Cerebral Cortex Advance Access published online on February 5, 2007
Cerebral Cortex, doi:10.1093/cercor/bhl171
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
When Does the Brain Inform the Eyes Whether and Where to Move? an EEG Study in Humans
1 Department of Psychology and Neuroscience, Bio-Imaging Research Center, University of Georgia, Athens, GA 30602, USA, 2 Department of Psychology, Washington University, St Louis, Missouri, USA, 3 Center for Cognitive Medicine, University of Illinois, Chicago
Address correspondence to Brett A. Clementz, PhD, Psychology Department, Psychology Building, University of Georgia, Athens, GA 30602, USA. Email: clementz{at}uga.edu.
 |
Abstract |
|---|
 Top Abstract IntroductionMaterials and MethodsResultsDiscussionConclusionReferences |
|---|
The current study addressed when in the course of stimulus processing, and in what brain areas, activity occurs that supports the interpretation of cues that signal the appropriateness of different and competing behaviors. Twelve subjects completed interleaved no-go–, pro-, and antitrials, whereas 64-channel electroencephalography was recorded. Principle component and distributed source analyses were used to evaluate the spatial distribution and time course of cortical activity supporting cue evaluation and response selection. By 158 ms poststimulus, visual cortex activity was lower for no-go trials than it was for both pro- and antitrials, consistent with an early sensory filter on the no-go cue. Prefrontal cortex (PFC) activity at 158 ms was highest during antitrials, consistent with this brain region's putative involvement in executive control. At 204 ms poststimulus, however, PFC activity was the same for pro- and antitrials, consistent with an ostensible role in response selection. PFC activity at 204 ms also was robustly inversely correlated (r = –0.75) with visual cortex activity on antitrials, perhaps indicating top-down modulation of early sensory processing that would decrease the probability of an error response. These data highlight how a distributed neural architecture supports the evaluation of stimuli and response choices.
Key Words: antisaccades • EEG • executive control • no go • prefrontal cortex • response selection • saccades
|
Introduction |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionConclusionReferences |
|---|
The saccadic eye movement control system is ideal for studying the neural correlates of response selection (Merriam et al. 2001
; Schall 2001; Fecteau et al. 2004; Nakamura et al. 2004; Everling and DeSouza 2005) as saccades can be precisely measured and variation in task demands are easily manipulated. The neural circuitry supporting saccadic performance is similar for simple and complex saccades (Leigh and Zee 2006), although additional neural regions are often recruited to support performance when saccades are initiated based on an analysis of contextual cues instead of simple sensorimotor control. Measurements of strength of activity across time in brain regions comprising saccadic circuitry, therefore, provide information about "where" and "when" decisions are made that will guide subsequent behavior.
The importance of specific cortical regions for saccadic control during both simple and complex cognitive tasks has been highlighted by single unit and lesion studies of nonhuman primates and studies of humans with specific brain lesions (e.g., Funahashi et al. 1993; Pierrot-Deseilligny et al. 2002; Lynch and Tian 2005; Ploner et al. 2005). Numerous blood flow–based neuroimaging studies of saccade control in normal humans have been largely consistent with the nonhuman primate and human lesion data (for recent reviews, see Hutton and Ettinger 2006; Sweeney et al. 2007). Neocortical circuitry common to all voluntary saccades includes the parietal eye field (PEF), frontal eye field (FEF), and supplementary eye field (SEF) and precuneus in superior parietal cortex. Striate and extrastriate cortices are necessarily involved in visual stimulus processing to guide visuomotor responses. Prefrontal cortex (PFC) is involved in more complex tasks when participants learn arbitrary stimulus–response (S–R) pairings to select the correct response, inhibit an unwanted response, and/or remember information relevant for proper task performance over time (Sweeney et al. 1996; Merriam et al. 2001; Miller and Cohen 2001; McDowell et al. 2002; Munoz and Everling 2004; Everling and DeSouza 2005; Ford et al. 2005).
Assessing strength of activity in cortical regions supporting response selection requires placing participants in situations where they must evaluate a cue and select an appropriate response from a set of alternatives. Understanding how the brain determines the appropriate response during such tasks requires collecting data with high temporal resolution, which facilitates comparison of neural activity functions in different brain regions and helps clarify whether such activity is a cause or a consequence of a particular response. Electroencephalography (EEG) and magnetoencephalography (MEG) allow for direct measurement of distributed neural activity with a high temporal (millisecond) resolution (e.g., Dale and Halgren 2001; Liu et al. 2002; Wang and Kaufman 2003). Several findings of note can be culled from EEG/MEG investigations of saccade generation and inhibition. There is activity over largely contralateral visual cortex associated with stimulus registration beginning 100–120 ms after stimulus onset (Matthews et al. 2002; Richards 2003; Tzelepi et al. 2004; McDowell et al. 2005; Tendolkar et al. 2005). When executive control is theoretically required for correct performance, there is neural activity over frontal cortex sensors about 150–200 ms after cue onset (Evdokimidis et al. 1996; Matthews et al. 2002; McDowell et al. 2005). EEG/MEG studies are also consistent with a preresponse role for involvement of posterior parietal cortex in the response planning required especially for successful antisaccade performance (Matthews et al. 2002; Richards 2003) and for the role of the frontal eye fields in especially prosaccade preparation and generation (Vanni and Uutela 2000; McDowell et al. 2005).
For a number of reasons, however, there is uncertainty about the time course and spatial locations of neural activities that support response selection in ocular motor tasks (e.g., Merriam et al. 2001; Munoz and Everling 2004; Pouget et al. 2005). First, study designs have been suboptimal for addressing this specific issue. For instance, no-go–, pro-, and antiparadigms are often used to assess cortical activations associated with response generation and inhibition. Pro- and antitasks both require response generation, and no-go– and antitasks both theoretically require response inhibition. Some studies that relied on these tasks, however, used blocked trial types, which does not allow for evaluation of on line response selection. Even those studies that randomly interleaved trial types tended to include only 2 alternatives (save Wauschkuhn et al. 1998) and/or heavily weighted cue probability toward a particular trial type (typically toward trials requiring an overt response). Brain activity differences in such studies could be associated with cognitive phenomena other than response selection, including predetermined response set, expectancy, and/or novelty. Second, few EEG/MEG studies on this topic have used either dense sensor arrays or source estimation algorithms. Such methodological and analytical advancements provide important means for determining the time course and spatial locations of neural activities supporting response selection in ocular motor tasks (e.g., Srinivasan et al. 1996; Picton et al. 2000).
