Research report Spatio-temporal dynamics of top-down control: directing attention to location and/or color as revealed by ERPs and source modeling Heleen A. Slagter a, * , Albert Kok a , Nisan Mol b , J. Leon Kenemans b a Department of Psychonomics, University of Amsterdam, Roeterstraat 15, 1018 W.B. Amsterdam, The Netherlands b Departments of Psychonomics and Psychopharmacology, Utrecht University, The Netherlands Accepted 8 September 2004 Available online 18 October 2004 Abstract This study investigated the nature and dynamics of the top-down control mechanisms that afford attentional selection using event-related potentials (ERPs) and dipole-source modeling. Subjects performed a task in which they were cued to direct attention to color, location, a conjunction of color and location or no specific feature on a trial-by-trial basis. Overall, similar ERP patterns were observed for directing attention to color and location, suggesting that spatial and non-spatial attention rely to a great extent on similar control mechanisms. The earliest attention-directing effect, at 340 ms, was localized to ventral posterior cortex and may reflect processes by which the cue is linked to its associated feature. Only late in the cue-target interval, differences in ERP were observed between directing attention to color and location. These originated from anterior and ventral posterior areas and may represent differences in, respectively, maintenance and perceptual biasing processes. The ventral posterior sources estimated for these late effects of directing attention to location and color were located posterior to those estimated for the modulatory effects of, respectively, spatial and non-spatial attention. This suggests that the precise neural populations involved in perceptual biasing and attentional modulation may differ. Conjunction cues initially elicited less posterior positivity than color and location cues, but evoked greater central positivity from 540 ms on. This central effect may reflect feature integration or ongoing processes related to cue-symbol translation. These results extend our understanding of the spatio-temporal dynamics of top-down attentional control. D 2004 Elsevier B.V. All rights reserved. Theme: Neural basis of behavior Topic: Cognition Keywords: Spatial; Non-spatial; Attentional control; Attentional selection; Event-related potentials; Dipole modeling 1. Introduction Functional neuroimaging studies have shown that stimuli presented at attended positions in space (e.g., Ref. [18]) or with an attended non-spatial stimulus feature, such as color (e.g., Ref. [5]), elicit enhanced activation in sensory brain areas corresponding to the attended stimulus dimension. This attention-related sensory facilitation of target process- ing enables us to respond faster and more accurately to important external events. Advance knowledge of both spatial and non-spatial stimulus characteristics has been shown to improve behavior [29,30]. Nevertheless, results from event-related potential (ERP) studies indicate that the temporal dynamics of the neural mechanisms underlying attentional modulation of target processing differ between spatial and non-spatial attention. Whereas visuospatial attention results in enhanced amplitudes of the exogenous components P1 and N1 evident in the ERP to stimuli at both attended and unattended locations as early as 80–90 ms post-stimulus (e.g., Refs. [8,41]), selection based on non- spatial visual stimulus features, such as color or form, is reflected by effects starting at around 150 ms post-stimulus, which are super imposed on the evoked components and have a very different morphology (e.g., Refs. [16,20]). Thus, results from ERP studies indicate that modulation 0926-6410/$ - see front matter D 2004 Elsevier B.V. All rights reserved. doi:10.1016/j.cogbrainres.2004.09.005 * Corresponding author. Fax: +31 20 6391656. E-mail address: [email protected] (H.A. Slagter). Cognitive Brain Research 22 (2005) 333 – 348 www.elsevier.com/locate/cogbrainres
16
Embed
Spatio-temporal dynamics of top-down control: directing ...brainimaging.waisman.wisc.edu/~slagter/SlagterCBR05.pdf · attention to color and location, suggesting that spatial and
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
Transcript
www.elsevier.com/locate/cogbrainres
Cognitive Brain Research
Research report
Spatio-temporal dynamics of top-down control: directing attention to
location and/or color as revealed by ERPs and source modeling
Heleen A. Slagtera,*, Albert Koka, Nisan Molb, J. Leon Kenemansb
aDepartment of Psychonomics, University of Amsterdam, Roeterstraat 15, 1018 W.B. Amsterdam, The NetherlandsbDepartments of Psychonomics and Psychopharmacology, Utrecht University, The Netherlands
Accepted 8 September 2004
Available online 18 October 2004
Abstract
This study investigated the nature and dynamics of the top-down control mechanisms that afford attentional selection using event-related
potentials (ERPs) and dipole-source modeling. Subjects performed a task in which they were cued to direct attention to color, location, a
conjunction of color and location or no specific feature on a trial-by-trial basis. Overall, similar ERP patterns were observed for directing
attention to color and location, suggesting that spatial and non-spatial attention rely to a great extent on similar control mechanisms. The
earliest attention-directing effect, at 340 ms, was localized to ventral posterior cortex and may reflect processes by which the cue is linked to its
associated feature. Only late in the cue-target interval, differences in ERP were observed between directing attention to color and location.
