Subjective and physiological reactivity to chocolate
images in high and low chocolate cravers
Sonia Rodrı ´gueza, Marı ´a Carmen Ferna ´ndeza,
Antonio Cepeda-Benitob, Jaime Vilaa,*
aUniversity of Granada, Granada, Spain
bTexas A&M University, TX, USA
Received 23 April 2004; accepted 29 October 2004
Available online 9 March 2005
Cue-reactivitytochocolateimages was assessedusingself-report andphysiologicalmeasures. Fromapre-screening sample of454,young
women were selected and assigned to high and low chocolate craving groups (N = 36/group). The experimental procedure consisted in the
elicitation and measurement of the cardiac defense and startle reflexes while viewing chocolate and standard affective images selected from
the International Affective Picture System. In response to chocolate images, high cravers reported more pleasure and arousal but less control
than lowcravers. In highcravers, viewing chocolate images inhibited the cardiac defense but potentiated the startle reflex,as compared to low
cravers. The results confirmed at the physiological level that the motivational state that underlies the experience of chocolate craving include
both appetitive (inhibition of the defense reflex) and aversive (potentiation of the startle response) components. The findings supported a
motivational conflict theory of chocolate craving.
# 2005 Elsevier B.V. All rights reserved.
Keywords: Craving; Chocolate; Startle reflex; Eyeblink; Defense reflex; Heart rate; Emotion modulation; Motivational conflict
Women report that the food they crave most often is
chocolate (Hill et al., 1991; Rogers and Smit, 2000).
Chocolate consumers frequently describe their passion
towards chocolate as an ‘‘addiction’’ and may label
themselves as ‘‘chocoholics’’ (Rogers and Smit, 2000).
Several reviews of the literature concur in noting the
behaviors of chocolate addicts and drug addicts share many
features (e.g., Benton, 2001; Hetherington, 2001; Rogers
find chocolate irresistible and eat it in excess, report positive
mood and pleasure during chocolate consumption, but may
also report negative mood in response to chocolate cues
(Hetherington, 2001; Rogers and Smit, 2000).
Baker et al. (1987) conceived drug cravings as affects
organized within two distinct, mutually exclusive, motiva-
tional pathways consisting of a ‘‘positive affect’’ and a
‘‘negative affect’’ network. The model follows Lang’s
theory of motivation and attention, which suggests that
positive and negative affects are associated with mutually
opposed appetitive (consummatory) and aversive (defen-
sive) motivational systems, respectively (e.g., Lang, 1995;
Lang et al., 1997). The model predicts that the intensity of a
defensive reflex (e.g., eye-blink startle) depends on the
affective valence (positive mood versus negative mood) and
the level of activation (calm versus activated) of the
individual at the time of the stimulus-response event. An
independently evoked defensivereflex should be potentiated
or augmented when the individual is processing stimuli
associated with negative affect and inhibited or reduced
when the individual is processing stimuli associated with
positive affect. Level of activation is presumed to enhance
Biological Psychology 70 (2005) 9–18
* Corresponding author. Present address: Departamento de Personalidad,
Evaluacio ´n y Tratamiento Psicolo ´gico, Facultad de Psicologı ´a, Campus
Universitario de la Cartuja, 18011 Granada, Spain. Tel.: +34 958 243753;
fax: +34 958 243749.
E-mail address: firstname.lastname@example.org (J. Vila).
0301-0511/$ – see front matter # 2005 Elsevier B.V. All rights reserved.
both potentiation and inhibition of responding, whichever is
primed at the time. To the extent that cravings can be
conceptualized as affectively laden experiences (e.g., Baker
et al., 1987), the methodological paradigms derived to test
Lang’s model of emotion can be applied to investigate the
motivational mechanisms or psychophysiology underlying
chocolate and other food cravings (see also Drobes et al.,
2001; Mauler et al., 1997).
