Abstract

When people are required to switch from performing one task to another, response times are typically slower than when they are required to repeat the same task twice in succession. This switch cost remains even when the need to switch is predictable or advanced preparation is possible. The mechanisms underlying such switch costs were the focus of this research. Rogers and Monsell (1995), using the "alternating runs" paradigm which they devised, provided evidence favoring the idea that switch costs are due to mental preparation for the task being switched to , that is, to the task set reconfiguation required by the upcoming task. Wylie & Allport (2000, Experiment 1), using the alternating runs paradigm embedded in a special three stage research design, provided evidence favoring the idea that switch costs are due to the carry over of inhibition present in the task being switched away from , that is, to a form of task set inertia stemming from the trial just completed. These two sets of researchers drew different conclusions as to what underlies switch costs; it is possible, however, that their different outcomes were due to the fact that they used very different stimuli and tasks. In particular, Rogers and Monsell's task set reconfiguration result was obtained when the two tasks involved were very similar to one another in terms of overall difficulty and how automatically they could be executed. On the other hand, Allport and Wylie's task set inertia result was obtained when the two tasks involved were asymmetrical in terms of overall difficulty, and when one task could be executed more automatically than the other. The present research investigated whether such differences underlie these different outcomes. Experiment 1 used the three stage research design advocated by Allport and Wylie for differentiating whether switch costs are due to what one is switching from versus switching to , but with stimuli and tasks of the type used by Rogers and Monsell. Unlike what Allport and Wylie would have predicted, the results supported Rogers and Monsell's task set reconfiguration hypothesis (what matters is what one is switching to ). Experiment 2 repeated the previous experiment with one task being made more difficult than the other by involving an episodic rather than a semantic judgment. Again, support was found for the task set reconfiguration hypothesis. Experiment 3 again repeated the main design of the first experiment, but one task involved more automatic processing than the other. This time, support was found for the task set inertia hypothesis (what matters is what one is switching from ). The results are discussed in terms of how the complex processing attentional and memory demands of switching between tasks can determine whether task set reconfiguration or carryover of inhibition plays a significant role in determining switch costs.