These originated from anterior and ventral posterior areas and may represent differences in, respectively, maintenance and perceptual biasing
processes. The ventral posterior sources estimated for these late effects of directing attention to location and color were located posterior to
those estimated for the modulatory effects of, respectively, spatial and non-spatial attention. This suggests that the precise neural populations
involved in perceptual biasing and attentional modulation may differ. Conjunction cues initially elicited less posterior positivity than color and
location cues, but evoked greater central positivity from 540 ms on. This central effect may reflect feature integration or ongoing processes
related to cue-symbol translation. These results extend our understanding of the spatio-temporal dynamics of top-down attentional control.
shown to improve behavior [29,30]. Nevertheless, results
from event-related potential (ERP) studies indicate that the
temporal dynamics of the neural mechanisms underlying
attentional modulation of target processing differ between
spatial and non-spatial attention. Whereas visuospatial
attention results in enhanced amplitudes of the exogenous
components P1 and N1 evident in the ERP to stimuli at both
attended and unattended locations as early as 80–90 ms
post-stimulus (e.g., Refs. [8,41]), selection based on non-
spatial visual stimulus features, such as color or form, is
reflected by effects starting at around 150 ms post-stimulus,
which are super imposed on the evoked components and
have a very different morphology (e.g., Refs. [16,20]).
Thus, results from ERP studies indicate that modulation
22 (2005) 333–348
H.A. Slagter et al. / Cognitive Brain Research 22 (2005) 333–348334
effects are not only of longer latency when attention is
directed to a non-spatial stimulus feature, but they are also
qualitatively different for spatial and non-spatial attention.
Given these dissociations observed with ERP, one may
ask whether the control processes that direct the focus of
attention and may produce attentional modulation of
sensory responses differ between spatial and non-spatial
attention. Only fairly recently, research has turned to address
this question (for review, see Refs. [46,64] ). A straightfor-
ward way to investigate attentional control processes is to
examine brain activity in the period before the test stimulus
is presented, that is, when subjects direct their attention to a
relevant stimulus feature in response to an attention-
directing cue. Recent studies using functional magnetic
resonance imaging (fMRI) have revealed a network of
activated brain areas in the period between attention-
directing cue and test stimulus, encompassing both frontal
and parietal regions for spatial [6,22,25,26,60] as well as
non-spatial [37,48,49,58] attention. However, some domain-
specificity appears to be present within this network, with
dorsal frontal and parietal areas and ventral occipito-
temporal regions being more strongly activated by, respec-
tively, spatial and non-spatial attention-directing cues
[13,50]. In addition, several studies have observed increased
activation in visual areas not only in response to target
stimuli, but also in the period preceding the presentation of
the target stimulus (e.g., Refs. [13,22,26]). The common
interpretation of these findings is that higher order areas in
frontal and parietal cortex send biasing signals to function-
ally specialized sensory areas, so that they in turn can
selectively process target information [7].
