Content uploaded by José A Hinojosa
Author content
All content in this area was uploaded by José A Hinojosa
Content may be subject to copyright.
Usually, two main types of visual emotional stimuli have
been employed in the study of affect-related processes:
verbal (e.g., single words or sentences) and pictorial (e.g.,
facial expressions or emotional scenes). An important and
recurrent question is whether both types of visual stimuli
are equally capable of inducing emotional reactions. Some
authors have pointed out that pictorial stimuli are associ-
ated with higher levels of emotional arousal than is ver-
bal emotional material (Carretié et al., 2008; Keil, 2006;
Kissler, Assadollahi, & Herbert, 2006; Mogg & Bradley,
1998). On the one hand, these assertions are based in part
on theoretical views that deny any role for verbal material
in the evolution of emotion-related neural systems and
assume that the processing of verbal material is culture
mediated. On the other hand, such assertions are based on
indirect evidence provided by the results of several stud-
ies that have dealt with the processing of verbal or picto-
rial emotional stimuli separately, across a wide variety of
tasks and experimental parameters.
The situation becomes still more problematic when dif-
ferences in the stimulus channel during the processing of
verbal and pictorial stimuli are considered. Some authors
have claimed that meaning is represented in a functional
unitary system that is directly accessed by both visual ob-
jects and words (Caramazza, 1996). However, an alterna-
tive theoretical perspective (e.g., Glaser, 1992; Glaser &
Glaser, 1989) postulates a distinction between a seman-
tic system involved in the perception of images, which
contains only semantic knowledge, and a lexicon that is
responsible for language perception, which includes only
linguistic knowledge. On this view, pictures have a privi-
leged access to all nodes of the semantic system, because
language perception comprises additional processing be-
fore accessing the semantic system. On the basis of these
models, some theories of affect state that affective infor-
mation is stored as a sort of tag, associated with the con-
cept node within a network similar to the semantic system
(Bower, 1981; Fazio, Sanbonmatsu, Powell, & Kardes,
1986; Fiske & Pavelchak, 1986). According to this view,
pictures would show a facilitated access to emotional as-
pects (De Houwer & Hermans, 1994).
Studies Comparing Words and Pictures
Only two previous studies have directly compared the
processing of emotional words and pictures (see also
Vanderploeg, Brown, & Marsh, 1987, for a comparison
between words and schematic drawings of faces). In the
first of these studies, De Houwer and Hermans (1994)
used a word–picture affective Stroop task and found that
emotional pictures, but not words, produced interference
173 © 2009 The Psychonomic Society, Inc.
Electrophysiological differences in the processing
of affective information in words and pictures
Jo s é A . Hi n o J o s A
Universidad Complutense de Madrid, Madrid, Spain
Lu i s CA r r e t i é
Universidad Autónoma de Madrid, Madrid, Spain
A n d
MA r í A A . V A L C á r C e L , Co n s t A n t i n o Mé n d e z -Bé r to L o , A n d Mi g u e L A . Po z o
Universidad Complutense de Madrid, Madrid, Spain
It is generally assumed that affective picture viewing is related to higher levels of physiological arousal than
is the reading of emotional words. However, this assertion is based mainly on studies in which the processing of
either words or pictures has been investigated under heterogenic conditions. Positive, negative, relaxing, neutral,
and background (stimulus fragments) words and pictures were presented to subjects in two experiments under
equivalent experimental conditions. In Experiment 1, neutral words elicited an enhanced late positive component
(LPC) that was associated with an increased difficulty in discriminating neutral from background stimuli. In Ex-
periment 2, high-arousing pictures elicited an enhanced early negativity and LPC that were related to a facilitated
processing for these stimuli. Thus, it seems that under some circumstances, the processing of affective informa-
tion captures attention only with more biologically relevant stimuli. Also, these data might be better interpreted
on the basis of those models that postulate a different access to affective information for words and pictures.
Cognitive, Affective, & Behavioral Neuroscience
2009, 9 (2), 173-189
doi:10.3758/CABN.9.2.173
J. A. Hinojosa, hinojosa@pluri.ucm.es
174 Hi n o j o s a , Ca r r e t i é , Va l C á r C e l, Mé n d e z -Bé rt o l o , a n d Po z o
Bradley, Elbert, & Lang, 2001; Schupp, Junghöfer, Weike,
& Hamm, 2003, 2004) and words (Kissler, Herbert, Peyk,
& Junghöfer, 2007), as compared with neutral stimuli. It
is not clear whether this component is sensitive to valence,
arousal, or both aspects. In this regard, some studies have
reported that high-arousing stimuli elicited larger am-
plitudes than did low-arousing stimuli (Junghöfer et al.,
2001; Kissler, Herbert, et al., 2007), whereas others have
shown that positive stimuli were associated to enhanced
amplitudes, as compared with negative stimuli (Schupp
et al., 2004). On the basis of these inconsistent results, the
early negativity has been thought to reflect general initial
phases of attention and evaluative processes during access
to affective information.
The Present Study
Overall, it seems that words and pictures with emotional
content are able to significantly modulate brain activity.
However, the lack of studies comparing the processing
of affective information in words and pictures makes it
difficult to figure out how electrophysiological data can
accommodate any of the alternative theoretical views. The
aim of our study was thus to directly compare the process-
ing of emotional words and pictures through the analysis
of electrophysiological (ERPs) and behavioral (reaction
times [RTs]) indices. Since words and pictures are dif-
ficult to equate in their physical attributes, we decided to
present them in two separate experiments with equivalent
parameters, instead of merging words and pictures in the
same experiment.
The experimental design selected for our purposes is a
critical issue, since temporal and topographical features of
the ERPs depend strongly on such issues as task require-
ments and stimuli duration. We opted to use an indirect
task and a high rate of stimulation. In our task, subjects
had to discriminate between senseless and nonsenseless
stimuli. Nonindirect tasks (i.e., subjects are asked to per-
form an affective categorization of the stimuli) might lead
subjects to consider that some stimuli are more important
than others (i.e., emotional, salient stimuli might be more
important than neutral ones for an experiment on emo-
tional reactions). This cognitive interference overlaps with
affective effects (Carretié, Hinojosa, Albert, & Mercado,
2006; Carretié, Hinojosa, Martín-Loeches, Mercado, &
Tapia, 2004). Indeed, the ERPs elicited by emotional
stimuli in direct tasks are different from those elicited in
indirect tasks (Carretié, Iglesias, García, & Ballesteros,
1997). Also, the stimuli were presented at high rates with
the purpose of exploring the early emotional processes
indexed by the early posterior negativity.
We expected that both emotional words and pictures
elicit two components: an early negativity and a late posi-
tivity. This allowed us to answer two important questions.
First, can affective information be accessed equally by
words and pictures? Second, are pictures more “affec-
tively powerful” stimuli than words? Regarding the first
question, according to those models that assume a dis-
tinct access to affective information for words and pic-
tures (Fiske & Pavelchak, 1986), a different modulation
effects. Also, naming times were reduced for negative
pictures, but not for negative words. These authors con-
cluded that pictures have privileged access to emotional
information. In an f MRI study, Kensinger and Schacter
(2006) presented positive, negative, and neutral words
and pictures that subjects had to rate as to whether they
described something animate or inanimate or something
common or uncommon. They found that the process-
ing of both emotional words and pictures enhanced the
activity of the amygdala and several regions of the pre-
frontal and anterior temporal cortex. In general, these ef-
fects were more pronounced in the case of pictures. Even
though these studies reveal interesting aspects of the
contribution of distinct brain areas to the processing of
the emotional aspects of words and pictures and address
other general questions, they say little about the tempo-
ral aspects involved in the processing of the affective
information in words and pictures, due to the temporal
limitations of the f MRI and behavioral measures. In the
present study, our aim was to resolve this issue by using
event-related potentials (ERPs), which are particularly
well suited for studying the temporal characteristics of
emotional processes.
ERP Studies of Affective Processing
in Words and Pictures
Studies with ERPs that have used either words or im-
ages with affective content as stimuli have revealed that
the processing of emotional information modulates brain
activity at different temporal stages and scalp locations.
One of the most consistent findings has been the enhance-
ment of late positivities at centro-parietal locations (the
P300, or late positive component [LPC]), by positive and
negative stimuli, as compared with neutral ones, by either
images (e.g., Cuthbert, Schupp, Bradley, Birbaumer, &
Lang, 2000; Delplanque, Silvert, Hot, Rigoulot, & Se-
queira, 2006; Keil et al., 2002; Olofsson & Polich, 2007;
see Olofsson, Nordin, Sequeira, & Polich, 2008, for a re-
view) or words (e.g., Carretié et al., 2008; Dillon, Cooper,
Grent-’t-Jong, Woldorff, & LaBar, 2006; Herbert, Kissler,
Junghöfer, Peyk, & Rockstroh, 2006; Kanske & Kotz,
2007; Naumann, Bartussek, Diedrich, & Laufer, 1992).
However, this modulation seems to be not unequivocal
in the case of words, since several studies failed to re-
port amplitude differences (Naumann, Maier, Diedrich,
Becker, & Bartussek, 1997; Schapkin, Gusev, & Kuhl,
2000). The amplitude of the LPC has been thought to re-
flect the functional mobilization of attentional resources
and the activation of the motivational circuits in the brain
that mediate emotional engagement (Bradley & Lang,
2007; Schupp et al., 2007). The LPC is especially sensi-
tive to the level of arousal of the stimulus, since high-
arousing stimuli are associated to enhance amplitudes,
as compared with low-arousing stimuli (Olofsson et al.,
2008).
