Elsevier

NeuroImage

Volume 35, Issue 2, 1 April 2007, Pages 940-948
NeuroImage

Separate conflict-specific cognitive control mechanisms in the human brain

https://doi.org/10.1016/j.neuroimage.2006.11.061Get rights and content

Abstract

To ensure optimal task performance, the human brain detects and resolves conflict in information processing via a cognitive control system. However, it is not known whether conflict resolution relies on a single central resource of cognitive control, or on a collection of independent control mechanisms that deal with different types of conflict. In order to address this question, we assessed behavioral and blood–oxygen-level-dependent (BOLD) responses during the simultaneous detection and resolution of two sources of conflict in a modified color-naming Stroop task: conflict stemming from incompatibility between the task-relevant and an irrelevant stimulus feature (stimulus-based or Stroop conflict), and conflict stemming from incompatibility between an irrelevant stimulus feature and response features (response-based or Simon conflict). Results show that control mechanisms recruited by stimulus-based conflict resolve stimulus-based conflict, but do not affect the resolution of response-based conflict, and vice versa. The resolution of response-based conflict was distinguished by modulation of activity in premotor cortex, whereas resolution of stimulus-based conflict was distinguished by the modulation of activity in parietal cortex. These results suggest that the human brain flexibly adopts, and independently controls, conflict-specific resolution strategies, biasing motor programming to resolve response-based conflict, and biasing stimulus representations to resolve stimulus-based conflict. We propose a non-centralized, modular architecture of cognitive control, where separate control resources operate in parallel, and are recruited in a context-sensitive manner.

Introduction

Human performance regulation involves mechanisms that detect and resolve conflict in information processing (Botvinick et al., 2001). A classic example of conflict is provided by the color-naming Stroop task: subjects are required to name the ink color (e.g., red) of a printed color–word, the meaning of which can be either compatible (RED) or incompatible (e.g., GREEN) with that color (Stroop, 1935, MacLeod, 1991). When ink color and word meaning are semantically incompatible, color and word processing streams lead to conflicting representations, resulting in slowed performance (Cohen et al., 1990). For optimal performance, such conflict must be detected and resolved by cognitive control mechanisms (Botvinick et al., 2001).

One neural strategy for conflict resolution is to bias stimulus processing in sensory pathways, where cortical representations of task-relevant stimulus features (e.g., ink color) may be amplified relative to task-irrelevant ones (e.g., word meaning) (Cohen et al., 1990, Desimone and Duncan, 1995, Kastner and Ungerleider, 2000). In tasks such as the Stroop protocol, where conflict stems from incompatibility between task-relevant and task-irrelevant stimulus features (stimulus-based conflict), such a “stimulus bias” conflict resolution mechanism has indeed been documented; specifically, conflict resolution has been found to be associated with enhanced processing of task-relevant stimulus information in sensory cortices (Egner and Hirsch, 2005a).

A second strategy for overcoming conflict is to bias the response selection process (Nieuwenhuis and Yeung, 2005), for instance by inhibiting the influence of task-irrelevant information on motor output (Sturmer et al., 2002, Sturmer and Leuthold, 2003). It is thought that such a “response bias” conflict resolution strategy is employed to resolve conflict in the Simon task (Simon, 1969), where conflict occurs due to incompatibility between a task-irrelevant stimulus feature and response features (response-based conflict) (Stoffels, 1996, Praamstra et al., 1999, Ridderinkhof, 2002, Sturmer et al., 2002, Sturmer and Leuthold, 2003). Here, subjects categorize the ink color of a stimulus presented to either the left or right of a central fixation, by pushing response buttons with their left (e.g., for green) or right hand (e.g., for red). Conflict occurs when the position of the stimulus is spatially incompatible with the position of the correct response effecter (e.g., a red stimulus presented on the left side) resulting in slowed performance (Simon, 1969, Lu and Proctor, 1995).

Two questions regarding the nature of cognitive control mechanisms arise from these findings: first, it is not known whether control mechanisms involved in resolving stimulus- versus response-based conflict are engaged independently of each other, or whether they rely on shared central resources. We addressed this question by independently manipulating sources of conflict, between stimulus-based (Stroop) and response-based (Simon) conflict, and assessing whether control mechanisms recruited by one type of conflict would affect the resolution of the other type of conflict. Secondly, if there were independent conflict resolution mechanisms, it is not known whether these would conform to the stimulus-biasing versus response-biasing strategies implied by the literature. We tested this proposal by acquiring functional magnetic resonance imaging (fMRI) data during task performance, which allowed us to contrast neural activity related to resolving stimulus-based conflict to that associated with resolving response-based conflict.

