Brain, Vol. 122, No. 6, 1093-1106,
June 1999
© 1999 Oxford University Press
A large-scale distributed network for covert spatial attention
Further anatomical delineation based on stringent behavioural and cognitive controls
1 Cognitive Neurology and Alzheimer's Disease Center, 2 Departments of Neurology and 3 Radiology, Northwestern University Medical School, 4 Chicago V.A. Healthcare System Lakeside Division, Chicago, USA and 5 Department of Experimental Psychology, University of Oxford,Oxford, UK
Correspondence to:
Darren R. Gitelman, MD, Northwestern University Medical School, 320 E. Superior St., Searle 11–470, Chicago, IL 60611, USA E-mail: d-gitelman{at}nwu.edu
 |
Abstract |
|---|
 Top Abstract IntroductionMethodsResultsDiscussionReferences |
|---|
Functional MRI was used to examine cerebral activations in 12 subjects while they performed a spatial attention task. This study applied more stringent behavioural and cognitive controls than previously used for similar experiments: (i) subjects were included only if they showed evidence of attentional shifts while performing the task in the magnet; (ii) the experimental task and baseline condition were designed to eliminate the contributions of motor output, visual fixation, inhibition of eye movements, working memory and the conditional (no-go) component of responding. Activations were seen in all three hypothesized cortical epicentres forming a network for spatial attention: the lateral premotor cortex (frontal eye fields), the posterior parietal cortex and the cingulate cortex. Subcortical activations were seen in the basal ganglia and the thalamus. Although the task required attention to be equally shifted to the left and to the right, eight of 10 subjects showed a greater area of activation in the right parietal cortex, consistent with the specialization of the right hemisphere for spatial attention. Other areas of significant activation included the posterior temporo-occipital cortex and the anterior insula. The temporo-occipital activation was within a region broadly defined as MT+ (where MT is the middle temporal area) which contains the human equivalent of area MT in the macaque monkey. This temporo-occipital area appears to constitute a major component of the functional network activated by this spatial attention task. Its activation may reflect the `inferred' shift of the attentional focus across the visual scene.
attention; hemispheric dominance; networks; spatial orientation
BA = Brodmann area; fMRI = functional MRI; FST = floor of the superior temporal visual area; MST = medial superior temporal area; MT = middle temporal area; SPM = statistical parametric map
|
Introduction |
|---|
|
TopAbstractIntroductionMethodsResultsDiscussionReferences |
|---|
The flexible orientation of attention to behaviourally relevant events is a function of great importance for adaptive interaction with the environment. We have suggested that spatial attention is mediated by a large-scale neural network, including three monosynaptically interconnected areas in the frontal, parietal and cingulate cortices (Mesulam, 1981
). Physiological and behavioural observations in monkeys and humans suggest that each of these regions makes a differential contribution to spatial attention (Chedru et al., 1973; Mesulam, 1981, 1985, 1990; Posner et al., 1984; Fox et al., 1985; Goldberg and Segraves, 1987; Zipser and Andersen, 1988; Galletti et al., 1993). The frontal eye fields and the surrounding premotor cortex provide access to a motor map for the distribution of overt exploratory behaviours, the posterior parietal cortex to a spatially addressed sensorimotor representation of the extrapersonal space and the cingulate cortex to a map for the spatial distribution of expectancy and motivational relevance (Mesulam, 1981). The effective apportionment of attention across the extrapersonal space requires the integrity of all three of these cortical areas, as well as their connections with each other and with specific subcortical regions in the thalamus and striatum. Lesions in any component of this network can lead to hemispatial neglect.
Hemispheric asymmetry is another characteristic feature of the attentional network: severe contralesional neglect is common after right hemisphere injury but rare after left hemisphere injury (Mesulam, 1981; Heilman et al., 1985; Weintraub and Mesulam, 1987; Spiers et al., 1990). The asymmetry of neglect-causing lesions has been ascribed to hemispheric specialization for the control of spatial attention, such that the right hemisphere regulates the distribution of spatial orientation across the entire extrapersonal landscape whereas the left hemisphere's influence is confined predominantly to the contralateral hemispace (Mesulam, 1981; Heilman et al., 1985).
Several functional imaging studies have examined the anatomical organization of spatial attention (Corbetta et al., 1993; Gitelman et al., 1996a; Nobre et al., 1997; Corbetta, 1998). However, with one exception (Nobre et al., 1997) the previous studies did not have the resolution to analyse activations in individual subjects, and none of them consistently controlled for response-related motor activity, response inhibition, central fixation and working memory. The present functional MRI (fMRI) study was designed to address some of these limitations with the help of a stringent control task.
|
Methods |
|---|
|
TopAbstractIntroductionMethodsResultsDiscussionReferences |
|---|
Subjects
Twelve volunteers participated in the study. There were seven men and five women, and none had any history of significant medical, neurological or psychiatric illness. Their mean (± standard deviation) age was 32.5 ± 7.2 years. All subjects were right-handed by self-report and the mean score on the modified Edinburgh handedness scale was +75.4 ± 30.2 (Ransil and Schachter, 1994
). Subjects gave their informed consent, and this study was approved by the Institutional Review Board at Northwestern University.
Behavioural tasks: general design
The behavioural task (Fig. 1) was a variant of one developed by Posner for examining endogenously triggered shifts of spatial attention in the absence of eye movements (Posner, 1980). Centrally (foveally) presented arrow cues were used to generate lateralized expectancy for subsequent target appearance. This task therefore emphasized the sensory–representational rather than the motor–exploratory components of spatial attention.
|
The background display for all tasks consisted of a small central diamond (1° wide) and two peripheral squares (1.5° wide and 7.5° eccentric in each visual field). Each active-task trial started when one side of the central diamond brightened, forming an arrow which cued the side for subsequent target appearance (Fig. 1
, top). The cue stayed on during a cue-to-target interval of 200, 400 or 800 ms. By maintaining the presence of the cue, potential working memory effects during the cue-to-target interval were minimized. A target then appeared for 100 ms, either on the same side as the cue 80% of the time (valid) or on the opposite side 20% of the time (invalid). The target stimulus was a diagonal cross (x) while the foil was an upright cross (+). The subject's task was to discriminate the target stimuli covertly (without eye movements) and to respond as quickly and accurately as possible by pressing a response button only when the target was a diagonal cross. Each trial lasted 2 s. Reaction times were recorded in the magnet as well as in a preceding separate pilot session in the cognitive laboratory.
