Abstract
Perceptual Load theory’s chief hypothesis proposes that easy tasks (low-load) curb efficient selection of targets, whereas difficult tasks (high-load) allow for efficient selection. In its flagship experiment (Lavie, 1995) low-load displays comprised a single-letter target and a single-letter distractor. Low-load trials (defined by fast RTs) elicited poor target selection. In high-load blocks, targets were embedded within a string of six letters. The task became difficult and selection improved. We propose that load has been confounded with target-distractor discriminability: in high-load blocks targets could be easily distinguished from distractors because the target was always within a large string of letters, whereas the distractor was always a single letter. Such discriminability was absent in low-load blocks, where targets and distractors shared the same size. Experiment 1 (Figure 1) comprised a low-load replication compared to four low-load conditions where distractors could be discriminated by four different patterns; a huge distractor, a 5-letters distractor, a different-color distractor and a distractor whose position relative to the target was fixed along each block. Load effects were reversed in all non-replicant conditions: selection drastically improved while load levels dropped even lower than the replication (Figure 2). In experiment 2, a high-load replication was compared to a high-load condition where targets and distractors shared the same size. Load predictions were reversed: the non-replicant condition resulted in deteriorated selection while load was higher than the replication. Experiment 3 comprised medium-load and low-load conditions. Selection was modulated by discriminability levels, irrespective of load levels. Contradicting the perceptual load hypothesis, nine independent conditions converged into a robust positive correlation between load and distractors' interferences (Figure 3). By manipulating discriminability, the correlation between load and efficiency of selection was propelled and reversed. Therefore, previously observed relationships between task difficulty and selection efficiency do not reflect an inherent characteristic of the perceptual system.
Meeting abstract presented at VSS 2012