The dissipating task-repetition benefit in cued task-switching : exploring the role of task-set decay vs.temporal distinctiveness in episodic task retrieval

摘要

In everyday life, we are constantly confronted with changing demands in our environment and constantly happen to perform a multitude of tasks. Current theories describe cognitive control (Logan, 1985; Wood & Grafman, 2003) as the functional underlying mechanism which enables flexible adaptations of behavior. The task-switching paradigm allows to study cognitive control processes when switching between two different tasks with unpredictable order. The basic idea assumes that contrary to well practiced tasks, switching between unpracticed, poorly structured, or unpredictable tasks is linked to mental effort, called “switch costs” (higher reaction times and error rates) (Monsell, 2003). The critical manipulation of the present work concerned the interval between a response and the subsequent cue (i.e., the response-cue interval, RCI). Previous studies have shown that performance in a task repetition is worse after a long RCI compared to a short RCI. This result is traditionally interpreted as follows. Benefits of repeating a task (compared to switching tasks) depend on the activation of task representations (task set) in short-term memory, since this activation passively decays with time unless the task set is maintained (Meiran, Chorev, & Sapir, 2000). Against the background of rather mixed empirical results for theories on performance in task switching (e.g., Altmann, 2005), the work at hand presents the following result. Effects of timing manipulations in task switching, especially in repeating a task, cannot be explained by assuming automatic and passive decay of task sets. The theoretical contribution of the work at hand lies in offering an alternative account for RCI effects on the task-repetition benefit. It is assumed that performance when switching between tasks, among other factors, is influenced by the temporal distinctiveness of episodes (tasks). More specific, the present work proposes that RCI effects in task switching depend on the process of episodic retrieval. This episodic retrieval is modulated by the temporal distinctiveness of episodes, which is defined as the ratio between the current RCI and the preceding RCI. Further, the data imply that episodic retrieval during the RCI is linked to stimulus-related components of the task set. In seven experiments subjects switched between a color- and shape task. The RCI was manipulated in different ways. In Experiments 1 und 2 different ranges of RCIs were used. Increasing RCI decreased the task-repetition benefit, but the slope of the RCI function depended on the range of RCIs rather than on the absolute duration of the RCI. In Experiment 3, the RCIs were blocked or random, and in Experiment 4, trial-wise predictability of RCIs was manipulated. RCI influenced the task-repetition benefit only when RCI changed from the previous to the current trial. Experiment 5 used two cues for each task and dissociated cue-repetition priming from task-repetition priming, suggesting that perceptual cue repetition priming appears to decay, whereas episodic task-set retrieval is influenced by temporal distinctiveness. Experiments 6 und 7 examined which part of the task set (Meiran, 2000) plays a role in episodic retrieval during the RCI. In Experiment 6, response valence (bivalent vs. univalent) was manipulated. The results showed that lengthening the RCI leads to a loss of repetition benefit, mainly when the RCI changed from the previous trial to the current trial, but this was comparable for bivalent (response keys relevant for both tasks) and univalent responses (each response assigned to a separate key). In Experiment 7, stimulus valence was manipulated. The results revealed stronger RCI effects with bivalent stimuli compared to univalent stimuli. Taken together, the data suggest that the influence of RCI in task switching is linked to retrieval of stimulus-related task components (Stimulus-Set) rather than to response-related components (Response-Set). The data of the present work can best be explained by the account of temporal distinctiveness in episodic task retrieval.