The primary methodology used to test Lang’s theory
has examined the potentiation/inhibition of the startle
reflex while emotionally charged stimuli are being
presented. Most often, the emotional stimuli consist of
pictures from the International Affective Picture System
(IAPS; Lang et al., 1995; Molto ´ et al., 1999; Vila et al.,
2001), which is a collection of images that can be rated
according to the valence (pleasure/displeasure), activation
(calm/aroused), and dominance (controlling/controlled)
In addition to the startle reflex, the cardiac defensive
reflex (CDR) is also modulated by IAPS images (Sa ´nchez,
2000; Sa ´nchez et al., 2002). Intense auditory stimulation
produces a complex pattern of heart rate changes consisting
of accelerative and decelerative components of short and
long latency that appear in sequential order (Turpin, 1986;
Pe ´rez et al., 1998; Vila, 1998). This response patterntends to
habituate rapidly after the first stimulus presentation. The
pattern is modulated when the auditory stimulation is
presented during the visualization of affective IAPS images:
unpleasant images increase the accelerative components
whereas positive and neutral images attenuate them
(Sa ´nchez, 2000).
In concordance with Lang’s theory of motivational
priming and Baker’s et al. affective theory of cravings, the
presence of inhibition or potentiation of defensive reflexes
while chocolate images are presented should provide an
unbiased indicator of the presence of appetitive or avoidant
mechanisms underlying chocolate craving. Thus, we
examined the effect of viewing chocolate images on two
defensive reflexes (CDR and eye-blink startle) in high and
low chocolate cravers. In addition, we assessed psycho-
physiological activation (skin conductance, SC) and self
report (affect, control, and arousal).
Given that individuals report both positive and negative
feelings following the consumption of chocolate (Benton,
2001), we did not venture to hypothesize whether exposure
to chocolate images would support appetitive or avoidant
motivational processes. However, since the theories of Lang
and Baker et al. predict that appetitive and aversive
mechanisms are mutually exclusive, we hypothesized that
presenting chocolate images in a neutral, controlled
environment would either inhibit or potentiate the defensive
reflexes, but not both. In congruence with self-reported
levels of craving, we predicted that high chocolate cravers
would bemore reactive,
subjectively, to chocolate images than low chocolate
The participants were 72 female students from the
Universidad de Granada, Spain. The participants were
selected from an initial pool of 454 students who reported
chocolate cravings using the Food Craving Questionnaire-
Trait (FCQ-T; Cepeda-Benito et al., 2003). Students who
scored, respectively, within the top and bottom tentiles of
Chocolate Craving groups. Of those who agreed to
participate, 15 students were excluded from the sample
due to artifacts in the physiological recordings. Most
participants (N = 36 per group) were in their 20s (M = 21.0;
S.D. = 2.61) and their body mass index (BMI) was within
normal range (High Cravers: M = 22.4; S.D. = 3.37; Low
Cravers:M = 21.91;S.D. = 3.23;
p > 0.52). Participants were healthy without visual or
auditory deficits. There were no between group differences
with regards to number of days since their last menstruation
and to length of time since their last food intake.
F[1, 70] = 0.40,
2.2. Psychophysiological measures
The VPM 11 software program (Cook, 1994) controlled
the sequence of stimuli presentations and the acquisition and
analysis of physiological data, both of which were
accomplished using the Advantech-PCL812PG A/D (12
bits) converter and input–output data card. Pictures were
presented using a Kodak Ektapro (model 9020) slide
projector. The projector presented 145 cm ? 95 cm images
3 m away from the participant. A Coulbourn Stimulator
(modules S81-02, S84-04, S82-24, and S22-18) was used to
produce the sounds that were presented to the subject
through earphones (Telephonics-TDH49P). Sound intensity
was calibrated with a sonometer (Bru ¨el and Kjaer-2235) and
anartificial ear (Bru ¨elandKjaer-4153).Theauditory stimuli
were a burst of 105 dB white noise with either a 500 or a
50 ms duration, instantaneous rise/fall time, that elicited the
cardiac defense reflex (CDR) or the startle reflex,
A Grass polygraph (Rps 7c 8b) was used to record the
heart rate (HR), which was derived from the EKG
(derivation II) using a 7P4 preamplifier. R–R intervals were
measured in milliseconds and transformed into average HR
every second using a weighted averaging method (Reyes del
Paso and Vila, 1998). The cardiac defense response was
obtained following the procedures outlined by Vila et al.