The goal of the current study was to understand when in the course of stimulus processing, and in what brain areas, differential activities occur that correlate with the ability to properly perform no-go–, pro-, and antitrials. The appropriate, equally probable, response was randomly determined on individual trials, and highly similar stimuli (peripheral cues of the same size and spatial location which differed only in their color, with color indicating task type) were used to initiate all 3 trial types. Sixty-four EEG sensors were used to measure neural activity during performance, and a distributed source estimation algorithm (L2 minimum norm; Hämäläinen and Ilmoniemi 1984; Wang and Kaufman 2003) was used to estimate cortical regions from which this activity emanated.
The following 3 issues were of particular interest. First, how early and where in neural processing do the trial types differ in strength of neural activity? We predicted that pro- and antitrials would deviate in occipital cortex by 100–150 ms poststimulus (e.g., McDowell et al. 2005). Differences this early in visual processing are expected given that attention modulates extrastriate cortex activity (Hillyard and Anllo-Vento 1998) and that these 3 trial types differ in extent of attentional demands (antitrials most because of necessary additional processing in relation to the cue for correct responding, and no-go trials least because the peripheral cue is irrelevant for subsequent behavior). Second, does PFC support executive control (i.e., do not look at the cue) on both no-go– and antitrials (Konishi et al. 1998; Liddle et al. 2001; Merriam et al. 2001; Matthews et al. 2002; Ford et al. 2005)? If so, early PFC activity should not differentiate between no-go– and antitrials. Third, does PFC bias neural activity early in the course of cue evaluation? If so, this would indicate that PFC can affect distant neural activity supporting sensory and perceptual processing in the face of changing situational demands.
|
Materials and Methods |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionConclusionReferences |
|---|
Participants
Twelve normal right-handed females (age range 18–22 years), drawn from the Psychology Department research pool, participated in the study after providing written informed consent. Participants had normal or corrected-to-normal vision, had no evidence of neurological impairment, were free of psychiatric or substance use disorders (by self-report), and were given class credit and monetary compensation ($10 US) for their participation. This project was approved by the relevant institutional review board.
Procedure
Subjects were taken to a quiet, darkened room. They were seated in a comfortable armchair, and a wraparound neck pillow was used to reduce head movements during the experiment. For stimulus presentation, a Zenith 1792 flat-panel color monitor was located 67 cm directly in front of a subject's eyes. Subjects were asked to remain still during the course of the experiment so that movement artifacts could be minimized.
Trials (see Fig. 1) took an average of 4000 ms and began with a precue period (2000–3000 ms; rectangular distribution) consisting of a central square box (2° on a side), within which was centered a small square (0.5° on a side). Peripheral circles (2° in diameter) were located ±8° from the central square, within which smaller circles were centered (each 0.5° in diameter; the small circles provided the desired response location). This means that the response location, regardless of pro- or antitrial, was always visible and was the same for both pro- and antitrial types. This stimulus approach was used to increase the similarity of response locations between pro- and antitrial types and to focus on neural activity associated with response selection separate from the need for participants to calculate spatial coordinates for antiresponses (although they would still need to compute the appropriate coordinate transformation to properly generate the response into the opposite visual field on antitrials). Immediately following the precue period came the cue period (200 ms) during which one of the large circles turned red (4.7 cd/m2), green (4.6 cd/m2), or blue (4.3 cd/m2). A peripheral cue for trial type was used to accentuate the salience of the peripheral target in an effort to stimulate more response errors (although in practice, subjects quickly learned the task and generated too few errors for analysis purposes; see Results). If the cue was red, subjects were to remain fixated on the central square (no-go trial). If the cue was green, subjects were to saccade toward the colored stimulus (protrial). If the cue was blue, subjects were to saccade toward the mirror image of the colored stimulus (antitrial). Presentation side and color were varied randomly. The cue was extinguished contemporaneously with presentation of the response period screen (1000 ms). The response period was followed by the postresponse period (300 ms), during which the small square (no-go trial) or small circle (pro- and antitrials) at the proper eye position was illuminated to reinforce the accuracy component of the task.
|
To ensure that subjects understood the task, they were given practice consisting of 12 no-go–, 12 pro-, and 12 antitrials using the stimulus procedures described above. Upon completion of the practice, subjects were presented with interleaved 300 no-go–, 300 pro-, and 300 antitrials. In each condition, 150 cues were presented in each visual field. Stimuli were presented in six 10-min blocks of 150 trials each, with brief rest intervals provided after every block.
Electrophysiological Recording and Screening
EEG data were collected using a 64-channel SA Instrumentation (San Diego, CA) Isometric Bioelectric amplifier system and a low-profile EasyCap (Falk Minnow Services, Herrsching-Breitbrunn, Germany) with evenly spaced sensor locations using a forehead ground. Impedances were kept below 10 kohms. EEG data were referenced to Cz, sampled at 1000 Hz, and analog bandpass filtered from 0.1 to 200 Hz. Eye movements were detected using sensors at the outer canthi of the 2 eyes (horizontal electro-oculography) and above and below the eyes (vertical EOG). Following task presentation, 3-dimensional (3D) sensor and fiducial locations (nasion and left and right preauricular points) were localized using a Polhemus Fastrak 3D digitization wand (Polhemus Inc., Colchester, VT) for later registration of the sensor locations with respect to the brain by matching the digitized fiducial locations to the corresponding locations on the head.
Data were digitally filtered (using Matlab, Release 12, the Mathworks, Inc., Natick, MA) from 0.5 to 50 Hz with a third-order Butterworth filter. Bad channels were identified by visual inspection (no more than 4 for any subject), and their voltage values were set to zero. Trials with EEG activity greater than 100 µV were automatically eliminated. The remaining trials were visually inspected, and those with blinks and movement artifacts were eliminated. Useable pro- and antitrials were then scored for saccade direction and latency (Dyckman and McDowell 2005). Stimulus-locked averages for correct trials were generated for no-go–, pro-, and anticonditions by cue direction (left or right). Individual subject averages for each condition and cue direction were then read into EEG analysis software (BESA Version 5.1, MEGIS Software, Gräfelfing, Germany). To preserve the maximum channel information for all subjects, bad channels were interpolated using BESA's spherical spline interpolation method. The data were then transformed to a common reference.
