Birkbeck ePrints: an open access repository of the
research output of Birkbeck College
Holmes, Amanda; Kiss, Monika and Eimer, Martin (2006).
Attention modulates the processing of emotional
expression triggered by foveal faces. Neuroscience
Letters 394 (1) 48-52.
This is an author-produced version of a paper published in Neuroscience
Letters (ISSN 0304-3940). This version has been peer-reviewed, but does not
include the final publisher proof corrections, published layout, or pagination.
The published version is available to subscribers at
Copyright © 2005 Elsevier Ireland Ltd All rights reserved.
All articles available through Birkbeck ePrints are protected by intellectual
property law, including copyright law. Any use made of the contents should
comply with the relevant law.
Citation for this version:
Holmes, Amanda; Kiss, Monika and Eimer, Martin (2006). Attention
modulates the processing of emotional expression triggered by foveal faces.
London: Birkbeck ePrints. Available at:
Citation for the publisher’s version:
Holmes, Amanda; Kiss, Monika and Eimer, Martin (2006). Attention
modulates the processing of emotional expression triggered by foveal faces.
Neuroscience Letters 394 (1) 48-52.
Contact Birkbeck ePrints at firstname.lastname@example.org
Attention modulates the processing of emotional
expression triggered by foveal faces
*School of Human and Life Sciences, Roehampton University, London, UK
**School of Psychology, Birkbeck University of London, UK
Corresponding author’s address: Dr. Amanda Holmes, School of Human and Life
Sciences, Roehampton University, Whitelands College, Holybourne Avenue, London
SW15 4JD, England. Tel: 0044 020 8392 3449. E-Mail: email@example.com
Number of text pages: 14
Number of figures: 2
Acknowledgements: This research has been supported by a grant from the
Biotechnology and Biological Sciences Research Council (BBSRC), UK. The authors
thank Heijo Van de Werf for technical assistance. M.E. holds a Royal Society-
Wolfson Research Merit Award.
KEYWORDS: Emotional expression; Event related potentials; Fearful faces; Foveal
vision; Spatial attention
To investigate whether the processing of emotional expression for faces presented
within foveal vision is modulated by spatial attention, ERPs were recorded in
response to stimulus arrays containing one fearful or neutral face at fixation, which
was flanked by a pair of peripheral bilateral lines. When attention was focused on the
central face, an enhanced positivity was elicited by fearful as compared to neutral
faces. This effect started at 160 ms post-stimulus, and remained present for the
remainder of the 700 ms analysis interval. When attention was directed away from the
face towards the line pair, the initial phase of this emotional positivity remained
present, but emotional expression effects beyond 220 ms post-stimulus were
completely eliminated. These results demonstrate that when faces are presented
foveally, the initial rapid stage of emotional expression processing is unaffected by
attention. In contrast, attentional task instructions are effective in inhibiting later,
more controlled stages of expression analysis.
Emotional information is prioritized for processing: it is analyzed rapidly [3,6],
and has the ability to engage preferential attention [1,9,10,19]. This preferential
processing of emotionally significant events may reflect a variety of mechanisms,
including modulatory influences from the amygdala on sensory processing ,
modulation of frontoparietal cortices by lateral orbitofrontal cortex , and
cholinergic enhancement of these brain systems .
An important and controversial issue is whether the encoding and analysis of
emotionally salient events can occur independently of attention [13,16]. Recent fMRI
studies have provided conflicting results. In one study , amygdala responses to
threat-related facial expressions were unaffected by spatial attention, and emotion-
specific activity in some frontal and parietal regions was observed for unattended
faces. This suggests that facial emotion can be discriminated pre-attentively, and that
information concerning the affective value of irrelevant stimuli is still available even
when attention is allocated to other events. Results from another study , however,
contradict these findings. In this study, elevated neural activation levels in response to
attended fearful versus neutral faces were found within a number of distinct brain
regions, including the amygdala. In contrast, when spatial attention was diverted away
from the faces, this differential activation was no longer apparent. These data clearly
challenge the hypothesis that the detection and analysis of emotional events is
independent of the current focus of attention.
