Sequential Task Predictability in Task Switching
Koch, Iring, Psychonomic Bulletin & Review
Many studies of task switching have found that a prolonged preparation time reduces switch costs. An alternative manipulation of task preparation is based on sequential task predictability, rather than preparation time. In Experiments 1 and 2 of the present study, participants performed explicitly instructed task sequences (i.e., AABB) and were then transferred to a random sequence. The observed benefit of predictability-based task preparation was not switch specific. In Experiment 3, the participants changed from random to predictable tasks. The observed predictability benefit again was not switch specific. The data thus suggest that task switching does not necessarily require a switch-specific reconfiguration process. Rather, task-specific control processes may be needed in both task switches and repetitions.
A vital question in cognitive psychology concerns the mechanisms of action control for changing task requirements. This control function has been ascribed to the configuration of cognitive processes, also termed task set. Recently, the dynamics of task set have been studied using task switching. In task-switching experiments, performance when participants switch between tasks is compared with that when the task is repeated. Typically, reaction time (RT) is longer with switches than with repetitions, indicating switch costs (see Monsell, 2003, for a review).
The goal of the present study was to explore task preparation processes in task switching. Many studies have shown that having a long preparation interval reduced RTs in task switches more than in repetitions, thus reducing switch costs. This effect of preparation on switch costs has been attributed to a task set reconfiguration process (e.g., Meiran, 1996; Rogers & Monsell, 1995). According to reconfiguration accounts, reconfiguration is required only in a task switch, but not in a task repetition, so that preparation benefits should be predicted to have a substantial switch-specific component.
However, the results of several studies of task preparation do not conform to this prediction. In some of these studies, preparation has been manipulated by varying task predictability, using an incidental sequence-learning methodology, instead of providing either a long or a short preparation time (Gotler, Meiran, & Tzelgov, 2003; Heuer, Schmidtke, & Kleinsorge, 2001; Koch, 2001). For instance, in Gotler et al.'s study, participants performed an eight-trial sequence of two different tasks. The participants were not informed about the predictable sequence, and each task was indicated by an instruction cue. Gotler et al. found an increased RT when participants were transferred to random tasks, indicating that the predictable sequence, at least partially, had been learned. Importantly, as in previous studies (Heuer et al., 2001; Koch, 2001), the preparation benefit was the same for switches and repetitions, so that it did not affect the size of the switch costs, unlike manipulations of preparation time.
Furthermore, other studies in which quite different experimental manipulations have been used to induce task predictability have reported the same surprising equivalence of the predictability benefit for task switches and repetitions. For example, Sohn and Carlson (2000) had participants perform trials consisting of pairs of tasks. Each task was unambiguously cued by the color of the bivalent stimuli, but for half of the participants it was also the case that the task transition was predictable. Predictability produced a large performance benefit relative to unpredictable tasks, and this benefit was equal for switches and repetitions. Likewise, Ruthruff, Remington, and Johnston (2001) induced task expectancy by presenting univalent stimuli, which unambiguously cue only one task, in a predictable sequence and occasionally introducing violations of the sequence. These violations were equally harmful for switches and repetitions. That is, these two studies reported the same data pattern as incidental task sequence learning studies (see also Dreisbach, Haider, & Kluwe, 2002). …