In order to directly contrast stimulus-based and response-based conflict processes, we factorially combined the classic color-naming Stroop and Simon tasks into a single experimental protocol (cf. Simon and Berbaum, 1990, Kornblum, 1994, Hommel, 1997). Note that in the current paper we employ the terms ‘stimulus-based’ and ‘response-based’ to refer to the origin of conflict in the Stroop and Simon tasks, respectively. Specifically, ‘response-based’ conflict should not be confounded with ‘response conflict’: we use the former to describe conflict that originates with an overlap between an irrelevant stimulus dimension and the response dimension, as is the case in the Simon task (Kornblum et al., 1990, Kornblum and Lee, 1995, Zhang et al., 1999), while the latter is used to refer to a co-activation of mutually incompatible response pathways. Therefore, both the Stroop and the Simon task entail ‘response conflict’ in the sense that incompatible trials are associated with incompatible response tendencies; however, the genesis of response conflict differs between the two tasks, in that it is response-based in the Simon task, and stimulus-based in the Stroop task. Below, these conflicts are described in more detail.

In the color-naming Stroop task, conflict arises from a dimensional overlap between the relevant stimulus dimension (ink color) and an irrelevant stimulus dimension1 (word meaning) (Kornblum et al., 1990, Kornblum and Lee, 1995, Zhang et al., 1999). As attentional selection of the task-relevant stimulus dimension (ink color) is not perfect, involuntary processing of the word meaning on incompatible trials leads to ‘conceptual conflict’ (often referred to as ‘stimulus conflict’) between color (e.g., red) and word (e.g., ‘GREEN’) processing streams. The processing of word meaning is thought to interfere particularly strongly with ink color naming because it constitutes a more highly practiced process (MacLeod and Dunbar, 1988). If the two processing streams are furthermore associated with two eligible but incompatible responses, the conceptual (stimulus) conflict will additionally result in simultaneous activation of mutually incompatible response pathways, thus also producing response conflict (Kornblum et al., 1990, Kornblum and Lee, 1995, Zhang et al., 1999, Milham et al., 2001, De Houwer, 2003, van Veen and Carter, 2005). While conflict in a typical Stroop protocol can thus be argued to reflect an additive effect of stimulus (conceptual) and response conflict (Milham et al., 2001, De Houwer, 2003, van Veen and Carter, 2005), we are here not concerned with this distinction. Instead, our concern is the origin of both these conflicts, which lies with the semantic incompatibility between relevant and irrelevant stimulus dimensions; therefore, Stroop conflict represents stimulus-based conflict (Kornblum et al., 1990, Kornblum and Lee, 1995, Zhang et al., 1999).

In the Simon task (Simon, 1969), on the other hand, conflict is due to an overlap between an irrelevant stimulus dimension (stimulus location) and the response dimension (left/right button press) (Kornblum et al., 1990, Kornblum and Lee, 1995, Zhang et al., 1999). This effect is thought to be due to an unintentional or ‘direct’ route of activation of the spatially corresponding response effecter by the (irrelevant) stimulus location (Kornblum et al., 1990, Hommel, 1993, De Jong et al., 1994). In other words, there appears to be an inherent propensity of the motor system to react towards the source of stimulation (Simon, 1969). This fast, direct route of response activation (or response priming) competes with a slower, ‘indirect’ route, represented by the intentional processing of the task-relevant color information in relation to the instructed stimulus–response mappings (Kornblum et al., 1990, Hommel, 1993, De Jong et al., 1994). On incompatible trials, direct route response activation conflicts with the response selection derived from processing the relevant color stimulus feature, resulting in response conflict (Kornblum et al., 1990, De Jong et al., 1994, Zhang et al., 1999). Therefore, like Stroop conflict, Simon conflict is associated with conflicting response tendencies; however, unlike Stroop conflict, it does not originate with incompatibility between stimulus dimensions, but with a direct interference of an irrelevant stimulus dimension with the response selection process (Acosta and Simon, 1976, Simon, 1982). Simon conflict therefore is held to represent response-based conflict (Kornblum et al., 1990, Kornblum and Lee, 1995, Zhang et al., 1999).

We combined the classic Stroop and Simon tasks by presenting color–word stimuli (RED and GREEN), printed in either red or green ink, either to the left or right of a central fixation cross, and requiring subjects to identify the ink color of a given stimulus by pressing a response button with their left index finger for stimuli of green ink color and with their right index finger for stimuli of red ink color (Fig. 1). Each stimulus could thus be compatible or incompatible with respect to the color–word, and with respect to the spatial location, resulting in a 2 (Stroop compatibility: compatible versus incompatible) × 2 (Simon compatibility: compatible versus incompatible) factorial design, while the task-relevant stimulus feature (ink color) was held constant. Behavioral studies employing similar designs have supported the notion that Stroop and Simon compatibility effects involve different, independent processing resources, as Stroop and Simon compatibilities both produce main effects, but do not interact (Simon and Berbaum, 1990, Kornblum, 1994, Hommel, 1997).

