Common and distinct neural substrates of attentional control in an integrated Simon and spatial Stroop task as assessed by event-related fMRI

Abstract

The purpose of this experiment was to directly examine the neural mechanisms of attentional control involved in the Simon task as compared to a spatial Stroop task using event-related fMRI. The Simon effect typically refers to the interference people experience when there is a stimulus–response conflict. The Stroop effect refers to the interference people experience when two attributes of the same stimulus conflict with each other. Although previous imaging studies have compared the brain activation for each of these tasks performed separately, none had done so in an integrated task that incorporates both types of interference, as was done in the current experiment. Both tasks activated brain regions that serve as a source of attentional control (dorsolateral prefrontal cortex) and posterior regions that are sites of attentional control (the visual processing stream–middle occipital and inferior temporal cortices). In addition, there were also specific brain regions activated to a significantly greater degree by one task and/or only by a single task. The brain regions significantly more activated by the Simon task were those sensitive to detection of response conflict, response selection, and planning (anterior cingulate cortex, supplementary motor areas, and precuneus), and visuospatial–motor association areas. In contrast, the regions significantly more activated by the Stroop task were those involved in biasing the processing toward the task-relevant attribute (inferior parietal cortex). These findings suggest that the interference effects of these two tasks are caused by different types of conflict (stimulus–response conflict for the Simon effect and stimulus–stimulus conflict for the Stroop effect) but both invoke similar sources of top-down modulation.

Introduction

Regardless of one's theoretical outlook, attentional control involves selection. This selection process can occur at multiple stages of processing—at the perceptual stage with regards to a specific attribute (e.g., the red item) or a particular location (e.g., the right hand item), at a more central stage concerning an abstract attribute (e.g., fruits but not vegetables), or at the response stage (e.g., pressing a right-hand key not a left-hand one). There are several paradigms that have been used extensively to examine attentional control, the Stroop task being one and the Simon task being another.

In the standard color–word Stroop task (Stroop, 1935), which was described by MacLeod (1992) as the “gold standard” of attentional measures, individuals see a colored word. The task is to identify a task-relevant dimension, such as the item's ink color, while ignoring a task-irrelevant dimension, such as the word's meaning or the response to which it leads. Typically, performance on incongruent trials in which the word and its ink color conflict (e.g., the word “red” in blue ink) is compared to that on neutral trials in which the word's meaning is not color-related (e.g., the word “lot” in blue ink). The former requires more attentional control than the latter as the color word has the ability to interfere with selection of a response based on ink color. Numerous neuroimaging studies have indicated that incongruent Stroop trials activate a series of brain regions involved in selection at the central and response stages. These regions include the dorsolateral prefrontal cortex (DLPFC), anterior cingulate cortex (ACC), and posterior parietal cortex (PPC) Banich et al., 2000a, Banich et al., 2000b, Banich et al., 2001, Barch et al., 2001, Bench et al., 1993, Brown et al., 1999, Bush et al., 1998, Carter et al., 1995, Leung et al., 2000, MacDonald et al., 2000, Milham et al., 2001, Milham et al., 2002, Milham et al., 2003, Pardo et al., 1990, Peterson et al., 1999. Similar regions are activated by spatial variants of the Stroop task. In a typical spatial Stroop task, individuals are told to respond to the location of a word and responses are generally slower on the incongruent trials in which the word is incompatible with its location (e.g., the word “above” positioned below a box) than on the neutral trials in which the word (e.g., “hope”) does not denote any spatial location (Banich et al., 2000b).

Another well-studied behavioral paradigm of attentional selection, the Simon task (Simon and Small, 1969), examines competition at the stimulus–response level (for a review, see Proctor and Reeve, 1990). In this task, performance of the incompatible trials on which one must give a response that is spatially incompatible with the stimulus (e.g., respond with the right hand to a left-hand stimulus) is compared to that of the compatible trials (e.g., respond with the right hand to a right-hand stimulus). Generally, responses are longer for the incompatible as compared to compatible stimuli. The neural basis of this phenomenon has not been well studied Bush et al., 2003, Dassonville et al., 2001, Fan et al., 2003, Iacoboni et al., 1998, Peterson et al., 2002.

Although both the Simon and spatial Stroop tasks share many similar properties (e.g., the location of the stimulus being the task-irrelevant dimension), they have been studied in distinct manners to examine different theoretical issues (Lu and Proctor, 1995). Behavioral research on these two paradigms has focused on diverse theoretical issues, in spite of the similarity among the task characteristics. For the Simon effect, researchers have focused on elucidating how the spatial stimulus–response conflict arises Hasbroucq and Guiard, 1991, Hommel, 1995, Proctor and Reeve, 1990. For the spatial Stroop effect, researchers have focused on how the stimulus–stimulus conflict between the task-relevant (e.g., location) and task-irrelevant (e.g., word meaning or response mapping) attributes is resolved by attentional selection (MacLeod, 1991).

