Abstract
Visual working memory has been implicated in several models of visual search performance, and of visual attention in general. Recent studies have investigated the involvement of working memory in search by introducing spatial and non-spatial working memory loads during visual search (Woodman, Vogel, & Luck, 2001; Oh & Kim, 2004; Woodman & Luck, 2004), demonstrating that although both types of working memory load increase overall reaction times (RTs), only a spatial working memory load interferes with search efficiency. In the present work, we set out to replicate these studies using eye-tracking to further resolve where and how spatial and non-spatial working memory loads impact search performance. For both spatial and non-spatial working memory loads, we find an increase in the first saccade latency, likely to accommodate encoding, as well as an increase in the time between fixating the target and making a response (target identification time), which may reflect either a disruption in the identification process itself or interference from a preparation for the memory test. Excluding first saccade latency and target identification time, the remaining time in search RTs is not impacted by spatial load, but incurs an additional cost under non-spatial load that is constant across set size. This result, in addition to an increase in both memory errors and overall RT intercepts under non-spatial load as compared to spatial load suggests that joint interference between search and memory tasks may in fact be greater for non-spatial than for spatial loads.
This work was supported by NSERC with a grant to DS and a graduate scholarship to GJFS.