Cerebral Cortex Advance Access published online on August 28, 2007
Cerebral Cortex, doi:10.1093/cercor/bhm147
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
Decomposing the Neural Correlates of Antisaccade Eye Movements Using Event-Related fMRI
1 Centre for Neuroimaging Sciences, Institute of Psychiatry, King's College London, UK, 2 Department of Psychology, Institute of Psychiatry, King's College London, UK, 3 Institut für Psychologie, Humboldt-Universität Berlin, Germany
Address correspondence to Ulrich Ettinger, Centre for Neuroimaging Sciences, PO Box 89, Institute of Psychiatry, King's College London, DeCrespigny Park, London SE5 8AF, UK. Email: u.ettinger{at}iop.kcl.ac.uk.
 |
Abstract |
|---|
 Top Abstract IntroductionMaterials and MethodsResultsDiscussionFundingReferences |
|---|
The antisaccade task is a model of the conflict between an unwanted reflexive response (which must be inhibited) and a complex volitional response (which must be generated). The present experiment aimed to investigate separately the neural correlates of these cognitive components using a delayed saccade paradigm to dissociate saccade inhibition from generation. Seventeen healthy volunteers completed event-related functional magnetic resonance imaging at 1.5 T during saccades to and away from a peripheral visual target (prosaccades and antisaccades, respectively). Saccades were requested in response to an auditory go signal on average 12 s after peripheral target appearance. It was found that the right supramarginal gyrus showed significantly greater activation during the inhibition phase than the generation phase of the paradigm for both antisaccade and prosaccade trials, suggesting a role in saccade inhibition or stimulus detection. On the other hand, the right lateral frontal eye field and bilateral intraparietal sulcus showed evidence of selective involvement in antisaccade generation. Ventrolateral and dorsolateral prefrontal cortices showed comparable levels of activation in both phases of the task. These areas likely fulfill a more general supervisory role in the volitional control of eye movements, such as stimulus appraisal, task set, and decision making.
Key Words: cognition • frontal • human • infrared oculography • neuroimaging • oculomotor • parietal
|
Introduction |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionFundingReferences |
|---|
In a typical antisaccade task, the participant first fixates on a central visual stimulus. When the stimulus is then abruptly moved to a peripheral location, the participant is required to look immediately in the opposite direction (Hallett 1978
). This intensely studied task has emerged as an excellent model of the conflict between a prepotent response (i.e., a reflexive saccade to the stimulus), which must be inhibited, and a volitional, spatially complex response (i.e., an antisaccade), which must be generated instead (Munoz and Everling 2004; Hutton and Ettinger 2006).
Successful negotiation of the incongruent stimulus–response (S-R) mapping required in antisaccades involves a number of cognitive processes, including reflexive response inhibition and volitional response generation. Reflexive response inhibition demands the suppression of the visual grasp reflex that occurs when a peripheral visual target appears. Antisaccade generation involves encoding the visuospatial signal of the target through a covert attentional shift and converting it into a motor vector triggering an antisaccade. There is evidence that the ability to suppress a reflexive response is intimately linked to the ability to generate a volitional antisaccade response and that the 2 processes take place in parallel (Massen 2004; Reuter and Kathmann 2004).
The neural correlates of antisaccade performance mirror these complex cognitive requirements. Neuroimaging studies of brain function during blocks of antisaccades (compared with blocks of reflexive saccades or fixation) have shown activation in a fronto-parieto-subcortical network of frontal eye fields (FEFs), supplementary eye fields (SEFs), dorsolateral prefrontal cortex (DLPFC), ventrolateral prefrontal cortex (VLPFC), posterior parietal cortex, supramarginal gyrus (SMG), striatum, thalamus, and cerebellum (O'Driscoll et al. 1995; Sweeney et al. 1996; Müri et al. 1998; McDowell et al. 2002; Matsuda et al. 2004; Tu et al. 2006). However, the specific contributions of these different areas to response inhibition and response generation are not well understood.
Studies of acquired brain lesions have the potential to address this question. Pierrot-Deseilligny et al. (2002) have argued that patients with lesions to the DLPFC (and relevant white matter connections to the striatum; Ploner et al. 2005) display increased rates of erroneous reflexive saccades to the target, whereas patients with FEF lesions display prolonged antisaccade latencies. These studies suggest that the DLPFC mediates reflexive saccade inhibition, whereas the FEFs generate volitional antisaccades.
Event-related functional magnetic resonance imaging (fMRI) studies also have the potential to dissociate the neural mechanisms of specific antisaccade components. These studies are at least partly compatible with the lesion literature, suggesting that increased activation in the FEF underlies successful antisaccade preparation and execution (Connolly et al. 2002; Cornelissen et al. 2002). A similar role has been attributed to the presupplementary motor area, an area located anterior to the SEF (Curtis and D'Esposito 2003). Concerning the cortical correlates of reflexive saccade inhibition, a recent fMRI study demonstrated a significant role of the right superior frontal sulcus, right SMG, and posterior cingulate sulcus (Brown et al. 2006).
The present study employed a "delayed antisaccade" task (Reuter and Kathmann 2004) to gain a better understanding of the neural correlates of antisaccade subcomponents. Whereas in the standard antisaccade task as outlined above, the process of reflexive response inhibition and volitional response generation are likely to take place in parallel (Massen 2004), the delayed antisaccade task introduces a temporal gap between the presentation of the peripheral target and an auditory go signal to perform a volitional antisaccade. The task thus allows the temporal dissociation of the inhibition of an unwanted response on the one hand (phase 1: inhibition) and the generation of a volitional saccadic response (phase 2: generation) on the other hand; using fMRI, the neural correlates underlying these 2 phases can thus be studied separately.
In this study, brain activation was first measured using event-related fMRI during performance of a standard antisaccade task. This task requires an antisaccade immediately following peripheral target appearance and temporally confounds response inhibition and generation. Therefore, the standard antisaccade task is expected to yield maximal antisaccade-related activation. Activation levels in brain areas identified using this task were then extracted from the 2 phases of the delayed antisaccade task (inhibition and generation). We also used a corresponding delayed prosaccade task (Hutton et al. 2002; Reuter et al. 2007). This task is identical to the delayed antisaccade task but requires a saccade toward the peripheral target after the delay. The key difference to the delayed antisaccade task is the lower demands on visuospatial remapping due to the existence of a visual saccade target in the delayed prosaccade task.