In order to dissociate control- from conflict-related processes associated with Stroop and Simon stimulus dimensions, we assessed the “conflict adaptation effect”, a sequential trial effect that has been argued to reflect the workings of the conflict-monitoring/cognitive control loop (Botvinick et al., 1999, Botvinick et al., 2001, Kerns et al., 2004, Egner and Hirsch, 2005a, Egner and Hirsch, 2005b), and which characterizes performance patterns both on the Stroop task (Kerns et al., 2004, Egner and Hirsch, 2005b, Notebaert et al., 2006) as well as on the Simon task (Stoffels, 1996, Praamstra et al., 1999, Ridderinkhof, 2002, Sturmer et al., 2002, Sturmer and Leuthold, 2003, Wuhr, 2005, Wuhr and Ansorge, 2005). The conflict adaptation effect is reflected in the finding that the degree to which task-irrelevant information interferes with the processing of task-relevant information varies as a function of trial sequence: conflict is reduced following incompatible trials compared to compatible trials (Gratton et al., 1992). According to the conflict-monitoring model, this previous by current trial compatibility interaction effect arises because high conflict on an incompatible trial leads to an up-regulation in cognitive control, resulting in improved selection of target information on the next trial, which is reflected in faster responses to incompatible trials (reduced interference) and slower responses to compatible ones (reduced facilitation) (Botvinick et al., 2001). Alternative accounts of this effect (Mayr et al., 2003, Hommel et al., 2004) are addressed in the Discussion section.

Here, we analyzed trial-to-trial conflict adaptation effects with respect to Stroop and Simon compatibility factors, in order to assess whether the cognitive control processes that resolve conflict are specific to the original source of the conflict (see also Wendt et al., 2006). For this purpose, the aforementioned 2 × 2 factorial design was expanded to incorporate previous stimulus type into the analysis, resulting in a 4-way 2 × 2 × 2 × 2 factorial design, with the factors of previous and current trial compatibility (compatible versus incompatible), current trial stimulus dimension (Stroop versus Simon) and previous trial stimulus dimension (same versus different) (a trial exemplar is displayed in Fig. 1). Based on our a priori hypotheses, adaptation effects were assessed with respect to the Stroop compatibility factor, to reveal whether stimulus-based conflict on the previous trial results in superior conflict resolution of stimulus-based conflict on the current trial, as documented in previous research (Kerns et al., 2004, Egner and Hirsch, 2005b, Notebaert et al., 2006). Adaptation effects were also assessed with respect to the Simon compatibility factor, to reveal whether response-based conflict on the previous trial results in superior conflict resolution of response-based conflict on the current trial, as documented in previous research (Stoffels, 1996, Praamstra et al., 1999, Ridderinkhof, 2002, Sturmer et al., 2002, Sturmer and Leuthold, 2003, Wuhr, 2005, Wuhr and Ansorge, 2005). Most importantly, however, the current design allowed us to test whether control processes triggered by stimulus-based conflict on the previous trial would affect the resolution of response-based conflict on the current trial, and vice versa, by assessing conflict adaptation across the Stroop and Simon compatibility factors.

If a single central cognitive control resource were responsible for resolving both stimulus-based and response-based conflict, one would expect the recruitment of one type of resolution mechanism to impair the resolution of the other type of conflict, that is, control processes triggered by stimulus-based conflict on the previous trial would impair the resolution of response-based conflict on the current trial, and vice versa. On the other hand, if there were independent resources for dealing with stimulus-based and response-based conflicts, two alternative predictions could be made. Either, any type of conflict may trigger both stimulus- and response-related control processes, in which case control processes triggered by stimulus-based conflict on the previous trial would enhance the resolution of response-based conflict on the current trial, and vice versa. Or, alternatively, the two conflict resolution strategies may be recruited in a conflict-specific manner, such that control processes triggered by stimulus-based conflict on the previous trial would have no effect on the resolution of response-based conflict on the current trial, and vice versa.

As noted above, both the theoretical and the empirical literature suggest that Stroop conflict is resolved via a stimulus-biasing strategy (Cohen et al., 1990, Botvinick et al., 2001, Egner and Hirsch, 2005a), whereas Simon conflict is resolved via a response-biasing strategy (Stoffels, 1996, Praamstra et al., 1999, Sturmer et al., 2002, Sturmer and Leuthold, 2003). If these assumptions were true, we would expect our fMRI data to show different brain regions to be implicated in the two resolution processes. Specifically, we would expect Simon conflict resolution to be associated with differential activation in premotor and/or motor cortices, while Stroop conflict resolution would be expected to be associated with differential activity in areas implicated in top-down stimulus biasing, such lateral frontal and superior parietal cortices.