Principle Component Analysis
Tasks requiring complex cognitive operations result in multiple sources of distinct neural activities that may overlap in time. The averaged event-related potential (ERP) signal recorded at the scalp at any one sensor and at any single time point, therefore, may contain activity from multiple neural generators (Dien et al. 2005). A common method for addressing this issue is to use statistical procedures that capture specific ERP components with more distinct neural source configurations (e.g., Carretié et al. 2004). For the present case, spatial principle component analysis (PCA) was calculated (PCA Toolbox, V1.09; Dien et al. 2003). PCA identifies the percentage of variance accounted for by linear combinations of sensors (factors) with distinct spatial distributions but variable time courses (see, e.g., Spencer et al. 2001). The percentage of variance accounted for provides a basis by which meaningful components may be distinguished from more negligible ones.
PCA analysis was performed separately for left and right visual field cues largely following the recommendations of Dien et al. (2005). The primary interest here was the identification of between-trial type differences in brain activities associated with stimulus registration and response selection, so brain activity time locked to response generation was not considered here. The first 300 ms after the initiation of the color cue were analyzed (as this was the time before generation of the earliest saccades at 300 ms after cue onset in this response selection task). In sum, the data matrix consisted of 65 sensors by 300 time points by 3 conditions by 12 subjects (65 sensors x 10 800 total observations). Calculating the PCA using all conditions simultaneously may reduce sensitivity for subtle differences but will accentuate the identification of primary differences in activity that are commonly observed between conditions (which was the major focus here). The PCA solution was rotated to simple structure using the Varimax rotation (with Kaiser normalization). For both the left and right visual field cues, 4 spatial factors accounted for more than 90% of the total variance, with no additional factors accounting for more than 2.5% of the total variance. The first 4 factors, therefore, were used in subsequent analyses.
Global Field Power of Factor Scores
After transforming the data from factor scores back to microvolts (see, e.g., Spencer et al. 2001), the global field power (standard deviation [SD] of microvolts across sensors at each time point) was calculated for each subject for each condition and for each factor. Global field power is a measure of variation of signal at each point in time such that high values correspond to greater signal strength. The time courses of the global field power plots (whose peaks were within a few milliseconds for the left and right visual field cues for each factor), therefore, reveal the time of maximal activity for each spatial factor (see Fig. 2). These time courses were then used to order the data analysis sequence. Only peaks in the global field power plots that were greater in magnitude than 3 times baseline activity (average of activity during the 100 ms before cue presentation) for every subject were considered to identify significant activations. Times of maximal activity for each factor were calculated from the onset of the cue. Factor 4 had the earliest postcue peak (158 ms), followed by Factor 1 (178 ms), Factor 2 (204 ms), a repetition of Factors 1 and 4 (250 ms), and finally Factor 3 (294 ms).
|
Distributed Source Analyses
The spatial distribution of ERP voltage for the 4 factors as a function of cue direction was then estimated using distributed source analysis. Locations of sources in the brain cannot be directly inferred from the spatial distribution of ERPs, and activity at any individual sensor does not unambiguously clarify from what region of the brain that activity emanated. Algorithms are available, however, that allow for neural sources to be estimated from the distribution of ERP signals recorded from multiple sensors on the scalp (see Hauk et al. 1999). Source analyses in the present manuscript used the L2 minimum norm approach to solving this computation problem (Hämäläinen and Ilmoniemi 1984; Wang and Kaufman 2003).
Cortical sources were estimated for each of the 4 spatial factors as a function of condition and cue direction following surface Laplacian transformation (Babiloni et al. 2000). The Laplacian transformation attenuates deep, nonfocal sources that are a cause of diffuse noise when examining cortical activity and accentuates the cortical sources best measured by surface ERP recordings (Nunez and Srinivasan 2006), leading to more focal solutions in source estimation algorithms (for an example of a similar approach, see also Moores et al. 2003). L2 minimum norm solutions (Hämäläinen and Ilmoniemi 1984) were calculated using BESA 5.1. For the minimum norm approach, the source configuration is fixed a priori (fixed source locations are specified on the surface from which ERP signals emanated; e.g., the cerebral cortex). Given the measured data, dipolar strength is estimated for each source at each time point. In BESA 5.1, 713 locations are evenly distributed on the surface of a smoothed standard cortical surface. For each subject and each factor, the minimum norm solutions for the first 300 ms across all sources were transformed to standard scores (mean 0 and unit variance).
Quantification of Brain Activations
For all factors, there were distinct regions of source activity that could be identified via visual inspection. These regions were consistent with those identified and described in previous functional magnetic resonance imaging (fMRI; Sweeney et al. 2007) and EEG/MEG studies (McDowell et al. 2005). Labels for these regions were based on knowledge of the functional anatomy of ocular motor control (Leigh and Zee 2006) and/or the general spatial location of the region (see, e.g., McDowell et al. 2005; Sweeney et al. 2007). The source location with the largest standard score value within a region of interest, averaged over all conditions, defined the "center of mass" for that region. Source locations that were immediately adjacent to this center of mass were added to a region definition if their averaged standardized values were at least 75% that of the center of mass standard score value and had a standard score value of at least 2.0. Source locations that were immediately adjacent to the new sources were tested in the same way until source locations no longer entered the definition. Thus, the number of adjacent sources used to quantify regions of interest ranged from a low of 7 for the right frontal eye field region seen in Factor 1 (at 178 ms) to a high of 39 for the temporo–parietal junction region seen in Factor 3 (at 294 ms). For each subject, condition, and cue direction, strength of activation for each region of interest was quantified by averaging over all sources that defined that region. These values were then used in subsequent statistical analyses.
|
Results |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionConclusionReferences |
|---|
Behavioral Results
No differences were found in response accuracy across cue types. Participants remained fixed at the central location 94% of the time on no-go trials, generated a saccade toward the cue 91% of the time on protrials, and generated a saccade to the mirror image location of the cue without first making a proerror 91% of the time on antitrials. Given the high proportion of correct behavior for all conditions, only correct trials were used in data analyses. Latencies of responses for correct protrial (M = 363.0 ms, SD = 23.3) and antitrial (M = 369.7 ms, SD = 25.8) did not differ significantly, which is consistent with other studies that used similar response selection tasks (e.g., Wauschkuhn et al. 1998).