In the baseline condition (Fig. 1, bottom) the central diamond and both side boxes were displayed every 2 s. The central diamond was either bold or not bold. The side boxes contained either a diagonal or an upright cross on both sides. Subjects were told to press the response button during the baseline condition if and only if the central stimulus was bold, regardless of the shape of the peripheral targets. Therefore the maintenance of central fixation to detect a bold diamond (baseline) or central arrow (active-task) and to avoid lateralized saccades to targets were features of both the baseline and the experimental conditions. The baseline task also controlled for the motor response, the conditional (no-go) aspect of responding and the visual display. However, the baseline task did not require shifts of spatial attention for successful performance. The fMRI experimental run itself was designed as a series of paired active and baseline conditions each lasting 30.5 s and alternating a total of eight times.
Behavioural tasks: cognitive laboratory
In order to train subjects and derive some measure of eye movement during the task, all subjects performed an initial set of studies in the cognitive psychophysics laboratory outside the magnet. In the cognitive laboratory we assessed (i) whether subjects could maintain central fixation throughout all phases of the task, and (ii) whether subjects showed evidence of the invalidity effect by having a larger reaction time when the cue and target were in different locations. The presence of an invalidity effect proved that the cue induced a covert shift of the attentional focus. Stimuli were presented using Superlab software (Cedrus, Phoenix, Ariz., USA) running on a Macintosh PowerPC computer (Apple, Cupertino, Calif., USA). Subjects viewed the stimuli 40 cm away from a 21-inch monitor. Eye movements were recorded using an ISCAN® infra-red monitoring system (ISCAN, Burlington, Mass., USA). The eye movement data were examined using the ILAB software (D. Gitelman), which displayed a map of eye position overlaid on the stimuli the subjects were viewing (Fig. 2). The reaction time data were analysed by a repeated measures ANOVA (analysis of variance) with factors for side and validity. Subjects who did not show the invalidity effect or those who could not maintain fixation within 2° of the centre in >90% of the trials did not undergo fMRI.
|
Behavioural tasks: functional imaging
Reaction times for the subjects' performance in the magnet were also analysed; subjects had to show an invalidity effect of at least 30 ms for their data to be included in the analysis. This requirement ensured that subjects were performing the task as requested during the imaging and that the task was producing the desired attentional shifts. For the fMRI group, reaction time data were analysed by a repeated measures ANOVA with factors for side and validity.
Stimuli were presented using software and hardware similar to that described above. The images were projected by a Proxima active matrix LCD (liquid crystal display) projector onto a custom-designed, non-magnetic rear-projection screen. Subjects viewed the screen located ~170 cm from their eyes via a mirror.
MRI: anatomical scans
T1-weighted anatomical images were obtained on a 1.5 T Siemens Vision scanner, using a 3D FLASH (fast low angle shot) sequence with the following parameters [repetition time/echo time (TR/TE) 15 ms/6 ms, flip angle 20°, field of view (FOV) 220 mm, matrix 256 x 256, slice thickness 1.0 mm]. All anatomical and functional scans were obtained in transaxial planes parallel to the anterior commissure–posterior commissure (AC–PC) line.
MRI: functional scans
Functional MRI volumes were acquired as thirty-two 4 mm slices using a single-shot EPI (echo planar imaging) method (TR/TE 4350/40 ms, flip angle 90°, matrix 64 x 64, FOV 220 mm). For all functional runs, the signal was allowed to equilibrate over four initial volumes that were excluded from the analysis.
Subjects had their head immobilized with a vacuum pillow (Vac-Fix, Bionix, Toledo, OH, USA) and the restraint callipers built into the head coil. They were given a non-magnetic button which enabled recording of their responses. A vitamin E capsule was taped to the left temporal region to mark laterality for image processing.
MRI: data analysis
Analysis of fMRI data was performed using SPM-96 software (Wellcome Department of Cognitive Neurology, London, UK) running under the MATLAB environment (Mathworks, Sherborn, Mass., USA) (Friston et al., 1995a, b, 1996; Worsley and Friston, 1995) on a HP-UX workstation (Hewlett-Packard, Palo Alto, Calif., USA). Application Visualization System software (Advanced Visual Systems, Waltham, Mass., USA) was used for scalp editing, volume rendering and image display. The Application Visualization System also enabled us to localize activations with respect to sulcal landmarks by using custom-designed software that projected a mark made on the `surface' of a three-dimensional ray-traced image to the corresponding two-dimensional slice, and vice versa (Damasio and Frank, 1992).
The MRI data were subjected to a number of preprocessing steps. All functional images were realigned to the image taken proximate to the anatomical study (Friston et al., 1995a) using affine transformation routines built into SPM-96. Subject movement was always <0.44 mm in any direction during each run (Parrish et al., 1998). The realigned scans were then normalized using the template image supplied with SPM-96. This template conforms to the space defined by the International Consortium for Brain Mapping project (ICBM) (NIH P-20 grant), and closely approximates the space described in the atlas of Talairach and Tournoux (Talairach and Tournoux, 1988). Functional data were then smoothed with a 7 mm isotropic Gaussian kernel.
SPM-96 uses the technique of statistical parametric mapping and references the probabilistic behaviour of a stationary Gaussian field for defining an image's probability density function. It employs the general linear model for testing hypotheses at each voxel. Contrasts were set up to test for voxelwise effects of signal differences between the baseline and active conditions, and SPM{t} and SPM{Z} maps were calculated (Friston et al., 1995b). We examined contrasts both for individual subjects and for the group as a whole.
