Intention deactivation: effects of prospective memory task similarity on aftereffects of completed intentions

For exceptional cases of aftereffects in conditions of changing ongoing tasks, however exclusively in older but not younger adults, see Scullin et al. (2012, 2011).

Besides commission errors, we analyzed ongoing-task errors on PM_REPEATED and oddball trials, which include misses (i.e., RTs > 3000 ms) as well as erroneous digit categorizations. Due to the complexity of the overall results and to allow readability, we refrained from including these errors in the manuscript. However, analyses of ongoing-task errors could be found in the online Supplemental Materials.

Note that we will refrain from the distinction between commission errors_PREVIOUS and commission errors_CURRENT in Experiments 3 and 4, because the designs do not allow to unambiguously separate between these error types.

Over the course of the repeated cycles of the experiment, participants might have noticed the appearance of PM-block targets in subsequent test blocks and thus adopted strategies in shielding PM_REPEATED and oddball trials. This might have been especially true in the PM-cue-category-switch condition in which symbols could be ignored because they were completely irrelevant for the current PM_WORD task. To disregard a strategic explanation due to repeated cycles we aimed at demonstrating that the difference between the PM-cue-category-similarity conditions is observable already in the first cycle (without repetition) and does not change over the course of the experiment. We compared RT and commission error aftereffects in the very first cycle between participants starting with the PM-cue-category repetition condition (n = 13, M = 178 ms, M = 11.5 %) and those starting with the PM-cue-category switch condition (n = 13, M = 81 ms, M = 0 %), which descriptively mirrored the overall analyses. Not surprisingly, and due to the immense loss in statistical power, the corresponding t tests on RTs, t(24) = 0.967, p = 0.343, d = 0.39, and commission errors, t(24) = 1.90, p = 0.074, d = 0.77, did not reach significance. In addition, analyses for the first session revealed that differences between aftereffects did not vary as a function of experiment half between conditions, both for RTs, F(1, 25) = 1.90, p = 0.180, nor commission errors, F(1, 25) = 2.56, p = 0.121. Similar analyses confirmed this pattern for the other experiments as well.

A further experiment (N = 25) conducted in our lab allowed to test for the modality dependence of this effect (Experiment 2B). Note that an erroneous activation of the current PM response was observed for response switches within different manual responses (Experiment 1) and for response switches from verbal to manual responses (Experiment 2). Experiment 2B included a switch from a manual to a verbal PM task. In this setting, the PM_REPEATED trials did not erroneously trigger the currently relevant verbal PM response. We interpret this finding as increased control over response activation for verbal than manual responses. Therefore, before erroneous verbal responses could be executed, participants might have already performed the relatively fast manual ongoing-task response (see online Supplemental Materials for methods and results of Experiment 2B).

Note that whereas the terms focal/non-focal are typically used to refer to the processing requirements of PM cues relative to the ongoing task (i.e., focal = PM cue is automatically processed during the ongoing task; non-focal = PM cue identification requires additional processing), here the terms refer to the question whether performing the PM task in the test block automatically leads to processing (focal) of PM_REPEATED and oddball trials or not (non-focal).

We have no explanation for the unexpectedly high oddball RT. As such deviations affect the size of the aftereffect, one might wonder whether oddballs as baseline are justified at all. Although we believe that oddballs represent the best possible baseline, we re-analyzed all modulations of aftereffects when taking standard trial RTs as baseline. In all experiments results remained identical and thus support our preferred interpretation of the data pattern of this study.