Electrophysiological Data
Figures 3 and 4 provide the topographic maps (voltages and surface Laplacians) and source analyses (minimum norm results plotted on an averaged cortical surface), respectively, for the 4 spatial factors. The source analysis results for the spatial factors will be presented in relation to their temporal order of occurrence. Repeated measures analyses of variance (ANOVAs) with Huynh–Feldt adjusted degrees of freedom were used to test for differences in neural activity by region of interest between conditions and visual field of cue presentation. For regions of interest with bilateral activations (middle occipital gyrus, occipito–parietal junction, and frontal eye fields), the ANOVA also included a level capturing whether the activity was contra- or ipsilateral to the visual field of cue presentation. Table 1 lists the anatomical regions of interest and the standardized means (SDs) of activity in those regions as a function of condition.
|
|
|
Factor 4 (158 ms)
For this factor, there were 2 identifiable regions of interest: bilateral middle occipital gyrus and right PFC. For middle occipital gyrus, there was a significant main effect of condition, F2,22 = 20.1, P < 0.001, and 
= 0.917. Pro- and anticonditions, which did not differ significantly, had higher middle occipital gyrus activity than the no-go condition. There was also a significant main effect of hemisphere, F1,11 = 37.2 and P < 0.001, with contralateral being higher than ipsilateral middle occipital gyrus activity (relative to visual field of cue presentation). The condition by hemisphere interaction was also significant, F2,22 = 29.9, P < 0.001, and = 1.00. Analysis of simple main effects revealed that in ipsilateral middle occipital gyrus, both the pro- and anticonditions, which did not differ significantly (P > 0.05), had higher activity than the no-go condition (P values < 0.05). In contralateral middle occipital gyrus, all 3 conditions differed on strength of neural activity (anti > pro > no go).
For right PFC, there was a significant main effect of condition, F2,22 = 6.2, P = 0.022, and = 0.616. Post hoc comparisons demonstrated that the anticondition had higher activity than both the pro- and no-go conditions (P values < 0.05). The pro- and no-go conditions did not differ significantly on PFC activity at 158 ms following cue onset (P values > 0.05).
Factor 1 (178 ms)
For this factor, there were 2 regions of interest: bilateral medial inferior parietal lobe and bilateral frontal eye fields. For medial inferior parietal lobe, there was a significant main effect of condition, F2,22 = 20.4, P < 0.001, and = 0.705, with the pro- and anticonditions, which did not differ significantly (P > 0.05), having higher activity than the no-go condition (P values < 0.05). There was also a significant main effect of hemisphere, F1,11 = 14.5 and P = 0.003, with contralateral being higher than ipsilateral medial inferior parietal lobe activity (relative to visual field of cue presentation). The condition by hemisphere interaction was also significant, F2,22 = 8.8, P = 0.003, and = 0.862. Post hoc analyses revealed that the 2-way interactions comparing antiactivity to separate no-go– and proactivities in medial inferior parietal lobe were both significant, F1,11 values > 9.4 and P values < 0.011. The 2-way interaction comparing no-go– and proactivities, however, was not significant, F1,11 = 0.7 and P > 0.05. This pattern indicated that medial inferior parietal lobe showed a similar hemisphere effect for both no-go– and protrials (contralateral activity greater than ipsilateral activity, P < 0.05), but this same pattern was not observed for antitrials (contralateral activity not different from ipsilateral activity).
For frontal eye fields, there was a significant main effect of condition, F2,22 = 11.1, P < 0.003, and = 0.668, with the anticondition having higher activity than both the no-go– and proconditions (P values < 0.05), which did not differ significantly (P > 0.05). There was also a significant main effect of hemisphere, F1,11 = 14.5 and P = 0.003, with contralateral being higher than ipsilateral frontal eye field activity (relative to visual field of cue presentation). The condition by hemisphere interaction was also significant, F2,22 = 14.4, P < 0.001, and = 1.0. Post hoc analyses revealed that the 3 conditions did not differ significantly on ipsilateral frontal eye field activity (P values > 0.05) and that the anticondition had higher contralateral frontal eye field activity than both the no-go– and proconditions (P values < 0.05), which did not differ significantly (P > 0.05).
Factor 2 (204 ms)
For this factor, there were 3 regions of interest: bilateral lateral middle occipital gyrus, superior parietal lobe, and right PFC. For middle occipital gyrus, there was a significant main effect of condition, F2,22 = 33.1, P < 0.001, and = 0.965, with the pro- and anticonditions, which did not differ significantly (P > 0.05), having higher activity than the no-go condition (P values < 0.05). There was also a significant main effect of hemisphere, F1,11 = 298.7 and P < 0.001, with contralateral being higher than ipsilateral middle occipital gyrus activity (relative to visual field of cue presentation). The condition by hemisphere interaction was also significant, F2,22 = 29.9, P = 0.003, and = 1.0. Post hoc analyses revealed that the groups did not differ significantly on ipsilateral middle occipital gyrus activity (P values > 0.05), so the main effect of condition was primarily a function of the contralateral activity.
For superior parietal lobe, there was a significant main effect of condition, F2,22 = 4.9, P = 0.017, and = 0.923. Post hoc comparisons revealed that the anticondition had higher superior parietal lobe activity than both the no-go– and proconditions (P values < 0.05), which did not differ significantly (P > 0.05). For right PFC, there was also a significant main effect of condition, F2,22 = 21.9, P < 0.01, and = 0.657. Post hoc comparisons demonstrated that the anti- and proconditions, which did not differ significantly (P > 0.05), had higher right PFC activity than the no-go condition (P values < 0.05).
