Mechanisms of feature binding in visual working memory are stable over long delays

No Fixed Limit for Storing Simple Visual Features: Realistic Objects Provide an Efficient Scaffold for Holding Features in Mind

“…However, previous studies used much more subtle manipulations of perceptual distinctiveness. For example, Brown et al (2021) tested how well participants bind two visual features-color and orientation-and found higher memory performance when the cued dimension at retrieval was more distinct and more binding errors when they were less distinct, in line with previous work finding that binding errors may be the main result when items become less distinct from each other (Emrich & Ferber, 2012;Oberauer & Lin, 2017). Were binding errors the main cause of the current results?…”

“…Assuming memories are organized in a hierarchically structured way, where an individual memorandum comprises bundles of both feature and object-level representations , real-world objects could provide a particularly useful structure for lower-level visual features: Realistic objects connect to higher-order conceptual knowledge, introducing additional dimensionality and potentially reducing interference between the otherwise lower-level simple features (Cohen et al, 2014;Wyble et al, 2016). Thus, similar to how stimuli that are more distinct in location or orientation have less interference between them and lead to improved color memory performance (Brown et al, 2021;Emrich & Ferber, 2012;Oberauer & Lin, 2017), it is plausible that simple feature memory could benefit from being encoded as part of meaningful objects. This would be particularly important in the context of meaningful objects because it would also suggest that existing studies have consistently underestimated memory for simple visual features: In the real world, colors appear on meaningful realworld objects and not in isolation; thus, from an ecological perspective, it is crucial to characterize and understand feature memory and its capacity in more naturalistic conditions.…”

Statistical learning biases working memory maintenance.

“…The incidental feature manipulated in the task, location, may, however play a special role in scaffolding working-memory representations. According to some accounts (Brown et al, 2021;Emrich & Ferber, 2012;Schneegans & Bays, 2017), non-spatial features are anchored to their spatial framework, so that for colour cues to access orientation information, it must do so via location, which binds colour and orientation together. It remains an open question whether location regularity improved performance through a privileged role in supporting working memory or simply as acting as a helpful incidental feature.…”

Stochastic attractor models of visual working memory

“…In the continuous colour-report task illustrated in Fig 1 , participants are shown a stimulus array in an encoding phase, the array disappears during a maintenance interval, and in a decision phase participants are cued with a location and asked to rotate a colour wheel to match the colour of the item at that location in the stimulus array. Similar experimental paradigms vary as to whether the cued or reported attributes are colours, orientations or locations [ 8 , 19 , 20 ]. The use of continuous reports, in which the reported variable is continuous rather discrete, has changed our understanding of the nature of working memory.…”