(1992). The 80 second-by-second HR values after the onset
of the defense evoking stimulus were expressed as
differential scores from a baseline and reduced to the
medians of 10 intervals.
The eyeblink component of the startle response was
measured by recording EMG activity from the orbicularis
oculi region beneath the left eye using small Sensormedic
S. Rodrı ´guez et al./Biological Psychology 70 (2005) 9–1810
electrodes filled with electrolyte paste. The raw EMG signal
was amplified and integrated using Coulbourn bioamplifiers
V75-04 and V76-23, respectively. The signal was filtered
using a frequency band of 90–1000 Hz and a time constant
of 75 ms. The startle reflex magnitude was defined as the
difference in microvolts between the peak of the integrated
response and the response onset occurring between 20 and
120 ms following the initiation of the startle evoking
Skin conductance (SC) was registered from two standard
Sensormedic electrodes filled with isotonic electrolyte paste
(0.29 g NaCl per 100 ml water) placed on the left
hypothenar eminence. The signal was recorded using the
Grass bioamplifier 7P1. The SC response to the slides was
defined as the average changes in microsecond every half
second, during 7 s after initiation of the slide, expressed as
deviations from a baseline period of 1 s preceding the slide.
2.3. Self-report measures
Food Craving Questionnaire-Trait and State([FCQ-T
T (39 items) and the FCQ-S (15 items) measure the intensity
of 9-trait and 5-state dimensions of food cravings,
respectively (Cepeda-Benito et al., 2003). The instructions
and the items of the instruments ask respondents about
cravings, urges or desires for one or more specific types of
food.Participantsare askedtothinkofthefoodorfoods they
usually crave (for the FCQ-T) or the specific foods they are
currently craving while completing the questionnaire (for
the FCQ-S).Thus, participants were asked to report cravings
how frequently each statement ‘would be true for you in
general’ using a 6-point scale that ranges from 1 (‘Never’ or
‘Not Applicable’) to 6 (‘Always’). For the FCQ-S,
participants were asked to indicate the extent to which they
agreed with each statement ‘right now, at this very moment’
using a 5-point Likert scale that ranged from 1 (‘Strongly
Agree’) to 5 (‘Strongly Disagree’). Full-scale and factor-
scale totals for both instruments can be calculated by simply
adding the corresponding item scores. In the present sample,
the overall alphas for the FCQ-T and the FCQ-S were 0.98
and 0.96, respectively. Evidence of the construct validity of
both instruments includes findings of predictive,convergent,
and discriminant validity in both Spanish and English
speaking samples (Cepeda-Benito et al., 2000a,b, 2003).
Self-Assessment Manikin (SAM; Lang, 1980; Hodes
et al., 1985). The SAM consists of human-like figures that
embody the dimensions of valence, arousal, and dominance,
with five figures representing five intensity levels within
each dimension. SAM’s valence rating ranges from a figure
with a large smile to a figure with a pronounced frown,
arousal rating ranges from a figure that appears agitated to a
rating ranges from a very large figure to a very small figure.
This subjective assessment method has been extensively
validated and is widely used in cue reactivity research (e.g.,
Bradley and Lang, 2000; Coffey et al., 2002).
Eating Attitude Test (EAT; Garner and Garfinkel, 1979).