All SPM{Z} values were thresholded at P 
0.001 (Z = 3.09). The threshold for reporting activations in areas of no anatomic hypothesis was P 0.05 (corrected for multiple comparisons over the entire brain volume). In regions of a priori hypothesis, this correction was made using the unified approach described by Worsley and colleagues (Worsley et al., 1996) for the specific volumes of interest. The a priori regions of hypothesis included the bilateral frontal eye fields, posterior parietal cortex, supplementary eye fields, anterior cingulate cortex, basal ganglia and posterior thalami.
In order to compare hemispheric differences in activation on a regional level, we defined regions of interest in homologous areas of each hemisphere based on our a priori hypotheses, and counted the number of activated voxels in these regions of interest at a mapwise threshold of P 0.001. A lateralization index (LI) was computed for each subject according to the formula LI = (VR – VL)/(VR + VL), where VL and VR are the number of activated voxels in each hemisphere in the left and right regions of interest, respectively. A sign test was then performed across subjects for those with positive (i.e. rightward) versus negative (i.e. leftward) lateralization indices (Desmond et al., 1995; Binder et al., 1996).
|
Results |
|---|
|
TopAbstractIntroductionMethodsResultsDiscussionReferences |
|---|
Behavioural data
In the cognitive laboratory, all subjects included in the analyses were able to maintain gaze within 2.0° of the centre on >90% of the trials. In addition there was no significant difference in eye movements between the rest and active conditions (P > 0.3, two-tailed on a paired t test for any movements beyond the specified region of interest). The eye movements of one subject in the cognitive laboratory are shown in Fig. 2
. The positions of the central regions of interest and the target boxes are also illustrated. Nine subjects showed a significant effect of validity in the magnet (P < 0.05, one-tailed) with 33–217 ms (mean 81.2 ± 58.0 ms) slower response times to invalid trials. One additional subject showed 34 ms of slowing to invalid trials and a trend towards a significant difference between the valid and invalid conditions (P = 0.075). His data were therefore included in the analysis. Two other subjects were excluded from fMRI analysis as they failed to show an invalidity effect in the magnet (P > 0.1). Thus, for the entire group of 10 subjects that were retained, the mean reaction times were 470 ± 66.1 ms for valid trials and 553 ± 109.0 ms for invalid trials (P < 0.001, Fig. 3). Reaction times did not show an interaction with target hemispace. During the rest condition, subjects had a mean reaction time of 431 ± 71.2 ms to the bold central diamond.
|
fMRI data
Parietal activations
The group activation results are shown in Fig. 4A
, and coordinates of representative activations are given in Table 1. The greatest area of parietal activation in the group was over the interparietal sulcus/superior parietal lobule junction. The other frequent region of parietal activation in the group was the precuneus.
|
|
Analysis of the individual subjects demonstrated activations in all 10 subjects in the posterior parietal cortex (Tables 1 and 2
and Fig. 4A, B, D and E). The activation of subject 6 is shown in Fig. 4B. In order to localize more precisely the activations relative to parietal lobe anatomy, the interparietal sulcus was traced on the cortical surface of each subject's three-dimensional brain image, and then projected onto the two-dimensional horizontal slices. The interparietal sulcus was defined according to the guidelines of Critchley (Critchley, 1953). The interparietal sulcus included a horizontal ramus (the interparietal sulcus proper), an anterior vertical limb (the inferior postcentral sulcus) and a posterior vertical limb (the descending interparietal sulcus including its junction with the interoccipital sulcus up to the transverse occipital sulcus). An activation was labelled as originating from the banks of the interparietal sulcus if its coordinates fell within a full width at half-maximum (FWHM 
8 mm) of the interparietal sulcus as determined by each subject's traced interparietal sulcus line. Figure 4E demonstrates the tracing results in subject 3. Most of the parietal activations in this subject involved the interparietal sulcus cortex as defined above. There was also a single area of superior parietal lobule/precuneus activation best seen on section ii of Fig. 4E.
|
The other nine subjects also showed activations within the interparietal sulcus region. In addition, activations were seen in the superior parietal lobule in seven subjects and in the precuneus in six subjects. Two subjects showed very small activations (13 and 10 voxels, respectively) in the inferior parietal lobule (supramarginal gyrus). Figure 5
plots the lateralization index for voxels in the parietal lobe for each subject. Eight out of 10 subjects showed a rightward bias of this index [P < 0.05 by the sign test for a rightward (positive) versus leftward (negative) lateralization index].
|
Frontal activations
Group analysis demonstrated activation of premotor cortex (frontal eye fields) along the precentral sulcus involving the posterior part of the middle frontal gyrus and, less frequently, the posterior part of the superior frontal gyrus bilaterally (Table 1
and Fig. 4A). This activation site was consistent with identified sites of frontal eye field activation in previous studies (Paus, 1996; Nobre et al., 1997). Supplementary motor cortex activation was also seen in the group analysis. Nine of 10 individuals displayed activation of the frontal eye fields. Consistent hemispheric asymmetry was not evident at this site, as most subjects showed approximately symmetrical activations. In the medial frontal cortex, small areas of activation were seen in the supplementary motor region (6/10 subjects). Other small, individual frontal areas of activation included the inferior and middle frontal gyri bilaterally in four individuals. The locations of these prefrontal activations would correspond to the inferior aspects of Brodmann areas (BA) 44, 45 and 46. The prefrontal regions of activation were not seen in the group analysis, probably because of the interindividual variability and small size of the activations.
Cingulate activations
Activation of the anterior cingulate gyrus was seen both in the group analysis and in 6/10 individuals. This activation was primarily located over the dorsal aspect of the gyrus and within the cingulate sulcus itself (Fig. 4B). No hemispheric asymmetry was noted for this activation.