Although fMRI provides detailed information about the
localization of neural processes, its temporal resolution is
still in the order of hundreds of milliseconds to a few
seconds at best [44]. This is much longer than the time it
takes to fully direct attention [39]. The high temporal
resolution of the event-related potential technique makes it
an excellent tool for the study of control processes and
preparatory states, as it can relate specific differences in
brain activation to changes in specific stages of information
processing. This temporal information is essential for a full
understanding of the attentional control mechanisms
reflected in fMRI activations. Even though fMRI studies
have shown involvement of roughly the same network of
brain regions in the directing of attention to spatial and non-
spatial stimulus attributes [13,50], the temporal sequence of
activation within these regions may be dependent on the
nature of the to-be-attended stimulus material.
Several ERP studies have previously investigated the
directing of attention to a location in space [9–11,15,
17,40,47,61,62] and non-spatial features [28,63]. However,
a comparison between results from these studies is at present
restrained by the fact that most studies of spatial attentional
control subtracted ERP responses to cues directing attention
to the left from ERP responses to cues directing attention to
the right hemifield [9–11,17,21,47,61,62] (but see Refs.
[15,40]). This comparison has revealed a sequence of effects
related to directing attention to a specific location in space,
consisting of an early directing attention negativity (EDAN)
at posterior parieto-occipital electrodes between 200 and 400
ms post-cue, an anterior directing attention negativity
(ADAN) at frontal electrodes between 300 and 500 ms
post-cue, and a late directing attention positivity (LDAP)
over lateral ventral occipito-temporal scalp regions starting at
around 500 ms post-cue. Yet, these effects cannot easily be
compared to results from studies of non-spatial attentional
control in which such an attend-left versus attend-right
comparison is obviously not possible. In addition, one may
ask whether these cue-direction-related effects reflect the full
temporal pattern of spatial attentional control (see also Ref.
[57]). Several studies of spatial top-down control have
reported behavioral cueing effects and attentional modulation
effects (i.e., P1, N1) in the absence of these cue-direction-
related ERP effects (i.e., EDAN [10,11], ADAN [15] and
LDAP [40,47]). This suggests that some attentional control
processes that may be mandatory for the establishment of an
attentional bias are not lateralized and, thus, do not show up in
the left–right subtraction. This further complicates an
integrative interpretation of the results from ERP studies of
spatial and non-spatial top-down control. Thus, in order to
adequately isolate the complete pattern of spatial or non-
spatial attentional control, one needs to compare the
attention-directing condition with a reference condition that
controls for processes that are not specific to the actual
initiation and directing of attention, such as cue-identification
and motor preparation processes, but that does not call upon
attentional control mechanisms.
In the present study, we examined the extent to which top-
down control processes are stimulus material-unspecific (i.e.,
general) or depend on the nature of the to-be-attended
stimulus feature (i.e., domain-specific) using a within-subject
design and a reference cue condition. ERPs elicited by
location and color attention-directing cues were compared to
ERPs elicited by reference cues to isolate processes related to
directing attention to location and color. The underlying
neural source configurations of the observed spatial and non-
spatial attention-directing effects were compared against each
other to reveal possible differences in the configuration and/
or timing of activated brain areas. Based upon position-
special models of attention [31,33,34,53,55,56], it can be
hypothesized that pre-target biasing effects of spatial
attention result from an initial activation in dorsal posterior
areas, which maintain location representations, which is
followed in time by activation in ventral posterior areas,
which maintain spatially corresponding feature representa-
tions. Biasing effects of non-spatial attention, on the other
hand, would be reflected by a reversed pattern of activation,
with ventral posterior areas being activated first and then
dorsal posterior areas, or activation of only ventral posterior
areas that hold representations of the non-spatial feature [19].
In LaBerge’s model of attention, for example, when the
location of the stimulus is predictable, parietal areas involved
H.A. Slagter et al. / Cognitive Brain Research 22 (2005) 333–348 335
in coding spatial information can modulate featural informa-
tion of an object in the occipito-temporal lobe by constricting
the effective receptive fields of cells within this area, thereby
aiding in the selection of the object with the attended feature
[32,33].
