Elsevier

NeuroImage

Volume 18, Issue 2, February 2003, Pages 483-493
NeuroImage

Regular article
Practice-related effects demonstrate complementary roles of anterior cingulate and prefrontal cortices in attentional control

https://doi.org/10.1016/S1053-8119(02)00050-2Get rights and content

Abstract

The purpose of this study was to test the hypothesis that the dorsolateral prefrontal cortex (DLFPC), not the anterior cingulate cortex (ACC), plays the predominant role in implementing top-down attentional control. To do so, we used fMRI to examine practice-related changes in neural activity during a variant of the Stroop task. The results indicated that the DLPFC’s activity decreased gradually as the need for control was reduced (as indexed by behavioral measures), while the ACC’s activity dropped off rapidly. Such a pattern is consistent with the DLPFC taking a leading role in implementing top-down attentional control and the ACC being involved in other aspects of attentional control, such as response-related processes. In addition, with practice, there was a reduction in activity within cortical systems handling the processing of task-irrelevant information capable of interfering with task performance. This finding suggests that with practice the brain is capable of identifying and strategically inhibiting such processing.

Introduction

While the advent of PET and fMRI rapidly advanced our ability to identify the network of neural structures involved in attentional control, the task of determining the respective contributions of these structures has proven to be far more challenging. Among the most controversial of topics is the determination of which region(s) implements attentional control (often referred to as sources of control; Frith et al., 2001). Despite general agreement that the frontal lobes are primarily responsible for implementing control, the relative contributions of specific regions such as the dorsolateral prefrontal cortex (DLPFC) and the anterior cingulate cortex (ACC) are a continuing source of debate Banich et al 2000a, Gehring and Knight 2000, MacDonald et al 2000, Milham et al 2001, Pardo et al 1990, Paus et al 1993, Posner and DiGirolamo 1998, Taylor et al 1997.

The implementation of control involves: (1) modulation of neural activity within posterior processing regions in accord with task demands (i.e., amplifying neural activity within posterior processing systems handling task-relevant information while dampening that of posterior processing systems handling task-irrelevant information) and (2) the biasing of working memory processes such that the selection, maintenance, and manipulation of task-relevant representations is favored over that of task-irrelevant representations (see Milham et al 2002, Milham 2002, for a longer discussion of the neural mechanisms involved in these two processes).

Based on findings of increased ACC activity during tasks requiring response inhibition, selective attention, target selection, or novel responses, it has been suggested that the ACC is responsible for implementing attentional control Bush et al 1998, Peterson et al 1999, Posner and Dehaene 1994, Posner and DiGirolamo 1998. However, reports of variability in ACC activity during the Stroop task, the gold standard of attentional selection tasks, cast doubt on such a conclusion Bench et al 1993, Taylor et al 1997. Furthermore, recent studies have found that when attentional demands are manipulated but response competition and/or response conflict is held constant, no modulation in the ACC’s activity is noted Banich et al 2000b, Botvinick et al 1999, Carter et al 2000, MacDonald et al 2000, Milham et al 2001, Milham et al 2002. In contrast, DLPFC’s activity does vary with attentional demand. As we have noted (Milham et al., 2001), at a minimum, the failure to detect changes in the ACC’s activity during manipulations of attentional selection when response-related factors are controlled restricts the situations under which the ACC can be posited to be responsible for implementing attentional control.

An alternative perspective that has emerged from this latter body of work posits that the DLPFC and the ACC play complementary roles in attentional control. Consistent with others, we have argued that the ACC is primarily involved in response-related processes Paus et al 1993, MacDonald et al 2000, Milham et al 2001, Paus 2001, such as conflict monitoring Carter et al 1998, Carter et al 2000, MacDonald et al 2000, error detection (Gehring et al., 1993), and response facilitation/inhibition Paus et al 1993, Paus 2001. In contrast, we have argued that the dorsolateral (BA 9 and BA 46) and posterior inferior PFC (BA 44) are responsible for the implementation of attentional control. Regions of the dorsolateral prefrontal cortex impose an “attentional set” for task-relevant information by modulating neural activity within posterior regions of the cortex responsible for processing such task-relevant information (e.g., V4 when color is the attribute to which an individual must respond) (for a similar perspective, see Carter et al 2000, MacDonald et al 2000. We have also argued that these regions facilitate the selection of task-relevant representations within working memory. To the degree that such top-down control is compromised, as in aged individuals, performance suffers and increases in activity are noted within both posterior processing regions that handle task-irrelevant information and inferior prefrontal regions involved in the maintenance of information within working memory Banich et al 2000a, Banich et al 2000b, Milham et al 2001, Milham et al 2002.