Temporal and insular activations
Activations were seen in the posterior temporal lobes bilaterally. While activation at this site was not hypothesized a priori, the height and size of the activations did survive correction for multiple comparisons. The largest region of posterior temporal activation was seen overlying the middle and inferior temporal gyri in BA 21 and 37 bilaterally (indicated by TO in Fig. 4A, B, D and E). Nine subjects showed activation in this area, and there was no consistent hemispheric asymmetry.
A smaller region of temporal activation was located near the intersection of the posterior superior temporal gyrus (BA 22) and the inferior parietal lobule (BA 40). Although activations were seen in this area in four subjects on the left and five subjects on the right (seven individuals in total), the group analysis showed this superior temporal activation as significant only on the right side (ST in Fig. 4A).
Figure 6 plots the primary focus of the group temporo-occipital activation, along with foci of activations previously identified as human area MT (middle temporal area), and a left hemisphere region found to subserve movement-related knowledge (Martin et al., 1995). As shown in Fig. 6, the maximum focus of temporo-occipital activation in the current study was located more inferiorly and laterally to the area identified as MT. It had a vector distance of 18 mm from the average of the MT regions on either side. The overall area of the temporo-occipital activation would overlap a region designated as MT+ (DeYoe et al., 1996; Tootell et al., 1998).
|
Bilateral activations were seen in the anterior insula in the group analysis (Fig. 4C
). Only one subject had a suprathreshold activation (accounting for multiple comparisons) in this area in the individual analyses. However, an additional three subjects showed subthreshold activations of the insula (Z scores ranging from 4.05 to 4.62).
Visual activations
Activations in visual cortex survived correction for multiple comparisons in six subjects. In the group analysis, the left-sided visual cortex activation was seen overlying the inferior occipital gyrus corresponding to BA 19, whereas right-sided activation was located in the fusiform gyrus (Table 1). In individual subjects, these activations occurred in regions corresponding to BA 18 and 19 and included the inferior occipital and fusiform gyri (FG in Fig. 4D).
Subcortical activations
Subcortical regions of activation included the thalamus in five subjects and the basal ganglia in seven subjects. Thalamic activation was difficult to localize precisely owing to the limited resolution of the functional data. The group analysis showed right thalamic activation possibly overlapping the ventral lateral nucleus (x, y, z = 12, –3, 3; Z = 4.73). However, individual activations appeared to overlap variably the mediodorsal, anterior pulvinar and ventral lateral or intralaminar thalamic nuclei.
Basal ganglia activation showed no consistent hemispheric asymmetry and was located in the caudate in two subjects and the putamen in five. In the group analysis, basal ganglia activations were seen in the left putamen (x, y, z = –18, –6, 15; Z = 5.62) and the right putamen/globus pallidus, with respective coordinates of x, y, z = 27, 6, –6; Z = 5.27 and 18, –3, 12; Z = 5.00 (Fig. 4C).
|
Discussion |
|---|
|
TopAbstractIntroductionMethodsResultsDiscussionReferences |
|---|
We have previously proposed a network approach to the anatomical organization of spatial attention (Mesulam, 1981
, 1990). This model was based on three major assumptions: (i) spatial attention is co-ordinated by a network of interconnected cortical areas, the epicentres of which are located in the frontal eye fields, posterior parietal cortex and cingulate gyrus; (ii) the frontal eye field has a relative specialization for the overt motor–exploratory aspect of spatial attention, the posterior parietal cortex for the sensory–representational and sensorimotor aspects, and the cingulate gyrus for the limbic–motivational aspect; and (iii) the right hemisphere co-ordinates attentional deployment across the entire extrapersonal space, whereas the left hemisphere primarily influences contraversive orientation in the right hemispace. These hypotheses were addressed in this study, based on a task that emphasized the sensory–representational aspect of spatial attention. In our task, foveally presented visual cues triggered covert shifts of the attentional focus, without any head or eye movements. Activations were seen in all major components of the cortical network originally hypothesized to control spatial attention: the posterior parietal cortex, the frontal eye fields and the cingulate gyrus. Although the task required symmetrical shifts of attention to both sides of space, 80% of subjects showed a greater area of right-sided parietal activation, consistent with the hypothesized specialization of this hemisphere for spatial attention.
Similar sites of activation (frontal, parietal and cingulate) have been seen in several previous studies of spatial attention. These studies were based on tasks that included visual targets, auditory targets, spatial priming, expectancy shifts and manual exploration (Corbetta et al., 1993; Gitelman et al., 1996a; Pugh et al., 1996; Kim et al., 1997; Nobre et al., 1997; Tzourio et al., 1997). The substantial overlap in activations among these studies of attention suggests that these three cortical regions provide a flexible large-scale substrate for all spatial attentional behaviour.
The current experiment was designed to address a number of conceptual and methodological issues that were incompletely examined in previous studies. The issues we considered to be of particular importance included (i) in-scanner task performance, (ii) achieving a better anatomical characterization of the activations in individual subjects, and (iii) controlling for task requirements that could potentially confound the interpretation of frontal eye field activations. These confounds included (a) the requirement for push-button responses, (b) the conditional (no-go) aspect of the required response, (c) the working memory load of the task, (d) the need for intense central fixation, and (e) the need to inhibit eye movements. Each of these considerations is addressed below.
All 10 subjects showed a slowing of reaction time of >30 ms to invalid cues during their performance of the task in the magnet. Our requirement that reaction times to validly cued stimuli be faster than to invalidly cued stimuli ensured that the cues shifted the focus of attention.
The role of the frontal eye fields in the attentional network has generated controversy. Conceivably, its activation in tasks of covert attention could be contingent on the requirement for responding (Corbetta et al., 1993), the undetected occurrence of eye movements (Sweeney et al., 1996; Bodis-Wollner et al., 1997; Luna et al., 1998), the intensity of central fixation (Anderson et al., 1994; Law et al., 1997), the need to inhibit eye movements (Corbetta et al., 1993; Law et al., 1997), the no-go contingency of the response (Kawashima et al., 1996b) or the working memory load of the task (Sweeney et al., 1996).
