Skip to main content
eScholarship
Open Access Publications from the University of California

Reevaluation of Formal Model Comparison Between Slot and Resource Models of Visual Working Memory

  • Author(s): Cappiello, Marcus
  • Advisor(s): Zhang, Weiwei
  • et al.
Abstract

Visual working memory actively maintains information over brief periods in service of other mental activities. Unlike other memory systems, visual working memory is highly limited in the amount of information that can be retained, however the nature of this limitation is still widely debated. Historically, research on working memory limitation was focused on the number of items that can be simultaneously maintained, but recently limitations in the precision of working memory representations has also been explored. Two theoretical models have been proposed to account for working memory limitations including precision that will be investigated in depth. The slot model theorizes a limited number of slots for working memory representations. When there are more items than there are slots, then some items will be forgotten entirely. Conversely, the resource model theorizes all items are retained in visual working memory, and any behavioral limitations stem from low precision of memory representations. That is, high error responses are accounted for by a capacity limit in the slot model, and accounted for by low precision representations in the resource model. Previous formal model comparison between the measurement models of these two theoretical models has found some support for the resource model, however important aspects have been overlooked. Support for the slot model is found in the current dissertation by expanding on the previous formal model comparison in four ways. First, a new model comparison analysis is created that includes the flexibility of model parameters; the inclusion of which shows support for the slot model at high set sizes when the capacity limit is exceeded. Second, using the measurement model for the resource model, a capacity limit is still observed. Third, using a non-parametric model fitting approach that does not assume any model, a capacity limitation is again found. Finally, a new measurement model is created that better matches the slot theoretical model. This new model outperforms the variable precision model showing further support for the existence of a capacity limit in visual working memory.

Main Content
Current View