Abstract:

Visual working memory (VWM)is a temporary buffer that maintains a limited set of items (i.e., 3-4 objects) in an “online” state. Luck and Vogel (1997) demonstrated that the capacity was limited by the number of the integrated objects, but not by the number of the individual features. However, with the conjunction of homogenous features served as memory materials, follow-up studies showed that the subjects could retain fewer items than single-feature objects. Some researchers considered that the lower accuracy for conjunction of homogenous features may result from the difference of precision requirement, which was manipulated by the similarity between sample and test array. However, some experiment results showed that higher precision could only affect the accuracy of change detection task, but not the capacity. In present study, we carried out two experiments using change detection paradigm to examine the storage mechanism of multi-feature objects in VWM. In experiment 1, multi-feature and single-feature objects were defined by double-color objects and single-color objects, respectively. To manipulate the precision requirement, we further divide double-color objects into two conditions, that is, a single-change condition in which only one of the two colors changed (high-precision), and a double-change condition in which both colors changed (low-precision). In experiment 2, we focused on the single-change condition, and investigated the units of storage in VWM by manipulating weather there was an arrow indicating the changed color in the test phase. The result would further explain whether the storage mechanism of multi-feature objects is in an integrated manner or in an individual features manner. In experiment 1, we first compared the results of double-color objects and single-color objects to assess the influence of the feature number on the allocation of VWM capacity. Behavioral results showed that double-color objects were responded less accurately than single-color objects. The ERPs results showed that the amplitude of CDA for double-color objects was significantly higher than for single-color objects. Second, we compared the double-change condition and single-change condition to explore the effect of precision requirement. Lower accuracy was found for the single-change condition. CDA amplitudes did not differ between the two conditions. Instead, the amplitude of LPC for the single-change condition was more positive than for the double-change condition. In experiment 2, the accuracy was not modulated by weather there was an arrow or not. These results reveal that the number of features, but not the precision requirement, has an effect on the allocation of VWM capacity. And the units that limit our VWM capacity are the integrated objects instead of the individual features.

Key words: visual working memory, capacity resource allocation, CDA, LPC, storage mechanism