Interference from a spatially incompatible salient distractor on target location probability learning

Locations that frequently contain a search target are prioritized, demonstrating experience-guided attention. Yet spatial attention is also guided by stimulus salience and task goals. Can people acquire location probability learning when they first attend to a salient distractor at a different location? Using a T-among-L search task, we asked participants to find a T target among L distractors. Unbeknownst to them, the target T’s location was biased toward one “T-rich” quadrant in the training phase and was unbiased in a testing phase. A salient distractor – a C-shape in a distinct color – more frequently appeared in a different quadrant, the “C-rich” quadrant. When the C-shape was task-irrelevant (Experiment 1), participants successfully ignored it, acquiring the target’s location probability learning that lasted through the unbiased testing phase. This finding shows that the target’s location probability learning is robust in the face of a spatially incompatible but task-irrelevant salient distractor. To test whether location probability learning was disrupted by a spatially incompatible task goal, in Experiment 2 participants have to first saccade to the C-shape before identifying the T. This requirement did not disrupt learning: participants were faster finding the T in both the T-rich and the C-rich quadrants, relative to the other two quadrants. Removing the C-shape from the testing phase revealed a persisting spatial bias toward the T-rich quadrant, without a spatial bias toward the C-rich quadrant. Thus, the target’s location probability learning is robust in the face of a spatially incompatible and task-relevant distractor, demonstrating independence between experience-guided attention and goal-driven attention. In addition, the lack of a persisting bias toward the C-rich quadrant showed that frequently shifting attention to a region was insufficient to induce a durable attentional bias. Location probability learning may depend instead on reinforcement learning from successful target detection.