Event-related potential (ERP) studies have also been used to examine
interactions between attention and emotion processing [7,11]. In one study ,
stimulus arrays consisting of two faces and two houses positioned along horizontal
and vertical axes were presented. Participants had to attend either to the two vertical
or to the two horizontal locations (as indicated by a cue presented at the start of each
trial), in order to detect occasional target events at these locations. On trials where
faces were presented at cued (attended) locations, an enhanced positivity was
triggered in response to arrays containing fearful as compared to neutral faces. This
emotional expression effect started at anterior electrodes at about 120 ms after
stimulus onset, had an initial frontal distribution, and was more broadly distributed
beyond 200 ms post-stimulus (see also , for similar findings). In marked contrast,
when faces were presented at uncued locations, and attention was directed to the
location of houses instead, these emotional expression effects were entirely eliminated.
Similar findings were obtained in a second study . In different blocks, participants
directed their attention either towards a pair of peripheral faces in order to judge their
emotional expression, or towards a centrally presented pair of vertical lines in order to
judge their length, and ignored the other stimulus in the display (lines in the faces task,
and faces in the lines task). Different facial expressions (anger, disgust, fear,
happiness, sadness, surprise) were presented in separate blocks. When emotional
faces were task-relevant, an enhanced positivity to emotional relative to neutral faces
was elicited. Although starting slightly later (at about 160 ms post-stimulus), the
distribution of this emotional positivity was similar to previous studies [6,11].
However, in blocks where participants directed their attention to the central lines
while ignoring lateral faces, these emotional expression effects on ERP waveforms
were no longer present.
These ERP results are largely consistent with the fMRI findings of Pessoa et al.
 by revealing a strong interaction between emotion and attention. One possible
explanation for the discrepancies between these findings and those of Vuilleumier et
al.  is that the attentional filtering of task-irrelevant information may have been
more efficient in the former studies, leading to the elimination of emotion-specific
effects. By contrast, in the Vuilleumier et al.  study, the gating of information may
have been incomplete, and thus emotional expression effects of task-irrelevant
information remained present. Pessoa, Kastner, & Ungerleider  have suggested
that the attentional task in the Vuilleumier et al.  study may have been less
demanding than in their own experiments. It would therefore seem likely that to-be-
ignored emotional information can still be processed, albeit in attenuated fashion,
when the performance of an attentional task is relatively easy, and spare processing
capacity remains available, thus resulting in incomplete attentional gating of task-
irrelevant information (see [13,14]).
The aim of the present ERP study was to further investigate the role of spatial
attention for the processing of emotional faces under conditions where these faces
should be hard to ignore. More specifically, we investigated whether ERP emotional
expression effects are still affected by attentional task instructions when fearful or
neutral faces are presented at fixation. The design of the present experiment was
similar to the task employed by Eimer et al. , in that participants were required to
focus their attention either on faces while ignoring simultaneously presented pairs of
vertical lines (Faces task), or to attend to these lines while ignoring faces (Lines task).
The main difference was that faces were not presented peripherally, but at fixation,
and were flanked by a pair of bilateral vertical lines that were either identical or
different in length. In the Faces task, participants had to detect immediate repetitions
of identical faces (which could be either fearful or neutral) across successive trials. In
the Lines task, they had to detect immediate repetitions of identical line pairs across
trials, and to ignore the central fearful or neutral faces.
ERPs triggered on trials containing a fearful face were compared to ERPs
elicited on trials where a neutral face was present, separately for the Faces task and
the Lines task. Since faces were always located at central fixation, they should be
harder to ignore even when participants are instructed to monitor the line stimuli, as
compared to our previous studies where faces were presented at peripheral locations
[7,11]. The critical question was whether there would still be systematic differences in
emotional expression effects on ERP waveforms between the Faces and the Lines task.
If the emotional expression of faces at fixation was always processed fully regardless
of attentional task instructions, emotional expression effects should be equivalent for
both tasks. In contrast, if the instruction to direct attention away from these faces
towards the laterally presented lines interfered with the processing of emotional
expression, these effects should be attenuated or possibly even completely absent in
the Lines task.
Materials and Methods
Subjects: Twelve paid volunteers (4 male, 8 female), aged 18-41 years (mean
age: 31 years) participated in the experiment. All subjects were right-handed and had
normal or corrected-to-normal vision.
Figure 1 about here
cabin, with response buttons under their left and right hands. All stimuli were presented
on a computer screen in front of a black background at a viewing distance of 70 cm.
