Elsevier

Acta Psychologica

Volume 141, Issue 2, October 2012, Pages 214-221
Acta Psychologica

Effects of deviant trials on precue-based versus memory-based switching among two or four tasks

https://doi.org/10.1016/j.actpsy.2012.08.002Get rights and content

Abstract

In two experiments, the effects of an introduction of deviant trials on precue-based and memory-based task switching were investigated. Deviant trials were trials that deviated from task foreknowledge as induced either by precues or memory of the task sequence. The experiments differed with respect to the number of tasks to be switched among (two in Experiment 1, four in Experiment 2). In Experiment 1, the introduction of deviant trials had a detrimental effect on precue-based task preparation, but had little impact on memory-based preparation. The latter observation was due to the fact that there was little evidence for memory-based preparation at the outset. In Experiment 2, introducing deviant trials had little effect on either precue-based or memory-based preparation. In contrast to Experiment 1, in Experiment 2 memory-based preparation was as effective as precue-based preparation.

Highlights

► Efficiency of memory vs. precues based task switching depends on the number of tasks. ► The effect of deviant trials on task switching depends on the number of tasks. ► Findings from two-task switching cannot be generalized to situations with more tasks.

Introduction

One of the main controversies in the task switching literature (for reviews, cf. Kiesel et al., 2010, Vandierendonck et al., 2010) relates to the question inasmuch preparation effects in terms of a reduction of switch costs that can be observed with sufficiently long preparation intervals are attributable to endogenous preparation (cf. Altmann, 2004, Meiran, 1996, Monsell and Mizon, 2006), or can be reduced to priming effects induced by the presentation of task cues (cf. Allport and Wylie, 2000, Logan and Bundesen, 2003). A substantial body of evidence in favour of the latter possibility consists of observations of virtually absent reductions of switch cost by preparation in cases in which participants switch tasks in a purely memory-based fashion without the aid of task cues, whereas switch costs are reliably reduced when task cues are presented in advance of the imperative stimulus (e.g., Abegg et al., 2011, Barton, Greenzang, et., 2006, Barton, Kuzin, et al., 2006, Koch, 2003, Sohn and Carlson, 2000). However, the evidence is not unequivocal in this respect. For example, Kleinsorge and Gajewski (2008) reported clear-cut evidence of endogenous preparation in the absence of external precues when participants switched among four tasks and task foreknowledge was not entirely reliable because in some cases the actual task differed from the one that was expected on the basis of memory or precues (deviant trials). However, switching among more than two tasks as well as unreliable foreknowledge represent two significant deviations from ordinary task switching procedures, the effects of which cannot be disentangled on the basis of their study. In the meantime, Kleinsorge and Apitzsch (2012) established that switching among four rather than two tasks makes a crucial difference with respect to the efficiency of purely memory-based task preparation. When switching among only two tasks, their participants generally exhibited much more efficient task preparation with the aid of precues compared to switching in a purely memory-based fashion, which is in line with most of the literature. However, this superiority of precue-based preparation was absent when participants switched among four tasks.

Kleinsorge and Apitzsch (2012) suggested that the interaction of the number of tasks with the source of task foreknowledge (memory versus precues) was due to differences in the underlying task representations. Specifically, they argued that switching among more than two tasks requires more elaborated task codes in order to determine the currently relevant task than switching among only two tasks. This is because reducing the number of potentially relevant tasks to two affords the implementation of strong antagonistic constraints among the two tasks in that any evidence against one of the tasks necessarily provides evidence for the only alternative task. However, with four tasks evidence against one of the tasks does not automatically favour one of the three remaining tasks, calling for the establishment of a more elaborated representation of the set of tasks. Because memory-based performance in general is aided by more elaborated and distinctive memory representations, the relative advantage of memory-based compared to precue-based task switching with four rather than two tasks may follow quite naturally from the need to adopt a more elaborated task representation in order to be able to individuate the four tasks.

Whereas memory-based preparation may be more suitable for switching among four tasks due to the need to establish a more elaborated task representation, precue-based preparation may be more suitable for switching among two tasks because of the antagonistic constraints inherent in a two-task environment. With only two tasks, all one needs to know is whether the currently relevant task is the same as in the preceding trial or not. With only one cue per task, this may reduce to a simple same-different judgement on the level of the cues, but even with more than one cue per task cue disambiguation seems to proceed on a level that is requires only relatively superficial processing that is functionally different from the level on which the actual retrieval of task rules takes place (cf. Kleinsorge, 2012). Thus, encoding the currently relevant task on a relatively superficial level may be sufficient to initiate a retrieval process that exploits the antagonistic constraints imposed by a selection among only two tasks.

The present study aimed at elucidating the role of unreliable foreknowledge with respect to the relative efficiency of precue-based and memory-based task switching. Based on the aforementioned considerations, this was done in the context of either two or four tasks. Therefore, in Experiment 1, participants switched among two tasks, whereas they switched among four tasks in Experiment 2.

