EUVIMEDEUVIMED entdecken
Open Access · CC BY 4.0

Modeling individual differences in working memory: Subject-level parameter recovery within the Memory Measurement Model framework (M3)

Jan Göttmann, Gidon T. Frischkorn, Klaus Oberauer, Simon B. Schaefer, Anna-Lena Schubert

Journal of Mathematical Psychology · 2026-12 · 131 · 103015

DOI: 10.1016/j.jmp.2026.103015 ↗

Kurz gefasst

Fachartikel in der Zeitschrift Journal of Mathematical Psychology, 2026-12. Verfasst von Jan Göttmann und 4 weiteren Autor:innen. Die Arbeit nennt 88 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0). Gefördert von: Baden-Württemberg Foundation.

AbstractBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

The memory measurement model (M 3 ) is a flexible modeling framework designed to measure parameters associated with different processes in working memory. It assumes that memory representations receive different levels of activation at retrieval, which arise from the sum of the strengths of item and binding memory. Transforming the activation of representations in different response categories into their respective recall probabilities then allows estimating the contributions of different memory processes to working memory performance. So far, parameter recovery has been reported only for group-level parameters of the M 3 . In contrast to experimental research, individual differences research focuses on variation in subject-level parameters. Adequate recovery of subject-level parameters is therefore a requirement for their valid measurement. To address this gap, we ran a parameter recovery simulation to assess subject-level parameter recovery as function of different experimental design choices. Our Bayesian hierarchical M 3 implementation recovers subject parameters acceptably to very well, as demonstrated both through simulations and empirical parameter recovery using a complex span dataset. Based on the results, we recommend that researchers ensure that participants select not-presented lures (i.e., items not in the memory set) at least sometimes (> 2%) and collect a minimum of 200 retrieval responses per subject (40 trials at set size 5). When these conditions are met, the simple and complex span models specified in the M 3 framework exhibit robust parameter recovery, well-calibrated posteriors, and satisfactory reliability. These challenges primarily arose from the substantial amount of data required to achieve sufficiently precise parameter estimates. In contrast, M 3 parameters modeling free-time effects such as extended encoding and removal processes could not be recovered with sufficient precision and showed poor reliability. Posteriors of the M 3 -CS-EE model were well calibrated, while the M 3 -MU model showed systematic miscalibration of the subject-level parameters.

Förderung

  • Baden-Württemberg Foundation

Zitieren

APA

Göttmann, J., Frischkorn, G. T., Oberauer, K., Schaefer, S. B., & Schubert, A. (2026). Modeling individual differences in working memory: Subject-level parameter recovery within the Memory Measurement Model framework (M3). Journal of Mathematical Psychology, 131, 103015. https://doi.org/10.1016/j.jmp.2026.103015

Vancouver

Göttmann J, Frischkorn GT, Oberauer K, Schaefer SB, Schubert A. Modeling individual differences in working memory: Subject-level parameter recovery within the Memory Measurement Model framework (M3). Journal of Mathematical Psychology. 2026;131:103015. doi:10.1016/j.jmp.2026.103015

BibTeX
@article{Gttmann2026,
  author = {Göttmann, Jan and Frischkorn, Gidon T. and Oberauer, Klaus and Schaefer, Simon B. and Schubert, Anna-Lena},
  title = {Modeling individual differences in working memory: Subject-level parameter recovery within the Memory Measurement Model framework (M3)},
  journal = {Journal of Mathematical Psychology},
  year = {2026},
  volume = {131},
  pages = {103015},
  doi = {10.1016/j.jmp.2026.103015}
}
Kostenlos starten →