Theoretical correlations and measured correlations: relating recognition and recall in four distributed memory models

Area of Science:

• Cognitive Psychology
• Computational Neuroscience
• Memory Research

Background:

• Item recognition and associative recall are fundamental memory tasks.
• Understanding the relationship between these tasks offers insights into underlying memory processes.
• Existing models often struggle to predict the observed correlation between recognition and recall performance.

Purpose of the Study:

• To investigate the common memory processes shared by item recognition and associative recall.
• To evaluate the capacity of different distributed memory models to account for the stable correlation between these memory tasks.
• To develop a theoretical framework for interpreting the contingencies observed in successive memory tests.

Main Methods:

• Assessed four classes of distributed memory models.
• Analyzed the contingency between successful recognition and successful recall.
• Introduced variability in goodness-of-encoding, response criteria, and output encoding within models.
• Examined performance in two mixed-list experiments.

Main Results:

• Basic memory models either under- or overpredicted the intertask correlation between recognition and recall.
• Models incorporating variability in encoding and response criteria successfully reproduced the moderate intertask correlation (approximately .5).
• These enhanced models also accounted for the observed increase in correlation in mixed-list experiments.

Conclusions:

• Distributed memory models, when accounting for encoding and response variability, can successfully explain the relationship between item recognition and associative recall.
• The findings provide a robust theoretical framework for understanding the interplay between different memory retrieval processes.
• This research advances computational approaches to memory research by refining model predictions.