Also, a common finding of those studies in which images
have been presented briefly and at high rates of stimula-
tion has been an enhancement of the amplitude of an early
posterior negativity for emotional pictures (Junghöfer,
af f e C t i V e Pr o C e s s i n g o f Wo r d s a n d Pi C t u r e s 175
720 words were divided into three sets due to the great amount of
time that would be necessary to assess all words as a single sample.
In the end, the words that were employed in Experiment 1 were
selected according to several criteria contrasted with one-way
ANOVAs and post hoc analyses with the Bonferroni correction
(p , .05). Positive and negative words were matched in arousal
and differed in valence. Relaxing words were matched to positive
words in valence but differed in arousal and differed in both valence
and arousal from negative words. Neutral words differed from each
other word type in both valence and arousal. All the nouns were
matched in frequency of use in Spanish [Alameda & Cuetos, 1995;
92 positive words, 86 negative words, 98 neutral words, and 91
relaxing words; F(3,57) 5 0.02, p . .05]. In addition, all the nouns
were equated in word length [3.2 syllables for positive words, 3.2
syllables for negative words, 3 for neutral words, and 2.7 for relax-
ing words; F(3,57) 5 1.2, p . .05] and had similar ratings on the
concreteness scale [6.4 for positive words, 6.7 for negative words,
6.5 for neutral words, 6.7 for relaxing words; F(3,57) 5 0.22, p .
.05]. Table 1 summarizes mean ratings for each word emotion cat-
egory on the arousal and valence dimensions, as well as the results
of the ANOVAs.
Procedure
. The subjects performed an indirect task. They had to
discriminate meaningful words embedded in a sequence of nonsense
stimuli by pressing a button as quickly as possible every time a word
was detected.
Figure 1A exemplifies the procedure. The stimuli were arranged
in eight sequences and were presented according to the rapid stream
stimulation procedure (Hinojosa, Martín-Loeches, et al., 2001;
Rudell, 1992). This procedure speeds up stimuli processing by pre-
senting them at a very high rate (with a stimulus onset asynchrony
of 250 msec). In each of these sequences, the computer displayed
mainly background stimuli, and after 4–7 of these stimuli (this num-
ber being randomized), a test stimulus was presented. Each sequence
included 20 test stimuli (5 positive words, 5 negative words, 5 neutral
words, and 5 relaxing words), together with the proportional amount
of background stimuli. Test stimulus order was pseudorandomly de-
termined, with the constraint that no more than 2 of the same type
occurred consecutively. Each type of test stimulus was presented
twice during an experimental session and was never repeated within
the same sequence. A practice sequence was presented prior to the
first sequence. The subjects were instructed to minimize blinking
during stimulus presentation.
Data acquisition
. Electroencephalographic data were recorded
using an electrode cap (ElectroCap International) with tin elec-
trodes. A total of 58 scalp locations homogeneously distributed over
the scalp were used. All scalp electrodes, as well as one electrode
at the left mastoid (M1), were referenced to one electrode placed at
the right mastoid (M2). A bipolar horizontal and vertical electro-
oculogram was recorded for artifact rejection purposes. Electrode
impedances were kept below 3 kΩ. The signals were recorded con-
tinuously with a bandpass from direct current 0.1 to 50 Hz (3-dB
of the early component by emotional words and pictures
would be expected. However, a similar modulation of the
early negativity would suggest that affective information
is accessed equally by words and pictures. This finding
would be in line with the proposal of a unitary semantic
system (Caramazza, 1996). Also, distinct modulation of
the pattern of RTs by emotional words and pictures could
be taken to support the functional segregation perspec-
tive, whereas a similar modulation of RTs by both types of
stimuli would be interpreted better in light of the unitary
system proposal.
Regarding the second question, the early negativity has
been shown not to be unequivocally modulated by arousal
manipulations (Schupp et al., 2004). However, the late
positivity has repeatedly been shown to be especially sen-
sitive to arousal aspects of the stimuli (Olofsson & Polich,
2007). A greater modulation of the amplitude of this com-
ponent by affective pictures, as compared with emotional
words, would support the statement that images are as-
sociated to higher degrees of physiological arousal (Lang,
Bradley, & Cuthbert, 1998).
EXPERIMENT 1
Method
Subjects
. Twenty native Spanish speakers (19 women) partici-
pated in the experiment as volunteers. Their mean age was 21 years
(ranging from 19 to 27 years), and they all were right-handed accord-
ing to the Edinburgh Handedness Inventory (Oldfield, 1971). All the
subjects reported normal or corrected-to-normal visual acuity.
Stimuli
. For the purpose of exploring both valence (negative to
positive) and arousal (calming to arousing) dimensions, stimuli be-
longed to four different categories: negative, positive, neutral, and
relaxing. Eighty nouns (20 positive, 20 negative, 20 relaxing, and
20 neutral), together with 80 background stimuli, were used for
experimental purposes. Background stimuli were made by cutting
the 80 words into several portions. These portions were combined
randomly, resulting in nonsense stimuli resembling the physical at-
tributes of the words (size, color, brightness, etc.). All the stimuli
were presented black-on-white on a computer monitor, controlled by
the Gentask module of the STIM2 package (NeuroScan Inc.).
In a previous phase, a 720-word list that included positive, nega-
tive, relaxing, and neutral nouns, divided into three sets (240 words
each) was assessed by 45 (15 for each set) subjects (different from
those who participated in Experiment 1), who rated the valence,
arousal, and concreteness of each word on a 9-point scale (9 being
very positive, very activating, and very concrete, respectively). The
Table 1
Means for Arousal and Valence Assessments Given by the Independent Samples
of Subjects for Each Word Type and Means for Arousal and Valence Assessments
According to the IAPS for Each Picture Type, With the Results of the
Statistical Analyses for Each of These Variables
Type One-Way ANOVA
Positive Negative Neutral Relaxing (Affect Type)
Words
Valence 7.6 2.1 5.1 7.4 F(3,57) 5 618.6, p , .0001
Arousal 7.1 7.2 5.1 2.9 F(3,57) 5 984.7, p , .0001
Pictures
Valence 7.6 2.1 5.1 7.4 F(3,57) 5 121.9, p , .0001
Arousal 7.1 7.2 5.1 2.9 F(3,57) 5 293.1, p , .0001
Note—For arousal, 1 5 calming, 9 5 arousing; for valence, 1 5 negative; 9 5 positive.
Post hoc Bonferroni analyses are reported in the text.
176 Hi n o j o s a , Ca r r e t i é , Va l C á r C e l, Mé n d e z -Bé rt o l o , a n d Po z o
B
Time
250 msec
250 msec
250 msec
250 msec
(Background)
(Background)
(Background)
(Negative)
(Positive)
250 msec
A
Time
(Background)
(Background)
(Negative)
250 msec
250 msec
250 msec
(Background)
(Positive)
250 msec
250 msec
Figure 1. Schematic illustration of the stimulation paradigm for
both experiments: (A) Experiment 1 (cárcel 5 jail; fiesta 5 party).
(B) Experiment 2.
points for 26-dB octave roll-off) and a digitization sampling rate
of 250 Hz.
Data analysis
. Trials with omissions and false alarms were ex-
cluded from the analyses. Also, those trials with RTs longer than
1,500 msec or shorter than 200 msec were not analyzed. Average
ERPs from 2200 to 800 msec after the presentation of every type of
test stimulus were computed separately. Those epochs with artifacts
other than eye movements were removed after visual inspection.
Offline correction of eye movement artifacts was made, using the
method described by Semlitsch, Anderer, Schuster, and Presslich
(1986). For the entire sample of electrodes, data originally M2-
referenced were rereferenced using the average of the mastoids.
Overall repeated measures ANOVAs were f irst performed to
compare amplitudes between the ERP pattern elicited by positive,
negative, neutral, and relaxing words. Amplitude was measured as
the mean voltage within a particular time interval. To avoid the loss
of statistical power when repeated measures ANOVAs are used to
quantify a large number of electrodes (Oken & Chiappa, 1986),
11 regions of interest were computed out of 58 electrodes. These
regions are shown in Figure 2.
Three within-subjects factors were included in the ANOVA with
the purpose of exploring possible topographic variation: affect type
(four levels: positive words, negative words, neutral words, and relax-
ing words), anteriority (three levels: anterior, middle, and posterior),
and laterality (three levels: left, middle, and right). Frontopolar and
occipital regions were analyzed separately with the within- subjects
factor of affect type. The Greenhouse–Geisser epsilon correction
was applied in order to adjust the degrees of freedom of the F ratios
where necessary. In a second step and in order to explore possible
interactions involving topographical factors, further ANOVAs were
conducted for each particular region of interest, with affect type as a
within-subjects factor. The p values were adjusted to the Bonferroni
correction for multiple comparisons. Finally, Bonferroni-corrected
post hoc comparisons (p , .05) were made for determining the sig-
nificance of pairwise contrasts.
Results
Behavioral data
. There were 3,200 epochs (40 aver-
ages for each of the 4 stimulus types in 20 subjects), and
7.4% were rejected because of artifacts, 0.7% because of
false alarms, and 0.2% because of premature or delayed
responses. Also excluded were those trials with omis-
sions, which represented 1.2% for positive words, 1.1%
for negative words, 1.3% for neutral words, and 0.9% for
relaxing words. Table 2 displays the mean and standard
deviation of the RTs for every stimulus condition. Mean
RTs were 363 msec for positive words, 359 msec for neg-
ative words, 364 msec for neutral words, and 365 msec
for relaxing words. Statistical analyses revealed that
these differences were not signif icant [F(3,57) 5 2.6,
p . .05].