It is important to note that while top-down control favors the activation of motor actions associated with task-relevant information, the activation of responses associated with task-irrelevant information is not necessarily abolished, especially when the processing of task-irrelevant information is relatively automatic. As such, even in the face of top-down control, some degree of facilitation and/or inhibition may be required at the level of response. We posit that these response processes are distinct from the top-down control imposed by the DLPFC and are supported by the ACC, a viewpoint consistent with others (e.g., Paus et al., 1993).

Here, we further test our hypothesis that the DLPFC and not the ACC plays the predominant role in top-down attentional control. Studies attempting to dissociate the DLPFC’s and ACC’s role in control typically compare patterns of neural activity across conditions with qualitative differences either in the need for control or in how it is produced (e.g., response conflict vs. semantic conflict). In the present work, we took a different approach, comparing practice-related changes in the DLPFC’s and ACC’s activity during performance of an attentionally demanding Stroop-like interference task.

We felt that an examination of practice-related effects would be an effective means of testing if one of these regions (DLPFC, ACC) is more crucial to top-down attentional control than the other. Regions involved in implementing top-down control (i.e., imposing an attentional set for task-relevant information) should remain active as long as task-irrelevant information is a potent source of interference. Over the course of our experiment, the need for top-down control during our interference condition gradually decreased with practice, but was never abolished, as indexed by behavioral measures. A region involved in implementing such control should show a similar pattern of results—gradually decreasing, but showing some degree of sustained activity. Thus, we examined the impact of practice on the pattern of activity for the ACC and DLPFC separately to determine if one of the two areas exhibited a pattern of neural activity more consistent with a predominant role in implementing top-down control than the other. Although prior studies have demonstrated practice-related changes in the activity of the ACC and/or DLPFC for a variety of paradigms (e.g., verb-generation task, counting-word Stroop task, complex finger-tapping tasks) (e.g., Bush et al 1998, Peterson et al 1999, none have focused specifically on contrasting the relative impact of practice on these two brain areas during performance of a single task.

Our hypothesis that the DLPFC implements top-down attentional control whereas the ACC is involved in response-related processes predicts that with practice, the DLPFC’s activity should be more sustained than that of the ACC. To test this prediction, we used a variant of the Stroop task introduced by MacLeod and Dunbar (1988) in which behavioral changes in attentional control can be observed over a relatively short period of time. We considered the standard Stroop paradigm ill-suited for the present investigation because it generally requires thousands of trials for significant practice effects to be observed behaviorally (MacLeod, 1991). In this modified Stroop paradigm, individuals were first taught to associate a unique color name to each of three previously unfamiliar shapes. Then we used fMRI to compare neural activity observed during blocks in which the shape’s ink color (task-irrelevant dimension) conflicted with its name (interference blocks) with blocks in which it did not (i.e., the shape appeared in white) (control blocks). Practice-related effects were detected by identifying changes in the comparison of interference versus control blocks over time.

Our paradigm also enabled us to study the mechanisms by which the processing of task-irrelevant information is suppressed. Although models of attention typically posit that neural activity within processing systems containing task-irrelevant information should be dampened, our prior neuroimaging work has demonstrated increased activity within these systems when the irrelevant information is related to the task-relevant dimension compared to when it is not. For example, when the task was to identify a word’s ink color, increased activity was observed within word-processing regions when the word named a color (e.g., the word “red” in blue ink) compared to when it did not (e.g., the word “lot” in blue ink). Likewise, when the task was to identify an object’s ink color, increased activity was observed within object-processing regions, when the object had a standard color with which it is associated (e.g., a frog is highly associated with the color green) compared to not (e.g., a car can be green but also red, blue, brown, tan, yellow, gray, white, black, and so forth) Banich et al 2000a, Banich et al 2000b, Banich et al 2001. We have speculated that this effect may occur for one of two reasons. First, it may be that top-down control is more “general,” activating all regions processing information related to color rather than ink color specifically. This general tuning may be adaptive to avoid an initial “tunnel vision” regarding task-relevant information. With experience or practice, however, this tuning may become more refined. Another possibility is that activation spreads from brain regions that are processing task-relevant information. Thus, when V4 becomes activated, regions that process associated information about color may be primed. Regardless of which mechanism (or possibly both) accounts for activation in regions processing task-irrelevant information, we would expect that with practice top-down control should dampen activation in brain regions processing task-irrelevant information. The paradigm in the present study allows us to examine such a prediction.