Face stimuli were photographs of faces of ten different individuals, all taken from a
standard set of pictures of facial affect . Facial expression was either fearful or
neutral. Faces covered a visual angle of about 8.6° x 5.7°, and were presented at
fixation together with a pair of grey vertical lines (0.2° width). These lines were
presented to the left and right of the central face at an eccentricity of 4.0° (see Figure
1). Two different line lengths were used. Short and long lines covered a visual angle
of 1.3° and 2.8°, respectively. All four possible line arrangements (bilateral short lines,
bilateral long lines, short left / long right, long left / short right) were presented with
Two task conditions were run, each consisting of two successive experimental
blocks. In the Faces task, participants were instructed to monitor the centrally
presented faces, to respond with a right-hand button press whenever the face
presented on the preceding trial was shown again on the current trial, and to ignore the
lateral lines. In the Lines task, participants were instructed to monitor these lines, to
respond with a right button press when the line arrangement presented on the
preceding trial reappeared on the current trial, and to ignore the centrally presented
faces. Each block consisted of 92 trials. Stimuli were presented for 300 ms, and were
separated by intertrial intervals of 1200 ms. Twelve trials per block were target trials,
which contained an immediate repetition of an identical face in the Faces task, and an
immediate repetition of an identical line pair in the Lines task. There were no
immediate repetitions of task-irrelevant stimuli (faces in the Lines task, and line pairs
in the Faces task) across trials. The remaining 80 trials were non-repetition trials (with
a fearful or neutral face each presented in random order on 40 trials). Participants were
instructed to respond as quickly as possible only to immediate repetitions of the task-
relevant stimulus and to maintain central fixation. Short practice blocks were delivered
for both task conditions.
ERP recording and data analysis: Recordings were made from twenty-three Ag-
AgCl electrodes, referenced to linked earlobes. Horizontal EOG (HEOG) was recorded
bipolarly from electrodes at the outer canthi of both eyes. Electrode impedance was kept
below 5 kΩ. EEG was DC-recorded with a sampling rate of 200 Hz and an upper cut-off
frequency of 40 Hz. EEG was averaged relative to a 100 ms pre-stimulus baseline for all
combinations of task (Faces task vs. Lines task) and facial expression (fearful vs.
neutral). To avoid any contamination with movement-related artefacts, ERP analyses
were restricted to non-repetition trials where no manual response was recorded. Trials
with eyeblinks (Fpz exceeding ±60 μV) and lateral eye movements (HEOG exceeding
±30 μV) were also excluded prior to analysis. Artefact rejection led to the exclusion of
24.9% of all non-repetition trials in the Faces task, and of 21.3% of all non-repetition
trials in the Lines task. Repeated measures analyses of variance (ANOVAs) were
performed on ERP mean amplitudes obtained within successive latency windows (120-
160 ms, 160-220 ms, 220-300 ms, and 300-700 ms post-stimulus, respectively). These
analyses were conducted separately for lateral electrode pairs F3/4, C3/4, and P3/4, and
for midline electrodes Fz, Cz, and Pz, for the factors task, facial expression, electrode
site (frontal vs. central vs. parietal) and hemisphere (left vs. right, for lateral electrodes
Stimuli and Procedure: Subjects were seated in a dimly lit sound attenuated
repetitions were faster in the Faces task than in the Lines task (666 ms vs. 749 ms;
t(11)=4.4; p<.001). Reaction times in the Lines task were not significantly affected by
the emotional expression of the task-irrelevant face. Participants missed more immediate
stimulus repetitions in the Lines task than in the Faces task (22.2% vs. 9.7%; t(11)=2.9;
p<.02). False Alarms on non-repetition trials occurred on less than 1% of these trials for
both task conditions.
