cognitive Psychology
- Psycholinguistics
- Computational Neuroscience
Background:
- Models of word recognition vary in their assumptions about decision-making processes.
- Discrete-threshold models propose distinct cutoffs for recognition, while continuous-strength models suggest criteria applied to graded activation levels.
Purpose of the Study:
- To empirically test discrete-threshold models against continuous-strength models of letter and word recognition.
- To determine which class of models better accounts for behavioral data in recognition tasks.
Main Methods:
- Experiment 1: Two-alternative forced-choice task with confidence ratings and whole-report, manipulating neighborhood frequency, lexical bias, word superiority, and pseudoword advantage.
- Experiment 2: Same-different task with confidence ratings to generate receiver operating characteristics (ROCs).
- Model predictions from the Dual-Readout Model (DROM) and a revised Activation-Verification Model were compared to empirical data.
Main Results:
- Experiment 1 showed discrepancies between DROM predictions and observed data; continuous strength model predictions for response distributions were superior.
- Experiment 2 ROC curves were more consistent with continuous strength assumptions than discrete threshold models.
Conclusions:
- Findings support models where decision criteria are applied to continuous measures of word strength over discrete-threshold models.
- The continuous strength assumption provides a more accurate account of human word recognition processes.