Electrophysiological data
. Figure 3 illustrates the
grand- averages corresponding to all types of test stim-
uli at every region of interest. A visual inspection of the
grand-averaged ERPs revealed that effects were notice-
af f e C t i V e Pr o C e s s i n g o f Wo r d s a n d Pi C t u r e s 177
positive and neutral words differed at the occipital, right-
parietal, right-frontal, frontopolar, and middle-frontal re-
gions. Finally, relaxing words differed from neutral words
at the frontopolar region.
Discussion
The main finding of the first experiment was that neutral
words elicited an enhancement of the amplitude of an LPC
as compared with positive, negative, and relaxing words in
the 350- to 425-msec time interval. No differences were
noticeable in either an early negativity or RTs.
Regarding the absence of differences in the RTs, the re-
sults from the behavioral literature are rather inconsistent.
Some studies have reported facilitation effects, whereas
others have shown inhibition for emotional words (Car-
retié et al., 2008; Fazio, 2001; Kanske & Kotz, 2007; Keil
& Ihssen, 2004; Rossell & Nobre, 2004; Williamson, Har-
pur, & Hare, 1991), using a variety of tasks, including
Stroop, lexical decision, priming, and speed of pronuncia-
tion. Also, the results of some studies resemble our find-
ings, since they have reported no effects at all (Klinger,
Burton, & Pitts, 2000; Spruyt, Hermans, Pandelaere, De
Houwer, & Eelen, 2004). Two explanations have been
proposed for this lack of effects. First, some authors have
pointed out that robust facilitation effects in behavioral
tasks are observed only when subjects are required to cat-
egorize stimuli on the basis of their valence (Spruyt, De
Houwer, Hermans, & Eelen, 2007). Second, it has been
suggested that facilitation effects may be diminished in
tasks that require the processing of a high number of
words in a short amount of time (Harris & Pashler, 2004;
Keil, 2006). Our findings support both views, since, in our
experiment, words were presented at high rates of stimu-
lation for short periods of time and the subjects were not
asked to assess the emotional connotation of the stimuli.
A common finding of those studies that have presented
words at high rates of stimulation has been the presence
of a posterior negative deflection around 250 msec (Hi-
nojosa, Martín-Loeches, Muñoz, Casado, & Pozo, 2004;
Martín-Loeches, Hinojosa, Gómez-Jarabo, & Rubia,
2001; Pulvermüller, 2001). This component has been re-
lated to early semantic processing—namely, lexical de-
cision processes (Hinojosa, Martín-Loeches, & Rubia,
2001) or word form analyses (Martín-Loeches, 2007).
Recently, Kissler, Herbert, et al. (2007) have shown that
these processes are modulated by the emotional conno-
tation of words. Several studies have also revealed that
the amplitude of an LPC is enhanced for negative and
positive words (Kanske & Kotz, 2007; Keil, Ihssen, &
Heim, 2006). This positivity has been thought to reflect
an enhancement of top-down attention to the ongoing
task (Cuthbert et al., 2000; Schupp et al., 2000).
Contrary to these findings, we found that the emotional
content of the words did not modulate the early negativ-
ity and that both positive and negative words showed re-
duced amplitudes of the LPC, as compared with neutral
stimuli. A number of reasons might account for this diver-
gent pattern of results. First of all, interestingly, several
studies have failed to report ERP differences between the
able in two time windows. The first of these effects was
a parieto- occipital negativity, peaking at about 250 msec,
with a positive counterpart at frontal electrodes. The
second effect was a parietal positivity with an onset of
about 350 msec, with a negative counterpart at frontal
electrodes. For statistical purposes, the amplitude of the
early effect was measured in the 225- to 275-msec time
interval, and the amplitude of the parietal positivity in the
350- to 425-msec time interval. Table 2 shows amplitude
means and standard deviations of these effects as a func-
tion of stimulus category.
225- to 275-msec effects. The results of the overall
ANOVA showed no significant effects involving the factor
of affect type. These results did not allow performing further
analyses.
350- to 425-msec effects. The overall ANOVA revealed
a significant effect for the affect type factor [F(3,57) 5
4.3, p , .05] and a statistical trend for the affect type 3
anteriority interaction [F(6,114)5 2.7, p , .1]. Table 3
summarizes the results of the ANOVAs for each region
of interest. They reached significance at the right-parietal
[F(3,57) 5 4.9, p , .01], frontopolar [F(3,57) 5 2.9, p ,
.05], and middle-frontal [F(3,57) 5 3, p , .05] regions.
Statistical trends were found at the occipital [F(3,57) 5
2.8, p , .1] and right-frontal [F(3,57) 5 2.2, p , .1] re-
gions. Differences between neutral and emotional stimuli
that were evident at several regions of interest, as revealed
by the post hoc analyses, are also reported in Table 3. Neu-
tral words always showed larger amplitudes than did emo-
tional words. The difference between negative words and
neutral words reached significance at the occipital, right-
parietal, right-frontal, and frontopolar regions. Similarly,
OC
RP
MP
LP
RC
MC
LC
RF
LF
FP
MF
Figure 2. Scalp regions in which ERPs were grouped for sta-
tistical contrasts. FP, frontopolar, LF, left frontal; MF, middle
central; RF, right frontal; LC, left central; MC, middle central;
RC, right central; LP, left parietal; CP, central parietal; RP, right
parietal; OC, occipital.
178 Hi n o j o s a , Ca r r e t i é , Va l C á r C e l, Mé n d e z -Bé rt o l o , a n d Po z o
Negative stimuli
Neutral stimuli
Positive stimuli
Relaxing stimuli
0–100
+6 µV
–6 µV
Frontopolar
100 200 300 400 500 600 700 800 msec
Left Frontal Right FrontalMiddle Frontal
Left Central Right CentralMiddle Central
Left Parietal Right ParietalMiddle Parietal
Occipital
Figure 3. Grand-averaged ERPs elicited by negative, neutral, positive, and relaxing words at every region of interest. Scales are
represented at the frontopolar region.
af f e C t i V e Pr o C e s s i n g o f Wo r d s a n d Pi C t u r e s 179
pear (Naumann et al., 1992; Naumann et al., 1997). More-
over, Fischler and Bradley (2006) have shown that effects
of word emotionality on late positivities are consistently
found when the task requires semantic processing of the
words, but not otherwise. Finally, it has been suggested
that word class might produce somewhat different effects
processing of emotional content and neutral words (Nau-
mann et al., 1997; Vanderploeg et al., 1987). Some au-
thors have attributed this result to task requirements. In
this regard, when subjects are not explicitly instructed to
perform an affective categorization of words, differences
in the processing of emotional and neutral words disap-
Table 2
Means and Standard Deviations for Behavioral and
Electrophysiological Recordings for Words and Pictures
Negative Positive Neutral Relaxing
M SD M SD M SD M SD
Words
Behavioral
RT (msec) 359.45 24.65 363.00 24.58 363.95 24.76 364.50 23.54
Electrophysiological
225- to 275-msec amplitudes (μV)
Frontopolar 1.27 1.28 1.25 1.62 1.03 1.40 1.64 1.70
Left frontal 1.11 1.44 1.07 1.29 1.10 1.33 1.36 1.46
Middle frontal 1.61 0.78 1.69 0.62 1.51 1.10 1.91 1.27
Right frontal 1.64 0.83 1.49 0.90 1.21 1.24 1.63 1.23
Left central 20.08 1.02 0.06 1.09 0.05 0.98 0.01 1.03
Middle central 1.58 1.22 1.66 1.08 1.54 1.33 1.72 1.43
Right central 0.63 0.90 0.38 0.89 0.34 0.96 0.42 0.92
Left parietal 22.96 1.18 22.74 1.08 22.67 1.41 23.13 1.34
Middle parietal 20.17 1.04 20.14 1.39 0.19 1.67 20.33 1.54
Right parietal 22.86 1.45 22.99 1.33 22.80 1.25 23.15 1.55
Occipital 23.36 1.65 23.25 1.59 22.87 1.85 23.79 2.09
350- to 425-msec amplitudes ( μV)
Frontopolar 21.67 2.58 21.58 2.73 22.57 2.66 21.71 2.71
Left frontal 21.90 2.36 21.80 2.21 22.44 2.35 22.13 2.13
Middle frontal 22.63 2.19 22.16 1.88 22.94 2.13 22.67 2.03
Right frontal 21.07 2.03 21.08 1.88 21.69 1.90 21.27 2.15
Left central 20.67 1.21 20.62 0.97 20.63 0.97 20.85 0.93
Middle central 20.54 1.88 20.32 1.80 20.30 2.15 20.46 1.88
Right central 20.26 1.13 20.49 0.98 20.22 1.19 20.25 0.96
Left parietal 2.83 1.90 2.62 1.49 3.19 1.44 2.79 1.85
Middle parietal 3.01 2.35 2.89 2.46 3.55 2.60 3.18 2.69
Right parietal 1.57 2.09 1.27 1.95 2.17 2.25 1.93 1.75
Occipital 3.32 2.80 3.09 2.72 4.10 2.80 3.58 2.97
Pictures
Behavioral
RT (msec) 406.07 38.68 408.54 39.11 433.86 48.31 449.00 54.46
Electrophysiological
175- to 275-msec amplitudes (μV)
Frontopolar 0.20 1.37 0.44 1.31 20.76 1.90 20.84 1.65
Left frontal 20.46 1.11 0.14 1.35 20.63 1.50 20.37 1.16
Middle frontal 20.30 1.13 0.16 1.31 20.72 1.03 20.56 1.17
Right frontal 0.20 0.98 0.39 0.96 20.50 1.16 20.56 1.25
Left central 20.42 0.66 0.08 0.78 20.07 0.80 0.20 0.66
Middle central 20.11 0.91 0.36 1.23 0 1.16 0.07 1.10
Right central 0.02 0.82 20.06 0.74 20.06 0.71 20.20 0.76
Left parietal 0.37 1.34 0.32 1.53 1.12 1.21 1.19 1.34
Middle parietal 0.58 1.41 0.40 1.36 1.05 1.59 0.86 1.47
Right parietal 0.47 1.38 20.58 1.62 0.71 1.54 0.54 1.36
Occipital 20.19 2.21 21.08 2.26 0.45 2.70 0.37 2.40
450- to 550-msec amplitudes ( μV)
Frontopolar 23.39 2.50 23.17 2.95 22.65 2.91 21.86 2.69
Left frontal 23.11 1.78 22.55 2.02 22.69 2.04 22.07 1.90
Middle frontal 20.92 2.28 20.82 2.41 21.42 2.30 21.14 1.92
Right frontal 22.08 2.22 22.02 2.43 21.44 2.02 21.20 1.85
Left central 20.56 1.09 20.18 1.04 20.76 1.09 20.51 1.13
Middle central 1.99 1.93 1.79 2.01 1.03 2.25 0.75 2.01
Right central 20.03 1.00 20.18 1.21 0.20 1.05 20.01 1.14
Left parietal 2.63 2.39 2.75 2.62 2.41 2.13 2.21 1.96
Middle parietal 3.93 1.95 3.65 2.62 3.45 2.16 2.86 2.21
Right parietal 1.37 1.86 0.82 2.00 1.50 2.07 1.00 1.80
Occipital 2.96 2.79 2.65 3.03 2.92 2.83 2.27 2.59
180 Hi n o j o s a , Ca r r e t i é , Va l C á r C e l, Mé n d e z -Bé rt o l o , a n d Po z o
(neutral words from background stimuli). Consequently,
it seems that the emotional content of words is not always
able to influence language processing. In the following
experiment, we tried to elucidate whether this would also
be the case with pictures.
