Abstract
When people “tune” attention for a pre-specified feature (i.e., when they establish an “attentional set”), distractors containing that feature become particularly strong elicitors of reflexive attention shifts, whereas distractors that typically capture attention reflexively - but which do not match attentional set - lose their power to do so. This is referred to as the “contingent capture” of attention (Folk et al., 1992). Evidence suggests that working memory (WM) is critical for maintaining target representations in mind (Desimone & Duncan, 1995), and that WM contents may help guide attention (Pashler & Shiu, 1999). However, the dependence of contingent capture on WM availability has not been formally tested. One possibility, given that WM load can increase stimulus-driven capture (Lavie & de Fockert, 2005), is that such load will diminish the influence of attentional set. In three experiments, we directly tested this hypothesis: participants performed a typical contingent attentional capture task while under high or low WM load. In Experiment 1, WM was occupied by a concurrent digit string task, previously shown to increase stimulus-driven capture (Lavie & de Fockert, 2005). In Experiment 2, visual WM was taxed through a concurrent change detection task involving 1 (low load) or 3 (high load) abstract shapes. Finally, because some participants reported using verbal strategies to encode these shapes, in Experiment 3 participants performed a change detection task on a 5-X-5 grid containing 4 filled-in grid locations. Under low load, the initial grid always contained the same symmetrical filled-in locations; under high load the filled-in locations were randomized (participants reported using no verbal strategies during this manipulation). In all experiments, contingent capture remained robust regardless of WM load. These results suggest that once attentional set is configured, target properties may not need to be maintained in WM in order to affect attentional capture.