Abstract
Visual short-term memory (VSTM) is severely limited in its capacity. Some researchers argue that VSTM capacity limits reflect a central “fixed slot” buffer that holds 4+/−1 objects (e.g., Luck and Vogel, 1997). Others suggest that capacity is variable and depends on the division of attentional resources across items. Recent evidence indicates that attention is mediated by independent resources in the left and right visual hemifields (Alvarez and Cavanaugh, 2005). In VSTM, hemifield independence might occur in terms of (additive) fixed slot capacity or (multiplicative) attentional resources effects on performance. Here, we investigated the “fixed slot,” “independent capacity,” and “independent resource” models by investigating how VSTM capacity varied between full-field and hemifield stimulus configurations.
Subjects performed a change detection task using oriented bars (50% chance of one bar changing orientation). Stimuli were either restricted to a single (random) hemifield or appeared across the whole visual field. The set size was varied across blocks. By using distractors, the number of presented stimuli remained constant across set size. A second study was performed without distractors to control for attentional demands of distractor suppression.
VSTM capacity was higher when stimuli were presented across the whole screen than when restricted to a hemifield, contrary to the central fixed slots model prediction. However, full field capacity was less than would be predicted from hemifield data by the “independent capacities” model. Instead, full field data were best predicted from hemifield data by a multiplicative “independent resource” model.
Our results demonstrate that VSTM capacity is optimal when objects are presented in both visual hemifields. They do not support either the fixed-slot or independent capacity forms of the capacity model. Rather, the results support the view that VSTM capacity is limited by attentional resources which in turn exhibit a high degree of hemispheric independence.
This research was supported by National Science Foundation BCS-0726061 to DCS.