EXPERIMENT 2
Method
Subjects
. Twenty-eight subjects (21 females), ranging in age
from 19 to 29 years (M 5 21.9), participated in this study as vol-
unteers. All had normal or corrected-to-normal vision, and all were
right-handed, according to the Edinburgh Handedness Inventory
(Oldfield, 1971).
Stimuli
. As in the first experiment, the stimuli consisted of 80
emotional stimuli (20 positive, 20 negative, 20 relaxing, and 20 neu-
tral pictures) and 80 background stimuli. The reasons for selecting
these four emotional categories were as explained previously. Emo-
tional pictures were taken from the IAPS database (Lang, Bradley,
& Cuthbert, 2001).1 Background stimuli were made by randomly
recombining portions of the 80 emotional pictures, resulting in non-
sense images that resembled the physical attributes of the emotional
pictures (color, brightness, etc.). Positive, negative, relaxing, and
neutral pictures were chosen according to the same criteria as those
followed in Experiment 1. As Table 1 illustrates, these criteria were
also contrasted with one-way ANOVAs and post hoc analyses with
the Bonferroni correction (p , .05). Again, positive and negative
pictures were matched in arousal and differed in valence. Relaxing
pictures were matched to positive images in valence but differed in
arousal and had different valence and arousal than did negative pic-
tures. Finally, neutral pictures differed from the other picture types
in both valence and arousal. Table 1 displays the mean valence and
arousal for every emotional picture category.
As in Experiment 1, stimuli were presented black-on-white on a
computer monitor, controlled by the Gentask module of the STIM2
package (NeuroScan Inc.).
Procedure
. Figure 1B illustrates the stimulation procedure. The
subjects were instructed to press a button as quickly as possible
whenever they detected a meaningful image. The stimuli were pre-
sented according to the rapid stream stimulation (Hinojosa, Martín-
Loeches, et al., 2001; Rudell, 1992) in eight sequences. A practice
sequence was also allowed. Each sequence contained 20 test stimuli
(5 positive pictures, 5 negative pictures, 5 neutral pictures, and 5 re-
laxing pictures), together with the proportional amount of back-
ground stimuli. The stimuli were presented following the criteria
described in the Procedure section in Experiment 1.
Data acquisition
. Electroencephalographic recording proce-
dures were exactly the same as in the first experiment, including the
use of the same electrodes and the same parameters.
Data analysis
. Data were analyzed in the same way as described
in the Data analysis section of Experiment 1.
(Kissler et al., 2006). For instance, Naumann et al. (1997)
failed to replicate their previous findings (Naumann et al.,
1992) of an enhancement of the amplitude of the LPC for
emotional adjectives when they used nouns as stimuli. In
our study, subjects were presented with nouns and had
to perform a perceptual discrimination task that did not
explicitly require either the affective categorization or the
semantic processing of words. Therefore, all the above-
mentioned reasons might explain the lack of differences
in the early negativity.
Although these arguments might also account in part
for the enhancement of the amplitude of the LPC elicited
by neutral, as compared with emotional, words, this result
deserves further consideration. A possible explanation
has to do with the proportion of neutral stimuli. In our
task, most of the stimuli had no emotional valence (neutral
and background stimuli), a situation that contrasts with
other studies in which more emotional than neutral words
have been presented (usually, two thirds emotional words
and one third neutral words). Under these circumstances,
words with emotional content could be more easily dis-
criminated from background stimuli than could neutral
words that shared the lack of emotional content with the
background stimuli. It could be the case that the low per-
centage of emotional words made these stimuli not salient
enough to disrupt the processing of the more frequent neu-
tral stimuli, resulting in the enhancement of the LPC. In
line with this argument, Naumann et al. (1997) found that
differences between emotional and neutral words in the
LPC vanished when the probability of occurrence of an
emotional word diminished.
An alternative possibility is that since emotional stimuli
were not able to attract additional attentional resources,
the repetition of neutral stimuli (both neutral words and
background) resulted in an enhanced LPC, due to an
“emotional” repetition effect. Several studies have shown
that these effects result in increased amplitudes of late
positivities (e.g., Henson, Rylands, Ross, Vuilleumier, &
Rugg, 2004; Rugg, Mark, Gilchrist, & Roberts, 1997).
In conclusion, the results of this experiment show that
at least under some circumstances, the emotional conno-
tation of words is not able to influence early processing.
Moreover, it seems that at later stages of processing, atten-
tion is directed to those words that require more effort to
be discriminated from stimuli that have a similar valence
Table 3
F Values Corresponding to the ANOVAs Performed at Each Region of Interest in the 350- to 425-msec Time Window for Words,
With the Results of the Post Hoc Bonferroni Analyses on These Contrasts
Fronto- Left Middle Right Left Middle Right Left Middle Right
polar Frontal Frontal Frontal Central Central Central Parietal Parietal Parietal Occipital
ANOVA (affect type)
on each factor 2.9*n.s. 3*2.21n.s. n.s. n.s. n.s. n.s. 4.1** 2.81
Post hoc neu . neg n.s. neu . pos neu . neg n.s. n.s. n.s. n.s. n.s. neu . neg neu . neg
neu . pos neu . pos neu . pos neu . pos
neu . rel
Note—Bonferroni analyses are reported by indicating the direction of the amplitude effect in those pairwise comparisons that reached significance
(p , .05). Neu, neutral words; neg, negative words; pos, positive words; rel, relaxing words; n.s., nonsignificant. df 5 3,57. *p , .05. **p ,
.01. 1Statistical trend, p , .1.
af f e C t i V e Pr o C e s s i n g o f Wo r d s a n d Pi C t u r e s 181
anteriority 3 laterality interaction [F(12,324) 5 2.5, p ,
.05]. Table 4 summarizes the results of the ANOVAs for
each region of interest and the post hoc analyses. The
ANOVAs at each region of interest reached significance
at the left-frontal [F(3,81) 5 7.6, p , .001], right-frontal
[F(3,81) 5 4.5, p , .01], frontopolar [F(3,81) 5 7.5, p ,
.001], middle-central [F(3,81) 5 16.5, p , .001], and
middle- parietal [F(3,81) 5 6.4, p , .01] regions. They
revealed that positive pictures elicited higher amplitudes
than did the neutral pictures at the middle-central region.
Positive pictures also differed from relaxing stimuli at the
frontopolar, right-frontal, middle-central, and middle-
parietal regions. Negative pictures elicited enhanced am-
plitudes, as compared with neutral images, at the middle-
central region. Finally, negative pictures were associated
with higher amplitudes than were relaxing pictures at
the frontopolar, left-frontal, middle-central, and middle-
parietal regions.
Discussion
Both behavioral and electrophysiological data show
that the emotional information significantly modulated
the processing of pictures. The results showed that posi-
tive and negative pictures elicited shorter RTs than did
neutral pictures. Neutral stimuli elicited shorter RTs than
relaxing pictures. Also, the activity of an early negativity
and an LPC was influenced by the emotional content of
pictures.