In addition, the present study examined the relation
between perceptual biasing and attentional modulation
effects for spatial and non-spatial attention separately by
comparing the neural source configurations underlying these
effects. Based upon results of event-related fMRI studies,
which have shown increased baseline activity in the same
visual areas that were modulated by spatial attention (e.g.,
Refs. [22,26]), we expected to obtain similar source
solutions for pre-target perceptual biasing and post-target
attentional modulation effects.
Lastly, the present study investigated the direction of
attention in a condition where attention was to be directed
simultaneously both to a location in space and to a color. If
the two types of attention rely on completely different
control structures, no interaction, but pure additive effects of
directing attention to location and directing attention to
color are expected. If, on the other hand, the two types of
attentional control rely on similar mechanisms, simulta-
neously directing attention to location and color should
place greater demands on these general control mechanisms
as reflected by enhanced or prolonged attention directing-
related ERP effects. Another possibility would be that
directing attention to a conjunction of location and color
calls upon entirely new processes specific to the conjoining
of the two stimulus attributes [54].
Fig. 1. Examples of cues used in the location (most left panel), color
(second panel), conjunction (third panel) and no-feature (most right panel)
conditions. Horizontal lines denote the letter-symbol(s) to be used to direct
attention (L=attend left, R=attend right, G=attend to yellow, B=attend to
blue). When presented next to the fixation cross (i.e., no-feature cue), no
specific color or location had to be attended.
2. Method
2.1. Subjects
Sixteen healthy volunteers participated in the study. Two
subjects were discarded from the analyses because of poor
eye fixation in the interval between cue and test stimulus or
excessive blink activity during EEG recordings. Thus, 14
subjects (7 men, mean age of 23.2 years) remained in the
sample. All subjects were students at the University of
Amsterdam, were right-handed, had no history of mental or
sustained physical illness, and had normal or corrected-to-
normal vision by self-report. Subjects received credits as
part of an introductory course requirement at the University
of Amsterdam.
2.2. Stimuli and procedure
Each trial began with a 100-ms presentation of a cue
(0.928 in width and 2.88 in height) that was located at
fixation. After a random interval between 800 and 1500
ms (rectangular distribution), during which only the
fixation cross (0.318 in width and 0.208 in height) was
shown at the center of the screen, the cue was followed by
a test stimulus (38 in height, 38 in width). This test
stimulus was a blue or yellow square and appeared 7.138to center from fixation in either the left or the right visual
field and 1.738 to center above the horizontal meridian.
The interval between test stimulus offset and onset of the
next trial was varied randomly between 1400 and 2100 ms
(rectangular distribution). During this interval, the fixation
cross remained on the screen. All stimuli were presented
on a black background. Within a run, subjects were
randomly cued to attend to (a) a color (blue or yellow;
color condition (COL)), (b) a location (left or right;
location condition (LOC)), (c) a color and a location
(e.g., blue and left; conjunction condition (CONJ)) or (d)
to dnothingT (no-feature condition (N); see below).
Each cue consisted of four white uppercase letters (all
equal in width (0.368) and height (0.518)) presented around
the fixation cross in a vertical array: dBT, dGT, dLT and dRT(see Fig. 1). Each letter corresponded to a stimulus feature:
dBT to blue, dGT to yellow (dgeelT in Dutch), dLT to left and
dRT to right. Letter order was counterbalanced across
subjects with the restriction that the two blocationQ letters(dLT and dRT) and the two bcolorQ letters (dBT and dGT) werealways grouped together, resulting in eight possible combi-
nations of letters: BGLR, BGRL, GBLR, GBRL, LRBG,
RLBG, LRGB and RLGB. In the color and location
conditions, the color or location to which attention was to
be directed, was indicated by two short, horizontal lines
(0.208 in width, 0.088 in height), one on each side of a given
letter (e.g., when presented next to dLT, attention had to be
directed to the left (see Fig. 1)). In the conjunction
condition, two letters, one representing a color, the other a
location, were flanked by horizontal lines (0.108 in width,
0.088 in height) indicating that those both had to be used to
direct attention. In the no-feature condition, the two
horizontal lines (0.208 in width, 0.088 in height) were
presented next to the fixation cross. Conjunction cues were
presented on 40%, and color, location and no-feature cues
each on 20% of the trials.