Factor 3 (296 ms)
For this factor there were 2 regions of interest: right lateral temporo–parietal junction and supplementary eye field. For temporo–parietal junction activity, there was a significant main effect of condition, F2,22 = 7.3, P = 0.004, and = 0.004. Post hoc comparisons revealed that the no-go– and anticonditions, which did not differ significantly (P > 0.05), both had higher temporo–parietal junction activity than the procondition (P values < 0.05). There were no significant effects on supplementary eye field activity, suggesting consistent increases in activity this brain region across all 3-task conditions.
Correlations between Regions of Interest on Neural Activity
In a preliminary exploration of the relationship between brain regions on strength of neural activity, correlations were calculated within specific time points (158, 178, 204, and 250 ms) between variables that showed significant condition effects in the above analyses. Bootstrapped confidence intervals (CIs) were used to determine whether a specific correlation was statistically significant and whether the strength of that correlation differed as a function of condition. First, 12 observations were selected, sampling with replacement, from the original 12 subjects. A Pearson correlation was then calculated between the relevant variable pairs, separately for each condition (no-go, pro, and anti), using this randomly generated sample of 12 observations. This procedure was repeated 2000 times. The median of these 2000 correlations within a condition was used to estimate the population correlation coefficient, and the 95% CI for this correlation was estimated from the distribution of these values.
Only 2 relationships had a 95% CI that did not include zero (Fig. 5). First, there was an inverse relationship for the anticondition between contralateral middle occipital gyrus and right PFC activities at 204 ms postcue (median r = –0.75; 95% CI = –0.91 to –0.34) such that higher PFC activity was associated with lower activity in middle occipital gyrus. This correlation significantly differed from the same no-go condition (median r = +0.32; 95% CI = –0.30 to +0.69) and procondition (median r = –0.21; 95% CI = –0.73 to +0.33) correlations, both of which did not differ significantly from either zero or each other. Second, there was a relationship for the anticondition between contralateral medial inferior parietal cortex and right PFC activities at 250 ms postcue (median r = +0.86; 95% CI = +0.13 to +0.95), such that higher medial inferior parietal cortex activity was associated with higher right PFC activity. This relationship differed significantly from the same procorrelation (median r = –0.12; 95% CI = –0.58 to +0.48) but not from the same no-go condition correlation (median r = +0.43; 95% CI = –0.06 to +0.81). The no-go– and procondition correlations did not differ significantly from either zero or each other.
|
|
Discussion |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionConclusionReferences |
|---|
The present study investigated where and when brain activity differences occurred that differed as a function of cue evaluation and response selection during the performance of interleaved no-go–, pro-, and antitasks. In general, activity patterns in posterior brain regions showed more similarity in preparation for performing pro- and antitrials, with no-go trials associated with decreased activity from very early in stimulus evaluation. Differences between pro- and antitrials on activity in posterior brain regions may provide important information about where and when neural processing supports the response requirements of these different behaviors. Frontal brain regions, however, tended to show relatively greater activity associated with antitrials. Moreover, PFC and middle occipital gyrus activities were related only during antitrials, suggesting the presence of top-down and/or bottom-up control mechanisms that helped support performance of the correct antibehavior. Specific issues concerning the nature of the distributed neural circuitry supporting task performance are discussed below.
Early Between-Task Differences in Neural Activity
The first specific issue of interest was that early differences were expected between trial types in visual cortex. The data showed that the decision about whether movement was required on a trial occurred by 158 ms postcue, and neural activity associated with this decision was present at least 40 ms earlier (see the temporal course of activity for Factor 4 in Fig. 2). At this point in stimulus processing, middle occipital gyrus activity for no-go trials was significantly less than for both pro- and antitrials. This finding is consistent with the thesis that subjects' initial node in the decision tree was dichotomous: do not move versus prepare to move.
PFC and Response Inhibition during No-go– and Antitasks
The second specific issue of interest was whether PFC supported executive control processes (i.e., do not look at the cue) necessary for correct no-go– and antiperformances (Konishi et al. 1998; Liddle et al. 2001; Merriam et al. 2001; Matthews et al. 2002; Ford et al. 2005). Early right PFC activity (at 158 ms poststimulus) had a specific relationship to antiperformance: it was larger in response to anticues as opposed to no-go– and procues. Activation of PFC on this time scale is consistent with nonhuman primate (Hasegawa et al. 2000) and other EEG (Foxe and Simpson 2002) studies. The timing and cue-related pattern of this PFC response is concordant with it mediating executive control processes necessary for correct performance on antitrials (Munoz and Everling 2004).
The results showing increased PFC activity during correct antitrials but not on no-go trials are inconsistent with the thesis that the no-go cues elicited a trial-by-trial (phasic) inhibitory (stop) signal in PFC. Indeed, in the present task, the only similarity between no-go– and antitrial activities was at 296 ms after cue onset in temporo–parietal junction, part of the right hemisphere attention system devoted to sensory processing of relevant and novel information (Corbetta and Shulman 2002). No-go– and antitrials, which have more atypical behavioral requirements than do protrials, may be expected to require greater voluntary attentional control. Consistent with this explanation, temporo–parietal junction activity was similarly high during correct no-go– and antitrials.
The finding that no-go cues did not elicit substantial cue-related PFC activity may indicate that inhibition-related signals are not necessary for proper performance of no-go tasks. There are at least 2 matters to consider before drawing such a conclusion. In the first place, most studies of no-go–related brain activity heavily weight trial probabilities toward those that necessitate generation of an overt response, but in the present study, we used equally probably no-go–, pro-, and anticues. Although the former paradigms could lead to brain activity differences that are associated with cognitive phenomena other than cue evaluation and response selection (which were the phenomena of interest here), they may provide the most effective means for eliciting no-go–related inhibitory activities.
In the second place, the reduced middle occipital gyrus activity by 158 ms for no-go cues compared with pro- and anticues is consistent with an early sensory filter on the no-go cue. It is likely that this early sensory filter was imposed on a "controlled" brain region (middle occipital gyrus in the present case) by a "controlling" region (perhaps PFC; e.g., Miller and Cohen 2001; see also, Trappenberg et al. 2001; Moore and Armstrong 2003). If so, it is likely that this early sensory filter (if the cue color is red, no further action is required), imposed by PFC, would be instituted on the time scale of the task (not time locked to stimulus presentation) rather than on the time course of individual trials (and time locked to stimulus presentation). Activity not time locked to stimulus presentation would be undetected with EEG/MEG in studies like the present one but could be readily observed in fMRI and PET studies. EEG/MEG, however, would detect stimulus-locked activity differences that could be a consequence of a tonic filtering operation (i.e., lower early middle occipital gyrus activity in relation to the no-go cue).