In line with previous literature (Calvo & Nummenmaa,
2007; Öhman, Flykt, & Esteves, 2001; Zhang, Lawson,
Guo, & Jiang, 2006), behavioral data suggest the existence
of a guided by the level of arousal gradient, since stimuli
that induced the highest levels of activation (negative and
positive) showed the shortest RTs and those that induced
the lowest amounts of activation (relaxing) showed the
longest RTs.
Emotional content also modulated electrophysiological
activity in an early negativity (175–275 msec) at several
scalp locations. Two main effects occurred during this
period of time. First, a specific effect was observed for
positive pictures that elicited enhanced amplitudes at right
parieto-occipital electrodes, as compared with all other
types of stimuli. Second, an arousal-dependent effect was
observed at left parieto-occipital locations, since both pos-
itive and negative pictures showed larger amplitudes than
did neutral and relaxing pictures. This effect replicates the
findings of the studies that have presented pictures at high
rates of stimulation (Junghöfer et al., 2001; Schupp et al.,
2004; Schupp et al., 2007) and has been related to a privi-
leged processing of emotional information in perceptual
areas (Schupp et al., 2003; Schupp et al., 2007). Interest-
ingly, Schupp et al. (2004) found an unexpected specific
effect for positive pictures that differed from negative and
neutral stimuli in the amplitude of this early negativity.
The authors speculated on the possibility that this effect
could be reflecting gender differences in affective pro-
cessing, since all the subjects in their study were females.
It seems that, as Schupp and collaborators’ and our own
data suggest, some brain circuits might be specifically en-
Results
Behavioral data
. There were 4,480 epochs (40 aver-
ages for each one of the 4 stimulus types in 28 subjects),
and 6.2% were rejected because of artifacts, 0.5% because
of false alarms, and 0.2% because of premature or delayed
responses. Those trials with omissions, which represented
1.4% for positive pictures, 1.4% for negative pictures,
1.3% for neutral pictures, and 1.2% for relaxing pictures,
were also excluded. Table 2 displays the mean and stan-
dard deviation of the RTs for every stimulus condition.
Mean RTs were 409 msec for positive pictures, 406 msec
for negative pictures, 433 msec for neutral pictures, and
449 msec for relaxing pictures. These differences were
statistically significant [F(3,81) 5 42.1, p , .001]. Bon-
ferroni post hoc analyses ( p , .05) revealed that RTs
were shorter for positive pictures than for neutral pictures
and relaxing pictures. RTs for negative pictures were also
shorter than those for neutral and relaxing pictures. Fi-
nally, RTs for relaxing pictures were larger than those for
neutral pictures.
Electrophysiological data
. Figure 4 shows the wave-
forms for all stimulus types at every region of interest.
Again, after a visual inspection of the grand averages,
ERP effects were observed in two time windows: a parieto-
occipital negativity peaking at about 225 msec with a posi-
tive counterpart at frontal electrodes and a parieto-occipital
positivity with an onset of about 450 msec and a negative
counterpart at frontal electrodes. The amplitude of these
effects was measured in the 175- to 275-msec and 450- to
550-msec time intervals, respectively, for statistical pur-
poses. Table 2 displays amplitude means and standard de-
viations of these effects for every stimulus condition.
175- to 275-msec effects. The overall ANOVA reached
significance for the affect type factor [F(3,81) 5 4.4, p ,
.05], and for the affect type 3 anteriority [F(6,162) 5 10.8,
p , .001], affect type 3 laterality [F(6,162) 5 3.7, p ,
.05], and affect type 3 anteriority 3 laterality [F(12,324) 5
4.8, p , .01] interactions. Table 4 shows the results of the
ANOVAs conducted at each region of interest. The ANOVAs
at each region of interest were significant at the occipital
[F(3,81) 5 11.2, p , .001], left-parietal [F(3,81) 5 10.6,
p , .001], right-parietal [F(3,81) 5 19.9, p , .001], and
right-frontal [F(3,81) 5 10.3, p , .001] regions. Differ-
ences between high-arousal and low-arousal pictures, as
revealed by the Bonferroni-corrected post hoc analyses, are
also reported in Table 4. Positive pictures showed enhanced
amplitudes, as compared with both neutral and relaxing
pictures, at the occipital, left- and right-parietal, and right-
frontal regions. Also, positive pictures were associated with
higher amplitudes than were negative pictures at the right-
parietal and occipital regions.
Negative pictures elicited enhanced amplitude, as com-
pared with both neutral and relaxing pictures, at the right-
frontal and left-parietal regions.
450- to 550-msec effects. The overall ANOVA revealed
a significant effect in the affect type factor [F(3,81) 5
11.7, p , .001], the affect type 3 anteriority interaction
[F(6,162) 5 3.5, p , .05], the affect type 3 laterality in-
teraction [F(6,162) 5 9.8, p , .001], and the affect type 3
182 Hi n o j o s a , Ca r r e t i é , Va l C á r C e l, Mé n d e z -Bé rt o l o , a n d Po z o
Negative stimuli
Neutral stimuli
Positive stimuli
Relaxing stimuli
0–100
+6 µV
–6 µV
Frontopolar
100 200 300 400 500 600 700 800 msec
Left Frontal Right FrontalMiddle Frontal
Left Central Right CentralMiddle Central
Left Parietal Right ParietalMiddle Parietal
Occipital
Figure 4. Grand-averaged ERPs elicited by negative, neutral, positive, and relaxing pictures at every region of interest. Scales are
represented at the frontopolar region.
af f e C t i V e Pr o C e s s i n g o f Wo r d s a n d Pi C t u r e s 183
early and late effects in both words and pictures and for
each emotional stimulus after subtracting the activity elic-
ited by neutral stimuli.
The brain activity corresponding to the two ERP com-
ponents elicited by words and pictures in all the emo-
tional conditions was subjected to an ANOVA. This
analysis included the within-subjects factors of affect
type (four levels), anteriority (three levels), and laterality
(three levels) and the between-subjects factor of stimulus
type (two factors: words and pictures). Only the interac-
tions involving the affect type and stimulus type factors
were relevant for the purposes of this study. Additional
ANOVAs involving the within-subjects factor of affect
type (four levels) and the between-subjects factor of stim-
ulus type (two factors) were carried out in each region of
interest in order to better characterize these interactions
whenever found. Finally, in those regions of interest that
showed significant interactions between affect type and
stimulus type, Bonferroni corrected ( p , .05) pairwise
tests comparing every emotional word category with its
picture counterpart were carried out (i.e., negative words
vs. negative pictures, neutral words vs. neutral pictures,
and so on). This procedure was an attempt to explore the
possibility that the differences observed between words
and pictures were due to an effect confined to a particular
emotional category.
In the early component, significant results were found for
the affect type 3 anteriority 3 stimulus type [F(6,276) 5
4.8, p , .01] and affect type 3 anteriority 3 laterality 3
gaged in the processing of positive information, as some
recent fMRI studies have shown (Hamann & Mao, 2002;
Kensinger & Schacter, 2006).
The emotional content of the pictures also influenced later
stages of processing as indexed by an LPC in the 450- to
550-msec time interval. Both positive and negative pictures
elicited more enhanced amplitudes than did low-arousing
pictures at centro-parietal and frontal electrodes. This result
suggests a modulation of the amplitude of the LPC accord-
ing to the level of arousal, which is similar to what occurs
with RTs. Similar findings have been also reported in pre-
vious studies (Cuthbert et al., 2000; Delplanque, Lavoie,
Hot, Silvert, & Sequeira, 2004; Olofsson & Polich, 2007).
These differences have been interpreted as reflecting the
engagement of the attentional system in a more complete
processing of affective stimuli (Carretié, Martín-Loeches,
Hinojosa, & Mercado, 2001; Cuthbert et al., 2000; Schupp
et al., 2007). Consequently, it seems that the processing of
high-arousing emotional pictures is able to attract top-down
modulated attention that reflects the activation of motiva-
tional systems in the brain even when they are embedded in
a stream of nonrecognizable stimuli.
Comparison Between Experiments 1 and 2
In order to allow a direct comparison between the ERPs
elicited by words and pictures, Figure 5 illustrates a se-
lection of those regions of interest in which differences
between emotional categories were more evident. In ad-
dition, Figure 6 shows the topographic distribution of the
Table 4
F Values Corresponding to the ANOVAs Performed at Each Region of Interest in the 175- to 275-msec and the 450- to 550-msec
Time Intervals for Pictures, With the Results of the Post Hoc Bonferroni Analyses on These Contrasts
Fronto- Left Middle Right Left Middle Right Left Middle Right
polar Frontal Frontal Frontal Central Central Central Parietal Parietal Parietal Occipital
175–275 msec
ANOVA (affect type)
on each factor n.s. n.s. n.s. 10.3*** n.s. n.s. n.s. 10.6*** n.s. 19.9*** 11.2***
Post hoc n.s. n.s. n.s. pos . neu n.s. n.s. n.s. pos . neu n.s. pos . neu pos . neu
pos . neg pos . neg
450–550 msec
ANOVA (affect type)
on each factor 7.5*** 7.6*** 16.5*** 4.5** n.s. 16.5*** n.s. n.s. 6.4** n.s. n.s.
Post hoc pos . rel pos . rel pos . rel pos . rel n.s. pos . rel n.s. n.s. n.s. n.s. n.s.
pos . neu pos . neu
Note—Bonferroni analyses are reported by indicating the direction of the amplitude effect in those pairwise comparisons that reached signifi-
cance ( p , .05). Neu, neutral pictures; neg, negative pictures; pos, positive pictures; rel, relaxing pictures; n.s., nonsignificant. df 5 3,81. **p ,
.01. ***p , .001.