In the current study, the item’s ink color provides the task-irrelevant information that is related to the task-relevant dimension (i.e., the item’s name that is a color). Thus, consistent with our prior studies, we predicted greater activity during interference blocks relative to control in those regions of the lingual and posterior fusiform gyri involved in color processing Beauchamp et al 1999, Corbetta et al 1991, Zeki et al 1991 at least early in task performance. However, these increases should be dampened over time, as participants become more experienced with selecting the shape’s color name over the task-irrelevant ink color.

Section snippets

Stimuli

The stimuli consisted of three previously unfamiliar nonsense shapes.

Apparatus

All stimuli were programmed using Mel V2.0 and presented using an IBM-PC-compatible computer.

Learning phase

Each of three previously unfamiliar nonsense shapes was shown in white along with one of three color words (i.e., “BLUE,” “YELLOW,” “GREEN”) with which it was uniquely paired. Each shape–word pairing was viewed eight times outside the magnet, with the shape–word pairings occurring in a random order. Participants were instructed to

Behavioral data

A repeated-measures ANOVA with the factors of Cycle (first two cycles, second two cycles, last two cycles) and Trial type (interference, control) yielded a main effect of Trial type (F(1, 10) = 21.25, P < 0.001), because responses were slower to interference than control trials, but no main effect of Cycle (F(2, 20) = 0.118, P < 0.889). A linear contrast [1 0 −1] of the interference effect across time (cycles 1 and 2, cycles 3 and 4, cycles 5 and 6) revealed a significant decrease in the

Discussion

Examination of practice-related changes in neural activity provided clear evidence of the differential involvement of the ACC and DLPFC in attentional control; practice-related decreases in the ACC’s activity were more rapid and more pronounced than those in the DLPFC. Findings of such a rapid drop-off in ACC activity (during cycles 3–6) despite the continued need for maintenance of an attentional set (as indicated by the presence of a behavioral Stroop interference effect) are not consistent

Acknowledgements

We acknowledge help from Stan Colcombe and Kirk Erickson.

References (39)

  • V. Van Veen et al.

    Anterior cingulate cortex, conflict monitoring, and levels of processing

    NeuroImage

    (2001)
  • M.T. Banich et al.

    Visionfrom neurons to cognition

  • M.T. Banich et al.

    fMRI studies of Stroop tasks reveal unique roles of anterior and posterior brain systems in attentional selection

    J. Cogn. Neurosci.

    (2000)
  • M. Beauchamp et al.

    An fMRI version of the Farnsworth-Munsell 100-hue test reveals multiple color-selective areas in human ventral occipitotemporal cortex

    Cereb. Cortex

    (1999)
  • M. Botvinick et al.

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

    Nature

    (1999)
  • G. Bush et al.

    The counting Stroopan interference task specialized for functional neuroimaging—validation study with functional MRI

    Hum. Brain Mapp.

    (1998)
  • C. Carter et al.

    Parsing executive processesstrategic vs. evaluative functions of the anterior cingulate cortex

    Proc. Natl. Acad. Sci. USA

    (2000)
  • C.S. Carter et al.

    Anterior cingulate cortex, error detection, and the online monitoring of performance

    Science

    (1998)
  • L.L. Chao et al.

    Cortical regions associated with perceiving, naming, and knowing about colors

    J. Cogn. Neurosci.

    (1999)
  • Cited by (266)

    • Validity of mental and physical stress models

      2024, Neuroscience and Biobehavioral Reviews
    • Cognitive flexibility in post-traumatic stress disorder: Sustained interference associated with altered modulation of cortical oscillatory activity during task-switching

      2023, NeuroImage: Clinical
      Citation Excerpt :

      In our study, Group 1 showed lower theta band activity in left DLPFC in switch trials compared to the other two groups, accompanied by strong suppression of beta-band activity in other brain regions including ACC. Since these findings are associated with relatively fast reaction times, they are analogous to the lower activity observed with fMRI in DLPFC and ACC with increased practice on a task (Milham et al., 2003), being likely indicative of low perceived task difficulty. Group 2 showed a marker of more effortful cognitive control (high theta band activity in left DLPFC) when compared to Group 1, accompanied by similarly strong suppression of beta-band activity in other brain regions including ACC.

    View all citing articles on Scopus

    This study was supported by the Beckman Institute for Advanced Science and Technology at the University of Illinois, Urbana-Champaign; Carle Clinic, Urbana, Illinois; and NIMH MD/PhD predoctoral National Research Service Award provided support to M.P.M. (MH12415-01).

    1

    Current address: Department of Psychology, University of Colorado at Boulder, E213-E Muenzinger Hall, 345 UCB, Boulder, CO 80309, USA.

    View full text