SALIENT TASK-IRRELEVANT STIMULI DISRUPT ORTHOGONAL INFORMATION BUT ENHANCE CORRELATED INFORMATION

Garner's speeded classification paradigm is the tool of choice for gauging the effect of irrelevant information on the perception of task information. The paradigm consists of three tasks. In Baseline, task-irrelevant stimulus dimensions are held constant and the participant classifies values on the task-relevant dimension. In Filtering, the participant again classifies values on the relevant dimension but values on the task-irrelevant dimension also vary from trial to trial in a random fashion. Finally, in Correlation the task-irrelevant values vary again but now in correspondence with values of the task-relevant dimension. In a series of experiments with the same stimuli, we manipulated the perceptual salience of the task-irrelevant dimension. The results showed that making the irrelevant stimuli salient impaired task performance in Filtering. However, the same manipulation improved task performance in Correlation. We conclude that attention-grabbing irrelevant information is not always detrimental to performance. Whether or not such information disrupts performance depends on its relationship with the task-information. Garner's speeded classification paradigm is a popular tool in the investigation of selectiv