Elsevier

Consciousness and Cognition

Volume 20, Issue 3, September 2011, Pages 556-567
Consciousness and Cognition

Unconscious activation of task sets

https://doi.org/10.1016/j.concog.2011.02.014Get rights and content

Abstract

Using an explicit task cuing paradigm, we tested whether masked cues can trigger task-set activation, which would suggest that unconsciously presented stimuli can impact cognitive control processes. Based on a critical assessment of previous findings on the priming of task-set activation, we present two experiments with a new method to approach this subject. Instead of using a prime, we varied the visibility of the cue. These cues either directly signaled particular tasks in Experiment 1, or certain task transitions (i.e., task repetitions or switches) in Experiment 2. While both masked task and transition cues affected task choice, only task cues affected the speed of task performance. This observation suggests that task-specific stimulus–response rules can be activated only by masked cues that are uniquely associated with a particular task. Taken together, these results demonstrate that unconsciously presented stimuli have the power to activate corresponding task sets.

Introduction

Even stimuli that we are not aware of can affect our behavior. This is the main message from many studies using the masked priming paradigm (Dehaene et al., 1998, Klotz and Neumann, 1999, Kunde et al., 2003, Kunde et al., 2005, Kunde, 2004, Vorberg et al., 2003; for reviews, see Kiesel et al., 2007, Kouider and Dehaene, 2007). In a typical masked priming paradigm, a visual target stimulus requires a speeded forced-choice response. The target is preceded by a visual prime stimulus that appears only briefly (usually about 20–30 ms) and it is pre- and/or post-masked by other, commonly irrelevant stimuli, or by the target itself, as with metacontrast masking (see Breitmeyer, 1984), which renders the prime essentially invisible. As the unconscious nature of the prime is crucial for interpreting results from masked priming studies, researchers commonly test whether the prime can be consciously reported, often by using a separate signal detection task.

The typical findings in masked priming studies are congruency effects: responding is faster and more accurate if prime and target are assigned to the same response (i.e., are congruent) than if they call for different responses (i.e., are incongruent). For instance, in the study of Dehaene et al. (1998), participants categorized numerals between 1 and 9 as smaller or larger than 5 by pressing one of two keys accordingly. The target number was preceded by a briefly presented and sandwich-like masked prime number. Performance was better if the prime number and the target number fell into the same response category (i.e., if they were both smaller or both larger than 5) than if they did not. This (often replicated) priming effect suggests that response selection can be affected by unconscious stimulus information, presumably by priming associated responses.

Recent studies have asked whether the impact of unconscious stimuli is restricted to activating response tendencies or whether they can also affect cognitive control processes. Cognitive control processes are traditionally conceptualized as strongly related to, and depending on consciousness, in the sense that these processes require and rely on conscious decision-making and awareness (Dehaene and Naccache, 2001, Jack and Shallice, 2001; for an overview, see Hommel, 2007). These “conscious” processes of cognitive control are oftentimes contrasted against “non-conscious” automatic actions. For example, Jack and Shallice emphasize that the underlying processes engaged by conscious action are different from those engaged by automatic action. Similarly, Dehaene and Naccache (2001) claim that, while processing is possible without consciousness, consciousness is required for specific cognitive control processes.

According to this view, masked stimuli that do not reach consciousness should not be able to influence cognitive control processes. In recent years, however, this view was challenged by a number of studies (e.g., Hughes et al., 2009, Lau and Passingham, 2007, Mattler, 2003, Mattler, 2005, Mattler, 2006, Mattler, 2007, van Gaal et al., 2008, van Gaal et al., 2009). These studies provided evidence that masked stimuli can trigger or at least affect cognitive control processes, such as the inhibition of unwanted responses (Hughes et al., 2009, van Gaal et al., 2008, van Gaal et al., 2009), shifting attention (Mattler, 2003, Exp. 3; Scharlau & Ansorge, 2003), and activating task sets (Lau and Passingham, 2007, Mattler, 2003, Exp. 5; 2006, Exp. 3, 2007, Exp. 3). Especially this last observation is surprising from a conscious-control point of view, as the implementation of task sets has been considered to represent one of the most central jobs of cognitive control (e.g., Meyer and Kieras, 1999, Monsell, 1996).