This design thus allows us to address differences between areas in their relative contribution to the antisaccade or prosaccade task, differences in relative contribution to response inhibition or response generation, and interaction effects (e.g., selective involvement in response generation for the antisaccade task only). In this work, we refer to the comparison of prosaccade and antisaccade tasks as a main effect of SaccadeTask and to the comparison between response inhibition and response generation as a main effect of DelayCondition. On the basis of the human lesion literature (Pierrot-Deseilligny et al. 2002), we hypothesized a role of the DLPFC in saccadic inhibition. On the basis of human lesion and fMRI studies, we expected a role of the FEF in antisaccade generation (Connolly et al. 2002; Pierrot-Deseilligny et al. 2002).
|
Materials and Methods |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionFundingReferences |
|---|
Participants
Nineteen participants underwent fMRI. Two participants had to be excluded, one due to scanner malfunction and one due to excessive movement, leaving a final sample size of N = 17 (7 females). Participants were right-handed (Oldfield 1971), healthy volunteers aged 20–40 years (mean = 27.82, SD = 5.15) recruited from amongst university staff and students. All participants provided written informed consent. The study was approved by the local research ethics committee.
fMRI Data Acquisition
Scanning was performed on a 1.5-T General Electric Signa Advantage MRI scanner. Participants wore headphones to reduce impact of scanner noise. Head movements were minimized using foam padding and a forehead band. A localizer scan for placing the volume of interest and a high-resolution structural scan for image coregistration were first acquired.
Twenty-two 
-weighted MR echo planar images of the whole brain showing the blood oxygenation level–dependent (BOLD) response were collected using interleaved slice acquisition. A quadrature headcoil was used for radio frequency transmission and reception. Images were aligned in parallel to the intercommissural plane (AC–PC line). Image parameters were as follows: repetition time = 2 s, echo time = 40 ms, field of view = 24 cm, flip angle = 80°, in-plane resolution = 3.75 x 3.75 mm, slice thickness = 5 mm, and inter-slice gap = 0.5 mm. Images (1236) were acquired in one continuous run with a total scanning time per subject of 41 min and 12 s. Four 2-s scans of dummy data acquisition were performed before each run to allow for the establishment of steady-state longitudinal magnetization; these scans were not used in the analysis.
Task Design and Stimuli
The saccade task employed an event-related design with an average epoch length of 12 s, allowing the BOLD signal to return to baseline following each change due to stimulus occurrence or saccade execution (Bandettini and Cox 2000; Donaldson and Buckner 2004). The beginning of each trial was triggered by the scanner. The task was written in Visual Basic.
The task interleaved standard and delayed trials of antisaccades and prosaccades with trials of fixation (Fig. 1). Participants were unaware of trial type (standard and delayed) at the beginning of the trial. Standard trials consisted of 2 epochs, with epoch durations jittered with 1-s increments. First, a central fixation target (0°) appeared for 11, 12, or 13 s (average = 12 s). Then a peripheral target appeared at ±8° for 11, 12, or 13 s (average = 12 s) while the central target remained on the screen. With appearance of the peripheral target, an auditory cue (a 200 ms enveloped 440 Hz sine wave) was sounded, signaling the requirement to make a saccade. Delayed trials consisted of 3 epochs, with epoch durations again jittered with 1-s increments. First, a central fixation target appeared for 11, 12, or 13 s (average = 12 s). Then a peripheral target appeared at ±8° for 23, 24, or 25 s (average = 24 s) while the central target remained visible (phase 1). The auditory cue occurred 11, 12, or 13 s (average = 12 s) after peripheral target onset, providing the signal to make a saccade (phase 2). Phase 1 and phase 2 lasted on average 12 s each. A standard trial was always 24 s and a delayed trial was always 36 s of duration. A fixation trial consisted of the central target presented for 24 s without any peripheral targets or auditory stimuli.
|
), is presented at the beginning of each from for 200 ms. Time flow is from top left to bottom right. The 3 panels on the right represent recordings with the infrared limbus tracker corresponding to the trials in the frames on the left (fixation, standard antisaccade, and delayed prosaccade).In antisaccade trials, the central fixation dot was red, in prosaccade trials it was green, and in fixation trials it was black (see e.g., Connolly et al. 2002
). The peripheral target was always black, and the background was gray. Participants were instructed to look to the mirror image location of the peripheral target when the auditory cue occurred on antisaccade trials and to look to the peripheral target when the auditory cue occurred on prosaccade trials. On fixation trials, participants were instructed to keep their eyes on the central target. There were 14 trials of each type. The number of trials in this experiment was chosen in order to minimize effects of boredom and fatigue on performance and brain activation. In a pilot study, we observed that a larger number of trials with shorter trial durations (and similar overall experiment duration) resulted in considerably reduced power to detect activation in oculomotor areas compared with the present design. Right and left targets were equally frequent for each saccade type. There were 26 fixation trials. Trials were randomly interleaved and presented in 8 blocks separated by 24 s of blank screen (rest condition).
Approximately 1 week before the fMRI scan, all participants performed an off-line delayed saccade task. The off-line task contained 80 trials of standard and delayed pro- and antisaccades (20 trials per trial type). The off-line task was faster paced than the fMRI task, in line with previous delayed saccade tasks (Reuter and Kathmann 2004). In standard trials, the central fixation point was presented for a random duration of 1.5–2.5 s, followed by a peripheral target for 1 s (with simultaneous onset of a 200-ms tone). In delayed trials, the central fixation point was presented for a random duration of 1.5–2.5 s, followed by a peripheral target for 1.5–2.5 s, and finally a 200-ms tone, whereas the peripheral target remained on for another 800 ms. Peripheral target locations (±6° and ±12°) were balanced across trial types.