Figure 2 about here
ERPs to stimulus arrays containing fearful versus neutral faces: Figure 2
shows ERPs in response to stimulus arrays containing a fearful face (dashed lines) or
a neutral face (solid lines) at fixation, separately for the Faces task (top panel) and the
Lines task (bottom panel). As predicted, a sustained positivity was elicited in response
to arrays containing emotional faces in the Faces task. This emotional expression
effect started at about 160 ms post-stimulus, overlapped with the P2 and N2
components, and remained present in a sustained fashion during the 700 ms interval
shown in Figure 2. When attention was directed away from the centrally presented
faces in the Lines task, longer-latency emotional expression effects were strongly
attenuated, or entirely absent. In contrast, the early phase of this effect (an emotional
positivity superimposed on the P2 component) appeared to remain present when faces
This was confirmed by statistical analyses. No significant emotional
expression effects were found in the earliest analysis window (120-160 ms post-
stimulus). In the 160-220 ms post-stimulus interval, significant main effects of facial
expression at lateral electrodes (F(1,11)=48.0; p<.001) and at midline electrodes
(F(1,11)=51.0; p<.001) reflected the fact that ERPs to fearful faces were more
positive than ERPs to neutral faces. Importantly, there was no indication of any task x
facial expression interaction, and subsequent analyses confirmed that early emotional
positivities were reliably present in the P2 time range not only in the Faces task
(F(1,11)=22.8 and 23.5; both p<.001, for lateral and midline sites), but also in the
Lines task (F(1,11)=5.0 and 6.4; both p<.05, for lateral and midline sites). At lateral
posterior electrodes T5/6 (not shown in Figure 2), where the face-specific N170
component is maximal, ERPs in response to fearful faces also tended to be more
positive than ERPs triggered by neutral faces during the 160 – 220 ms measurement
interval, although this difference failed to reach significance (F(1,11)=3.6; p<.09).
A very different pattern of results emerged during the subsequent
measurement interval (220-300 ms post-stimulus). Although main effects of facial
expression were still observed at lateral and midline electrodes (F(1,11)=16.1 and
13.3; both p<.005), these were now accompanied by significant task x facial
expression interactions (F(1,11)=5.5 and 6.0; both p<.05, for lateral and midline sites).
Follow-up analyses revealed that significant emotional expression effects were
present at both lateral and midline electrodes in the Faces task (F(1,11)=27.7 and 23.3;
both p<.001). In contrast, facial expression effects were completely eliminated in the
Lines task (both F<1). Essentially the same pattern of effects was also found for the
longer-latency analysis window (300-700 ms post-stimulus). Again, main effects of
facial expression at lateral and midline electrodes (F(1,11)=26.1 and 16.1; both
p<.002) were accompanied by task x facial expression interactions (F(1,11)=11.7 and
Behavioural performance: Mean reaction times to immediate stimulus
12.0; both p<.01). While highly significant facial expression effects were present in
the Faces task (F(1,11)=31.3 and 23.7; p<.001, for lateral and midline electrodes), no
reliable ERP difference between arrays containing fearful and neutral faces were
found for the Lines task (both F<1).
Findings from previous studies [7,11,15] have shown that the processing of
emotional faces is strongly dependent on focal attention. Effects of emotion, as
reflected in haemodynamic responses  or ERPs [7,11], were found to be entirely
eliminated when attention was focused away from critical emotional stimuli. The aim
of the present study was to investigate the limits of this attentional gating of
emotional information by studying whether attentional task instructions would even
affect the processing of emotional faces when these faces are presented centrally
within foveal vision, rather than in the periphery of the visual field, as in previous
experiments. ERPs were recorded to stimulus arrays containing centrally presented
fearful or neutral faces when these faces were either task-relevant and therefore
attended (Faces task), or when they were task-irrelevant because attention was
directed towards a pair of bilateral lines instead (Lines task).
When faces were attended (Faces task), emotional faces elicited an enhanced
positivity relative to neutral faces, which started at about 160 ms after stimulus onset,
and remained present throughout the 700 ms analysis interval. These ERP effects
were very similar to the result obtained in the Faces task of our previous study ,
where analogous experimental procedures were used, except that faces were presented
peripherally, while lines were located close to fixation. They are also in line with
other previous findings [3,6,11]. The critical question was whether these effects
would remain unchanged in the Lines task, where attention was directed away from
the foveal faces. The early phase of the emotional expression effect observed between
160 and 220 ms post-stimulus was indeed preserved in the Lines task. During this
time interval, an enhanced positivity was elicited for trials that contained a fearful
face relative to trials where a neutral face was presented, even though participants’
task was to monitor the lines. The absence of any task x facial expression interaction
during this time interval suggests that the initial processing of the emotional
expression of foveally presented faces is largely unaffected by attentional task
instructions. In marked contrast, emotional expression effects at latencies beyond 220
ms post-stimulus were completely eliminated in the Lines task, suggesting that
attentional task instructions had a strong impact on later stages of emotional face
These findings contrast with previous results, which demonstrated an absence
of emotional expression effects at early as well as late phases of facial expression
processing when attention was directed away from the location of these faces [7,11].
