Abstract
In many everyday situations, we bias our perception using task-dependent information. To explore the cost involved in shifting top-down attention to a new task, we adopted a task-switching paradigm, in which ‘switch’ and ‘repeat’ trials in mixed task blocks are contrasted with single task blocks. We use two visual tasks in our paradigm: object detection in cluttered gray-level natural scenes (‘animal’ vs. ‘non-animal’ and ‘vehicle’ vs. ‘non-vehicle’); and discriminating the color of the frame enclosing these images (‘orange’ vs. ‘purple’ or ‘blue’, and ‘blue’ vs. ‘orange’ or ‘purple’). We distinguished switch costs with (e.g. switching from detecting orange among purple/blue distracters to detecting animals in natural scenes) and without top-down attention shifts. We found significant switch costs in reaction time of 20ms for switching from a color task to an object detection task (p<0.05), and of 28ms for switching from an object detection task to a color task (p<0.0001). There are no significant switch costs for switches within a stimulus attribute, when no top-down attention shift is required. We conclude that deploying top-down attention to a different attribute incurs a significant cost in reaction time, but that biasing to a different feature value within the same stimulus attribute does not. ANOVAs of mixing and switch cost show significant (p<10−8) differences among individual subjects for mixing cost, but no such effect for switch cost: shifting top-down attention has a fixed processing duration among individuals.