In addition to this off-line delayed saccade task, all participants were shown approximately 10 trials of the actual fMRI task on the day of the fMRI scan to familiarize them with the task requirements.
Oculomotor Data Acquisition and Analysis
Horizontal movements of the left eye were recorded using an MRI-compatible infrared oculographic limbus tracker (MR-Eyetracker, CRS Ltd., Rochester, UK). The MR-Eyetracker uses fibre optic cables for guiding infrared light between the hardware in the control room and an eye piece fixed to the headcoil and placed beneath the participant's left eye. The system has a minimum spatial resolution of 0.2° and a horizontal range of ±20°. Signals were digitized using 12-bit analogue-to-digital converter (data translation DT9802) and sampled at 500 Hz. A 3-point calibration (±8° and 0°) was performed before the scan.
Oculomotor data during fMRI were collected to 1) obtain saccadic latencies that were used to model the BOLD response (see below) and 2) exclude error and artifact trials. Excluded trials were trials with recording artifacts, trials in which the participant did not respond, or trials where the participant committed an incorrect response (directional errors and early responses). Incorrect trials were modeled in the fMRI analysis but not analyzed separately as they were too infrequent to yield sufficient power for formal analysis. Saccades were identified in Eyemap (AMTech GmbH, Weinheim, Germany) using minimum latency (100 ms) and amplitude (1°) criteria.
The statistical analysis of oculomotor data focused on the latency of correct saccades in the delayed prosaccade and antisaccade tasks using paired samples t-tests. We also used paired samples t-tests to compare the frequency of excluded trials in the delayed antisaccade and prosaccade tasks to check whether similar numbers of trials were excluded from these 2 tasks.
fMRI Data Analysis
Analysis was carried out using SPM2 (http://www.fil.ion.ucl.ac.uk/spm/) implemented in Matlab 6. Images were first slice-time corrected to reflect the temporal sequence of image acquisition. Preprocessing consisted of 3 steps. First, motion correction was carried out by realigning each subject's images to the first image in the time series. Second, images were transformed into standard space using affine registration and nonlinear transformations and the Montreal Neurological Institute (MNI) template in SPM2. Third, images were spatially smoothed with an 8-mm Gaussian full width at half maximum filter and high-pass filtered.
Data were analyzed with a general linear model in SPM2, involving 2 steps. First, each event of interest was investigated at the single-subject level, by convolving the event with a canonical hemodynamic response function. In this step, a multiple regression analysis was performed within SPM2 at each voxel using a model consisting of BOLD signal as the dependent variable, individual events as predictors (regressors) and a constant term related to overall mean signal. The analysis produces parameter estimates for each class of event at each voxel, which can then be compared using linear contrasts to produce images of specific experimental questions. Events of interest were modeled separately for antisaccades and prosaccades and included the generation of a saccade in the standard task, the appearance of the peripheral target in delayed saccade trials (phase 1/inhibition), and the generation of a saccade in delayed saccade trials (phase 2/generation). The saccade onset was modeled for phase 2 in order to temporally optimize the analysis for the neural correlates of saccade generation. Similar results were obtained when our model was based on the auditory go signal onset, the difference being a slight reduction in activity within saccade-related areas together with a slight increase in activity within auditory-related areas. The second 12-s segment of fixation trials, that is, a period of pure fixation without peripheral stimuli or saccades during or immediately preceding it, was modeled for use as a baseline.
Contrast images of interest from the single-subject analysis were then combined at the group level in a random effects analysis to produce SPMs of group activation. A 1-sample t-test of the contrast standard antisaccade versus fixation baseline was carried out to identify the standard antisaccade network. As argued above, this condition compounds the processes of reflexive response inhibition and volitional response generation and is therefore expected to yield maximal brain activation in the corticosubcortical antisaccade network. Areas that showed significant activation (P < 0.05 at the corrected cluster level) were included in further analysis. Additionally, as activation in the key areas of left medial FEF (medFEF) and right DLPFC (but no other areas) was just below threshold (P < 0.1 at the corrected cluster level), these areas were also included.
MNI coordinates of identified areas given in SPM2 were converted to Talairach coordinates (Talairach and Tournoux 1988) using a nonlinear transformation algorithm (http://imaging.mrc-cbu.cam.ac.uk/imaging/MniTalairach). Brodmann areas (BAs) and anatomical regions were identified on the basis of the Talairach atlas (Talairach and Tournoux 1988).
Further analysis was then done on subject-specific parameter estimates extracted from the peak voxel within each region of interest (Table 2) using the MarsBAR toolbox in SPM2. This strategy, similar to previous work (Connolly et al. 2002; Curtis and D'Esposito 2003; Sauer et al. 2006), allows an anatomically focused analysis of the role of specific brain areas in delayed saccade task performance. Parameter estimates were entered into a 2 x 2 SaccadeTask-by-DelayCondition analysis of variance (ANOVA) in the statistical package for the social sciences (SPSS) Release 13 (SPSS Inc., Chicago, IL). The "SaccadeTask" factor consisted of 2 levels (antisaccade and prosaccade). The "DelayCondition" factor also consisted of 2 levels (phase 1/inhibition and phase 2/generation). Significant interactions were followed up by paired samples t-tests.
|
Results |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionFundingReferences |
|---|
Saccadic Performance
Descriptive statistics of saccadic variables during fMRI are presented in Table 1. The percentage of excluded trials did not differ between delayed antisaccade and prosaccade trials (t = 1.61, degrees of freedom [df] = 16, P = 0.13). The latency of delayed antisaccades was nonsignificantly longer compared with delayed prosaccades (t = –1.98, df = 16, P = 0.07).
|
fMRI
Standard Antisaccade Experiment
Table 2 and Figure 2 show the pattern of activation associated with performance of standard antisaccades. Activation was observed in bilateral medFEFs, right lateral FEF (latFEF), SEF, DLPFC, VLPFC, right SMG, bilateral intraparietal sulcus (IPS), right thalamus, left caudate, and bilateral visual cortex (BA17). Within the IPS, we identified anterior (aIPS) and posterior (pIPS) activation clusters bilaterally following previous studies (Corbetta et al. 1998; Brown et al. 2004). The locales of the areas identified here are compatible with previous analyses of saccade-related brain areas (Grosbras et al. 2005; Amiez et al. 2006).
