Article (Scientific journals)
Semantic Representations in Working Memory: A Computational Model
Kowialiewski, Benjamin; Oberauer, Klaus
2025In Psychological Review
Peer Reviewed verified by ORBi
 

Files


Full Text
Manuscript.pdf
Author postprint (1.53 MB)
Download

All documents in ORBi are protected by a user license.

Send to



Details



Keywords :
Working Memory; Serial Recall; Semantics; Semantic similarity; Computational modeling
Abstract :
[en] Verbal Working Memory (WM) is supported by semantic knowledge. One manifestation of this is the rich pattern of semantic similarity effects found in immediate serial recall tasks. These effects differ from the effects of similarity on other dimensions (e.g., phonological similarity), which renders them difficult to explain. We propose a comprehensive mechanistic explanation of semantic similarity effects by extending a standard connectionist architecture for modeling immediate serial recall to incorporate semantic representations. Central to our proposal is the selective encoding of categorical features shared among multiple list items. The selective encoding of shared semantic features is made possible via a tagging mechanism that enables the model to encode shared feature retrospectively. Through this mechanism, our model accounts for the majority of semantic similarity effects. Our results imply that working memory represents semantic information in a more restricted way than phonological information.
Disciplines :
Theoretical & cognitive psychology
Author, co-author :
Kowialiewski, Benjamin  ;  Université de Liège - ULiège > Département de Psychologie
Oberauer, Klaus;  UZH - University of Zürich > Cognitive Psychology
Language :
English
Title :
Semantic Representations in Working Memory: A Computational Model
Publication date :
2025
Journal title :
Psychological Review
ISSN :
0033-295X
Publisher :
American Psychological Association, Washington, United States - District of Columbia
Peer reviewed :
Peer Reviewed verified by ORBi
Available on ORBi :
since 28 March 2025

Statistics


Number of views
164 (6 by ULiège)
Number of downloads
323 (3 by ULiège)

Scopus citations®
 
2
Scopus citations®
without self-citations
2
OpenCitations
 
0
OpenAlex citations
 
2

Bibliography


Similar publications



Contact ORBi