Table 5
F Values Corresponding to the ANOVAs Performed at Each Region of Interest in the
Early and Late Effects for the Comparison of Words and Pictures
Fronto- Left Middle Right Left Middle Right Left Middle Right
polar Frontal Frontal Frontal Central Central Central Parietal Parietal Parietal Occipital
Early Effect
ANOVA (affect type)
on each factor 6.36** 2.6712.6714.46** n.s. n.s. n.s. 6.52** n.s. 7.14*** 6.26**
Late Effect
ANOVA (affect type)
on each factor 5.74** 4.12** n.s. 5.11** n.s. 9.17*** n.s. n.s. 4.53** n.s. n.s.
Note—Only the relevant results of the interaction between the affect type and stimulus type factors are shown. n.s., nonsignificant. df 5
3,138. **p , .01. ***p , .001. 1Statistical trend, p , .1.
184 Hi n o j o s a , Ca r r e t i é , Va l C á r C e l, Mé n d e z -Bé rt o l o , a n d Po z o
Negative stimuli
Neutral stimuli
Positive stimuli
Relaxing stimuli
AWords BPictures
Occipital
Middle Frontal Middle Frontal
Middle Central Middle Central
Right ParietalLeft ParietalRight ParietalLeft Parietal
Occipital
Frontopolar Frontopolar
0–100
+6 µV
–6 µV
100 200 300 400 500 600 700 800 msec
Figure 5. Comparison of the grand-averaged waveforms at a selected sample of regions of interest elicited by words (A) and pictures (B). Scales are represented at
the frontopolar region.
af f e C t i V e Pr o C e s s i n g o f Wo r d s a n d Pi C t u r e s 185
Post hoc analyses conducted in those regions in which the
affect type 3 stimulus type interaction reached significance
showed that negative, positive, neutral, and relaxing words
differed from negative, positive, neutral, and relaxing im-
ages at the same regions (occipital, left- and right-parietal,
left-, middle-, and right-frontal, and frontopolar locations).
The amplitude of the late component showed differences
as a function of the stimulus type, reflected in the signifi-
cant affect type 3 stimulus type [F(3,138) 5 7, p , .001],
affect type 3 anteriority 3 stimulus type [F(6,276) 5 3.3,
p , .05], and affect type 3 laterality 3 type of stimuli
stimulus type [F(12,552) 5 2.1, p , .05] interactions. A
statistical trend was observed for the interaction between af-
fect type and stimulus type [F(6,162) 5 3.7, p , .1]. Table 5
summarizes the results of these analyses. The ANOVAs
conducted at each region of interest reached significance
at the occipital [F(3,138) 5 6.3, p , .01], left- parietal
[F(3,138) 5 6.5, p , .01], right-parietal [F(3,138) 5 7.1,
p , .001], right-frontal [F(3,138) 5 4.5, p , .01], and
frontopolar [F(3,138) 5 6.4, p , .01] regions. A statis-
tical trend was found at the left-frontal [F(3,138) 5 2.7,
p , .1] and middle-frontal [F(3,138) 5 2.7, p , .1] regions.
A
B
175–275 msec
450–550 msec
Relaxing � Neutral
Relaxing � Neutral
Relaxing � Neutral
Relaxing � Neutral
Words
Pictures
Negative � Neutral
Negative � Neutral
Negative � Neutral
Negative � Neutral
Positive � Neutral
Positive � Neutral
Positive � Neutral
Positive � Neutral
–0.8 µV
–1 µV
1.1 µV
1.2 µV
225–275 msec
350–425 msec
–0.4 µV
–0.5 µV
0.4 µV
1 µV
Figure 6. Topographic difference maps of the distribution of the early posterior negativity and the late
positive component for every word and emotion picture category. The activity associated to neutral stimuli
has been subtracted from the activity elicited by affective stimuli.
186 Hi n o j o s a , Ca r r e t i é , Va l C á r C e l, Mé n d e z -Bé rt o l o , a n d Po z o
tailed discussions of this issue, see Caramazza, 1996; Hi-
nojosa, Martín-Loeches, Gómez-Jarabo, & Rubia, 2000;
Keil, 2006; Martín-Loeches, 2007), the processing of
both words and pictures was associated with two similar
components in our study: an early posterior negativity
and an LPC.
Emotional pictures elicited an enhancement of the am-
plitude of a posterior negativity between 175 and 275 msec
after the stimulus presentation. However, there were no
differences between affective categories, even if a narrow
time window from 225 to 275 msec was considered. Dif-
ferences between words and pictures were more evident
at frontal and parieto-occipital electrodes. Several studies
have shown facilitation for the processing of affective in-
formation at early stages reflected by similar negativities
for either words (Kissler, Herbert, & Junghöfer, 2007) or
pictures (Junghöfer et al., 2001). However, the results of
our study suggest that whereas this advantage is consistent
for pictorial information, it cannot be generalized for ver-
bal information. This finding could account for the results
of those studies that failed to report early modulations by
emotional information during word processing (e.g., Nau-
mann et al., 1997; Vanderploeg et al., 1987) and suggests
that special care should be taken when approaching the
study of emotion using words as stimuli. The different
modulation of the early negativity by words and pictures
supports those theoretical perspectives that postulate a
functional segregation of the access to affective informa-
tion for this type of stimuli and fits well with the idea of
a privileged access of images to emotional information
(Bower, 1981; Fiske & Pavechak, 1986).
The processing of emotional information also differed
between words and pictures at later stages, as reflected by
the different modulation of the LPC, a component that is
thought to reflect the functional mobilization of attentional
resources in affective processing (Delplanque et al., 2006).
High-arousing pictures elicited larger amplitudes than did
low-arousing images, whereas neutral words elicited larger
amplitudes than did emotional words. Such differences
were especially evident at frontal and centro-parietal scalp
regions and suggest that attention was engaged differently
in the processing of pictorial and verbal affective informa-
tion. It seems that it was task compromised in the case of
words, facilitating the processing of those stimuli that were
more difficult to discriminate (see the Discussion section
of Experiment 1). By contrast, attentional engagement was
emotionally driven in the case of pictures, facilitating the
processing of those stimuli that are more relevant from a
biological perspective (see the Discussion section of Ex-
periment 2). These results now provide direct evidence
supporting the view that emotional images are capable of
inducing a higher amount of physiological arousal than are
affective words, as has previously been suggested (Carretié
et al., 2008; Keil, 2006).
Another difference refers to the fact that LPC effects
were noticeable at different time windows, from 350 to
425 msec for words and from 450 to 550 msec for pic-
tures. Several studies have reported LPC effects at dif-
ferent time windows for the affective processing of either
[F(6,276) 5 3.9, p , .01] interactions. As can be appreci-
ated in Table 5, the interaction between affect type and
type of stimuli was significant in several regions of inter-
est. It was significant at left-frontal [F(3,138) 5 4.1, p ,
.01], right-frontal [F(3,138) 5 5.1, p , .01], frontopolar
[F(3,138) 5 5.7, p , .01], middle-central [F(3,138) 5
9.2, p , .001], and middle-parietal [F(3,138) 5 4.5, p ,
.01] regions. The results of the post hoc analyses indicated
that negative, positive, neutral, and relaxing words and
pictures differed at middle-central electrodes. Also, the
difference between negative words and pictures was sig-
nificant at the frontopolar and left-frontal regions.
GENERAL DISCUSSION
The results of this study showed that the processing
of emotional information operated in a different way for
words and pictures even under equivalent experimental
circumstances in which no explicit semantic processing
was required. Differences were evident at both behav-
ioral and electrophysiological levels and indicated that
affective information modulated the processing of pic-
tures and had little influence in word processing. Several
authors have previously suggested that stimuli that are
relevant from a biological point of view, such as emo-
tional images, are capable of inducing higher amounts
of physiological arousal in the viewer than are verbal
emotional material whose emotional meaning is acquired
by learning (Keil, 2006; Kissler et al., 2006; Mogg &
Bradley, 1998; Vanderploeg et al., 1987). However, such
a suggestion has been based mainly on indirect data. Our
study now provides direct electrophysiological evidence
supporting this view.
Reaction Times
A first difference in the processing of emotional aspects
of words and pictures emerged in the pattern of RTs. A
clear influence of the arousal dimension is noticeable in
the case of pictures. In this regard, positive and negative
high-arousing pictures elicited shorter RTs as compared
with neutral, mildly arousing pictures and low-arousing
relaxing images. No differences were apparent between
arousal categories in the case of words. This finding is
also in agreement with the results of the study by De Hou-
wer and Hermans (1994). These authors reported differ-
ences in several behavioral measures, including interfer-
ence effects and naming times, between the processing of
emotional and neutral pictures, but not between emotional
and neutral words. Taken together, these data support the
claim that there is a different access to affective informa-
tion by words and pictures (Fiske & Pavelchak, 1986).
ERP Data
Differences in the processing of affective information
between words and pictures were even more evident in
the pattern of ERPs. Despite the existence of some dif-
ferences in the processing of words and pictures, due to
the fact that activating an emotional network via lexical
information is different from object recognition (for de-
af f e C t i V e Pr o C e s s i n g o f Wo r d s a n d Pi C t u r e s 187
views by showing that, under some circumstances, those
emotional stimuli that are not as biologically prepared as
others might not always be able to surpass the threshold
and attract attention.