The activation of task sets is commonly investigated by means of the task switching paradigm. In a task switching experiment, participants perform one of two (or more) tasks in each trial (e.g., to categorize a target number as odd or even) but occasionally are to switch to the other task (e.g., to decide whether the target number is smaller or larger than 5; see Kiesel et al., 2006, Sudevan and Taylor, 1987). With the explicit task cuing procedure (Meiran, 1996) that was used in the present experiments, a task cue is presented at the beginning of each trial, informing the participant which task to perform in that trial. The common observation is that response times (RTs) are elevated after a task switch, suggesting that some kind of cognitive control processes, like for example reconfiguration of the cognitive system for processing another task, has to take place (e.g., Rogers & Monsell, 1995). Given that the same stimuli require different responses in the two tasks, participants need to figuratively “rewire” the relevant stimulus–response mappings. It is assumed that when performing different tasks, participants adopt so-called task sets, which, as a basic definition, are a representation of the task and its S–R-mappings. When one of the possible tasks is about to be carried out, the associated task set is activated to enable the participant to perform the task. There is no generally agreed upon definition for the term task set (see, for example, Kiesel et al., 2010, Rogers and Monsell, 1995). Within the given context, an operational definition is sufficient. Whenever participants perform a task, we assume that prior to task execution the corresponding task set was activated. There are theories that assume that a task set is not activated as a whole, but that the activation is composed of several single steps in which different aspects of the task set become activated (e.g., Koch and Allport, 2006, Mayr and Kliegl, 2000, Rogers and Monsell, 1995, Rubinstein et al., 2001). These models of task-set activation will be discussed in more detail later on when we discuss the results of our experiments.

Recently, Mattler, 2003, Mattler, 2005, Mattler, 2006, Mattler, 2007 reported evidence that task sets might be activated by unconscious stimulus information, suggesting that the activation of task sets might not rely on conscious decisions. However, as we will argue, this evidence is not as straightforward as it has been taken to be, so that a reevaluation is in order. In the following section, we justify this claim based on the examination of the central findings from two of Mattler, 2003, Mattler, 2006 studies. As we will show, Mattler’s method is likely to have invited artifacts that render strong conclusions from his observations premature. We then present results from two experiments that introduced a new methodological strategy that helps circumventing these problems.

Section snippets

Priming of task sets or perceptual priming?

To investigate whether masked primes can activate task sets, Mattler (2003, Exp. 5) presented masked task-set primes in a task cuing experiment with random task sequences (Meiran, 1996, Rogers and Monsell, 1995). The target stimuli were high- or low-pitched tones played by either a piano or a marimba. The two tasks required responding to either the pitch of the tone (high vs. low) or to its timbre (piano vs. marimba). Tasks were cued by presenting four stimuli at the corners of an imaginary

Experiment 1

In the first experiment, we presented masked and non-masked task cues on a trial-to-trial basis. Target stimuli were the numbers 1–9, excluding the 5, and participants made manual binary-choice responses to indicate either the magnitude (smaller vs. larger than 5) or the parity of the target number (odd vs. even). The two tasks were signaled by the letters w and b, which in the test language (German) bear no relation to the names of the tasks. In some trials, these task cues were non-masked and

Experiment 2

In standard task switching experiments, each task cue is associated with a specific task and thus tells the participant directly which task to apply. This leads to the possibility of the aforementioned stimulus compound strategy which allows correct responding in task switching experiments without actually switching between the tasks. One method to avoid this problem is to use transition cues instead of task cues. Transition cues are not associated with any particular task but instruct the

General discussion

The aim of the present study was to investigate whether subliminally presented stimuli can activate task sets under experimental conditions that rule out possible artifacts due to prime–cue interactions. Experiment 1 provided preliminary evidence that even masked, and presumably subjectively invisible task cues can trigger task-set activation. However, the design of this experiment could have led participants to make use of the cue–stimulus compound strategies considered by Logan and Bundesen

Acknowledgments

This research was funded through Deutsche Forschungsgemeinschaft Grants KI 1388/1-2 and KU 1964/3-2 awarded to Andrea Kiesel and Wilfried Kunde. We would like to thank Carsten Pohl for his support, and Iring Koch and an anonymous reviewer for their helpful and valuable comments on an earlier version of this article.

