Abstract
Soto and colleagues (Soto, Heinke, Humphreys and Blanco, 2005; Soto & Humphreys, 2007) claim that people reflexively and involuntarily fixate objects which match the contents of working memory (WM). To investigate this issue, we ran two experiments to test the limits of this hypothesis. The first experiment was a variation on a card trick; we used eye movements to attempt to identify a subject's card. Subjects picked three cards to remember and then viewed multiple displays, each consisting of three cards. When a memory card appeared in a display, subjects' first eye movements went directly to the card on 2/3 of the trials, providing evidence for an early bias towards items held in WM. However, this bias interacted with, rather than supplanted, a long-term bias toward viewing items from left to right; people made a majority of first eye movements to the left, even when the memory card appeared on the right. Results suggest that the WM effect interacts with rather than supersedes long standing biases. A second experiment investigated how WM bias and short-term biases interact. In this experiment, people held objects in memory while performing an anti-saccade task. We manipulated whether or not the signal to generate the anti-saccade matched the contents of WM. In aggregate, there was no difference in the number of pro-saccade errors between these conditions, suggesting that short-term biases due to explicit task demands supersede the bias to look towards items in WM. The WM bias, however, was evident within a subset of our subjects who performed poorly, in general, at the anti-saccade task. These results are consistent with the view that attention is biased towards items in WM; however, this bias is not completely automatic and reflexive. It interacts with long-term biases associated with visual routines and short-term biases associated with task set.