This thesis is speculative, but having such filtering operations, imposed by PFC but computed in sensory cortex, would confer a number of functional advantages, including reducing risk of an error if the attenuation signal did not reach sensory cortex promptly and freeing PFC to perform other operations for which it is well suited (see below). An early filtering operation, computed in sensory cortex even though initiated by PFC, would greatly reduce the risk of saccade generation to the peripheral no-go cue, trials on which neither additional computational effort nor cognitive operations were required. Upon detection of the pro- and anticues, however, additional executive control activity would be needed because, unlike on no-go trials, subjects still need to invest attentional resources in the peripheral location to successfully complete the computations necessary for proper behavioral performance.
PFC and Modulation of Visual Cortex Activity
The third specific issue was whether there was evidence that PFC provides an early bias signal on stimulus processing in visual cortex. Of relevance, after the initial antispecific increase in right PFC activity at 158 ms, there was a later increase in right PFC activity that peaked at 204 ms postcue. Between-conditions analyses indicated that this activity was of equally high magnitude in relation to both pro- and anticues. PFC is known to be involved in integrating perception and action planning (Fuster 1997) by, for instance, using learned rules to guide behavior in the appropriate situation (for a review, see Bunge 2004). In the present task, participants' initial decision seemed to be whether no movement was required on a particular trial (thus, the strongly attenuated extrastriate cortex activity to no-go cues at the beginning of cue evaluation). Past this initial decision, participants needed to select and initiate the appropriate response. One possibility is that this later PFC activity was associated with response selection: move toward versus move opposite the peripheral cue. Indeed, in the present task, there were no significant differences between pro- and antiresponse times (see also Wauschkuhn et al. 1998), a finding consistent with the possibility that the movement direction decision was made on the same time scale during both trial types.
Although this PFC activity at 204 ms was equal in magnitude for pro- and antitrials, its pattern of intracondition correlations with extrastriate cortex activity suggested a direct involvement with response inhibition on antitrials. At this time, there was a robust inverse relationship (r = –0.75) between right PFC and contralateral middle occipital gyrus activities only during antitrials. This effect suggests that PFC at this time was serving multiple functions. Although it may have been involved in response selection, it also appears that PFC was simultaneously modulating middle occipital gyrus activity, perhaps to suppress buildup of neuronal firing in relation to the peripheral cue that would otherwise bias output of sensorimotor systems to the generation of an unwanted response (Munoz and Everling 2004). Given the increased processing resources that such dual functions may have required, it is not surprising that superior parietal lobe activity was accentuated specifically in response to anticues at this same time (cf. Simon et al. 2002).
There was also a strong and statistically significant positive correlation (r = +0.86) between right PFC and contralateral medial inferior parietal cortex activity at 250 ms postcue. The same relationship did not differ from zero during no-go– and protrials, although the no-go correlation (r = +0.43) and anticorrelation did not differ significantly. The current task manipulations do not allow for an unambiguous interpretation of this relationship, but it may reveal a bottom-up influence of reentrant parietal cortex activity. If, for instance, attention-related inferior parietal cortex activity reached a level that was dangerously close to a response generation threshold (on the possible relationship between attention-related and saccade-generating structures, see Corbetta 1998), an inhibitory input, perhaps directly from PFC to superior colliculus, may be needed to decrease the probability of this unwanted occurrence. The PFC-inferior parietal cortex correlations suggest that such a mechanism may have been operative during both anti-, and, to a lesser extent, no-go trials.
S–R Mapping during Antiperformance
Early medial inferior parietal cortex activity (peaking at 178 ms, but starting at least 40 ms earlier; see Factor 1 in Fig. 2) had a unique role in antiperformance. In this study, brain regions with bilateral activations (middle occipital gyrus, medial inferior parietal cortex, and frontal eye fields) showed stronger responses contralateral to the visual field of cue presentation. Medial inferior parietal cortex activity was largely contralateral to cue location during no-go– and protrials, but during antitrials contra- and ipsilateral activities did not differ significantly. This finding suggests that parietal cortex activity contralateral to the response direction (but ipsilateral to the cue location) was specifically involved in correct antiperformance. Activation in the vicinity of this same inferior parietal cortex region was also identified in a previous MEG/EEG study (McDowell et al. 2005). The neural activity patterns across both studies are consistent with this medial inferior parietal cortex region being involved in the S–R mapping (or spatial remapping) process necessary for correct antiperformance (Zhang and Barash 2000; Merriam et al. 2003).
|
Conclusion |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionConclusionReferences |
|---|
The present data provide insights into how a distributed neural architecture supports the ability to manage cue evaluation and response selection and highlight the key role played by PFC in supporting such processes in the face of varying situational demands (Miller and Cohen 2001
). This study also highlights issues that should be addressed in subsequent investigations. First, pro- and antitrials did not differ on response times in our paradigm (see also Olk and Kingstone 2003), even though prosaccades are typically generated more promptly than are antisaccades when visually guided saccades are made reflexively rather than based on cue evaluation. Nevertheless, there were a number of brain activity differences consistent with expectations when comparing pro- and antitrial types (i.e., antispecific increases in early PFC activity, antispecific PFC-extrastriate cortex brain activity relationships, and evidence for a spatial remapping process in parietal cortex during antitrials). Second, brain activity differences of importance for this investigation cannot be assessed using a single imaging modality or research design. A comprehensive picture of how the brain supports response selection may require multimodal neuroimaging studies (fMRI and MEG/EEG data on the same subjects) and/or translational research involving collection of multimodal neuroimaging data on humans and single-unit data on nonhuman primates who performed the same tasks.
|
Acknowledgments |
|---|
This work was supported by grants for the United States Public Health Service (MH51129, MH001852). Conflict of Interest: None declared.
