Content area
Abstract
Individuals seem unable to avoid a performance decrement when switching between simple perceptual identification tasks, compared to when repeating a task. Even though the participant may fully understand both tasks and performance on either task alone may be fast and accurate, there is a cost for switching between them. Although some researchers have been able to reduce these costs, none have been able to eliminate them. The experiments conducted for this dissertation were designed to identify factors that might affect these task switching costs and to provide a coherent explanation for the underlying cost mechanisms. All experiments used color (red/green) and shape (circle/triangle) identification tasks.
Experiments 1 and 2 revealed that switching costs are not memory load effects; stimulus overlap leads to higher switching costs; and task predictability allows participants to avoid potential costs due to changing stimuli. Experiment 3 showed that providing cues about the upcoming trial affects the overall difficulty of the trial but does not affect switching costs. Experiment 4 found that equalizing residual motor response programming does not reduce switching costs. Experiments 5 and 6 showed that providing meaningful, task-irrelevant information between trials in a blocked format reduces, but does not eliminate, switching costs.
An n-back analysis on all six experiments looked at the effects of the two or three preceding trials on the current trial, revealing a build-up of facilitation over task repetitions, but no build-up of inhibition.
These results suggest that there are two components to switching costs. The first component, disengaging the old task, can be affected by manipulations such as providing task-irrelevant information between trials. Because this information is irrelevant, it should draw participants away from either task and should not help them to engage the new task. The second component, a priming effect, leads to the residual switching costs. The facilitation effect observed in the n-back analyses provides support for this component. The fact that performance continues to improve over the first two repetitions suggests that a priming benefit on repetition trials contributes to the performance differences between repetition and switch trials.