In the current study, subjects had to make push-button responses in both the active and the baseline conditions. Furthermore, responses had a conditional (no-go) component during both conditions, such that subjects had to inhibit reactions to the + sign during the active task, and to the 
(non-bold diamond) during the baseline task. The active and baseline conditions also had similar requirements for intense central fixation (i.e. to detect if the central diamond was wholly or partly bold) and for inhibition of eye movements in the presence of peripheral stimuli. Given these controls, the frontal activations in our subjects are unlikely to have been secondary to factors such as the motor response, go/no-go contingency or central fixation. However, the cognitive subtraction model used in this study cannot entirely rule out the possibility that the frontal eye field activation reflected an interaction between one of these factors and attention, rather than the effect of spatial attention alone.
The contribution of undetected eye movements to frontal eye field activations was a potential concern. There is currently no available method for monitoring eye movements of <5° during the acquisition of fMRI data without also introducing major artefacts in the imaging of the frontal lobes. Nevertheless, there are several reasons for proposing that the frontal eye field activation seen in this study should not be attributed to eye movements. First, as noted above, our subjects were highly trained to maintain central fixation. Secondly, in another study (Nobre et al., 1998) we specifically tested the possibility that eye movements could account for frontal eye field activations in a covert spatial attentional task similar to that used in the current study. The baseline task in that study required subjects to make vigorous horizontal saccadic eye movements of 4.2° or 24.3° every 400–1200 ms, with no specific emphasis on shifting spatial attention. When the activation elicited by this saccade task was subtracted from that of the covert attention shifting task, residual activity was still seen in the frontal eye fields, indicating that at least some of the frontal eye field activation during covert spatial attentional shifts could be attributed to the attentional rather than the oculomotor aspects of the task. A recent visual–motion study came to similar conclusions regarding attentional influences on frontal eye field activity (Buchel et al., 1998).
Some paradigms that have been used for the investigation of spatial attention contain a potentially confounding requirement for working memory. The study by Corbetta et al., for example, included cue-to-target delays as long as 2000 ms, requiring the subject to hold the spatial information in working memory during that interval (Corbetta et al., 1993). By contrast, in the present study the cue was visible until the target appeared, eliminating spatial working memory requirements. This is a very important consideration since the region of the frontal eye fields may also be activated by working memory (Cohen et al., 1997; Courtney et al., 1997; LaBar et al., 1998), a possibility that was minimized in this study by the design of the task.
Activation was also noted in visual association cortices such as the inferior occipital gyrus on the left and the fusiform gyrus on the right. These regions both potentially overlap areas of the visual cortex that appear to be active in processing object form (Tootell et al., 1998). Thus, in the current task, activity in these regions may have reflected the need to discriminate between the diagonal cross (x) and the vertical cross (+) during the active but not the baseline task.
An area that showed prominent activation despite the stringent controls was centred around the temporo-occipital junction, at the posterior part of the middle temporal gyrus (TO in Fig. 4A, B, D and E). This region does not overlap the location of areas activated by form discrimination (Van Essen and Drury, 1997). Its activation is thus unlikely to reflect the requirement to discriminate between the x and the +. Therefore, we conclude that the temporo-occipital activation is linked to the covert shifts of attention required by the experimental task. In some subjects, this activation extended into the adjacent parts of the superior and inferior temporal sulci. A similar region of activation was noted in the study by Nobre and colleagues (Nobre et al., 1997). The centroid for the temporo-occipital activation was in the vicinity of two motion-related areas: the movement-sensitive area V5 (MT) and an area activated by the evocation of knowledge related to action (Martin et al., 1995). Figure 6 shows the anatomical relationships among these activation sites.
The temporo-occipital activation does overlap the region designated as MT+ (DeYoe et al., 1996; Tootell et al., 1998). This area probably includes the human homologues of area MT as well as MST (medial superior temporal area), FST (floor of the superior temporal visual area) and other motion-related areas initially identified within the superior temporal sulcus of the macaque (Boussaoud et al., 1990; DeYoe et al., 1996). In the macaque monkey, area MST (and perhaps also FST) is interconnected with all three cortical epicentres of the attentional network: the frontal eye fields, the inferior parietal lobule–intraparietal sulcus and the cingulate cortex (Mesulam et al., 1977; Barbas and Mesulam, 1981; Mesulam, 1981; Boussaoud et al., 1990). Areas MST and FST are also interconnected with adjacent inferotemporal polysensory areas of the superior temporal sulcus and TEO (temporo-occipital area) (Boussaoud et al., 1990). This region therefore provides a synaptic bridge between the dorsal and ventral streams of visual processing, a level of integration that may be important for moving the focus of attention among objects in the extrapersonal space.
Neurones in MST and FST have larger receptive fields than those in MT. Furthermore, MST neurones have response properties which suggest that they use information about motion to generate signals for smooth pursuit eye movements (Dürsteler and Wurtz, 1988). The other three cortical components of the attentional network (posterior parietal cortex, frontal eye fields and cingulate cortex) play critical roles in the control of saccadic eye movements (Petit et al., 1996). Involvement in the generation of eye movements may therefore be an important common property of all cortical areas playing a critical role in spatial attention, even when the attentional shifts are done without actual eye movements (Nobre et al., 1998). This relationship emphasizes the tight interactions between the `perceptual' and `exploratory' components of spatial attention (Mesulam, 1981). It has also been hypothesized that area MST is sensitive to the inferred movement of occluded targets (Maunsell, 1995). Thus, it is possible that the activation of this area reflects an inferred movement of the attentional focus across the visual scene.