|
References |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionConclusionReferences |
|---|
Babiloni F, Babiloni C, Locche L, Cincotti F, Rossini PM, Carducci F. (2000) High-resolution electro-encephalogram: source estimates of Laplacian-transformed somatosensory-evoked potentials using realistic subject head model constructed from magnetic resonance images. Med Biol Eng Comp 38:512–519.[CrossRef][Web of Science][Medline]
Bunge SA. (2004) How we use rules to select actions: a review of evidence from cognitive neuroscience. Cogn Affect Behav Neurosci 4:4564–579.[Medline]
Carretié L, Tapia M, Mercado F, Albert J, Lopez-Martin S, de la Serna JM. (2004) Voltage-based versus factor score-based source localization analysis of electrophysiological activity: a comparison. Brain Topogr 17:2109–115.[CrossRef][Web of Science][Medline]
Corbetta M. (1998) Frontoparietal cortical networks for directing attention and the eye to visual locations: identical, independent, or overlapping neural systems? Proc Natl Acad Sci USA 95:831–838.
Corbetta M and Shulman GL. (2002) Control of goal-directed and stimulus-driven attention in the brain. Nat Rev Neurosci 3:3201–215.[Web of Science][Medline]
Dale AM and Halgren E. (2001) Spatiotemporal mapping of brain activity by integration of multiple imaging modalities. Curr Opin Neurobiol 11:2202–208.[CrossRef][Web of Science][Medline]
Dien J, Beal DJ, Berg P. (2005) Optimizing principle components analysis of event-related potentials: matrix type, factor loading weighting, extraction, and rotations. Clin Neurophysiol 116:81747–1753.[CrossRef][Web of Science][Medline]
Dien J, Spencer KM, Donchin E. (2003) Localization of the event-related potential novelty response as defined by principle components analysis. Cogn Brain Res 17:3637–650.[CrossRef][Medline]
Dyckman KA and McDowell JE. (2005) Behavioral plasticity of antisaccade performance following daily practice. Exp Brain Res 162:163–69.[CrossRef][Web of Science][Medline]
Evdokimidis I, Liakopoulos D, Constantinidis TS, Papageorgiou C. (1996) Cortical potentials with antisaccades. Electroencephalogr Clin Neuropsychol 98:5377–384.
Everling S and DeSouza JF. (2005) Rule-dependent activity for prosaccades and antisaccades in the primate prefrontal cortex. J Cogn Neurosci 17:91483–1496.[CrossRef][Web of Science][Medline]
Fecteau JH, Au C, Armstrong IT, Munoz DP. (2004) Sensory biases produce alternation advantage found in sequential saccadic eye movement tasks. Exp Brain Res 159:184–91.[Web of Science][Medline]
Ford KA, Goltz HC, Brown MR, Everling S. (2005) Neural processes associated with antisaccade task performance investigated with event-related FMRI. J Neurophysiol 94:1429–440.
Foxe JJ and Simpson GV. (2002) Flow of activation from V1 to frontal cortex in humans. A framework for defining "early" visual processing. Exp Brain Res 142:1139–150.[CrossRef][Web of Science][Medline]
Funahashi S, Chafee MV, Goldman-Rakic PS. (1993) Prefrontal neuronal activity in rhesus monkeys performing a delayed anti-saccade task. Nature 365:6448753–756.[CrossRef][Medline]
Fuster JM. (1997) Network memory. Trends Neurosci 20:10451–459.[CrossRef][Web of Science][Medline]
Hämäläinen MS and Ilmoniemi RJ. (1984) Interpreting measured magnetic fields of the brain: estimates of current distributions. Technical report TKK-F-A559(Helsinki University of Technology, Helsinki, Finland).
Hasegawa RP, Matsumoto M, Mikami A. (2000) Search target selection in monkey prefrontal cortex. J Neurophysiol 84:31692–1696.
Hauk O, Berg P, Wienbruch C, Rockstroh B, Elbert T. (1999) The minimum norm method as an effective mapping tool for MEG analysis. In Yoshimoto T, Kotani M, Kuriki S, Karibe H, Nakasato N (Eds.). Recent advances in biomagnetism (Proceedings of the 11th Conference on Biomagnetism)(Tohoku University Press, Sendai (Japan)) pp. 213–216.
Hillyard SA and Anllo-Vento L. (1998) Event-related brain potentials in the study of visual selective attention. Proc Natl Acad Sci USA 95:3781–787.
Hutton SB and Ettinger U. (2006) The antisaccade task as a research to in psychopathology: a critical review. Psychophysiology 43:302–313.[CrossRef][Web of Science][Medline]
Konishi S, Nakajima K, Uchida I, Kikyo H, Kameyama M, Miyashita Y. (1998) Common inhibitory mechanism in human inferior prefrontal cortex revealed by event-related functional MRI. Brain 122:981–991.
Leigh RJ and Zee DS. (2006) The neurology of eye movement. 4th ed (Oxford University Press, New York).
