Conceived and designed the experiments: CNB RMK MGW. Performed the experiments: CNB LCC. Analyzed the data: CNB LGA RMK LCC. Wrote the paper: CNB LGA RMK LCC MGW.
The authors have declared that no competing interests exist.
Inhibitory motor control is a core function of cognitive control. Evidence from diverse experimental approaches has linked this function to a mostly right-lateralized network of cortical and subcortical areas, wherein a signal from the frontal cortex to the basal ganglia is believed to trigger motor-response cancellation. Recently, however, it has been recognized that in the context of typical motor-control paradigms those processes related to actual response inhibition and those related to the attentional processing of the relevant stimuli are highly interrelated and thus difficult to distinguish. Here, we used fMRI and a modified Stop-signal task to specifically examine the role of perceptual and attentional processes triggered by the different stimuli in such tasks, thus seeking to further distinguish other cognitive processes that may precede or otherwise accompany the implementation of response inhibition. In order to establish which brain areas respond to sensory stimulation differences by rare Stop-stimuli, as well as to the associated attentional capture that these may trigger irrespective of their task-relevance, we compared brain activity evoked by Stop-trials to that evoked by Go-trials in task blocks where Stop-stimuli were to be ignored. In addition, region-of-interest analyses comparing the responses to these task-irrelevant Stop-trials, with those to typical relevant Stop-trials, identified separable activity profiles as a function of the task-relevance of the Stop-signal. While occipital areas were mostly blind to the task-relevance of Stop-stimuli, activity in temporo-parietal areas dissociated between task-irrelevant and task-relevant ones. Activity profiles in frontal areas, in turn, were activated mainly by task-relevant Stop-trials, presumably reflecting a combination of triggered top-down attentional influences and inhibitory motor-control processes.
Inhibitory motor control — i.e. the ability to suppress unwanted behavioral responses — provides crucial flexibility in goal-directed behavior, allowing individuals to quickly adjust to a changing environment and to overcome pre-potent responses when they are inadequate or inappropriate (see
One of the most prominent experimental paradigms designed to investigate response-inhibition capabilities is the Stop-signal task
Although it is very likely that the loop between the frontal cortex and the basal ganglia/thalamus described above is a core structure subserving response inhibition, it is increasingly recognized that other mechanisms play an important role leading up to response inhibition and in determining whether it will be successful or not. Specifically, it has been reported that selective attention to the task-relevant stimuli can play an important role in determining trial outcome in the Stop-signal task. Numerous studies have reported transient modulations of sensory processing of the relevant stimuli in the time-range of the sensory evoked N1 ERP component, which precedes the implementation of response inhibition, and that these modulations are predictive of the outcome of the process
Unfortunately, such conclusions are much more difficult to draw for activity at later time-ranges and in other brain areas, so that a separation of perceptual/attentional processes from those that are directly related to response inhibition has proven difficult. A case in point relates to the right IFG, which has received a lot of experimental support as a key structure in response inhibition. This area is reliably activated in human fMRI studies investigating response inhibition (for a recent comprehensive review, see
An important distinction in this context that has not yet been established (neither for the right IFG nor for other involved brain areas) is the degree of automaticity with which Stop-trial stimulation elicits neural activity. Specifically, in the existing attention literature it is appreciated that the presentation of rare, and/or physically salient, stimuli (note that Stop-trials meet both criteria) tend to automatically capture attention and activate at least parts of the ventral attention system
In the present report, we have carried out additional sets of analyses of the data from a recent study
(A) In Stop-relevant blocks, a choice-reaction stimulus (a green German traffic-light symbol oriented to the left or right) was either presented for the entire stimulus duration of 800 ms (Go-trial) or replaced by a red Stop-stimulus (Stop-trial) after a variable SOA set trial-to-trial by a tracking algorithm. The Stop-stimulus indicated that the response to the Go-stimulus was to be cancelled, yielding successful (SSTs) and unsuccessful Stop-trials (USTs). (B) In Stop-irrelevant blocks the visual stimulation was identical, but the Stop-stimuli were all irrelevant, i.e. responses were required for all the Go-trials regardless of whether they were followed by a Stop-stimulus. (C) Response times were slowest for Stop-relevant (SR) Go-trials but similar for unsuccessful Stop-trial, Stop-irrelevant (SI) Stop-trials, and Stop-irrelevant Go-trials. The Stop-signal reaction time (SSRT) was calculated to be 230 ms (grand-average data + standard error of the mean (SEM)).