In the monkey, areas MST and FST are located in the posterior part of the superior temporal sulcus, a region that was included in the `parietal' component of the large-scale network for spatial attention (Mesulam, 1990). Lesions in the inferior parietal lobule in the monkey elicit contralesional extinction and reaching deficits whereas lesions along the banks of the superior temporal sulcus (probably including MST and FST) elicit neglect even for unilaterally presented stimuli (Lynch and McLaren, 1989; Watson et al., 1994). In humans, posterior cortical lesions that yield extinction and other components of contralesional neglect tend to be centred within the posterior parietal cortex, and may extend into the region of the temporo-occipital activation (Bisiach et al., 1981; Friedrich et al., 1998). Hasselbach and Butter (Hasselbach and Butter, 1997) reported a 53-year-old patient with a lesion confined to the region corresponding to the temporo-occipital (TO) activation in Fig. 4. This patient had no visual field cut or extinction but did show moderate left-sided neglect in a target cancellation task, even 4 years after sustaining a stroke. In view of these considerations, it appears that the posterior part of the middle temporal gyrus (TO in Fig. 4) may play a prominent role in spatial attention. Additional work will be required to determine if the role of this temporo-occipital area is confined to visually guided attentional tasks or if it also has a multimodal scope, similar to that of the other three cortical components of the attentional network.
The literature based on focal lesions contains a controversy concerning the location of the parietal area which is critical for spatial attention. Some authors suggest that the critical area is in the superior parietal lobule (Posner et al., 1984; Vallar, 1993) while others emphasize a location in the inferior parietal lobule (Vallar, 1993; Karnath, 1997; Driver and Mattingley, 1998). This study, as well as the previous functional imaging experiment of Nobre et al. and a meta-analysis by Corbetta, shows that the `parietal' epicentre for the attentional network is centred along the posterior interparietal sulcus and that its extension into other parts of the posterior parietal cortex occurs predominantly in the direction of the superior parietal lobule (Nobre et al., 1997; Corbetta, 1998).
In the monkey, the interparietal sulcus contains a mosaic of discrete areas known as LIP (lateral intraparietal area), AIP (anterior intraparietal area), VIP (ventral intraparietal, MIP (medial intraparietal area) and PIP (posterior intraparietal area). Neurones of the LIP encode the head-centred coordinates of visual events and the intention to make them the targets of saccadic eye movements (Zipser and Andersen, 1988; Snyder et al., 1998). Furthermore, in a covert spatial attention task similar to the one we used, many interparietal sulcus neurones responded best when attention was directed towards the contralateral hemifield (Robinson et al., 1995). Areas AIP and MIP play important roles in visuosomatosensory integration, visuomotor control and visual reaching (Gallese et al., 1997; Johnson et al., 1997), and movement-sensitive neurones in area VIP participate in the encoding of optic flow and self-motion (Bremmer et al., 1997). In the human, the banks of the interparietal sulcus have been activated by tasks of visually guided saccadic eye movements and visually guided manual reaching (Clower et al., 1996; Kawashima et al., 1996a; Luna et al., 1998). The interparietal sulcus region thus displays physiological specializations that are consistent with its role as one of the most critical areas in the control of spatial attention.
Control tasks based on passive rest have suggested that covert shifts of spatial attention could lead to frontal eye field and posterior parietal activations of comparable magnitude (Gitelman et al., 1996b). The introduction of more stringent controls in the current experiment, however, shows that the activations in the banks of the interparietal sulcus and in the temporo-occipital region are far more prominent than those in the frontal eye fields, highlighting the greater involvement of posterior association cortices in tasks of covert spatial attention which emphasize the sensory–representational rather than the motor–exploratory aspects of the task. Additional work will be needed to show if the reverse relationship will be identified in tasks that emphasize the exploratory (i.e. overt) aspects of spatial attention.
In previous studies in which the baseline task consisted of passive viewing without motor responses, we detected prominent activation in the supplementary motor area and cerebellum during covert attentional shifts (Kim et al., 1997; Nobre et al., 1997). In the present set of experiments, covert shifts of attention did not elicit cerebellar activation and the activation in the supplementary motor area was quite modest. Thus, the use of stringent controls helped to show that the cerebellum is unlikely to play a major role in covert shifts of spatial attention and that the role of the supplementary motor area is not as prominent as might have been surmised on the basis of earlier work. In fact, the investigation of patients with focal lesions shows that cerebellar lesions do not impair the covert shifting of spatial attention (Yamaguchi et al., 1998).
Basal ganglia activations were seen in seven subjects: five in the putamen and two in the caudate. Activation of the thalamus occurred in the region of the ventral lateral nucleus. A recent functional imaging study has reported the activation of this nucleus in relation to attention and arousal (Portas et al., 1998). Its activation may therefore reflect sustained attentional effort rather than shifts of spatial attention. However, individual subjects also showed activation of the dorsomedial and pulvinar nuclei, areas known to be heavily interconnected with the cortical epicentres of the spatial attentional network and to participate in the control of visual attention (Mesulam, 1981; LaBerge and Buchsbaum, 1990). Additional experiments will be necessary to better define the localization of the subcortical activations associated with spatial attention.
Table 1 shows the presence of interindividual differences in the distribution of parietal activations. Such individual patterns may reflect variations in the strategies used to accomplish the experimental task (Mazziotta et al., 1982) or in the anatomical organization of the relevant neural network. Exploring individual differences in the functional organization of the brain will undoubtedly become one the most challenging directions for future research.
|
Acknowledgments |
|---|
We are indebted to Drs K. Friston, R. Frackowiak and their colleagues (Wellcome Department of Cognitive Neurology, London, UK) for providing and supporting the SPM software, and to Dr Roger Ray for assistance with programming. We are grateful to Drs D. Enzmann, P. Finn and E. Russell (Department of Radiology) for their support of this research. This work was supported by the McDonnell-Pew Program in Cognitive Neuroscience 96–23 (D.R.G.), NIH grant NS30863–04 (M.M.M.), a project grant from The Welcome Trust and a research fellowship from the McDonnell-Pew Centre in Cognitive Neuroscience (A.C.N.), a NARSAD young investigator award (K.S.L.) and a fellowship from KSEF/CBNU, Korea (Y.H.K.).
|
References |
|---|
|
TopAbstractIntroductionMethodsResultsDiscussionReferences |
|---|
Anderson TJ, Jenkins IH, Brooks DJ, Hawken MB, Frackowiak RS, Kennard C. Cortical control of saccades and fixation in man: a PET study. Brain 1994; 117: 1073–84.