Despite the similarities between the Simon and spatial Stroop effects as reviewed in Lu and Proctor (1995), Kornblum (1992) classified them into different categories of stimulus–response ensembles. In Kornblum's terms, ensembles are characterized according to the dimensional overlaps between (1) the relevant and irrelevant stimulus dimensions, (2) the relevant stimulus dimension and the response dimension, and (3) the irrelevant stimulus dimension and the response dimension. According to Kornblum, the classic Simon task is a Type 3 stimulus–response ensemble, in which the relevant stimulus dimension does not overlap with the response dimension while the irrelevant stimulus dimension does. In contrast, he considers the classic color–word Stroop task as a Type 8 ensemble, in which there is overlap not only between the relevant stimulus and response dimensions and between the irrelevant stimulus and response dimensions, but also between the relevant and irrelevant stimulus dimensions. Thus, the Stroop effect is distinct from the Simon effect in that it has the addition of overlap between the relevant and irrelevant stimulus dimensions. Moreover, we conceptualize these two effects as representing somewhat different types of response conflict. The Simon effect results from the direct stimulus–response conflict, due to the need to overcome the potent association between the stimulus and response of the same side (e.g., left hand response to left-side stimulus). In contrast, the Stroop effect results from the stimulus–stimulus conflict between the two attributes, which also lead to the conflicting responses (e.g., word “red” and blue ink lead to conflicting responses when both red and blue are potential responses).

Given the superficial and taxonomical similarities and differences between the Simon and Stroop effects, Lu and Proctor (1995) suggested, based on the analysis of behavioral performance that they share a common theoretical foundation: attentional and response selection. Several theoretical and computational models have been proposed that provide an integrated account for both the Simon and Stroop effects Hasbroucq and Guiard, 1991, Kornblum et al., 1999, Zhang et al., 1999. However, there is still a debate over whether there is a common source for both effects, especially with regards to their neural substrates. Lu and Proctor (1995) noted that “None of the accounts developed for the Simon effect or the spatial Stroop effect seems capable of handling the majority of findings from both task domains without significant modification. However, many of the central features of these accounts are supported by the existing evidence, and a model that combines several of these features in a principled manner would seem to be most promising.”

An examination of the findings of recent neuroimaging studies on each task in isolation suggests that these two measures of attentional control activate similar neural structures Bush et al., 2003, Peterson et al., 2002, although each appears to activate unique brain regions as well Banich et al., 2000b, Brown et al., 1999, Carter et al., 1995, Fan et al., 2003, Iacoboni et al., 1996, Iacoboni et al., 1998, Leung et al., 2000, MacDonald et al., 2000, Pardo et al., 1990, Praamstra et al., 1999. Typically, the Stroop effect activates broadly distributed brain areas, including DLPFC, ACC, inferior frontal, inferior parietal, and inferior temporal cortices, while the Simon effect usually activates dorsal premotor, posterior, and superior parietal areas. Across studies, regions activated by both tasks appear to be the DLPFC and dorsal ACC.

We are aware of only two studies to date in which the standard color–word Stroop and Simon effects have been directly compared Fan et al., 2003, Peterson et al., 2002. In one study, Peterson et al. (2002) conducted two event-related fMRI experiments on the same group of participants. In one experiment, participants were given a color–word Stroop task and in the other a Simon task. For the Stroop task, they were presented either a congruent (e.g., the word “red” in red color) or incongruent (e.g., the word “red” in blue color) color–word stimulus and asked to silently respond to the ink color of the word. For the Simon task, they were presented a white arrow pointing either to the left or right against a black background, either to the left or right of a central fixation cross. Participants were instructed to press a key to either the left-pointing or right-pointing arrow with the index (relative leftward) or middle (relative rightward) finger of their right hand, respectively. On congruent trials, the direction of the arrow was the same as the location of the arrow relative to the fixation (e.g., a rightward pointing arrow to the right of fixation), whereas on incongruent trials, the direction of the arrow was opposite to the location of the arrow relative to the fixation (e.g., a rightward pointing arrow to the left of fixation). In both the Stroop and Simon tasks, the incongruent trials were more infrequent than the congruent trials.