Effects of deviant trials should be intimately linked to the intensity of task preparation. The cost induced by a deviant trial should be larger the more prepared the originally expected task was. Therefore, one may expect that the introduction of deviant trials should strongly discourage task preparation. However, as argued by Kleinsorge and Gajewski (2008), this discouragement of task preparation may be less pronounced with foreknowledge based on memory than with foreknowledge based on precues. This is because foreknowledge based on memory encompasses more than single trials (i.e., a whole sequence of tasks) and therefore is not completely invalidated in case of a deviant trial. In contrast, precues only provide information regarding the upcoming trial and which is completely invalidated in case of a deviant trial. Thus, one may expect that an introduction of deviant trials compromises precue-based preparation more than memory-based preparation, provided that this preparation takes place at all. As already mentioned, there is a substantial body of evidence suggesting that memory-based preparation is lacking with two tasks. However, as argued in the preceding sections, switching among four tasks may represent a situation that is more favourable for supporting memory-based preparation because of the inherent requirement to establish a more elaborated task representation.

Any attempt to contrast precue-based with memory-based task switching is complicated by the following problem. If the same stimulus material is used for the two conditions, one cannot exclude that precue-based performance is affected by implicit or even explicit knowledge of the sequence of tasks that is to be memorized in the memory-based condition. Alternatively, one may use unpredictable task sequences in precue-based conditions, but this confounds the source of foreknowledge with the type of task sequences. The use of identical task sequences is probably less problematic when the effect of interest consists of a (relative) advantage of explicit knowledge compared to precue-based performance because an effect of implicit or explicit knowledge on the latter condition should diminish differences between conditions. Therefore, using the same predictable sequences in both conditions represents a conservative approach in this respect. More problematic are eventual advantages or, in case of deviant trials, disadvantages of precue-based compared to memory-based preparation because it is hard to disentangle whether these advantages would also accrue without the additional influence of implicit or explicit knowledge. With regard to the interaction of the source of foreknowledge and the number of tasks reported by Kleinsorge and Apitzsch (2012), an eventual benefit of precue-based preparation due to implicit or explicit knowledge may have underestimated the level of memory-based preparation with four tasks. In general, an unintended influence of knowledge of the task sequence on precue-based performance should reduce differences between memory-based and precue-based performance.

Section snippets

Experiment 1

Participants were assigned to one of two groups. Both groups performed the same tasks that were arranged into regular sequences of the type AABBB or AAABB.1 Only one group of participants (the Explicit Knowledge Group) was informed about this regularity.

The experiment was subdivided into three phases. The first phase served to familiarize participants

Experiment 2

Experiment 2 replicated the procedure of Experiment 1 with four tasks. In light of the results of Kleinsorge and Apitzsch (2012), it was expected to observe clear-cut preparation effects with the Explicit Knowledge Group in Phase 2. An open question was whether this memory-based preparation would be enhanced (or modulated otherwise) by the introduction of deviant trials in Phase 3. In addition, it was asked whether the breakdown of precue-based preparation due to deviant trials that was

General discussion

Processes of task-specific preparation are at the heart of task-switching research, and the question inasmuch preparation can be based on purely endogenous processes that operate without the aid of external stimuli has elicited considerable debate (cf. Barton, Kuzin et al., 2006, for a review). Substantial evidence suggests that the scope of endogenous preparation is rather limited, but this evidence stems almost exclusively from task-switching studies with only two tasks to switch among. The

Acknowledgments

Thanks to Marina Kandelberg and Felix Meyer for assistance in programming and running the experiments. The research reported in this article was supported by grant KL 1205/6-1 of the Deutsche Forschungsgemeinschaft.

References (20)

  • M. Abegg et al.

    Knowing the future: Partial foreknowledge effects on the programming of prosaccades and antisaccades

    Vision Research

    (2011)
  • J.J.S. Barton et al.

    The use of working memory for task prediction: What benefits accrue from different types of foreknowledge?

    Neuroscience

    (2006)
  • A. Allport et al.

    'Task switching', stimulus-response bindings and negative priming

  • E.M. Altmann

    Advance preparation in task switching: What work is being done?

    Psychological Science

    (2004)
  • J.J.S. Barton et al.

    Switching, plasticity, and prediction in a saccadic task-switch paradigm

    Experimental Brain Research

    (2006)
  • M. Gade et al.

    Cue-task associations in task switching

    The Quarterly Journal of Experimental Psychology

    (2007)
  • A. Kiesel et al.

    Control and interference in task switching—A review

    Psychological Bulletin

    (2010)
  • T. Kleinsorge

    Task switching with a 2:1 cue-to-task mapping: Separating cue disambiguation from task-rule retrieval

    Psychological Research

    (2012)
  • T. Kleinsorge et al.

    Task preparation based on precues versus memory: Precues lead to superior performance with two tasks but not with four tasks

    Journal of Cognitive Psychology

    (2012)
  • T. Kleinsorge et al.

    Task switching based on externally presented versus internally generated information

    Psychological Research

    (2008)
There are more references available in the full text version of this article.
View full text