In each task condition, the cue was followed by a test
stimulus, which was presented for either 50 ms (standard
duration; 75% of all trials in the attention-directing
conditions, 87.5% in the no-feature condition) or 150 ms
H.A. Slagter et al. / Cognitive Brain Research 22 (2005) 333–348336
(deviant duration; 25% of all trials in the attention-directing
conditions, 12.5% in the no-feature condition). In case of an
attention-directing cue, subjects were instructed to respond
as fast and accurately as possible to test stimuli with the
attended feature(s) that were presented slightly longer (i.e.,
150 ms). On 50% of all trials, the test stimulus possessed the
attended attribute. On 12.5% of all trials, therefore, target
test stimuli were presented (with the attended attribute(s)
and of longer duration). In case of a no-feature cue, subjects
were asked to respond as fast and accurately as possible to
test stimuli that were presented slightly longer (i.e., 150 ms),
regardless of their color or location. Subjects used their right
index finger to respond to targets.
The experiment consisted of two sessions: a practice
session and an EEG session. The aim of the practice session
was to make subjects familiar with the specific task
requirements and to make sure that they did not show
excessive eye blink activity. It consisted of 8 runs of 80
trials (approximately 3.5 min each). During the EEG
recording session, subjects sat in a comfortable chair with
a computer monitor placed 80 cm in front of their eyes and
positioned so that the vertical and horizontal straight-ahead
lines of sight were the same for all subjects. After the
electrode cap was placed, subjects practiced the task once
and subsequently performed 24 task runs of 80 trials each
while their EEG was recorded. Subjects were asked to
minimize eye and body movements and allowed to pause
between the runs if they wished to do so.
2.3. ERP recordings
Recordings were made with 60 Ag-AgCl-electrodes
mounted in an elastic cap: FP1, FP2, AF7, AF8, AF3,
and strength) were estimated and entered into ANOVA’s or
H.A. Slagter et al. / Cognitive Brain Research 22 (2005) 333–348338
paired t-tests. For a more detailed description of this
procedure, see Kenemans et al. [27].
3. Results
3.1. Behavioral results
There were no significant differences between the diffe-
rent task conditions in response latency (COL: 579, LOC:
576, CONJ: 567, N: 576 ms, relative to target onset) or the
number of omitted responses to target stimuli (COL: 12.1%,
LOC: 14.0%, CONJ: 13.8%, N: 15.5%). Neither were any
difference observed between the different cue conditions in
the number of false alarms to attended (COL: 1.4%, LOC:
1.7%, CONJ: 2.9%, N: 1.5%) or unattended (COL: 0.1%,
LOC: 0.1%, CONJ: 0.1%) test stimuli of short duration.
However, subjects made more false alarms to unattended
test stimuli of long duration in the CONJ condition (4.2%)
Fig. 2. (A) Grand average ERP waveforms to attended (Att) and unattended (Unat
(B, C) Grand average, average reference spline interpolated isopotential maps. Not
attended-unattended left test stimuli (left panel) and attended-unattended right test
(left panel) and right—no-feature cues (right panel) at 752 ms after cue onset.
than in the COL (0.9%) and LOC (1.8%) conditions
[F(2,26)=6.154, p=0.006].
