Residual switch costs are notoriously difficult to eliminate. Yet Hunt and Klein (2002) eliminated them in a task that required observers to alternate between 8 trials of prosaccades and 8 trials of antisaccades, as long as there was at least 1 sec between the task cue and the onset of the saccade target. It was proposed that the elimination of residual switch costs occurred because prosaccade responses are computed very rapidly. These so-called hypercompatible responses bypass memory retrieval stages of the response process, thereby eliminating the source of residual switch costs. Here we tested this hypothesis by requiring observers to alternate between responding with the finger that was vibrated (another task that meets the criteria for hypercompatibility) and responding with the finger of the opposite hand. Residual switch costs were not eliminated, suggesting that their elimination in Hunt and Klein (2002) was due to special properties of the prosaccade-antisaccade task.
Performance involving two alternating tasks is significantly worse than performance involving a single task (see, e.g., Allport, Styles, & Hsieh, 1994). The cost of switching between tasks can be reduced as a function of preparation time, but the switch cost is not completely eliminated, even when ample time to prepare for the impending change in task is provided (Meiran, 1996; Rogers & Monsell, 1995). Until recently, these "residual switch costs" were considered to be such a fundamental component of task switching that they could not be eliminated (see, e.g., Pashler et al., 2000). According to the widely accepted view, reconfiguration is complete only with the appearance of the target stimulus for the new task (Rogers & Monsell, 1995). There are, therefore, at least two components of the task-switching process: One is voluntary control of task set reconfiguration, and the other, which gives rise to residual cost, is exogenously driven by the arrival of the imperative stimulus.
Recently, however, we challenged the conventional idea that the residual switch cost is a necessary component of executive control in task switching (Hunt & Klein, 2002). In this study, participants switched between a prosaccade task (a spatially compatible eye movement) and an antisaccade task (an incompatible eye movement). Residual switch costs were eliminated in both reaction time (RT) and accuracy. In Hunt and Klein (2002), we suggested that four conditions contributed to the elimination of switch costs: (1) a sufficiently long preparatory interval; (2) a predictable target onset time; (3) an easily interpreted task cue; and, (4) most importantly, application of the simplicity, or reflexive nature, of the saccadic response system.
The task applied in our 2002 study was critically different from most other laboratory tasks used in task switching. Conventional tasks, such as the Stroop color-word task, require of the subject additional processing in order to translate the relevant attribute of the stimulus into the arbitrary response according to the experimenter's instructions. In Hunt and Klein (2002), the task was characterized by a hypercompatible stimulus-response (S-R) mapping, for which "the relationship between the stimulus and the response is sufficiently strong to require very little attention or memory to perform" (pp. 537-538). If the residual switch cost were associated with interference from memory retrieval processes, and thus constituted an additional process before response execution (Logan, 1988; Meiran, 2000; Wylie & Allport, 2000), hypercompatible S-R mappings, dissociated from a memory retrieval stage, would have an advantage, resulting in the elimination of residual costs.
The purpose of the present research was to examine the idea that the use of hypercompatible mappings is responsible for eliminating residual costs. We tested this idea using another hypercompatible task-namely, a tactile-manual …