Abstract
A central feature of visual working memory is that it is a highly capacity-limited system. Several studies have shown that, on average, subjects are highly accurate at maintaining about 3–4 simple objects in memory. Results such as these have been taken to suggest that memory capacity is primarily determined by a limit in the number of object representations (e.g., Luck & Vogel, 1997). Other research, however, has suggested that memory capacity may be best described as a flexible resource in which complex objects consume a greater amount of available memory capacity irrespective of the number of objects to be remembered (e.g., Alvarez & Cavanagh, 2004). Here, we recorded behavioral and electrophysiological data while subjects performed a change detection paradigm with simple and complex objects. Specifically, we examined the contralateral delay activity (CDA), which has been shown to be a highly sensitive online measure of the number of object representations that are maintained in memory during a given trial (e.g., Vogel & Machizawa, 2004). In a series of experiments, we found that change detection accuracy was considerably poorer for complex objects than for more simple objects. However, CDA amplitude was not modulated by the complexity of the objects despite being highly sensitive to the number of objects currently being held in memory. These results suggest that the total number of objects that can be held in memory is not determined by the complexity of the objects. Thus, poorer behavioral performance for complex objects may be the result of limited resolution representations that lead to errors in the comparison process at test rather than a reduction in the number of items that can be maintained.