3.2. ERPs
3.2.1. ERPs to test stimuli
As expected, P1 amplitudes were larger for stimuli
presented at attended compared to unattended locations in
the location condition between 112 and 140 ms post-
stimulus [6.2bF(1,12)b29.9, pb0.05] (see Fig. 2A). This
difference in attention-related activity was larger over
contralateral scalp regions as indicated by an interaction
between attention (attended, unattended), hemisphere (left,
right) and test stimulus feature (left, right) between 104 and
120 ms post-stimulus at electrodes P7 and P8
[5.6bF(1,12)b6.8, pb0.05]. Furthermore, compared to
stimuli of the unattended color, stimuli of the attended
color elicited a larger positive response at electrodes F3 and
F4 between 128 and 228 ms in the color condition
t) test stimuli presented left or right from fixation for electrodes P7 and P8
shaded: areas of positive amplitude. Shaded: areas of negative amplitude. B
stimuli (right panel) at 120 ms post-test stimulus. (C) Left—no-feature cues
.
:
H.A. Slagter et al. / Cognitive Brain Research 22 (2005) 333–348 339
[5.7bF(1,12)b19.4, pb0.05] (see Fig. 3). In addition, a
significant main effect of attention was observed at Oz
between 152 and 188 ms [4.9bF(1,12)b39.7, pb0.05]
reflecting a larger positive response to stimuli of the
attended versus the unattended color. This positive response
was followed by greater attention-related negativity, which,
however, never reached significance.
3.2.2. ERPs to cues
Fig. 4 shows the representative waveforms elicited by
each type of cue (COL, LOC, CONJ, N) and the grand
average difference waveforms for the color, location and
conjunction effects (i.e., SFLOC, SFCOL, CJLOC and
CJCOL). Potential distributions corresponding to the
attention-directing-related effects are shown in Fig. 5. Table
1 lists time intervals and F-value ranges for main effects of
COND and interactions between this factor and the factors
HEMI and/or SITE within 0–800 ms post-cue, for each
regional analysis. Each of the regional effects will be
discussed next.
The posterior analyses (electrode sites: P7/P8, P3/P4,
PO5/PO6) revealed the earliest significant effect of condition.
This effect was observed between 181 and 220ms post-cue at
Fig. 3. Grand average ERP difference waveforms (attended–unattended
color test stimuli) displaying the frontal selection positivity (FP) effect at
electrode Fz. The grand average, spline-interpolated isopotential map (two-
dimensional projection) shows the topographical distribution of this effect
at 144 ms post-test stimulus. The spacing between isopotentials in this map
is 0.2 AV. White areas denote areas of positive amplitude and dotted areas
denote areas of negative amplitude.
posterior sites. Post-hoc comparisons and inspection of the
data revealed that within this interval location cues elicited
less negativity than color, conjunction or neutral cues (main
effect of COND [2.9bF(3,39)b4.5, pb0.05]). This effect was
followed by a more pronounced effect of condition between
261 and 500 ms post-cue at posterior sites (main effect of
COND [4.6bF(3,39)b61.7, pb0.05]), which extended to
more central scalp locations [321–400 ms: 5.8bF(3,39)b7.9,
pb0.05]. Post-hoc comparisons and inspection of the grand
average ERP waveforms showed that, in this interval, single
feature cues elicited greater biphasic positivity than both no-
feature and conjunction cues and, furthermore, that no-
feature cues elicited greater positivity than conjunction cues
at posterior sites especially during the early part of this
biphasic positivity (261–400 ms post-cue). Scalp topogra-
phies of this early effect of cue condition show that it was
maximal over lateral parietal and occipital sites (CON-
D*SITE interaction [3.1bF(6,78)b7.8, pb0.05]) and that,
during the first phase, the effect was more prominent at right
compared to left posterior scalp locations, whereas during the
second phase, it was more pronounced at left compared to
right posterior scalp locations (COND*HEMI interaction
[3.8bF(3,39)b6.9, pb0.05]). In the later part of the cue-target
interval, starting at 661 ms, a third main effect of condition
was observed for posterior scalp sites [3.4bF(3,39)b12.4,
pb0.05]. This effect lasted until the end of the cue-target
interval and reflected larger positive voltage over dorsal
posterior sites for conjunction compared to single feature
cues and greater positivity to single feature cues than no-
feature cues over parieto-occipital sites (COND*SITE
interaction [3.1bF(6,78)b7.8, pb0.05]). This late posterior
effect spread to central scalp locations.
