Randomized trial assesses how language justifications predict recognition accuracy in memory tests, suggesting language reflects recollective experiences.
Key Points
This research aims to explore how the language used to justify memory decisions relates to recognition accuracy and recall performance.
Two experiments conducted where subjects justified recognition claims for hits and false alarms.
Trained a bag-of-words and BERT embeddings-based classifier to predict recognition outcomes from justifications.
Experiment 2 integrated numeric confidence ratings for comparison with language-based predictions.
Language models successfully explained recognition accuracy and predicted final recall of justified items (p<0.05).
Language predictions outperformed confidence ratings in predicting recall outcomes.
Support for the recollection sensitivity hypothesis was confirmed through findings regarding memorable recollective experiences.