Related Experiment Videos
ROC curves and confidence judgements in recognition memory.
1Department of Psychology, Johns Hopkins University, Baltimore, Maryland 21218, USA.
Summary
Recognition memory models assume confidence directly reflects familiarity. Experiments show confidence judgments are not solely based on familiarity, suggesting a more complex information accumulation process in memory.
Area of Science:
- Cognitive Psychology
- Neuroscience
- Memory Research
Background:
- Current recognition memory models often use a signal detection theory framework.
- These models assume confidence judgments are directly proportional to familiarity signals.
Purpose of the Study:
- To test the assumption that confidence in recognition memory is directly scaled from perceived familiarity.
- To investigate how response bias affects the relationship between confidence and accuracy.
Main Methods:
- Two experiments were conducted manipulating response bias through stimulus probabilities and payoffs.
- Confidence-based z-receiver operating characteristic (zROC) curves were analyzed.
Main Results:
- The shape of zROC curves varied significantly with induced response bias.
- This variation indicates that confidence is not a simple scaling of familiarity or likelihood.
Conclusions:
- The findings challenge the direct familiarity-scaling assumption in current recognition memory models.
- A novel information accumulation model is proposed to explain the observed confidence-accuracy relationship.