Is attention really biased toward the last target location in visual search? Attention, response rules, distractors,...

The visual search and target-target cueing literatures have reached opposite conclusions about whether a shift of attention is biased toward or away from, respectively, previously attended target locations. In this article, we aimed to figure out why. The main differences between the two experimental approaches concern (1) the stimulus-response translation rules ("what" identification keypresses vs. "where" localization responses), (2) the amount of attention required in order to identify the target, and (3) distractor presence or absence. Experiment 1 tested the role of stimulus-response translation rules by requiring both an eye movement "where" response and a keypress "what" response to each target, in a typical search paradigm. Eye movements showed a bias away from the vicinity of the previous target, whereas keypresses showed a bias toward the previous target location, but only when the keypress response repeated. Experiment 2 removed the keypress identification requirement, to test whether reducing the amount of attention to the target would alter the eye movement bias; it did not. Experiment 3 removed the distractors, to test whether eliminating the potential for distractor location effects would alter the eye movement bias; it did, by accentuating the eye movement bias against the last target location. Collectively, the findings revealed that different stimulus-response translation rules and distractor-processing requirements are the main reasons for the discrepancy, while demonstrating that shifts of attention intrinsically tend away from prior target locations. The findings are generally consistent with episodic-retrieval and inhibited spatial-reorienting theories.