Examining the signal-detection account of visual working memory

Area of Science:

• Cognitive Psychology
• Neuroscience
• Psychophysics

Background:

• Visual working memory research often contrasts signal-detection theory (SDT) with discrete-slots models.
• Existing debates rely heavily on unverified parametric assumptions and selective-influence manipulations.
• A general latent-variable framework, random-scale representation, offers a unifying approach to test these models.

Purpose of the Study:

• To investigate whether visual working memory judgments adhere to the structural constraints of a random-scale representation.
• To determine if this framework can differentiate between SDT and discrete-slots models.
• To examine the applicability of these constraints across different memory tasks, including item-location binding.

Main Methods:

• Employed multiple-alternative forced-choice judgments to assess adherence to random-scale constraints.
• Reconstructed single-item receiver operating characteristic (ROC) functions without response-bias manipulations.
• Tested random-scale constraints in change detection tasks involving item-location bindings.

Main Results:

• Judgments in Experiments 1a and 1b conformed to random-scale constraints, enabling ROC reconstruction.
• Reconstructed ROC functions exhibited curvature and asymmetry, contradicting discrete-slots predictions.
• Random-scale constraints failed when their theoretical conditions were unmet (Experiment 2).
• Item-location binding in change detection also produced asymmetric ROC functions, incompatible with SDT and discrete-slots models.

Conclusions:

• Visual working memory judgments satisfy the general assumptions of SDT.
• The findings undermine the core tenets of discrete-slots theories.
• A random-scale representation provides a more comprehensive framework for understanding visual working memory.