These researchers found remarkably similar results for both the Simon and Stroop tasks. The brain regions activated by the incongruent stimuli as compared to the congruent stimuli in both tasks included DLPFC, ACC, supplementary motor areas (SMA), visual association cortex, inferior temporal, inferior parietal, and inferior frontal cortices, as well as the caudate nuclei. In addition, the time courses of the brain activity were also very similar across the tasks. They concluded that the neural systems that subserve successful performance in both tasks are likely to be similar.

However, the Simon task employed in the Peterson et al. (2002) study is not a “pure” one. A typical Simon effect refers to the interference people experience when the response required by a task is spatially opposite to the location of the stimulus (e.g., right finger press to a stimulus left to the fixation), which creates a stimulus–response conflict. There is usually no conflict between the relevant and irrelevant stimulus dimensions, which is the critical distinction between the Simon and Stroop effects (Kornblum, 1992). By presenting the left-pointing or right-pointing arrow on either the left or right of the fixation, the investigators obscured this distinction. Hence, their so-called Simon effect was in fact a combination of both the Simon and Stroop effects, since there was not only conflict on the incongruent trials between the irrelevant stimulus attribute (e.g., the left of the fixation) and response (e.g., right key press), but also conflict between the relevant (e.g., right-pointing arrow) and irrelevant (e.g., the left of fixation) dimensions as well. Therefore, their findings showing the common activation patterns by both the Simon and Stroop effects cannot be unambiguously interpreted that these two effects indeed share a common neural basis.

The other study that compared the color–word Stroop task and the Simon spatial conflict task also employed the Eriksen flanker task (Fan et al., 2003) so that the neural substrates of conflict monitoring and resolution could be compared across the three tasks. They found that although three tasks shared a common attentional control and conflict resolution network involving the brain areas such as dorsal ACC and prefrontal cortex, these tasks activated other distinct brain regions within the posterior parietal cortex and visual processing areas. The interpretation of this study has the converse problem to that of the Petersen study as the Stroop task employed utilized different stimulus dimension (i.e., colors and words) than the Simon task (spatial dimensions). Hence, it is difficult to determine whether differences in the neural substrates activated by each task are due to the difference in stimulus attributes (color and word information vs. spatial information) or the nature of the selection of each task.

To resolve the issue of the extent to which the neural substrates of attentional control in the Simon and Stroop tasks are common or distinct, we utilized an event-related fMRI experiment that incorporated both types of interference within an integrated task (see Fig. 1). Participants were presented with an upward or downward arrow at the fixation and trained to respond to one arrow with the index finger and to the other with the middle finger of their right hand. The mapping of the upward and downward arrows to fingers was counterbalanced across participants. During the fMRI testing session, the arrows were presented in one of the four locations surrounding the central fixation cross. This display created two types of conflict within a single paradigm. The Simon effect was elicited when an arrow was presented in a location incompatible with the relative spatial position of the finger used to respond (e.g., an upward arrow placed to the right of the fixation, which required response of the index finger—the more leftward of the two fingers). The spatial Stroop effect was elicited when the direction of the arrow (i.e., the task-relevant dimension) was incompatible with its spatial location (i.e., the task-irrelevant dimension) (e.g., an upward arrow below the fixation), as compared to the compatible condition (e.g., an upward arrow above the fixation).

Given the distinction made by Kornblum (1992), we hypothesize that both tasks will yield highly similar patterns of activation in prefrontal regions as both tasks require the top-down control of attention (Posner and Petersen, 1990). We also hypothesize, however, that each task will uniquely activate specific brain regions because the two tasks differ in the nature of the conflict that is engendered by each task. More specifically, we predict that the brain regions previously theorized as the source of attentional selection such as DLPFC will be commonly activated more in the incongruent trials than in the congruent trials for both tasks, since both require people to selectively respond to the task-relevant information (i.e., the direction of the arrow) while ignoring the conflicting task-irrelevant information (i.e., the location of the arrow). We further predict similar brain areas of visual processing will be modulated by the top-down attentional control so that the conflict between the task-relevant and task-irrelevant attributes is resolved by biasing the processing toward the task-relevant attribute (i.e., the direction of the arrow). In contrast, according to the taxonomy of Kornblum (1992), the core and unique component of the Stroop effect is the stimulus–stimulus conflict while the nature of the conflict of the Simon effect is more stimulus–response. Therefore, we predict that the resolution of the conflict in these two effects will involve different neural substrates. In particular, we predict that parietal areas involved in multi-attribute processing and visual attention will be more activated in the Stroop task. In contrast, we predict that regions involved in stimulus–response processing, such as portions of the anterior cingulate and premotor regions, will be more activated in the Simon task.