|
|
All areas except DLPFC and left medFEF were significant at the corrected cluster level (P < 0.05). Activations for DLPFC (P = 0.07) and left medFEF (P = 0.06) reached trend level (P < 0.1) at the corrected cluster level. However, because these areas were expected on the basis of our hypotheses and previous data (McDowell and Clementz 2001
; Munoz and Everling 2004; Hutton and Ettinger 2006), both DLPFC and left medFEF were included in further analyses. No other areas reached trend level (P < 0.1) at the corrected cluster level. Delayed Saccade Experiment
The ANOVA results for all areas of interest are presented in Table 3. Results for SMG, latFEF, IPS, and DLPFC are presented in Figures 3–6.
|
|
Significant ANOVA main effects of DelayCondition (inhibition and generation) was found for right SMG, right medFEF, and left medFEF. These effects indicate greater activation during the inhibition than the generation phase of delayed saccades in SMG and medFEF. Additionally, a significant main effect of SaccadeTask was observed for right and left medFEF (but not SMG), indicating greater activation during antisaccade than prosaccade trials.
Right latFEF showed a significant ANOVA main effect of SaccadeTask, indicating greater activation during antisaccade than prosaccade trials, and a significant SaccadeTask-by-DelayCondition interaction (Fig. 4). The interaction indicated that activation during antisaccade generation was greater than during prosaccade generation (t = 4.20, df = 16, P = 0.001), whereas antisaccade and prosaccade inhibition did not differ (t = 0.97, df = 16, P = 0.35). Antisaccade generation showed greater activation than antisaccade inhibition (t = –2.52, df = 16, P = 0.02), an effect that was not seen on prosaccade trials (t = 0.94, df = 16, P = 0.36).
|
Similar to the right latFEF, IPS showed evidence of selective involvement in antisaccade generation (Fig. 5). In the ANOVA, there were significant main effects of SaccadeTask on left aIPS and left pIPS, indicating greater activation during antisaccade than prosaccade trials. There was a significant effect of DelayCondition on right pIPS and a trend-level effect on left pIPS, indicating greater activation during saccade generation than inhibition. Finally, there were significant interaction effects for right and left aIPS and right pIPS, but not for left pIPS. Significant interactions were followed up by paired-samples t-tests, yielding the following pattern.
|
For right aIPS, there were trends toward increased activation during antisaccade than prosaccade generation (t = 1.86, df = 16, P = 0.08) and during antisaccade generation compared with antisaccade inhibition (t = –1.99, df = 16, P = 0.06). There were no differences between antisaccade and prosaccade inhibition or between prosaccade inhibition and prosaccade generation (both P > 0.21).
For left aIPS, there was significantly greater activation during antisaccade than prosaccade generation (t = 3.73, df = 16, P = 0.002) and during prosaccade inhibition than prosaccade generation (t = 2.17, df = 16, P = 0.045). There were no differences between antisaccade and prosaccade inhibition or between antisaccade inhibition and antisaccade generation (both P > 0.23).
For right pIPS, there was significantly greater activation during antisaccade than prosaccade generation (t = 3.39, df = 16, P = 0.004) and during antisaccade generation than antisaccade inhibition (t = –5.18, df = 16, P < 0.001). There were no differences between antisaccade and prosaccade inhibition or between prosaccade inhibition and prosaccade generation (both P > 0.14).
No significant main or interaction effects were observed for DLPFC (Fig. 6), VLPFC, and right thalamus (all P > 0.13).
|
SEF and left caudate showed significant main effects of SaccadeTask (but no other main or interaction effects), indicating greater activation during antisaccade than prosaccade trials irrespective of DelayCondition.
For bilateral visual areas (BA17), there were highly significant effects of DelayCondition indicating greater activation during generation than inhibition. There were no significant effects of SaccadeTask or interactions.
|
Discussion |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionFundingReferences |
|---|
The present experiment aimed to decompose the neural correlates of the cognitive subcomponents of antisaccades, that is, it separated the volitional generation of antisaccades from reflexive response inhibition. To do so, we employed a modified antisaccade task, which introduces a delay interval between peripheral target appearance and saccadic response. Our a priori assumption was that the standard antisaccade task, which requires an antisaccade eye movement immediately in response to the peripheral stimulus appearance, compounds these processes and is thus associated with activation of all relevant antisaccade areas. Therefore, the standard antisaccade condition was used as an anatomic localizer task to extract coordinates of brain regions for analysis of the delayed saccade task.
The standard antisaccade task activated an extensive fronto-parieto-subcortical network, consisting of bilateral medFEF, right latFEF, SEF, right DLPFC, right VLPFC, right SMG, IPS, thalamus, caudate, and bilateral visual areas. This network is compatible with results from standard antisaccades in previous block-design fMRI or positron emission tomography experiments (O'Driscoll et al. 1995; Sweeney et al. 1996; Müri et al. 1998; Matsuda et al. 2004; Tu et al. 2006).
In order to provide an anatomically driven investigation of the roles of these areas in saccadic response inhibition and generation, activation levels were extracted and entered into statistical analysis of the SaccadeTask-by-DelayCondition design. The following discussion summarizes the key findings of this experiment by area and aims to detail the role of each area in response inhibition and response generation. We will first discuss findings relating to response inhibition (SMG), then response generation (latFEF and IPS), then areas that play a role in both functions (DLPFC, VLPFC, and thalamus), and finally other areas (medFEF, SEF, caudate, and visual areas).
Supramarginal Gyrus
Activation levels in the right SMG were significantly stronger in relation to response inhibition in the delayed saccade task than during response generation (main effect of DelayCondition).