Barbas H, Mesulam M-M. Organization of afferent input to subdivisions of area 8 in the rhesus monkey. J Comp Neurol 1981; 200: 407–31.[ISI][Medline]
Beauchamp MS, Cox RW, DeYoe EA. Graded effects of spatial and featural attention on human area MT and associated motion processing areas. J Neurophysiol 1997; 78: 516–20.
Binder JR, Swanson SJ, Hammeke TA, Morris GL, Mueller WM, Fischer M, et al. Determination of language dominance using functional MRI: a comparison with the Wada test. Neurology 1996; 46: 978–84.
Bisiach E, Capitani E, Luzzatti C, Perani D. Brain and conscious representation of outside reality. Neuropsychologia 1981; 19: 543–51.[ISI][Medline]
Bodis-Wollner I, Bucher SF, Seelos KC, Paulus W, Reiser M, Oertel WH. Functional MRI mapping of occipital and frontal cortical activity during voluntary and imagined saccades. Neurology 1997; 49: 416–20.
Boussaoud D, Ungerleider LG, Desimone R. Pathways for motion analysis: cortical connections of the medial superior temporal and fundus of the superior temporal visual areas in the macaque. J Comp Neurol 1990; 296: 462–95.[ISI][Medline]
Bremmer F, Duhamel J-R, Hamed SB, Graf W. The representation of movement in near extra-personal space in the macaque ventral intraparietal area (VIP). In: Thier P, Karnath H-O, editors. Parietal lobe contributions to orientation in 3D space. Berlin: Springer; 1997. p. 619–30.
Buchel C, Josephs O, Rees G, Turner R, Frith CD, Friston KJ. The functional anatomy of attention to visual motion. A functional MRI study. Brain 1998; 121: 1281–94.
Chawla D, Phillips J, Buechel C, Edwards R, Friston KJ. Speed-dependent motion-sensitive responses in V5: an fMRI study. Neuroimage 1998; 7: 86–96.[ISI][Medline]
Chedru F, Leblanc M, Lhermitte F. Visual searching in normal and brain-damaged subjects. Cortex 1973; 9: 94–111.[Medline]
Clower DM, Hoffman JM, Votaw JR, Faber TL, Woods RP, Alexander GE. Role of posterior parietal cortex in the recalibration of visually guided reaching. Nature 1996; 383: 618–21.[Medline]
Cohen JD, Perlstein WM, Braver TS, Nystrom LE, Noll DC, Jonides J, et al. Temporal dynamics of brain activation during a working memory task [see comments]. Nature 1997; 386: 604–8. Comment in: Nature 1997; 386: 559–60.[Medline]
Corbetta M. Frontoparietal cortical networks for directing attention and the eye to visual locations: identical, independent, or overlapping neural systems? [Review]. Proc Natl Acad Sci USA 1998; 95: 831–8.
Corbetta M, Miezin FM, Shulman GL, Petersen SE. A PET study of visuospatial attention. J Neurosci 1993; 13: 1202–26.[Abstract]
Courtney SM, Ungerleider LG, Keil K, Haxby JV. Transient and sustained activity in a distributed neural system for human working memory [see comments]. Nature 1997; 386: 608–11. Comment in: Nature 1997; 386: 559–60.[Medline]
Critchley M. The parietal lobes. London: Edward Arnold; 1953.
Damasio H, Frank R. Three-dimensional in vivo mapping of brain lesions in humans. Arch Neurol 1992; 49: 137–43.[Abstract]
Desmond JE, Sum JM, Wagner AD, Demb JB, Shear PK, Glover GH, et al. Functional MRI measurement of language lateralization in Wada-tested patients. Brain 1995; 118: 1411–9.
DeYoe EA, Carman GJ, Bandettini P, Glickman S, Wieser J, Cox R, et al. Mapping striate and extrastriate visual areas in human cerebral cortex. Proc Natl Acad Sci USA 1996; 93: 2382–6.
Driver J, Mattingley JB. Parietal neglect and visual awareness. Nat Neurosci 1998; 1: 17–22.[ISI][Medline]
Dupont P, De Bruyn B, Vandenberghe R, Rosier A-M, Michiels J, Marchal G, et al. The kinetic occipital region in human visual cortex. Cereb Cortex 1997; 7: 283–92.
Dürsteler MR, Wurtz RH. Pursuit and optokinetic deficits following chemical lesions of cortical areas MT and MST. J Neurophysiol 1988; 60: 940–65.
Fox PT, Fox JM, Raichle ME, Burde RM. The role of cerebral cortex in the generation of voluntary saccades: a positron emission tomographic study. J Neurophysiol 1985; 54: 348–69.
Friedrich FJ, Egly R, Rafal RD, Beck D. Spatial attention deficits in humans: a comparison of superior parietal and temporal–parietal junction lesions. Neuropsychology 1998; 12: 193–207.[ISI][Medline]
Friston KJ, Ashburner J, Frith CD, Poline J-B, Heather JD, Frackowiak RSJ. Spatial registration and normalization of images. Hum Brain Mapp 1995a; 3: 165–89.
Friston KJ, Holmes AP, Worsley KJ, Poline J-B, Frith CD, Frackowiak RSJ. Statistical parametric maps in functional imaging: a general linear approach. Hum Brain Mapp 1995b; 2: 189–210.
Friston KJ, Williams SR, Howard R, Frackowiak RS, Turner R. Movement-related effects in fMRI time-series. Magn Reson Med 1996; 35: 346–55.[ISI][Medline]
Gallese V, Fadiga L, Fogassi L, Luppino G, Murata A. A parietal-frontal circuit for hand grasping movements in the monkey: evidence from reversible inactivation experiments. In: Thier P, Karnath H-O, editors. Parietal lobe contributions to orientation in 3D space. Berlin: Springer; 1997. p. 255–70.