A further effect of condition was observed over fronto-
central scalp locations. Greater positive response was
revealed to no-feature cues compared to attention-directing
cues over central and anterior sites between, respectively,
401 and 640 ms [3.6bF(3,39)b15.7, pb0.05] and 441 and
600 ms [4.1bF(3,39)b17.3, pb0.05] after cue onset. This
effect reflects a more anterior distribution of the posterior
positivity in the no-feature compared to the other conditions.
Lastly, the anterior analyses revealed a further difference
between cue conditions: greater negativity to color cues
compared to location, conjunction and no-feature cues
[4.7bF(3,39)b10.5, pb0.05]. Between 641 and 760 ms,
color cues elicited a larger negative response than location
cues at frontal scalp locations. This effect was maximal over
midline frontal electrodes.
3.3. Source localization
A close correspondence in GFP peaks and the above-
described statistically significant ERP effects was observed.
At 184 ms post-cue, a small peak in GFP was only observed
for SFLOC, followed by a bigger peak at 340 ms post-cue.
The neural generators of these two effects (at 184 and 340
ms post-cue) were estimated first for the grand average
Fig. 4. Top part: Grand average, cue-locked ERP waveforms for the different cue conditions for a selected number of electrodes. Bottom part: Grand average
ERP difference waveform for the contrasts: conjunction–location cues (Conj-Loc; CJCOL), color–no-feature cues (Col-NF; SFCOL), conjunction–color cues
(Conj-Col; CJLOC) and location–no-feature cues (Loc-NF; SFLOC).
H.A. Slagter et al. / Cognitive Brain Research 22 (2005) 333–348340
difference waveform (SFLOC), and then for the individual
subject difference waveforms, where the grand average
solution parameters were used as a starting point (cf. Ref.
[27]). Fitting of one symmetric dipole pair localized both
effects to the ventral-lateral compartment of posterior cortex
[RV=4.0% (184 ms) and 1.6% (340 ms)]. No differences in
location or orientation parameters were observed between
the source models obtained at 184 and 340 ms for SFLOC.
The main results of the grand average and per-subject
estimation procedures for the 340 ms SFLOC effect are
Fig. 5. Grand average spline-interpolated isopotential maps (two-dimensional projections) for the different contrasts at 180, 340, 520, 700 and 800 ms post-cue.
Col=color cues, NF=no-feature cues, Loc=location cues and Conj=conjunction cues. The spacing between isopotentials is 0.3 AV. White: areas of positive
amplitude. Shaded: areas of negative amplitude.
H.A. Slagter et al. / Cognitive Brain Research 22 (2005) 333–348 341
summarized in Fig. 6A and B (second panel). Furthermore,
for both time points, no interaction was observed between
the attention-direction of the location cue (SFLEFT,
SFRIGHT) and hemisphere when dipole moments were
compared, suggesting that the two effects were not
lateralized with respect to the cued location.
Inspection of the grand average GFP functions revealed
that the second peak observed for SFLOC at 340 ms after
cue presentation was also present for the other contrasts
(SFCOL: 344 ms, CJLOC: 336 ms and CJCOL: 336 ms).
One bilateral dipole pair with mirror-symmetric locations
across hemispheres resulted in a model distribution explain-
ing more than 95% of the variance in each of the recorded
individual subject instantaneous source models for the
different contrasts, which were derived at the d340T ms
GFP peak latencies, are summarized in Fig. 6. Statistical
analyses revealed that these dipoles did not differ
significantly with respect to location across conditions.