The SMG is an area in the inferior parietal lobule. SMG activation was previously reported in relation to standard antisaccades (Matsuda et al. 2004) and during inhibitory tasks, such as prepulse inhibition (Kumari et al. 2003) and inhibition of return (Lepsien and Pollmann 2002). Other studies have suggested a role of inferior parietal lobule areas neighboring SMG in BA40 during an antisaccade task (Chikazoe et al. 2007) and during tasks involving working memory, interference, inhibition, and spatial processing (Colby and Goldberg 1999; Sylvester et al. 2003; Hester et al. 2004; Brass et al. 2005; Wager et al. 2005).
A recent fMRI study of antisaccades also provided evidence of a role of SMG in the inhibition of unwanted reflexive saccades (Brown et al. 2006). In that previous study, right SMG activation was significantly stronger during a no-go condition (comparable with phase 1 of our experiment) than a reflexive saccade condition.
However, it should be noted that the inhibition phase of the current delayed saccade experiment is likely to have required a significant attentional component. The peripheral target was not an irrelevant distractor; instead it was a salient target providing the necessary visuospatial input for a later saccade. Therefore, it may be argued that at least some of the activation seen in SMG in this condition may be due to stimulus detection and covert attentional processing of the peripheral target.
Evidence compatible with a role of SMG in attention comes from studies showing that covert attentional shifts activate frontoparietal networks including SMG (Corbetta et al. 1998; Perry and Zeki 2000). Additionally, patients with right SMG lesions show hemineglect, a failure to attend to the contralesional half of the visual environment (Posner et al. 1984). It has also been suggested that SMG might mediate a nonspatial attentional function, such as stimulus detection or alerting other areas to the appearance of a salient stimulus irrespective of precise spatial location (Perry and Zeki 2000). In the present design, we cannot dissociate the contributions of inhibition and stimulus detection to SMG activation. It is interesting to note however that inhibition of an unwanted reflexive response (as in the visual grasp reflex) is typically confounded with stimulus detection. Future work is needed to further separate these processes.
Activation in the left SMG did not reach statistical significance during the standard antisaccade task. This finding might be compatible with evidence from previous studies showing a preferential right hemisphere involvement in inhibitory tasks (Garavan et al. 2006).
LatFEF and IPS
Phase 2 (generation) of the delayed saccade paradigm allowed the investigation of the neural correlates of saccade execution in the absence of any additional processes that take place under standard saccade task conditions, such as the requirement to inhibit a reflexive response to a novel visual stimulus. The right latFEF and bilateral IPS showed activation patterns consistent with a role in generating a volitional antisaccadic response.
Human lesion studies have pointed to an important role of the FEF in antisaccade generation (Rivaud et al. 1994). Pierrot-Deseilligny et al. (1991) studied patients with a variety of lesions in the precentral sulcus, including the latFEF site identified in our study. FEF lesions in that study led to prolonged antisaccade latencies but not increased error rates (Pierrot-Deseilligny et al. 1991). Our latFEF data are also compatible with previous neuroimaging studies. Brown et al. (2004) found greater latFEF activation during a memory-guided saccade compared with a reflexive saccade in a task that involved a working memory delay interval between a brief target appearance and saccade generation.
The IPS, particularly in the right hemisphere, also showed activation patterns suggestive of a role in antisaccade generation. The IPS, which contains the human parietal eye field, was found to be a key structure in saccade generation in a meta-analysis of functional neuroimaging data (Grosbras et al. 2005). More recently, an event-related fMRI study observed IPS activation in relation to prosaccade and antisaccade generation (Brown et al. 2006). Importantly, in the study of Brown et al. (2006) (as in ours), IPS activation was greater during antisaccades than prosaccades. Brown et al. (2006) concluded that the activation observed during antisaccades may thus not be uniquely attributed to simply triggering a saccade but might reflect visual attention and visuospatial remapping of the target into an oculomotor response (Brown et al. 2006). Our IPS data are in agreement with the interpretation of Brown et al.(2006).
Our findings may also be reconciled with studies of saccadic eye movements in patients with chronic IPS lesions. Machado and Rafal (2004a) observed prolonged antisaccade latencies away from contralesional targets in patients with chronic IPS lesion. This effect showed a high level of behavioral specificity, as IPS lesion patients did not show increased rates of contralesional reflexive errors on the antisaccade task or prolonged latencies toward contralesional targets on a reflexive saccade task (Machado and Rafal 2004a, 2004b).
Oculomotor areas in the posterior parietal lobe are known to trigger saccades through direct projections to the superior colliculus (SC) (Leigh and Zee 1999). Additionally, the role of IPS in volitional saccade generation in our study may include processes such as visuospatial attention, access of a spatial code in working memory, and the transformation of a visual signal into motor action (Goodale and Milner 1992; Calton et al. 2002; Curtis et al. 2004). It is of interest to note in this context that in monkey, lateral intraparietal (LIP) neurons, corresponding to human IPS, appear to show a role in contralateral attentional (stimulus coding) processing rather than saccade generation per se (Gottlieb and Goldberg 1999). In the study by Gottlieb and Goldberg (1999), it was also found that LIP activation was greater for antisaccade than prosaccade generation.
It should be noted that antisaccade latencies were nonsignificantly longer than prosaccade latencies in the delayed saccade experiment (P = 0.07). The increased levels of brain activation during generation of delayed antisaccades compared with prosaccades in latFEF and IPS are thus paralleled by these differences. This difference is likely to be due to the greater attentional and visuospatial demands of antisaccades (Olk and Kingstone 2003), factors which may also underlie the differences in brain activation.
DLPFC and VLPFC and Thalamus
The standard antisaccade task revealed activation in the DLPFC and VLPFC. For both areas, activation was restricted to the right hemisphere, consistent with previous evidence of more pronounced right than left frontal involvement during antisaccades and other complex inhibitory tasks (Walker et al. 1998; McDowell et al. 2002; DeSouza et al. 2003; Ettinger et al. 2005; Garavan et al. 2006).
DLPFC is a key area thought to mediate antisaccade performance and antisaccade deficits in psychiatric patient groups such as schizophrenia (McDowell et al. 2002). In an attempt to specify the role of DLPFC in saccade tasks, lesion studies have shown that damage to the DLPFC (and relevant white matter connections) results in significantly increased rates of erroneous glances to the target on the antisaccade task (Pierrot-Deseilligny et al. 1991; Ploner et al. 2005). This result was interpreted as indicating a role of the DLPFC in inhibiting reflexive saccades in this task (Pierrot-Deseilligny et al. 2002).