Liddle PF, Kiehl KA, Smith AM. (2001) Event-related fMRI study of response inhibition. Hum Brain Mapp 12:2100–109.[CrossRef][Web of Science][Medline]
Liu AK, Dale AM, Belliveau JW. (2002) Monte Carlo simulation studies of EEG and MEG localization accuracy. Hum Brain Mapp 16:147–62.[CrossRef][Web of Science][Medline]
Lynch JC and Tian JR. (2005) Cortico-cortical networks and cortico-subcortical loops for the higher control of eye movements. Prog Brain Res 151:461–501.[Web of Science][Medline]
Matthews A, Flohr H, Everling S. (2002) Cortical activation associated with midtrial change of instruction in a saccade task. Exp Brain Res 143:4488–498.[CrossRef][Web of Science][Medline]
McDowell JE, Brown GG, Paulus M, Martinez A, Stewart SE, Dubowitz DJ, Braff DL. (2002) Neural correlates of refixation saccades and antisaccades in normal and schizophrenia subjects. Biol Psychiatry 51:3216–223.[CrossRef][Web of Science][Medline]
McDowell JE, Kissler JM, Berg P, Dyckman KA, Gao Y, Rockstroh B, Clementz BA. (2005) Electroencephalography/magnetoencephalography study of cortical activities preceding prosaccades and antisaccades. Neuroreport 16:7663–668.[CrossRef][Web of Science][Medline]
Merriam EP, Colby CL, Thulborn KB, Luna B, Olson CR, Sweeney JA. (2001) Stimulus-response incompatibility activates cortex proximate to three eye fields. Neuroimage 13:5794–800.[Web of Science][Medline]
Merriam EP, Genovese CR, Colby CL. (2003) Spatial updating in human parietal cortex. Neuron 39:2361–373.[CrossRef][Web of Science][Medline]
Miller EK and Cohen JD. (2001) An integrative theory of prefrontal cortex function. Annu Rev Neurosci 24:167–202.[CrossRef][Web of Science][Medline]
Moore T and Armstrong KM. (2003) Selective gating of visual signals by microstimulation of frontal cortex. Nature 421:6921370–373.[CrossRef][Medline]
Moores KA, Clark CR, Hadfield JL, Brown GC, Taylor DJ, Fitzgibbon SP, Lewis AC, Weber DL, Greenblatt R. (2003) Investigating the generators of the scalp recorded visuo-verbal P300 using cortically constrained source localization. Hum Brain Mapp 18:153–77.[CrossRef][Web of Science][Medline]
Munoz DP and Everling S. (2004) Look away: the anti-saccade task and the voluntary control of eye movement. Nat Rev Neurosci 5:3218–228.[CrossRef][Web of Science][Medline]
Nakamura K, Roesch MR, Olson CR. (2004) Neuronal activity in macaque SEF and ACC during performance of tasks involving conflict. J Neurophysiol 93:2884–908.
Nunez PL and Srinivasan R. (2006) Electric fields of the brain: the neurophysics of EEG. 2nd ed (Oxford University Press, New York).
Olk B and Kingstone A. (2003) Why are antisaccades slower than prosaccades? A novel finding using a new paradigm. Neuroreport 14:151–155.[CrossRef][Web of Science][Medline]
Picton TW, Bentin S, Berg P, Donchin E, Hillyard SA, Johnson R, Miller GA, Ritter W, Ruchkin DS, Rugg MD, et al. (2000) Guidelines for using human event-related potentials to study cognition: recording standards and publication criteria. Psychophysiology 37:127–152.[CrossRef][Web of Science][Medline]
Pierrot-Deseilligny CH, Ploner CJ, Muri RM, Gaymard B, Rivaud-Pechoux S. (2002) Effects of cortical lesions on saccadic: eye movements in humans. Ann N Y Acad Sci 956:216–229.[Web of Science][Medline]
Ploner CJ, Gaymard BM, Rivaud-Pechoux S, Pierrot-Deseilligny C. (2005) The prefrontal substrate of reflexive saccade inhibition in humans. Biol Psychiatry 57:101159–1165.[CrossRef][Web of Science][Medline]
Pouget P, Emeric EE, Stuphorn V, Reis K, Schall JD. (2005) Chronometry of visual responses in frontal eye field, supplementary eye field, and anterior cingulate cortex. J Neurophysiol 94:32086–2092.
Richards JE. (2003) Cortical sources of event-related potentials in the prosaccade and antisaccade task. Psychophysiology 40:6878–894.[CrossRef][Web of Science][Medline]
Schall JD. (2001) Neural basis of deciding, choosing and acting. Nat Rev Neurosci 2:133–42.[Web of Science][Medline]
Simon SR, Meunier M, Piettre L, Berardi AM, Segebarth CM, Boussaoud D. (2002) Spatial attention and memory versus motor preparation: premotor cortex involvement as revealed by fMRI. J Neurophysiol 88:42047–2057.
Spencer KM, Dien J, Donchin E. (2001) Spatiotemporal analysis of the late ERP responses to deviant stimuli. Psychophysiology 38:2343–358.[CrossRef][Web of Science][Medline]
Srinivasan R, Nunez PL, Tucker DM, Silbrstein RB, Cadusch PJ. (1996) Spatial sampling and filtering of EEG with spline Laplacians to estimate cortical potentials. Brain Topogr 8:4355–366.[CrossRef][Web of Science][Medline]
Sweeney JA, Luna B, Keedy SK, McDowell JE, Clementz BA. Forthcoming. (2007) fMRI studies of eye movement control. Neuroimage.
Sweeney JA, Mintun MA, Kwee S, Wiseman MB, Brown DL, Rosenberg DR, Carl JR. (1996) Positron emission tomography study of voluntary saccadic eye movements and spatial working memory. J Neurophysiol 75:1454–468.
Tendolkar I, Ruhrmann S, Brockhaus-Dumke A, Pauli M, Mueller R, Pukrop R, Klosterkotter J. (2005) Neural correlates of visuo-spatial attention during an antisaccade task in schizophrenia. Schizophr Res 73:2–3 pp. 297–310.[CrossRef][Web of Science][Medline]
Trappenberg TP, Dorris MC, Munoz DP, Klein RM. (2001) A model of saccade initiation based on the competitive integration of exogenous and endogenous signals in the superior colliculus. J Cogn Neurosci 13:2256–271.[CrossRef][Web of Science][Medline]
Tzelepi A, Lutz A, Kapoula Z. (2004) EEG activity related to preparation and suppression of eye movements in three-dimensional space. Exp Brain Res 155:4439–449.[CrossRef][Web of Science][Medline]
Vanni S and Uutela K. (2000) Foveal attention modulates responses to peripheral stimuli. J Neurophysiol 83:42443–2452.
Wang J-Z and Kaufman L. (2003) Magnetic source imaging: search for inverse solutions. In Lu Z-L and Kaufman L (Eds.). Magnetic source imaging of the human brain(Lawrence Erlbaum Associates, Inc., Mahwah (NJ)) pp. 101–134.
Wauschkuhn B, Verleger R, Wascher E, Klostermann W, Burk M, Heide W, Kompf D. (1998) Lateralized human cortical activity for shifting visuospatial attention and initiating saccades. J Neurophysiol 80:62900–2910.
Zhang M and Barash S. (2000) Neuronal switching of sensorimotor transformations for antisaccades. Nature 408:6815971–975.[CrossRef][Medline]
CiteULike 
Connotea 
Del.icio.us What's this?
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||