AUTHOR NOTE
This work was supported by Grant SEJ2005-08461-C02-01/PSIC
from the Ministerio de Educación y Ciencia of Spain. Correspondence
concerning this article should be addressed to J. A. Hinojosa, Instituto
Pluridisciplinar, Universidad Complutense de Madrid, 28040 Madrid,
Spain (e-mail: hinojosa@pluri.ucm.es).
REFERENCES
Alameda, J. R., & Cuetos, F. (1995). Diccionario de frecuencias de
las unidades lingüísticas del castellano. Oviedo: Universidad de
Oviedo.
Bower, G. H. (1981). Mood and memory. American Psychologist, 36,
129-148.
Bradley, M. M., & Lang, P. J. (2007). Emotion and motivation. In
J. T. Cacioppo, L. G. Tassinary, & G. G. Berntson (Eds.), Handbook
of psychophysiology (3rd ed., pp. 581-607). Cambridge: Cambridge
University Press.
Calvo, M. G., & Nummenmaa, L. (2007). Processing of unattended
emotional visual scenes. Journal of Experimental Psychology: Gen-
eral, 136, 347-369.
Caramazza, A. (1996). Pictures, words and the brain. Nature, 383, 216-
217.
Carretié, L., Hinojosa, J. A., Albert, J., López-Martín, S., de la
Gándara, B. S., Igoa, J. M., & Sotillo, M. (2008). Modulation of
ongoing cognitive processes by emotionally intense words. Psycho-
physiology, 45, 188-196.
Carretié, L., Hinojosa, J. A., Albert, J., & Mercado, F. (2006). Neu-
ral response to sustained affective visual stimulation using an indirect
task. Experimental Brain Research, 174, 630-637.
Carretié, L., Hinojosa, J. A., Martín-Loeches, M., Mercado, F., &
Tapia, M. (2004). Automatic attention to emotional stimuli: Neural
correlates. Human Brain Mapping, 22, 290-299.
Carretié, L., Iglesias, J., García, T., & Ballesteros, M. (1997).
N300, P300 and the emotional processing of visual stimuli. Electro-
encephalography & Clinical Neurophysiology, 103, 298-303.
Carretié, L., Martín-Loeches, M., Hinojosa, J. A., & Mercado, F.
(2001). Emotion and attention interaction studied through event- related
potentials. Journal of Cognitive Neuroscience, 13, 1109-1128.
Cuthbert, B. N., Schupp, H. T., Bradley, M. M., Birbaumer, N., &
Lang, P. J. (2000). Brain potentials in affective picture processing:
Covariation with autonomic arousal and affective report. Biological
Psychology, 52, 95-111.
De Houwer, J., & Hermans, D. (1994). Differences in the affective
processing of words and pictures. Cognition & Emotion, 8, 1-20.
Delplanque, S., Lavoie, M. E., Hot, P., Silvert, L., & Sequeira, H.
(2004). Modulation of cognitive processing by emotional valence
studied through event-related potentials in humans. Neuroscience
Letters, 356, 1-4.
Delplanque, S., Silvert, L., Hot, P., Rigoulot, S., & Sequeira, H.
(2006). Arousal and valence effects on event-related P3a and P3b dur-
ing emotional categorization. International Journal of Psychophysiol-
ogy, 60, 315-322.
Dillon, D. G., Cooper, J. J., Grent-’t-Jong, T., Woldorff, M. G., &
LaBar, K. S. (2006). Dissociation of event-related potentials index-
ing arousal and semantic cohesion during emotional word encoding.
Brain & Cognition, 62, 43-57.
Fazio, R. H. (2001). On the automatic activation of associated evalua-
tions: An overview. Cognition & Emotion, 15, 115-141.
Fazio, R. H., Sanbonmatsu, D. M., Powell, M. C., & Kardes, F. R.
(1986). On the automatic activation of attitudes. Journal of Personal-
ity & Social Psychology, 50, 229-238.
Fischler, I., & Bradley, M. M. (2006). Event-related potential studies
of language and emotion: Words, phrases, and task effects. Progress
in Brain Research, 156, 185-203.
words (Dillon et al., 2006; Herbert et al., 2006; Kanske &
Kotz, 2007) or pictures (Huang & Luo, 2006; Keil et al.,
2002; Schupp et al., 2007), ranging from 200 to 800 msec.
The results of some of these studies are favorable to
there being a temporal advantage for the processing of
words, whereas others show that pictures are processed
earlier. However, the existence of serious divergences in
the experimental procedures does not allow establishing
whether emotional words or pictures are processed more
quickly. Under the same experimental conditions, our data
show that LPC effects for words precede those reflecting
picture processing. Several reasons could account for this
discrepancy. For pictures, it might be the case that subse-
quent facilitated processing reflected by the LPC does not
occur until previous facilitation effects in the access to
affective information at early stages of processing (as re-
flected by the early negativity) have been resolved. Since
this previous facilitation did not occur for words, the pro-
cesses indexed by the LPC could be triggered earlier than
in the case of pictures. Alternatively, this difference might
be related to the distinct nature of the engagement of at-
tention (in a task-driven way for words vs. an emotion-
driven fashion for pictures).
It should be remarked that in addition to affective sig-
nificance, the LPC is sensitive to stimulus probability
and rate of target occurrence (Polich, 2007; Rosenfeld,
Biroschak, Kleschen, & Smith, 2005). Although LPC
amplitude might be reduced in rapid serial presentation
paradigms (Schupp et al., 2007), stimulus probability ap-
pears unlikely to explain the differential LPC findings
for words and pictures in this study. Above all, the target/
nontarget ratio was the same, at least for positive, nega-
tive, and relaxing stimuli. However, relaxing stimuli did
not show different LPC effects, as compared with either
emotional words or pictures. Furthermore, even though
stimulus probability for all the categories was the same in
both experiments, the results were quite the opposite.
Conclusion
Our data suggest that emotional information exerts
a different influence on the processing of pictures, as
compared with the processing of words, in accordance
with some theoretical proposals (Bower, 1981; Fiske &
Pavelchak, 1986). This seems to be true at least when
stimuli have to be processed quickly, without focusing
the attention on their emotional or semantic content, and
when they are embedded in a stream of nonemotional
stimuli. It has been suggested that emotional stimuli
need to exceed a critical threshold value before they are
able to capture attention (Koster, Crombez, Van Damme,
Verschuere, & De Houwer, 2004; Mogg & Bradley,
1998). Several factors have been claimed to influence
this threshold. Among them, the level of involvement in
the ongoing cognitive task seems to play a crucial role
(Schwartz et al., 2005). Also, it has been pointed out that
emotional stimuli are not insensitive to some regulatory
influences, including inhibition effects due to sensory
competition or high attentional load (Lavie, 2005; Vuil-
leumier, 2005). Our results are in accordance with these
188 Hi n o j o s a , Ca r r e t i é , Va l C á r C e l, Mé n d e z -Bé rt o l o , a n d Po z o
M. T. Balaban (Eds.), Attention and orienting: Sensory and motiva-
tional processes (pp. 97-135). Mahwah, NJ: Erlbaum.
Lang, P. J., Bradley, M. M., & Cuthbert, B. N. (2001). International
affective picture system (IAPS): Instruction manual and affective rat-
ings (Tech. Rep. A-5). Gainesville: University of Florida, Center for
Research in Psychophysiology.
Lavie, N. (2005). Distracted and confused?: Selective attention under
load. Trends in Cognitive Sciences, 9, 75-82.
Martín-Loeches, M. (2007). The gate for reading: Reflections on the
recognition potential. Brain Research Reviews, 53, 89-97.
Martín-Loeches, M., Hinojosa, J. A., Gómez-Jarabo, G., & Rubia,
F. J. (2001). An early electrophysiological sign of semantic processing
in basal extrastriate areas. Psychophysiology, 38, 114-124.
Mogg, K., & Bradley, B. P. (1998). A cognitive-motivational analysis
of anxiety. Behaviour Research & Therapy, 36, 809-848.
Naumann, E., Bartussek, D., Diedrich, O., & Laufer, M. E. (1992).
Assessing cognitive and affective information processing functions of
the brain by means of the late positive complex of the event-related
potential. Journal of Psychophysiology, 6, 285-298.
Naumann, E., Maier, S., Diedrich, O., Becker, G., & Bartussek, D.
(1997). Structural, semantic, and emotion-focused processing of neu-
tral and negative nouns: Event-related potential correlates. Journal of
Psychophysiology, 11, 158-172.
Öhman, A., Flykt, A., & Esteves, F. (2001). Emotion drives attention:
Detecting the snake in the grass. Journal of Experimental Psychology:
General, 130, 466-478.
Oken, B. S., & Chiappa, K. H. (1986). Statistical issues concerning
computerized analysis of brainwave topography. Annals of Neurol-
ogy, 19, 493-497.
Oldfield, R. C. (1971). The assessment and analysis of handedness:
The Edinburgh inventory. Neuropsychologia, 9, 97-113.
Olofsson, J. K., Nordin, S., Sequeira, H., & Polich, J. (2008). Af-
fective picture processing: An integrative review of ERP findings.
Biological Psychology, 77, 247-265.
Olofsson, J. K., & Polich, J. (2007). Affective visual event-related
potentials: Arousal, repetition, and time-on-task. Biological Psychol-
ogy, 75, 101-108.
Polich, J. (2007). Updating P300: An integrative theory of P3a and P3b.
Clinical Neurophysiology, 118, 2128-2148.
Pulvermüller, F. (2001). Brain reflections of words and their mean-
ing. Trends in Cognitive Sciences, 5, 517-524.