In our study, there were no significant main or interaction effect for DLPFC or VLPFC. This finding suggests that each area is involved to comparable extent in the 2 phases of the delayed saccade task. DLPFC and VLPFC are areas known to be involved in response inhibition but also functions such as the representation of task set (when 2 or more task sets are competing for execution), working memory, stimulus appraisal, memory retrieval, and related decision making processes (Markowitsch 1995; Konishi et al. 1999; Herath et al. 2001; Brass and von Cramon 2002; Honey et al. 2002; Aron et al. 2003; Hazeltine et al. 2003; Rubia et al. 2003; Brass et al. 2005; Petrides 2005; Wager et al. 2005; Garavan et al. 2006).
Our observation that DLPFC and VLPFC showed similar activation levels during saccade inhibition and generation in the delayed saccade task is compatible with these previous findings. Both phases of the current experiment involve supervisory functions such as stimulus appraisal, representation of task set, and the selection of an appropriate response. It is, therefore, possible that the mediation of these functions by DLPFC and VLPFC was associated with consistently elevated activation levels in these areas, making the dissociation of their specific roles in the current design difficult.
Patients with DLPFC lesions have been described as displaying increases in rates of antisaccade errors (Pierrot-Deseilligny et al. 2002), suggesting a specific role of the DLPFC in the inhibition of reflexive saccades. Our findings of a lack of difference between response inhibition and generation would also predict an effect of DLPFC lesions on antisaccade latencies (a measure of the ability to generate antisaccades). Early studies by Pierrot-Deseilligny et al. (1991) did not report antisaccade latencies in DLPFC patients, thus not allowing this hypothesis to be tested. A recent study (Ploner et al. 2005) investigated frontal lobe–lesioned patients with or without increased rates of reflexive errors and localized the lesion underlying reflexive errors to the DLPFC. However, the patients with DLPFC lesions and error rate increases in the study of Ploner et al. (2005) also had increased antisaccade latency (524 ms for ipsilesional and 562 ms for contralesional saccades) relative to patients with lesions outside this area (351 ms for ipsilesional and 311 ms for contralesional saccades) and controls (283 ms right, 292 ms left). These findings are compatible with our conclusions of an overall supervisory role of the DLPFC in antisaccade performance and involvement in both response inhibition and generation.
VLPFC activation during antisaccades is seen less consistently but has been observed in a standard antisaccade block-design study (Tu et al. 2006). Additionally, there is a report of a patient with right VLPFC lesions who displayed severe antisaccade abnormalities (Walker et al. 1998). Our data suggest that the VLPFC, like the DLPFC, is likely to play a role in antisaccade performance, which is not restricted to either response inhibition or generation.
The right thalamus showed a similar pattern of activation. The thalamus is an important structure in oculomotor circuitry (Alexander et al. 1990) both in its role in filtering and relaying visual inputs and through its direct connections with the prefrontal cortex. Activation in the right (Petit et al. 1993; Sweeney et al. 1996) and left (O'Driscoll et al. 1995) thalamus during voluntary saccades has been demonstrated previously. The present data confirm that the thalamus is under significant demand during standard antisaccades and suggest comparable involvement during the 2 phases of the delayed saccade task.
Medial FEF
Activation in the medial portion of the FEF (Amiez et al. 2006) was greater during antisaccade than prosaccade trials (main effect of SaccadeTask). Additionally, activation was greater during saccadic inhibition than generation (main effect of DelayCondition), but there was no SaccadeTask-by-DelayCondition interaction.
Our finding of a main effect of SaccadeTask, indicating greater activation during antisaccade than prosaccade trials, is consistent with previous observations. MedFEF activation during antisaccades is probably the most robust finding to emerge from block-design studies of healthy humans (O'Driscoll et al. 1995; Sweeney et al. 1996; Matsuda et al. 2004; Tu et al. 2006).
On the basis of event-related fMRI (Connolly et al. 2002) and human lesion studies (Pierrot-Deseilligny et al. 2002), the FEFs are thought to mediate saccade preparation and generation. Our data instead suggest a role in inhibiting an unwanted response to an abruptly presented peripheral stimulus. This finding might be compatible with nonhuman primate, human lesion, and fMRI studies demonstrating an inhibitory function of the FEF. For example, single-cell recording studies in nonhuman primates have demonstrated the importance of fixation and saccade neurons in FEF and SC in the antisaccade task (Everling et al. 1999; Everling and Munoz 2000). Whereas saccade neurons fire during saccade generation and show little or no discharge during rest, fixation neurons show the inverse pattern. The inhibition of reflexive saccades in the antisaccade task appears to require the inhibition of saccade neurons, a process which is thought to stem in part from FEF neurons (Burman and Bruce 1997; Munoz and Everling 2004). Concerning human lesion studies, Machado and Rafal (2004a) have demonstrated that chronic FEF lesions cause increased rates of reflexive errors to contralesional targets, supporting the role of FEF in the inhibition of reflexive saccades. Finally, bilateral FEF activation was recently found in a large study of inhibitory function using a go/no-go task (Garavan et al. 2006).
However, as argued above, the inhibition phase of the delayed saccade task is confounded with a significant covert attentional shift. Given that medFEF activation was greater during antisaccade than prosaccade trials (unlike the pattern observed for SMG), it is likely that the increase in activation during inhibition than generation reflects a covert attentional and sensorimotor transformations. As inhibitory requirements are likely to be similar for both tasks in the present design, a sole role of an area in inhibition (rather than generation) would not lead to a main effect of SaccadeTask (antisaccade > prosaccade), as was observed in SMG (see above). Target encoding (using attentional shifts) and transforming the signal into a motor output involves the FEF, and attentional and sensorimotor transformation processes are under greater demand during antisaccades than prosaccades (Krappmann et al. 1998; Olk and Kingstone 2003). As there was no significant SaccadeTask-by-DelayCondition interaction, it may be argued that the reason for this observation is not the operation of greater inhibitory requirements on medFEF during antisaccade trials but instead a greater recruitment of attentional processes and vector transformations in the incongruent S-R mapping posed by the antisaccade task mediated by the medFEF.
