Abstract
Introduction: Basing on the hierarchical working memory theory, the present study investigated the relation between representations of feature- and object-level. Participants needed to complete a color and a face recognition task in the same trial. Memory load of color task was manipulated. According to previous studies, increasing the colors load would reduce the fidelity of facial features competing for resources. We assumed that, if there was independence between these two levels, degradation of feature fidelity should have little influence on object-level representation (increasing the color load won’t affect the face performance). In contrast, the opposite result should be observed. Method: 24 students participated. The experiment comprised two sessions. Each session comprised three blocks (64 trials each) that set size of colors in color task was two, three, and four respectively. In the face-first session, each trial began with two visually-presented digits for rehearsing to prevent semantic encoding. Then, the study items of face task (an upright/inverted face, 50% each) and color task (color patches) were presented subsequently. The probes of face task (a face, 50% matched) and color task (a color patch, 50% matched) was then presented subsequently. Participants needed to match the probe with study item for each task and respond by key press. The trial end with two visually-presented digits and participants needed to match them with those had rehearsed. The task sequence was switched in the color-first session. Orders of sessions and blocks were randomized. VWM performance was measured by the sensitivity (d’). Main Results: Three-way analyses of variance with within-subjects factors of task order, color load, and face orientation were conducted on d’ of face task. Only the main effect of task order (color-first better) and face orientation (upright better) were significant. Conclusion: There is some level of independence between feature- and object-level representations in VWM.
Meeting abstract presented at VSS 2015