Table 1
Results from repeated measurement analyses at posterior, central and frontal elec
Posterior Central
COND 181–220 2.9bF(3,39)b4.5 321–800
261–500 4.6bF(3,39)b61.7
661–800 3.4bF(3,39)b12.4
COND*HEMI 241–380 3.8bF(3,39)b6.9 481–580
COND*SITE 281–780 3.1bF(6,78)b7.8 261–800
COND*HEMI 261–320 3bF(6,78)b4.1
*SITE 381–520 2.7bF(6,78)b4.6
561–800 2.6bF(6,78)b4.5
Time windows are given for each significant effect ( pb0.05 for two successive tim
condition with hemisphere (HEMI; left, right) and/or electrode site (SITE), along
This indicates that SFLOC, SFCOL, CJLOC and CJCOL
have equivalent dipole locations in the lateral ventral
posterior compartment of the cortex at around 340 ms post-
cue. The orientation of the dipoles, however, differed
across conditions [in the left hemisphere: x: F(3,39)=7.1,
y: F(3,39)=159.8, z: F(3,39)=20.0; in the right hemisphere:
x: F(3,39)=9.8, y: F(3,39)=72.0, z: F(3,39)=8.1]. The
dipole orientations for CJLOC and CJCOL were reversed
(flipped around the x-, y- and z-axes) relative to the
SFLOC and SFCOL dipole orientations. This effect reflects
the fact that color and location cues elicited greater
positivity over posterior scalp regions compared to both
no-feature and conjunction cues. The exact reversal in
orientation between the SFLOC and SFCOL, on the one
hand, and CJLOC and CJCOL, on the other hand,
illustrates the sensitivity of the modeling approach used
in the present study (cf. Ref. [27]).
For SFLOC and SFCOL, another GFP peak was
observed at 532 and 544 ms, respectively. In this time
window, a difference in positivity was observed over frontal
trode locations
Anterior
3.6bF(3,39)b15.7 421–800 2.9bF(3,39)b17.3
3.8bF(3,39)b8.1 501–580 3.1bF(3,39)b7
2.8bF(6,78)b20.6 281–640 2.8bF(6,78)b7.1
661–800 2.6bF(6,78)b5.1
e bins (i.e., 40 ms)) of condition (COND; SFLOC, SFCOL, CONJ, N), or of
with the minimum and maximum F-values for each effect.
Fig. 6. (A) Grand average source solutions at d340T ms post-cue for the contrasts: location–no-feature cues (Loc-NF; SFLOC), color–no-feature cues (Col-NF;
SFCOL), conjunction–color cues (Conj-Col; CJLOC) and conjunction–location cues (Conj-Loc; CJCOL). (B) Grand average (dark grey) and individual (black)
dipole solutions at GFP peak latency d340T ms displayed for each contrast (i.e., Loc-NF, Col-NF, Conj-Col and Conj-Loc) separately.
H.A. Slagter et al. / Cognitive Brain Research 22 (2005) 333–348342
and central scalp locations between single feature and no-
feature cues. One bilateral dipole pair with mirror symmetric
locations in posterior cortex gave a good fit for both
contrasts [RV(SFLOC)=2.4% and RV(SFCOL)=1.4%] (see
Fig. 7). Furthermore, their estimated source parameters did
Fig. 7. Grand average source solutions for color–no-feature cues (Col-NF;
SFCOL) and location–no-feature cues (Loc-NF; SFLOC) for the early
posterior (344 and 340 ms, respectively) and intermediate (544 and 532 ms,
respectively) effects.
not differ, indicating that, at around 540 ms post-cue, similar
areas of cortex were differentially activated by the two
single feature cues versus the no-feature cue. These sources
were located more medially ( p=0.015) and anteriorly
( p=0.004), and somewhat more dorsally than the sources
that were estimated for the early posterior effects at around
340 ms post-cue.
Fig. 8. Grand average source solutions for color–no-feature cues (Col-NF
(SFCOL); black dipoles) and location–no-feature cues (Loc-NF (SFLOC);
dark grey dipoles) at 740 and 752 ms post-cue, respectively.
Fig. 9. Grand average ventral posterior sources of the late-latency attention-
directing effects (left panel) and the first attentional modulation effects
(right panel) for both spatial (A) and non-spatial (B) attention. Abbrevia-