Rosenfeld, J. P., Biroschak, J. R., Kleschen, M. J., & Smith, K. M.
(2005). Subjective and objective probability effects on P300 ampli-
tude revisited. Psychophysiology, 42, 356-359.
Rossell, S. L., & Nobre, A. C. (2004). Semantic priming of different
affective categories. Emotion, 4, 354-363.
Rudell, A. P. (1992). Rapid stream stimulation and the recognition poten-
tial. Electroencephalography & Clinical Neurophysiology, 83, 77-82.
Rugg, M. D., Mark, R. E., Gilchrist, J., & Roberts, R. C. (1997).
ERP repetition effects in indirect and direct tasks: Effects of age and
interitem lag. Psychophysiology, 34, 572-586.
Schapkin, S. A., Gusev, A. N., & Kuhl, J. (2000). Categorization of
unilaterally presented emotional words: An ERP analysis. Acta Neuro-
biologiae Experimentalis, 60, 17-28.
Schupp, H. T., Cuthbert, B. N., Bradley, M. M., Cacioppo, J. T.,
Ito, T., & Lang, P. J. (2000). Affective picture processing: The
late positive potential is modulated by motivational relevance. Psy-
chophysiology, 37, 257-261.
Schupp, H. T., Junghöfer, M., Weike, A. I., & Hamm, A. O. (2003).
Emotional facilitation of sensory processing in the visual cortex. Psy-
chological Science, 14, 7-13.
Schupp, H. T., Junghöfer, M., Weike, A. I., & Hamm, A. O. (2004).
The selective processing of briefly presented pictures: An ERP analy-
sis. Psychophysiology, 41, 441-449.
Schupp, H. T., Stockburger, J., Codispoti, M., Junghöfer, M.,
Weike, A. I., & Hamm, A. O. (2007). Selective visual attention to
emotion. Journal of Neuroscience, 27, 1082-1089.
Schwartz, S., Vuilleumier, P., Hutton, C., Maravita, A., Dolan,
R. J., & Driver, J. (2005). Attentional load and sensory competition
in human vision: Modulation of f MRI responses by load at fixation
during task-irrelevant stimulation in the peripheral visual field. Cere-
bral Cortex, 15, 770-786.
Fiske, S. T., & Pavelchak, M. A. (1986). Category-based versus
piecemeal- based affective responses: Developments in schema-
triggered affect. In R. M. Sorrentino & E. T. Higgins (Eds.), Hand-
book of motivation and cognition (pp. 167-203). New York: Wiley.
Glaser, W. R. (1992). Picture naming. Cognition, 42, 61-105.
Glaser, W. R., & Glaser, M. O. (1989). Context effects on Stroop-like
word and picture processing. Journal of Experimental Psychology:
General, 118, 13-42.
Hamann, S., & Mao, H. (2002). Positive and negative emotional verbal
stimuli elicit activity in the left amygdala. NeuroReport, 13, 15-19.
Harris, C. R., & Pashler, H. (2004). Attention and the processing of
emotional words and names: Not so special after all. Psychological
Science, 15, 171-178.
Henson, R. N., Rylands, A., Ross, E., Vuilleumier, P., & Rugg,
M. D. (2004). The effect of repetition lag on electrophysiological and
haemodynamic correlates of visual object priming. NeuroImage, 21,
1674-1689.
Herbert, C., Kissler, J., Junghöfer, M., Peyk, P., & Rockstroh, B.
(2006). Processing of emotional adjectives: Evidence from startle
EMG and ERPs. Psychophysiology, 43, 197-206.
Hinojosa, J. A., Martín-Loeches, M., Casado, P., Muñoz, F.,
Fernández- Frías, C., & Pozo, M. A. (2001). Studying semantics
in the brain: The rapid stream stimulation paradigm. Brain Research
Protocols, 8, 199-207.
Hinojosa, J. A., Martín-Loeches, M., Gómez-Jarabo, G., & Rubia,
F. J. (2000). Common basal extrastriate areas for the semantic process-
ing of words and pictures. Clinical Neurophysiology, 111, 552-560.
Hinojosa, J. A., Martín-Loeches, M., Muñoz, F., Casado, P., &
Pozo, M. A. (2004). Electrophysiological evidence of automatic early
semantic processing. Brain & Language, 88, 39-46.
Hinojosa, J. A., Martín-Loeches, M., & Rubia, F. J. (2001). Event-
related potentials and semantics: An overview and an integrative pro-
posal. Brain & Language, 78, 128-139
Huang, Y.-X., & Luo, Y.-J. (2006). Temporal course of emotional nega-
tivity bias: An ERP study. Neuroscience Letters, 398, 91-96.
Junghöfer, M., Bradley, M. M., Elbert, T. R., & Lang, P. J. (2001).
Fleeting images: A new look at early emotion discrimination. Psy-
chophysiology, 38, 175-178.
Kanske, P., & Kotz, S. A. (2007). Concreteness in emotional words: ERP
evidence from a hemifield study. Brain Research, 1148, 138-148.
Keil, A. (2006). Macroscopic brain dynamics during verbal and picto-
rial processing of affective stimuli. Progress in Brain Research, 156,
217-232.
Keil, A., Bradley, M. M., Hauk, O., Rockstroh, B., Elbert, T. [R.],
& Lang, P. J. (2002). Large-scale neural correlates of affective picture
processing. Psychophysiology, 39, 641-649.
Keil, A., & Ihssen, N. (2004). Identification facilitation for emotionally
arousing verbs during the attentional blink. Emotion, 4, 23-35.
Keil, A., Ihssen, N., & Heim, S. (2006). Early cortical facilitation for
emotionally arousing targets during the attentional blink. BMC Biol-
ogy, 4, 23.
Kensinger, E. A., & Schacter, D. L. (2006). Processing emotional
pictures and words: Effects of valence and arousal. Cognitive, Affec-
tive, & Behavioral Neuroscience, 6, 110-126.
Kissler, J., Assadollahi, R., & Herbert, C. (2006). Emotional and
semantic networks in visual word processing: Insights from ERP stud-
ies. Progress in Brain Research, 156, 147-183.
Kissler, J., Herbert, C., & Junghöfer, M. (2007). Attention and emo-
tion in visual word processing: An ERP study [Abstract]. Psycho-
physiology, 44, S37.
Kissler, J., Herbert, C., Peyk, P., & Junghöfer, M. (2007). Buzzwords:
Early cortical responses to emotional words during reading. Psycho-
logical Science, 18, 475-480.
Klinger, M. R., Burton, P. C., & Pitts, G. S. (2000). Mechanisms
of unconscious priming: I. Response competition, not spreading ac-
tivation. Journal of Experimental Psychology: Learning, Memory, &
Cognition, 26, 441-455.
Koster, E. H. W., Crombez, G., Van Damme, S., Verschuere, B., &
De Houwer, J. (2004). Does imminent threat capture and hold atten-
tion? Emotion, 4, 312-317.
Lang, P. J., Bradley, M. M., & Cuthbert, B. N. (1998). Motivated
attention: Affect, activation, and action. In P. J. Lang, R. F. Simons, &
af f e C t i V e Pr o C e s s i n g o f Wo r d s a n d Pi C t u r e s 189
Zhang, Q., Lawson, A., Guo, C., & Jiang, Y. (2006). Electrophysi-
ological correlates of visual affective priming. Brain Research Bul-
letin, 71, 316-323.
NOTE
1. Positive images: 1710, 2160, 2216, 2345, 4599, 4607, 4608, 4641,
4652, 4660, 4670, 4680, 5480, 5621, 5623, 8034, 8370, 8400, 8470, 8501;
negative images: 3101, 3220, 3230, 3500, 3550, 6250, 6570.1, 7920,
9050, 9181, 9300, 9410, 9420, 9600, 9620, 9635.1, 9800, 9910, 9911,
9921; neutral images: 1935, 2410, 2575, 5395, 5455, 5531, 5532, 5533,
5535, 7491, 7495, 7496, 7500, 7504, 7510, 7560, 7620, 7820, 7830, 9472;
relaxing images: 1604, 1610, 1900, 1910, 2304, 2370, 2388, 5000, 5001,
5010, 5020, 5030, 5200, 5551, 5720, 5750, 5760, 5800, 5811, 5870.
(Manuscript received May 8, 2008;
revision accepted for publication November 3, 2008.)
Semlitsch, H. V., Anderer, P., Schuster, P., & Presslich, O. (1986).
A solution for reliable and valid reduction of ocular artifacts, applied
to the P300 ERP. Psychophysiology, 23, 695-703.
Spruyt, A., De Houwer, J., Hermans, D., & Eelen, P. (2007). Af-
fective priming of nonaffective semantic categorization responses.
Experimental Psychology, 54, 44-53.
Spruyt, A., Hermans, D., Pandelaere, M., De Houwer, J., &
Eelen, P. (2004). On the replicability of the affective priming effect
in the pronunciation task. Experimental Psychology, 51, 109-115.
Vanderploeg, R. D., Brown, W. S., & Marsh, J. T. (1987). Judgments
of emotion in words and faces: ERP correlates. International Journal
of Psychophysiology, 5, 193-205.
Vuilleumier, P. (2005). How brains beware: Neural mechanisms of
emotional attention. Trends in Cognitive Sciences, 9, 585-594.
Williamson, S., Harpur, T. J., & Hare, R. D. (1991). Abnormal pro-
cessing of affective words by psychopaths. Psychophysiology, 28,
260-273.
