Inhibition of task set: Converging evidence

domains depends critically on the performer’s ability to select and perform the proper task while ignoring other potential tasks. One popular approach to studying this ability, called cognitive control, is to have people repeat or switch tasks. In this paradigm, performance is typically slower following a task switch than following a task repetition (the switch cost; see, e.g., Allport, Styles, & Hsieh, 1994). Switch costs are robust, occurring even when people have ample time to prepare for the switch (e.g., Lien, Ruthruff, Remington, & Johnston, 2005; Meiran, 1996; Rogers & Monsell, 1995; Ruthruff, Remington, & Johnston, 2001). These residual switch costs suggest that the cognitive processes responsible for preparing a new task set are limited in their effectiveness. Recent studies have suggested that preparatory processes might include inhibiting the previous task set, in order to increase the relative accessibility of the new task (e.g., Mayr & Keele, 2000). To support this claim, Mayr and Keele examined how performance varied as a function of how recently the task was performed. They used an explicit cuing paradigm, where a visual cue indicated what task should be performed next. Critically, they found slower responses for the task performed two trials earlier (n22 repetition; e.g., ABA) than for less recent tasks (n22 switch; e.g., CBA). This n22 repetition cost, according to Mayr and Keele, reflects the need to overcome lingering task-set inhibition. They assumed that the inhibition is applied when switching away from a task and then decays over time. The bulk of the evidence for task-set inhibition comes from performance delays in ABA task sequences relative to CBA sequences (for a different approach, see Hübner, Dreisbach, Haider, & Kluwe, 2003; Kuhns, Lien, & Ruthruff, 2007). Such delays might reflect inhibition but could also reflect other factors (e.g., negative priming). Furthermore, most studies favoring task-set inhibition have used an explicit cue (e.g., a visual task label in Mayr & Keele, 2000; a symbolic cue in Schuch & Koch, 2003) in order to indicate the task to be performed next. Of the rare studies that looked for inhibition outside this paradigm, a few actually failed to find it (e.g., with tasks cued by location or by task sequence; Arbuthnott, 2005; Lien, Ruthruff, & Kuhns, 2006). It is conceivable that the processing delays in an ABA sequence reflect confusion over the fact that the cue seems very familiar, yet the task switches (but see Altmann, 2007). Thus, it is important to provide converging evidence for task-set inhibition and to show that it is a general phenomenon, not restricted to a particular paradigm. There is no obvious reason why inhibition should not occur in the absence of explicit task cues, as in many real-world scenarios in which one voluntarily chooses what to do next. Recently, Arrington and Logan (2004, 2005) developed a voluntary task-switching procedure, in which partici-