SEF and Caudate
The SEF was activated more strongly during antisaccade than prosaccade trials, irrespective of condition (main effect of SaccadeTask). Whereas the SEF was traditionally thought to mediate sequences of saccades (Gaymard et al. 1993; Müri et al. 1995), more recent evidence suggests that medial oculomotor cortex including the SEF is in fact also involved in the generation of single intentional saccades, including antisaccades (O'Driscoll et al. 1995; Everling et al. 1997; Müri et al. 1998; Coe et al. 2002; Curtis and D'Esposito 2003; McDowell et al. 2005). Nonhuman primate studies have shown that SEF neurons fire more frequently before and during saccade onset for antisaccades than prosaccades (Schlag-Rey et al. 1997; Amador et al. 2004).
The left caudate showed a similar pattern of activation. The caudate is a key node in the fronto-striato-collicular projections mediating saccadic control (Hikosaka et al. 1989). Previous neuroimaging studies have implicated the caudate in antisaccade performance (Crawford et al. 1996; Raemaekers et al. 2002; Ettinger et al. 2004).
The present findings show that SEF and caudate are generally more strongly involved during antisaccade than prosaccade trials without playing a specific role in either inhibition or generation.
Visual Areas
There was a strong activation in visual areas (BA17) during standard antisaccades compared with fixation. When entered into analysis of its role in the delayed saccade tasks, it was found that bilateral visual activation was greater during saccade generation than inhibition (main effect of DelayCondition). This effect was similar for prosaccade and antisaccade trials (no effect of SaccadeTask), and there was no significant interaction of these 2 factors.
It is likely that this effect relates to the significant visual input during the generation of saccades. Although no new visual stimulus is presented during that phase of the experiment, the visual environment shifts as a saccade is generated and a new eye position is assumed. It is interesting to note that this change in visual input, as reflected in bilateral BA17 activation during saccade generation, is highly significantly greater than that due to the appearance of a novel peripheral target during the inhibition phase of the paradigm.
Future Research
The present design provided the participant with information about SaccadeTask (prosaccade and antisaccade) at the outset of the trial. A possible implication of this design is that participants, who were thus aware of the requirement of saccade type before actually generating the saccade, may have started programming the saccade vector at the time of peripheral stimulus appearance in the delayed saccade task. However, saccade preparation is unlikely to account for the activation seen at stimulus appearance (phase 1). For example, antisaccades are known to require greater preparatory activation in frontal areas than prosaccades (Connolly et al. 2002; Curtis and D'Esposito 2003). In our study, there were no differences in SMG activation between prosaccade and antisaccade trials at phase 1, thus suggesting this activation was not of a predominantly preparatory nature.
In order to avoid possible confounds, future research might wish to apply modifications of this experiment. For example, the design could be modified to give no a priori information about trial type (prosaccade and antisaccade) at trial onset (i.e., the central fixation stimulus is black). The instruction of trial type could then be conveyed when the auditory go signal occurs, either by a change in central stimulus color to red or green or by introducing tones of different pitch (e.g., high pitch requires an antisaccade and low pitch requires a prosaccade). Alternatively, the instructional cue (antisaccade or prosaccade) could be conveyed by the introduction of another experiment phase between phase 1 and phase 2, such as a brief change in central target color to red or green. These design modifications would allow a more direct measurement of the brain processes underlying response inhibition in the absence of concurrent saccade vector programming or other confounding factors.
A further topic of interest concerns the effects of trial history on performance and brain activation. It was recently shown that saccade trials following an antisaccade are slower in latency and associated with reduced brain activation in the FEF and other brain areas compared with trials following a prosaccade (Manoach et al. 2007). Given these recent findings of the role of trial history in response variability, it will be important to investigate whether specific patterns of activation observed in this and other studies are influenced by effects of trial history. Although we found supportive evidence for such effects in our behavioral and imaging data (data not shown), the present study was not designed a priori to address this issue; for example, the inclusion of fixation and rest trials meant that only a small number of saccade trials could be used in an analysis of the effects of antisaccades versus prosaccades on subsequent trials. Future work is needed to fully investigate the influence of trial history on the results reported here.
Conclusions
The present study investigated the role of brain areas known to be activated during standard antisaccades in the 2 separable components of response inhibition and response generation in a delayed saccade paradigm. Different frontoparietal networks were involved in the 2 different stages of the experiment. Whereas right SMG showed significantly greater activation during inhibition than generation, right latFEF and bilateral IPS showed evidence of selective involvement in antisaccade generation. Medial bilateral FEF showed increased activation during antisaccade than prosaccade trials and during inhibition than generation. Interestingly, some areas showed no specific effects of experimental manipulation within the delayed saccade task. For example, SEF and caudate appeared to be more active during antisaccade than prosaccade trials irrespective of experimental phase. VLPFC, DLPFC, and thalamus showed comparable levels of activation in both phases of the delayed saccade task, suggesting a general supervisory and evaluative role in task performance, which might be required to similar extents during response inhibition and generation.
|
Funding |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionFundingReferences |
|---|
Leverhulme Trust (ECF/2004/0370) and ESRC/MRC (PTA-037-27-0002 to UE); Wellcome Senior Research Fellowship (VK) 052467; and Wellcome Clinician Scientist Fellowship (DF, 067427).
|
Acknowledgments |
|---|
Conflict of Interest: None declared.
|
References |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionFundingReferences |
|---|
Alexander GE, Crutcher MD, Delong MR. Basal ganglia-thalamocortical circuits: parallel substrates for motor, oculomotor, "prefrontal" and "limbic" functions. Prog Brain Res (1990) 85:119–146.[Medline]
Amador N, Schlag-Rey M, Schlag J. Primate antisaccade. II. Supplementary eye field neuronal activity predicts correct performance. J Neurophysiol (2004) 91:1672–1689.
