Automatic and intentional activation of task sets

Four experiments examined automatic and intentional activation of task sets in a switching paradigm. Experiment 1 demonstrated incidental task sequenc e learning that was not accompanied by verbalizable task sequence knowledge . This learning did not affect task shift cost and may be attributed to aut omatic task-set activation. In Experiment 2, both shift cost and learning e ffect increased when the response-cue interval was short indicating the inf luence of residual, persisting activation of the preceding task set. In Exp eriment 3, learning disappeared with a long cue-stimulus interval (CSI), wh ich resulted in a strong preparation effect. This preparation, however, red uced reaction time level but was not specific to task shifts. Finally, Expe riment 4 showed that a within-subject CSI variation also leads to reduced s hift costs. Together, the data suggest an activational account of task prep aration and may have relevant implications for inhibitory accounts.