The main difference between the present experiment and these earlier studies is that
faces were now presented centrally within foveal vision, rather than in the periphery
of the visual field. It would appear, therefore, that under conditions where emotional
facial stimuli are presented foveally, and are consequently more difficult to exclude
from attentional processing, early stages of facial expression analysis take place
irrespective of current task demands, even when the encoding of these stimuli is
irrelevant to the task at hand (i.e., judgments of line lengths). It is only at later, more
controlled, stages of facial expression processing that attentional task instructions are
effective in inhibiting the further analysis of foveally presented affective information.
In this context, it is interesting to note that Pessoa et al.  also presented faces
in foveal locations, and yet still found strong effects of attention on emotional expression
processing, as revealed by fMRI measures. The present findings suggest that these
results by Pessoa et al.  are unlikely to reflect the attentional modulation of very
early stages of facial expression processing, but rather attentional effects on longer-
latency, more sustained aspects of emotional processing, which are elicited in a
controlled and task-dependent fashion after an initial rapid analysis of emotional
information is completed. Results from another ERP study, which compared effects of
attention on foveal versus peripheral non-emotional visual stimuli , also suggest that
early stages in the processing of foveal information may remain unaffected by attention.
In this experiment, attentional modulations of visual P1 components were only found in
response to peripheral events, whereas effects of attention on the subsequent N1
component were present for both foveal and peripheral visual stimuli. These findings
highlight the importance of recording electrophysiological responses in addition to
haemodynamic activity, as ERPs can indicate the emergence of potentially subtle and
transient effects that may not be evident from fMRI results. Neuroimaging measures, by
contrast, have the capacity to reveal the precise spatial characteristics of longer lasting
While current experimental evidence suggests that the processing of emotional
stimuli is strongly dependent on the current focus of spatial attention, other aspects of
a task such as the need to explicitly process emotional content seem to be much less
important. For example, similar ERP effects of emotional expression processing have
been elicited across a number of studies irrespective of whether processing of emotion
was explicit, because observers had to judge emotional expressions , or incidental,
because other non-emotional stimulus attributes were relevant [6,11]. Functional
imaging data also reveal that the processing of signals associated with threat-related
facial expressions may be independent of the explicit monitoring of emotional content
(; see ).
Taken together, these strands of evidence would seem to suggest that the
allocation of selective spatial attention is critical for processing the emotional
significance of attended information, and that other non-spatial aspects of selective
attention are weaker modulators of emotional content. The importance of spatial
attention is readily observed in situations where it can be successfully directed away
from affective content, for example, when task-irrelevant emotional faces are
presented eccentrically, and outside of focal attention [7,11]. In such circumstances, it
is clear that affective value can be fully gated. In other cases, it may be more difficult
to divert spatial attention away from affectively valenced information, as, for example,
in the current study, where task-irrelevant expressive faces were presented inside
foveal vision. Here, late stages of emotion analysis were directly influenced by
manipulations of attention, revealing attention-dependent processing of motivationally
relevant stimuli, similar to results in previous studies [7,11]. By contrast, early stages
of emotion analysis were largely unaffected by instructions to attend away from the
faces and towards competing peripheral stimuli, revealing attention-independent
processing of affective information. These findings are consistent with the view that
attentional load, spare processing capacity , and ease of excluding irrelevant
information from the window of spatial attention, are important determinants of the
extent to which the emotional significance of perceived information is encoded in the
human brain. In conclusion, the present ERP results demonstrate that selective
spatial attention gates the processing of emotional facial expression for foveal stimuli
at latencies beyond 220 ms after stimulus onset. In contrast, earlier ERP effects of
emotional expression appear to be largely immune to manipulations of spatial
attention. These results provide new evidence for a special status of foveal stimuli in
relation to their ability to trigger early stages of emotional processing irrespective of
current task demands. However, they also underline the importance of attention for
subsequent controlled stages of emotional expression analysis.
