Abstract
When a group of single-feature stimuli is stored in visual working memory (VWM), stimuli with similar features form clusters. Based on this result, the present study investigated how stimuli with multiple features form clusters in VWM. In two experiments, participants remembered the orientations of five colored isosceles triangles and then estimated the orientation of a triangle cued to be recalled. Cluster structures of these triangles were independently manipulated on the color and orientation dimensions. Specifically, we made two clusters on the orientation dimension by forming a large cluster of four similar orientations and leaving dissimilar one alone as a single-item cluster. The triangles’ colors also formed two clusters (of four similar colors and a single dissimilar color) but made different junctions with the orientation-based clusters. In a congruent-junction condition, the color-based clusters corresponded to the orientation-based clusters such that four similar orientations had similar colors. In an incongruent-junction condition, however, one item from the large color cluster formed the single-item cluster on the orientation dimension while one item from the large orientation cluster formed the single-item cluster on the color dimension. We found that regardless of junction conditions, representations of the triangles formed clusters based on the orientation dimension, indexed by comparable orientation bias patterns between the two conditions. This indicated clustering of multi-feature stimuli based on a task-relevant feature dimension. However, in the incongruent-junction condition, the orientation of the single-item orientation cluster was falsely recalled as the orientations of the large color cluster whose colors were similar to its color. These swapping errors occurred more frequently when colors became more relevant to the task, by designating to-be-recalled item using colors (Experiment 2). These results suggest that stimuli with multiple features form clusters mostly based on a task-relevant feature dimension, but that an irrelevant dimension can interfere with the clustering.
Acknowledgement: This research was supported by the Brain Research Program through the National Research Foundation of Korea (NRF) funded by the Ministry of Science and ICT (NRF-2017M3C7A1029658).