References (61)

  • C.M. Arrington et al.

    Voluntary task switching: Chasing the elusive homunculus

    Journal of Experimental Psychology: Learning, Memory, and Cognition

    (2005)
  • G.E. Bodner et al.

    Masked priming of number judgments depends on prime validity and task

    Memory & Cognition

    (2005)
  • G.E. Bodner et al.

    Beyond spreading activation: An influence of relatedness proportion on masked semantic priming

    Psychonomic Bulletin & Review

    (2003)
  • G.E. Bodner et al.

    Masked priming of fixed and free-choice responses

    Experimental Psychology

    (2010)
  • B.G. Breitmeyer

    Visual masking: An integrative approach

    (1984)
  • S. Dehaene et al.

    Imaging unconscious semantic priming

    Nature

    (1998)
  • S.C. Draine et al.

    Replicable unconscious semantic priming

    Journal of Experimental Psychology: General

    (1998)
  • B.U. Forstmann et al.

    Methodological and empirical issues when dissociating cue-related from task-related processes in the explicit task-cuing procedure

    Psychological Research

    (2007)
  • B.U. Forstmann et al.

    Voluntary selection of task sets revealed by functional magnetic resonance imaging

    Journal of Cognitive Neuroscience

    (2006)
  • T.H. Gollan et al.

    Should I stay or should I switch? A cost-benefit analysis of voluntary language switching in young and aging bilinguals

    Journal of Experimental Psychology: Learning, Memory, and Cognition

    (2009)
  • A. Gotler et al.

    Nonintentional task set activation: Evidence from implicit task sequence learning

    Psychonomic Bulletin & Review

    (2003)
  • A.G. Greenwald et al.

    Three cognitive markers of unconscious semantic activation

    Science

    (1996)
  • A.G. Greenwald et al.

    Activation by marginally perceptible (“subliminal”) stimuli: Dissociation of unconscious from conscious cognition

    Journal of Experimental Psychology: General

    (1995)
  • J. Hoffmann et al.

    Task switches under Go/NoGo conditions and the decomposition of switch costs

    European Journal of Cognitive Psychology

    (2003)
  • B. Hommel

    Consciousness and control: Not identical twins

    Journal of Consciousness Studies

    (2007)
  • G. Hughes et al.

    Unconscious priming of a no-go response

    Psychophysiology

    (2009)
  • A. Kiesel et al.

    Evidence for task-specific resolution of response conflict

    Psychonomic Bulletin & Review

    (2006)
  • A. Kiesel et al.

    Mechanisms of subliminal response priming

    Advances in Cognitive Psychology

    (2007)
  • A. Kiesel et al.

    Control and interference in task switching—A review

    Psychological Bulletin

    (2010)
  • W. Klotz et al.

    Motor activation without conscious discrimination in metacontrast masking

    Journal of Experimental Psychology: Human Perception and Performance

    (1999)
  • Cited by (49)

    • Attentional modulation of masked semantic priming by visible and masked task cues

      2019, Cognition
      Citation Excerpt :

      As masked semantic priming is assumed to benefit from a sensitization of semantic pathways (Kiefer & Martens, 2010), we expected larger priming following a semantic induction task or semantic task cue than following a perceptual induction task or perceptual task cue. Using the research by Reuss et al. (2011) as starting point, the first experiment assessed, whether unconsciously perceived masked task cues (letters A and B) are able to modulate subsequent masked semantic priming. Similar to this earlier task cue experiment (Reuss et al., 2011), the task cue was only presented visibly at a long duration in induction task trials, but was briefly presented and masked in task cue-only trials.

    • Unconscious biases in task choices depend on conscious expectations

      2015, Consciousness and Cognition
      Citation Excerpt :

      Cues were actually presented in all trials, although their visibility (and therefore, participants’ awareness) varied according to the cue-mask interval. To maintain similarity with previous studies (e.g. Reuss et al., 2011) to a maximum, thresholds of consciousness were not measured initially for each participant. Although doing so would have been desirable, estimating the thresholds in advance would have arisen suspicions in participants regarding the presence of cues in every trial.

    View all citing articles on Scopus
    View full text