 A.K. Anderson, E.A. Phelps, Lesions of the human amygdala impair enhanced
perception of emotionally salient events, Nature 411 (2001) 305-309.
 J.L. Armony, R.J. Dolan, Modulation of spatial attention by fear-conditioned
stimuli: an event-related fMRI study, Neuropsychologia 40 (2002) 817-826.
 V. Ashley, P. Vuilleumier, D. Swick, Time course and specificity of event-related
potentials to emotional expressions, Neuroreport 15 (2003) 211-216.
 P. Bentley, P. Vuilleumier, C.M. Thiel, J. Driver, R.J. Dolan, Cholinergic
enhancement modulates neural correlates of selective attention and emotional
processing, NeuroImage 20 (2003) 58-70.
 H. Critchley, E. Daly, M. Phillips, T. Van Amelsvoort, D. Robertson, A. David,
D.G.M. Murphy, Explicit and implicit neural mechanisms for processing of social
information from facial expressions: a functional magnetic resonance imaging study,
Hum Brain Mapp. 9 (2000) 93-105.
 M. Eimer, A. Holmes, An ERP study on the time course of emotional face
processing, Neuroreport 13 (2002) 427-431.
 M. Eimer, A. Holmes, F. McGlone, The role of spatial attention in the processing
of facial expression: An ERP study of rapid brain responses to six basic emotions,
Cogn Affect Behav Neurosci. 3 (2003) 97-110.
 P. Ekman ,W.V. Friesen, Pictures of Facial Affect, Consulting Psychologists Press,
Palo Alto, CA, 1976.
 E. Fox, V. Lester, R. Russo, R.J. Bowles, A. Pichler, K. Dutton, Facial
expressions of emotion: Are angry faces detected more efficiently? Cogn Emo. 14
 A. Holmes, S. Green, P. Vuilleumier, The involvement of distinct visual channels
in rapid attention towards fearful facial expressions, Cogn Emo. (in press).
 A. Holmes, P. Vuilleumier, M. Eimer, The processing of emotional facial
expression is gated by spatial attention: evidence from event-related brain potentials,
Cogn Brain Res. 16 (2003) 174-184.
 C. Miniussi, A. Rao, A.C. Nobre, Watching where you look: modulation of
visual processing of foveal stimuli by spatial attention, Neuropsychologia 40 (2002)
 L. Pessoa, To what extent are emotional visual stimuli processed without
attention and awareness? Curr Opin Neurobiol. 15 (2005) 188-196.
 L. Pessoa, S., Kastner, L.G. Ungerleider, Attentional control of the processing of
neutral and emotional stimuli, Cogn Brain Res. 15 (2002a) 31-45.
 L. Pessoa, M. McKenna, E. Gutierrez, L.G. Ungerleider, Neural processing of
emotional faces requires attention, Proc Natl Acad Sci. USA 99 (2002b) 11458-11463.
 P. Vuilleumier, Facial expression and selective attention, Curr Opin Psyc. 15
 P. Vuilleumier, J.L. Armony, J. Driver, R.J. Dolan, Effects of attention and
emotion on face processing in the human brain: An event-related fMRI study, Neuron,
30 (2001) 829-841.
 P. Vuilleumier, M.P. Richardson, J.L. Armony, J. Driver, R.J. Dolan, Distant
influences of amygdala lesion on visual cortical activation during emotional face
processing, Nat. Neurosci. 11 (2004) 1271-1281.
 P. Vuilleumier, S. Schwartz, Emotional expressions capture attention, Neurology
56 (2001) 153-158.
FIG. 1. Example of a stimulus array used in both tasks. In the Faces task, participants
had to detect immediate repetitions of an identical face at fixation. In the Lines task, they
had to detect immediate repetitions of an identical lateral line pair across trials.
FIG. 2. Grand averaged ERPs elicited in the 700 ms after stimulus onset on non-
repetition (non-target) trials in response to stimulus arrays containing a neutral face
(solid lines) or a fearful face (dashed lines). Top: ERPs elicited in the Faces task where
faces were task-relevant (attended). Bottom: ERPs elicited in the Lines task where faces
were irrelevant (unattended).
Faces Attended: Faces Task
Faces Unattended: Lines Task