Eighteen participants took part in this study, two of which had to be excluded due to technical problems, and another one due to particularly poor behavioral performance. The 15 remaining participants (nine female) had a mean age of 22.9 years, all with correct or corrected-to-normal visual acuity, and none reporting a history of psychiatric or neurological disorders. All participants gave written informed consent and the study was approved by the Duke University Health System Institutional Review Board. Participants were compensated $20 per hour.
The present experiment entailed two variants of the typical Stop-signal task
Stop-relevant blocks used a standard Stop-signal task (using German traffic-light signs, see
A common approach for controlling performance is to titrate the Go-Stop SOA using an adaptive staircase procedure to yield approximately equivalent numbers of SST and UST for each participant. We implemented such a procedure here, increasing the SOA by 17 ms (one refresh screen) after SSTs and decreasing it by the same amount after USTs (starting SOA: 200 ms). This procedure allowed us to calculate the Stop-signal response time (SSRT), which is viewed as reflecting the mean amount of time that is required to implement the inhibition of a motor response and is derived by subtracting the mean Go-Stop SOA from the average Go-trial response time
During Stop-irrelevant blocks, visual stimulation was identical to the Stop-relevant ones (
MR data was acquired on a 3-Tesla GE Signa MRI system. Functional images were acquired with a reverse spiral imaging sequence (TR = 2000 ms, TE = 25 ms; flip angle = 75°; 32 slices with 3×3×3 mm resolution; AC-PC orientation providing coverage approximately from the top of the brain down to the pons). The first five functional images were excluded from the analysis, to allow the scanner to reach steady-state magnetization. For anatomical reference, a high-resolution structural T1 (3D Fast Spoiled Gradient Recalled (FSPGR); 1×1×1 mm resolution) was obtained. The fMRI data were analyzed using SPM5 (
The parameter estimates resulting from each condition/contrast and participant (first-level analysis) were entered into a second-level, random-effects group analysis using one-sample t-tests. In order to test for areas that were more active for Stop-irrelevant Stop-trials than Stop-irrelevant Go-trials on a brain-wide level, a voxel-wise analysis was performed. The respective group-level results were thresholded at T>3 (uncorrected) and a minimum cluster size of k = 10 contiguous voxels. Additionally, cluster-level correction for multiple comparisons was performed. Clusters surviving this correction (p<0.05) are highlighted in the Results tables, and strong inferences are limited to these areas. Despite the danger of false positives, we also report those activations that did not survive this correction. Such two-stage procedure was employed to meet our inferential goals to simultaneously not
Additionally, a region of interest (ROI) analysis was performed to compare activity elicited by the different conditions in the key regions involved in this task. In order to define ROIs that would allow for a comparison between Stop-trials from the Stop-relevant and the Stop-irrelevant task blocks, a t-contrast was employed that tested the average of these Stop-trial responses across the blocks against the average of all Go-trial responses across the blocks. Due to the very robust and widespread activations identified by this contrast, the group-level effects were thresholded comparatively conservatively (p<0.01; FDR-corrected corrected on the voxel level with an extent threshold k = 50 contiguous voxels; note that the resulting clusters also survived cluster-level multiple-comparison correction). Defining ROIs on the basis of this contrast enabled quantitative comparisons between the Stop-relevant and Stop-irrelevant Stop-trials in these regions, because the ROI selection was not biased in favor of either of those conditions
Participants performed very accurately during both the Stop-relevant and Stop-irrelevant task blocks. No significant differences in accuracy were observed for the three trial types that always required a response (i.e., Stop-relevant Go-trials [97.6%], Stop-irrelevant Go-trials [96.6%], and Stop-irrelevant Stop-trials [97.1%]; F(1.5,21.4) = 2.2, p = 0.15). Response times were slower on Stop-relevant Go-trials (520 ms) relative to unsuccessful Stop-trials (446 ms) and relative to Stop-irrelevant Go- and Stop-trials (436 and 439 ms; overall F-Test: F(1.8,25.3) = 32.3, p<0.001), but were similar between the latter three conditions (F(1.3,17.61) = 0.9, p = 0.37; see
In order to identify brain areas that respond differentially to Stop-stimuli, as compared to Go-stimuli, even if those stimuli are entirely task-irrelevant, we performed a voxel-wise comparison between Stop-trials and the Go-trials in the Stop-irrelevant blocks (T>3; k = 10; additionally, cluster-level correction for multiple comparison was employed, and clusters surviving this procedure are highlighted below and in
Activity differences were most prominent in occipito-temporal and parietal areas but were also present in the right IFJ and pre-SMA (note that only the large parietal and occipital clusters survived strict cluster-level correction for multiple comparisons).
