Decomposing the semantic processes underpinning veridical and false memories

Area of Science:

• Cognitive Psychology
• Neuroscience
• Computational Linguistics

Background:

• The precise semantic mechanisms underlying false memory formation remain unclear.
• Understanding these processes is crucial for memory research and applications.

Purpose of the Study:

• To investigate the semantic origins of false memories using data-driven semantic models.
• To differentiate the role of semantic memory in veridical versus false memory.

Main Methods:

• Utilized the Deese-Roediger-McDermott (DRM) task with word lists and recognition tests.
• Employed distributional semantic models to predict participant performance.
• Reanalyzed previous data and conducted an independent experiment for validation.

Main Results:

• Participant performance was better explained by local semantic strategies than global ones.
• False recognition increased with higher semantic similarity between lures and list words.
• Semantic memory involvement differs significantly between true and false memories.

Conclusions:

• False memories arise from specific semantic clusters activated by individual word lists.
• Memory performance follows a continuous semantic gradient influenced by item similarity.
• An empirically-driven framework for semantic processes in memory formation is proposed.