Amiez C, Kostopoulos P, Champod AS, Petrides M. Local morphology predicts functional organization of the dorsal premotor region in the human brain. J Neurosci (2006) 26:2724–2731.
Aron AR, Fletcher PC, Bullmore ET, Sahakian BJ, Robbins TW. Stop-signal inhibition disrupted by damage to right inferior frontal gyrus in humans. Nat Neurosci (2003) 6:115–116.[CrossRef][ISI][Medline]
Bandettini PA, Cox RW. Event-related fMRI contrast when using constant interstimulus interval: theory and experiment. Magn Reson Med (2000) 43:540–548.[CrossRef][ISI][Medline]
Brass M, Derrfuss J, von Cramon DY. The inhibition of imitative and overlearned responses: a functional double dissociation. Neuropsychologia (2005) 43:89–98.[CrossRef][ISI][Medline]
Brass M, von Cramon DY. The role of the frontal cortex in task preparation. Cereb Cortex (2002) 12:908–914.
Brown MR, DeSouza JF, Goltz HC, Ford K, Menon RS, Goodale MA, Everling S. Comparison of memory- and visually guided saccades using event-related fMRI. J Neurophysiol (2004) 91:873–889.
Brown MR, Goltz HC, Vilis T, Ford KA, Everling S. Inhibition and generation of saccades: rapid event-related fMRI of prosaccades, antisaccades, and nogo trials. Neuroimage (2006) 33:644–659.[CrossRef][ISI][Medline]
Burman DD, Bruce CJ. Suppression of task-related saccades by electrical stimulation in the primate's frontal eye field. J Neurophysiol (1997) 77:2252–2267.
Calton JL, Dickinson AR, Snyder LH. Non-spatial, motor-specific activation in posterior parietal cortex. Nat Neurosci (2002) 5:580–588.[CrossRef][ISI][Medline]
Chikazoe J, Konishi S, Asari T, Jimura K, Miyashita Y. Activation of right inferior frontal gyrus during response inhibition across response modalities. J Cogn Neurosci (2007) 19:69–80.
Coe B, Tomihara K, Matsuzawa M, Hikosaka O. Visual and anticipatory bias in three cortical eye fields of the monkey during an adaptive decision-making task. J Neurosci (2002) 22:5081–5090.
Colby CL, Goldberg ME. Space and attention in parietal cortex. Annu Rev Neurosci (1999) 22:319–349.[CrossRef][ISI][Medline]
Connolly JD, Goodale MA, Menon RS, Munoz DP. Human fMRI evidence for the neural correlates of preparatory set. Nat Neurosci (2002) 5:1345–1352.[CrossRef][ISI][Medline]
Corbetta M, Akbudak E, Conturo TE, Snyder AZ, Ollinger JM, Drury HA, Linenweber MR, Petersen SE, Raichle ME, Van Essen DC, et al. A common network of functional areas for attention and eye movements. Neuron (1998) 21:761–773.[CrossRef][ISI][Medline]
Cornelissen FW, Kimmig H, Schira M, Rutschmann RM, Maguire RP, Broerse A, den Boer JA, Greenlee MW. Event-related fMRI responses in the human frontal eye fields in a randomized pro- and antisaccade task. Exp Brain Res (2002) 145:270–274.[CrossRef][ISI][Medline]
Crawford TJ, Puri BK, Nijran KS, Jones B, Kennard C, Lewis SW. Abnormal saccadic distractibility in patients with schizophrenia: a 99mTc-HMPAO SPET study. Psychol Med (1996) 26:265–277.[ISI][Medline]
Curtis CE, D'Esposito M. Success and failure suppressing reflexive behavior. J Cogn Neurosci (2003) 15:409–418.
Curtis CE, Rao VY, D'Esposito M. Maintenance of spatial and motor codes during oculomotor delayed response tasks. J Neurosci (2004) 24:3944–3952.
DeSouza JF, Menon RS, Everling S. Preparatory set associated with pro-saccades and anti-saccades in humans investigated with event-related fMRI. J Neurophysiol (2003) 89:1016–1023.
Donaldson DI, Buckner RL. Effective paradigm design. In: An introduction to methods—Jezzard P, Matthews PM, Smith SM, eds. (2004) Oxford: Oxford University Press. 177–195. Functional MRI.
Ettinger U, Antonova E, Crawford TJ, Mitterschiffthaler MT, Goswani S, Sharma T, Kumari V. Structural neural correlates of prosaccade and antisaccade eye movements in healthy humans. Neuroimage (2005) 24:487–494.[CrossRef][ISI][Medline]
Ettinger U, Kumari V, Chitnis XA, Corr PJ, Crawford TJ, Fannon DG, O'Ceallaigh S, Sumich AL, Doku VC, Sharma T. Volumetric neural correlates of antisaccade eye movements in first-episode psychosis. Am J Psychiatry (2004) 161:1918–1921.
Everling S, Dorris MC, Klein RM, Munoz DP. Role of primate superior colliculus in preparation and execution of anti-saccades and pro-saccades. J Neurosci (1999) 19:2740–2754.
Everling S, Krappmann P, Flohr H. Cortical potentials preceding pro- and antisaccades in man. Electroencephalogr Clin Neurophysiol (1997) 102:356–362.[CrossRef][ISI][Medline]
Everling S, Munoz DP. Neuronal correlates for preparatory set associated with pro-saccades and anti-saccades in the primate frontal eye field. J Neurosci (2000) 20:387–400.
Garavan H, Hester R, Murphy K, Fassbender C, Kelly C. Individual differences in the functional neuroanatomy of inhibitory control. Brain Res (2006) 1105:130–142.[CrossRef][ISI][Medline]
Gaymard B, Rivaud S, Pierrot-Deseilligny C. Role of the left and right supplementary motor areas in memory-guided saccade sequences. Ann Neurol (1993) 34:404–406.[CrossRef][ISI][Medline]
Goodale MA, Milner AD. Separate visual pathways for perception and action. Trends Neurosci (1992) 15:20–25.