Section snippets

Subjects

Participants were 15 native English-speaking, healthy volunteers (mean age = 27 years, 7 females) with normal or corrected-to-normal vision, who provided informed consent conforming to institutional guidelines. Participants were screened via self-report for neurological or psychological conditions, use of psychiatric medication, color blindness, and dyslexia. Due to excessive movement artifacts in the MRI data, 2 subjects were excluded from all analyses.

Procedure

The task programming, stimulus delivery,

Behavioral results

Mean reaction times (RT) were calculated for each of the trial transitions of interest in each subject, excluding error trials, post-error trials, and condition-specific outlier values (> 2 SDs from the mean); group means are presented in Table 1. Analysis of variance (ANOVA) showed that there were significant main effects of both Stroop compatibility (compatible trial mean = 615 ms, incompatible trial mean = 644 ms; F(1,12) = 14.5, p < 0.005) and Simon compatibility (compatible trial mean = 622 ms,

Discussion

Cognitive control mechanisms are thought to resolve conflict in information processing in two principal ways, by biasing the processing of stimulus features, and by biasing response processes (Nieuwenhuis and Yeung, 2005). Our behavioral results document that these conflict resolution mechanisms appear to be recruited and implemented in a conflict-specific manner, and do not rely on a single central resource: stimulus-based (Stroop) conflict on the previous trial resulted in superior resolution

Acknowledgments

We thank Amit Etkin, Emily Stern, and Chris Summerfield for their helpful comments on the manuscript.

References (65)

  • G. Rizzolatti et al.

    The organization of the cortical motor system: new concepts

    Electroencephalogr. Clin. Neurophysiol.

    (1998)
  • J.R. Simon

    Effect of any auditory stimulus on the processing of a visual stimulus single- and dual-tasks conditions

    Acta Psychol.

    (1982)
  • J.R. Simon et al.

    Effect of conflicting cues on information processing: the ‘Stroop effect’ vs. the ‘Simon effect’

    Acta Psychol.

    (1990)
  • E.J. Stoffels

    On stage robustness and response selection: further evidence

    Acta Psychol.

    (1996)
  • V. van Veen et al.

    Separating semantic conflict and response conflict in the Stroop task: a functional MRI study

    NeuroImage

    (2005)
  • S. Yantis et al.

    Cortical mechanisms of space-based and object-based attentional control

    Curr. Opin. Neurobiol.

    (2003)
  • H.H. Zhang et al.

    A parallel distributed processing model of stimulus–stimulus and stimulus–response compatibility

    Cogn. Psychol.

    (1999)
  • E. Acosta et al.

    The effect of irrelevant information on the stages of processing

    J. Mot. Behav.

    (1976)
  • M. Botvinick et al.

    Conflict monitoring versus selection-for-action in anterior cingulate cortex

    Nature

    (1999)
  • M.M. Botvinick et al.

    Conflict monitoring and cognitive control

    Psychol. Rev.

    (2001)
  • C. Buchel et al.

    Modulation of connectivity in visual pathways by attention: cortical interactions evaluated with structural equation modelling and fMRI

    Cereb. Cortex

    (1997)
  • J.D. Cohen et al.

    On the control of automatic processes: a parallel distributed processing account of the Stroop effect

    Psychol. Rev.

    (1990)
  • M. Corbetta et al.

    Control of goal-directed and stimulus-driven attention in the brain

    Nat. Rev. Neurosci.

    (2002)
  • J. De Houwer

    On the role of stimulus–response and stimulus–stimulus compatibility in the Stroop effect

    Mem. Cognit.

    (2003)
  • R. De Jong et al.

    Conditional and unconditional automaticity: a dual-process model of effects of spatial stimulus–response correspondence

    J. Exp. Psychol. Hum. Percept. Perform.

    (1994)
  • R. Desimone et al.

    Neural mechanisms of selective visual attention

    Annu. Rev. Neurosci.

    (1995)
  • T. Egner et al.

    Cognitive control mechanisms resolve conflict through cortical amplification of task-relevant information

    Nat. Neurosci.

    (2005)
  • T. Egner et al.

    Where memory meets attention: neural substrates of negative priming

    J. Cogn. Neurosci.

    (2005)
  • M. Eimer

    Stimulus–response compatibility and automatic response activation: evidence from psychophysiological studies

    J. Exp. Psychol. Hum. Percept. Perform.

    (1995)
  • E. Fox

    Negative priming from ignored distractors in visual selection: a review

    Psychon. Bull. Rev.

    (1995)
  • K.J. Friston et al.

    Attentional modulation of effective connectivity from V2 to V5/MT in humans

    Proc. Natl. Acad. Sci. U. S. A.

    (2000)
  • G. Gratton et al.

    Optimizing the use of information: strategic control of activation of responses

    J. Exp. Psychol. Gen.

    (1992)
  • Cited by (233)

    View all citing articles on Scopus
    View full text