Galletti C, Battaglini PP, Fattori P. Parietal neurons encoding spatial locations in craniotopic coordinates. Exp Brain Res 1993; 96: 221–9.[ISI][Medline]
Gitelman DR, Alpert NM, Kosslyn S, Daffner K, Scinto L, Thompson W, et al. Functional imaging of human right hemispheric activation for exploratory movements. Ann Neurol 1996a; 39: 174–9.[ISI][Medline]
Gitelman DR, Nobre AC, Meyer JR, Parrish TB, Callahan C, Russell EJ, et al. Functional magnetic resonance imaging of covert spatial attention. Neuroimage 1996b; 3: S180.
Goldberg ME, Segraves MA. Visuospatial and motor attention in the monkey. Neuropsychologia 1987; 25: 107–18.[ISI][Medline]
Hasselbach M, Butter CM. Ipsilateral displacement of egocentric midline in neglect patients with, but not in those without, extensive right parietal damage. In: Thier P, Karnath H-O, editors. Parietal lobe contributions to orientation in 3-dimensional space. Berlin: Springer, 1997. p. 579–96.
Heilman KM, Watson RT, Valenstein E. Neglect and related disorders. In: Heilman KM, Valenstein E, editors. Clinical neuropsychology. 2nd ed. New York: Oxford University Press; 1985. p. 243–93.
Johnson PB, Ferraina S, Garasto MR, Battaglia-Meyer A, Ercolani L, Burnod Y, et al. From vision to movement: cortico-cortical connections and combinatorial properties of reaching-related neurons in parietal areas V6 and V6A. In: Thier P, Karnath H-O, editors. Parietal lobe contributions to orientation in 3D space. Berlin: Springer; 1997. p. 221–36.
Karnath HO. Spatial orientation and the representation of space with parietal lobe lesions. Philos Trans R Soc Lond B: Biol Sci 1997; 352: 1411–9.[ISI][Medline]
Kawashima R, Naitoh E, Matsumura M, Itoh H, Ono S, Satoh K, et al. Topographic representation in human intraparietal sulcus of reaching and saccade. Neuroreport 1996a; 7: 1253–6.[ISI][Medline]
Kawashima R, Satoh K, Itoh H, Ono S, Furumoto S, Gotoh R, et al. Functional anatomy of GO/NO-GO discrimination and response selection—a PET study in man. Brain Res 1996b; 728: 79–89.[ISI][Medline]
Kim YH, Gitelman DR, Nobre AC, Parrish TB, Russell EJ, Mesulam MM. The large-scale neural network for spatial attention displays multifunctional overlap [abstract]. Soc Neurosci Abstr 1997; 23: 1589.
LaBar KS, Gitelman DR, Parrish TB, Kim Y-H, Mesulam M-M. Overlap of frontoparietal activations during covert spatial attention and verbal working memory in the same set of subjects: an fMRI study. Soc Neurosci Abstr 1998; 24:
LaBerge D, Buchsbaum MS. Positron emission tomographic measurements of pulvinar activity during an attention task. J Neurosci 1990; 10: 613–9.[Abstract]
Law I, Svarer C, Holm S, Paulson OB. The activation pattern in normal humans during suppression, imagination and performance of saccadic eye movements. Acta Physiol Scand 1997; 161: 419–34.[ISI][Medline]
Luna B, Thulborn KR, Strojwas MH, McCurtain BJ, Berman RA, Genovese CR, et al. Dorsal cortical regions subserving visually guided saccades in humans: an fMRI study. Cereb Cortex 1998; 8: 40–7.
Lynch JC, McLaren JW. Deficits of visual attention and saccadic eye movements after lesions of parietooccipital cortex in monkeys. J Neurophysiol 1989; 61: 74–90.
Martin A, Haxby JV, Lalonde FM, Wiggs CL, Ungerleider LG. Discrete cortical regions associated with knowledge of color and knowledge of action. Science 1995; 270: 102–5.
Maunsell JH. The brain's visual world: representation of visual targets in cerebral cortex. [Review]. Science 1995; 270: 764–9.
Mazziotta JC, Phelps ME, Carson RE, Kuhl DE. Tomographic mapping of human cerebral metabolism: auditory stimulation. Neurology 1982; 32: 921–37.
Mesulam M-M. A cortical network for directed attention and unilateral neglect. [Review]. Annals of Neurology 1981; 10: 309–25.[ISI][Medline]
Mesulam M-M. Attention, confusional states and neglect. In: Mesulam M-M, editor. Principles of behavioral neurology. Philadelphia: F.A. Davis; 1985. p. 125–68.
Mesulam M-M. Large-scale neurocognitive networks and distributed processing for attention, language, and memory. Ann Neurol 1990; 28: 598–613.
Mesulam M-M, Van Hoesen GW, Pandya DN, Geschwind N. Limbic and sensory connections of the inferior parietal lobule (area PG) in the rhesus monkey: a study with a new method for horseradish peroxidase histochemistry. Brain Res 1977; 136: 393–414.[ISI][Medline]
Nobre AC, Sebestyen GN, Gitelman DR, Mesulam MM, Frackowiak RS, Frith CD. Functional localization of the system for visuospatial attention using positron emission tomography. Brain 1997; 120: 515–33.
Nobre AC, Dias EC, Gitelman DR, Mesulam MM. The overlap of brain regions that control saccades and covert visual spatial attention revealed by fMRI. Neuroimage 1998; 7: S9.
Parrish TB, Gitelman DR, Kim Y-H, LaBar KS, Hallam D, Mesulam M-M. Clinical fMRI: is patient motion really an issue? Neuroimage 1998; 7: S560.
Paus T. Location and function of the human frontal eye-field: a selective review. [Review]. Neuropsychologia 1996; 34: 475–83.[ISI][Medline]