Robust Color-Shape Binding Representations for Multiple Objects in Visual Working Memory

• Published in: Journal of experimental psychology. General 2019-05, Vol.148 (5), p.905-925
• Main Author: Saiki, Jun
• Format: Article
• Language: English
• Subjects: Color; Fads and Fashions; Female; Human; Male; Memory; Meta-analysis; Short Term Memory; Systematic review; Tracking; Visual Memory; Visual perception
• ISSN: 0096-3445; 1939-2222
• DOI: 10.1037/xge0000562

The nature of feature-bound object representations in visual working memory (VWM) remains unclear. Many studies claim that they are held by a resource-limited system and are fragile. Using a novel paradigm called the redundant feature reviewing task, the current study showed that color-shape binding representations for multiple objects are maintained and matched with perceptual representation in a robust fashion in VWM. A set of features was presented in a two-object memory display, followed by a linking display in which placeholders either moved or stayed. Then participants judged if a single object probe contained any features of the memory display, regardless of object correspondence. The advantage of color-shape conjunction relative to single features, as indexed by feature coactivation, was retained in the memory matching within the same object and between different objects, regardless of object motion. In contrast, the object-specific preview benefit (OSPB), an index of accessibility to memory representations, was reduced when objects moved, suggesting that motion disrupts the spatiotemporal continuity of memory representations. A meta-analysis of the magnitude of the OSPB in the moving condition across the experiments showed a significant OSPB, suggesting that object file representations bound to spatiotemporal locations are responsible for feature coactivation. Taken together, these findings suggest that the nonspatial features of multiple objects are bound and maintained in VWM, in a robust fashion, regardless of the objects' motion. The cost of object motion in the memory matching performance likely reflects the correspondence between identity and spatiotemporal location, rather than the binding of nonspatial features.