Do all subjects fit the same recognition memory model? Comparisons of continuous, discrete, and hybrid models using...

Area of Science:

• Cognitive Psychology
• Computational Neuroscience
• Psychometrics

Background:

• Traditional recognition memory models (e.g., two-high-threshold, signal detection theory) have limitations.
• Multinomial processing tree models enhance analysis of ordinal and continuous variables.
• Previous simulations showed response times (RTs) and confidence levels (CLs) improve parameter estimation and model selection accuracy.

Purpose of the Study:

• To investigate whether recognition memory is a continuous or discrete process.
• To propose and evaluate a hybrid model extending the Atkinson-Juola (AJ) model.
• To analyze recognition memory data using RTs and CLs.

Main Methods:

• Utilized data from a recognition memory experiment (Juola et al., 2019).
• Manipulated relative target frequencies and recorded RTs and CLs.
• Proposed and tested a hybrid AJ model against purely discrete and continuous models.

Main Results:

• The extended AJ model provided a superior fit for most participants.
• Patterns of CLs and RTs aligned with the AJ model's theoretical assumptions.
• No single model adequately fit all participants, highlighting individual variability.

Conclusions:

• A hybrid model offers a better representation of recognition memory than purely discrete or continuous approaches.
• Individual differences in cognitive strategies persist, indicating a need for more generalized models.
• Future research should focus on models accommodating diverse decision-making strategies in recognition memory.