Anatomical structure | Hemi-sphere | Cluster size [voxel] | T-Value | Peak coordinates MNI (mm)x y z |
|
||||
Inferior frontal junction (IFJ) | R | 124 | 4.73 | 42 8 36 |
Pre-SMA | R | 83 | 4.22 | 4 20 48 |
|
||||
Inferior parietal lobule (IPL) |
R | 1651 | 6.37 | 30 -56 50 |
Inferior parietal lobule (IPL) |
L | 647 | 5.06 | -54 -34 36 |
Supramarginal gyrus | L | 52 | 4.18 | -62 -18 26 |
|
||||
Middle occipital gyrus (MOG) |
L | 576 | 8.51 | -50 -76 0 |
Middle occipital gyrus (MOG) |
R | 750 | 7.15 | 46 -72 0 |
Main local maxima. Data are thresholded at T>3 (uncorrected), with a cluster-level of k = 10.
(*) denotes clusters that are significant after correction for multiple comparisons on the cluster level.
While the above analyses provide a formal brain-wide test for which areas are activated during Stop-trials (as compared to Go-trials) even when these stimuli are task-irrelevant, the relationship to activity triggered by task-relevant Stop-trials is hard to evaluate without direct reference to these other trial types. Importantly, activity in some areas might not be triggered in an all-or-none fashion by Stop-stimuli. Rather it is possible that some areas may be activated in a graded fashion, wherein a certain amount of activity is triggered even by task-irrelevant Stop-stimuli, which gets more pronounced if those stimuli are in fact task-relevant. A good example process for which such a pattern might be present is attentional capture, although other processes might also be engaged in a graded fashion. Specifically, attentional capture has an automatic component that does not depend on task-relevance. However, attentional capture effects can get enhanced if the capturing stimulus is furthermore relevant to the task. Accordingly, in order to provide a more detailed analysis of the contributions of different key areas to the processing of task-relevant and task-irrelevant Stop-trials, we performed additional analyses within ROIs that were delineated based on both kinds of Stop-trials. Another advantage of such an ROI analysis is that voxel-wise comparisons are necessarily quite conservative, whereas ROI-analyses can focus on the most relevant areas derived from orthogonal contrasts, thus ameliorating the multiple-testing problem. In order to allow for an unbiased comparison between the Stop-trials from the different task blocks, ROIs were selected on the basis of a contrast comparing all Stop-trials from the two task blocks (i.e., Stop-relevant and Stop-irrelevant) against all Go-trials from those tasks. The present ROI analysis is related to an ROI analysis of some of these data applied in our earlier paper
Due to the very robust and widespread activations that were identified by this contrast, we opted for a comparatively conservative voxel-level threshold (FDR corrected p<0.01; k = 50). This contrast yielded eight activation clusters (note that all clusters furthermore survived cluster-level correction for multiple comparisons; note also that the present set of areas is very similar to other studies that have compared Stop-trials with Go-trials, which presumably indicates that activity levels in Stop-relevant Stop-trials were sufficient to identify typical stopping-related areas even when averaged with Stop-irrelevant Stop-trials that may have failed to elicit substantial activity in some of these areas.). The respective maxima of seven of these were highly distinctive and were thus directly used for further analysis (see
Areas in the lateral occipital cortex displayed a pattern of activity that mostly reflected sensory stimulation (i.e., no significant difference between task-relevant [average of SST and UST] and task-irrelevant Stop-trials, along with substantial response to Go-trials; dark blue bars). Bilateral responses in the inferior parietal lobule appeared to mainly reflect attentional capture by the infrequent Stop-stimulus, irrespective of its task relevance (i.e., no significant difference between task-relevant [average of SST and UST] and task-irrelevant Stop-trials, accompanied by a weak response to Go-trials; light blue bars). The only area in the posterior part of the brain that reflected the task-relevance of Stop-stimuli was in the superior temporal gyrus (STG) close to the TPJ (red bars; significantly larger response to Stop-relevant Stop-trials [average of SST and UST] than Stop-irrelevant Stop-trials, along with a weak response to Go-trials). Error bars depict the SEM; activity estimates for Go-trials are represented without a fill color to set them apart from Stop-trials and to indicate that the ROI definition favored Stop-trials so that statistical comparisons including Go-trials were avoided.
None of the frontal areas displayed strong activity estimates for Go-trials. All frontal areas except for IFJ displayed a clear difference between the average response to the task-relevant Stop-trials and the response to the task-irrelevant ones (red bars). Additional significant differences between the individual Stop-trial types are indicated in the bar plots (*<0.05; **<0.01; ***<0.001; two-tailed; error bars depict the SEM.). Right IFJ displayed a somewhat different pattern, in that SST responses were larger than both the UST and Stop-irrelevant Stop-trial responses, but that the average response to Stop-relevant Stop-trials (i.e., averaged across SSTs and USTs) was not larger than that to Stop-irrelevant ones. Error bars depict the SEM; activity estimates for Go-trials are represented without a fill color to set them apart from Stop-trials and to indicate that the ROI definition favored Stop-trials so that statistical comparisons including Go-trials were avoided.
Anatomical structure | Hemi-sphere | Cluster size [voxel] | T-Value | Peak coordinates MNI (mm)x y z |
|
||||
Inferior frontal gyrus (IFG)/anterior insula | R | 393 | 8.16 | 44 18 -2 |
Inferior frontal junction (IFJ) | R | 259 | 7.31 | 50 14 34 |
Middle frontal gyrus (MFG) | R | 95 | 7.24 | 40 42 28 |
Pre-SMA | R | 300 | 6.46 | 4 20 48 |
Anterior insula | L | 274 | 6.15 | -36 20 -2 |
|
||||
Inferior parietal lobule (IPL) | R | 3613 |
8.62 | 36 -44 44 |
Inferior parietal lobule (IPL) | L | 1471 | 7.06 | -34 -48 42 |
|
||||
Temporo-parietal junction/ superior temporal gyrus (TPJ/STG) | R | 3613 |
8.21 | 50 -54 8 |
|
||||
Middle occipital gyrus (MOG) | L | 754 | 10.01 | -50 -76 0 |
Middle occipital gyrus (MOG) | R | 3613 |
9.06 | 46 -72 0 |
Main local maxima. Data are thresholded at p<0.01 (FDR-corrected), with a cluster-level of k = 50.
(#) the three main local maxima were taken from this larger cluster subtending the right occipito-temporal and parietal cortex.
Among these ROIs, we predicted finding three distinctive activity profiles for the different conditions: (1) Sensory-driven activity that would be present for Go-trials and further enhanced for Stop-trials (due to the extra sensory stimulation), but not differing significantly between Stop-relevant and Stop-irrelevant Stop-trials (dark blue bars in
Analyses of the five posterior ROIs revealed three different activity profiles that were largely symmetrical for the bilaterally activated areas (
Of the five frontal clusters identified, all areas except the right IFJ displayed qualitatively the same pattern of activity. More specifically, these areas did not respond strongly to Go-trials in either task block, nor to the Stop-trials from the Stop-irrelevant blocks, but responded strongly to Stop-relevant Stop-trials. In all these areas the Stop-relevant Stop-trials yielded significantly stronger activations than the Stop-irrelevant Stop-trials (all p<0.05; see
The present fMRI study aimed at delineating the neural processes that are involved in the context of response inhibition during the Stop-signal task and to distinguish different neural underpinnings of the various cognitive processes engaged during such tasks. In an attempt to identify areas that respond to the rare and salient sensory stimulation of Stop-trials in an automatic fashion, we found that occipital and inferior parietal areas respond more strongly to Stop-trials than to Go-trials, even if the Stop-stimuli are completely task-irrelevant. Interestingly, this analysis also identified clusters in the right IFJ and the right pre-SMA that responded in a similar fashion, albeit only on a comparatively lenient uncorrected significance level. An additional ROI analysis that focused on the comparison of neural responses to Stop-trials from task blocks in which the Stop-stimuli were versus were not task-relevant identified three major activity profiles for different cortical areas. These profiles indicate a hierarchy in which pure sensory processing is mostly restricted to occipital areas, whereas some degree of automatic attentional capture by rare Stop-stimuli regardless of their task-relevance occurs in the inferior parietal lobules. In contrast, in the third profile, activity in a wide range of frontal areas and the right TPJ/STG was prominent only for Stop-trials that were task-relevant. These findings provide an important step towards establishing a framework in which sensory processes, bottom-up attentional processes, and top-down control functions can be attributed to specific portions of the wider cortical network that is typically associated with response inhibition during the Stop-signal task.
At least three recent publications have highlighted the difficulty of distinguishing processes directly involved in response inhibition from those related to the attentive processing of the relevant stimuli, focusing on the role of right IFG (
In addition to an attentional account of the function of the IFG, the ubiquity of its activation across different tasks might also relate to recent accounts that assign more global control functions to this area (for recent reviews, see
Importantly, none of the studies mentioned above included a condition in which Stop-stimuli were entirely task-irrelevant, thus leaving open the question whether the right IFG and other areas would respond to Stop-stimuli even when they are entirely task-irrelevant. Such activation could arise by means of bottom-up attentional capture by the Stop-stimuli simply due to their rarity and physical salience. The present study identified activation by task-irrelevant Stop-stimuli bilaterally in occipital areas and in IPL, as well as in the right IFJ and pre-SMA (albeit on a more lenient, uncorrected significance threshold). Other frontal areas such as the IFG, however, did
Thus, the area that appeared to be most consistently activated by the mere rarity and salience of Stop-stimuli, irrespective of their task relevance, was the bilateral IPL, thus arguing against notions that have ascribed a direct involvement of this area in response inhibition (e.g.,
For areas that respond to Stop-trials only when they are task-relevant, however, response inhibition processes cannot be easily distinguished from those related to enhanced attentive processing of the Stop-stimuli, or those related to increased response control demands that are not inhibitory (e.g.,
While the right IFG and pre-SMA have commonly been discussed in the context of response inhibition, other frontal areas such as the right IFJ, MFG, and left anterior insula have also been frequently implicated in such tasks. Considerably less is known about the functional role of these other areas, however. For example, the right IFJ has been argued to play a role in detecting infrequent NoGo-trials in a Go-NoGo task, including because it has been shown to also respond to an additional type of Go-trials during a Go-NoGo task when they are presented as infrequently as the NoGo-trials
The current results also revealed a large cluster of activity in the right MFG during task-relevant Stop-trials, in line with results from earlier response-inhibition paradigms
Given recent reports of right IFG activation by control stimuli that do not require response inhibition (e.g.,