The EATis a 40-item instrument that was designed to assess
a broad range of eating disorder symptoms and is most
useful as a clinical screening device. Respondents rate each
item using a 6-point Likert scale ranging from Never to
Always. The score can range from 0 to 120, with higher
scores indicating higher symptom levels. Cutoffs of 30 and
50 characterize at risk and clinical populations (Mintz and
O’Halloran, 2000). The EAT-40 has been validated and
extensively used with Spanish speaking populations (e.g.,
Castro et al., 1991; Castro et al., 2002; Penas-Lliedo et al.,
2002; Rivas et al., 2001). In the present sample, the overall
alpha for the EAT was 0.87.
developed for use in the present experiment to assess how
participants hadfelt whileviewingthechocolateimages.The
in control, joyful, important, interested, eager, satisfied, and
relaxed) and 8 negative states (depressed, ashamed, anxious,
were asked to indicate on a scale of 0 (not at all) to 100 (very
much so) the extent towhich each of the adjectives described
how they felt. Thetotal score foreach scalewas calculated as
theaverage of theirrespective eightitems. Theoverallalphas
forthepositive andnegative scalesofthe CRQ were0.84and
Data were collected within a single session that lasted
90 min. Individual sessions could start at one of six times
(9:30,11:00 a.m.and12:30,4:00,5:30or7:00 p.m.),withthe
times equally balanced across the high and low chocolate
craving groups. Participants were given a brief description of
thestudypriortogiving theirwritten informedconsent.They
and to obtain information about time elapsed since their last
chocolate and food intake, and the date of their last
menstruation. For all participants, the psychophysiological
test consisted of a 10-min baseline adjustment period,
followed by 3 CDR trials and 32 subsequent startle-response
trials. When possible, the pictures were selected according to
the Spanish calibration of the IAPS (Molto ´ et al., 1999; Vila
et al., 2001). We selected and matched positive and negative
pictures that elicit high levels of arousal. Conversely, we
arousal. Given that the IAPS only has three chocolate-related
pictures, the set of chocolate pictures was completed using
five additional pictures found on the internet. These pictures
products were the main content of the pictures. That is, the
Each CDR trial began with the collection of 15 s baseline
followed by a 6 s picture presentation of one of three
S. Rodrı ´guez et al./Biological Psychology 70 (2005) 9–18 11
pictures (chocolate, neutral or unpleasant), with the order of
presentation counterbalanced across participants within
each craving group. Due to rapid habituation, CDR
comparisons across picture categories precludes the use
of a within subjects design. Thus, given our estimated
sample size, we were limited to compare the CDR across a
maximum of three picture categories (N = 12 per picture
category per craving group). The defense auditory stimulus
was presented at 3.5 s from picture onset and data collection
continued for 80 s after the picture presentation ended. The
length of the intertrial intervals varied randomly between 1
and 4.5 s.
After the 3 cardiac defense trials, the 32 startle-trials
ensued as follows: (a) 3 s of baseline data collection; (b) 6 s
randomlyatapointbetween2.5and4.5 s;(c)6 spost-picture,
extended data collection; (d) a randomly varied 1–4.5 s
intertrial interval. There were 8 pictures from each of the 4
categories (pleasant, neutral, unpleasant, and chocolate) for a
were presented in the same pseudorandom order for all
participants. That is, the 32 pictures were presented in 8, 4-
picture blocks, with each block including 1 picture of the 4
different image categories. The order of presentation of the
categories within each block was randomized.
Following the physiological recording, the sensors were
each of the images. The images were presented again during
6 s with an inter-image interval of 12 s to complete their
ratings. Participants then completed the CRQ to further
assess their subjective emotional response to the chocolate
images. Finally participants completed the FCQ-S and the
EAT to, respectively, measure levels of chocolate craving
and eating disorder symtomatology across the two groups.
2.5. Data analysis
The subjective measures were analyzed using either
simple between group ANOVAs (FCQ-T, FCQ-S and EAT)
or mixed between group with repeated measures ANOVAs
(CRQ and SAM). In each case, we tested whether high and
low cravers differed on the subjective measure. All
physiological data were analyzed by means of mixed
between group (high and low craving groups)ANOVAs with
one (image type) or two (image type and time) repeated
Analysis of the cardiac defense responsewas restricted to
the first defense trial, given the rapid habituation of the
response. Analysis of the startle reflex and the skin
conductance response was performed on the average of
the eight-startle trials for each image category. All analyses
were performed using the SPSS 11.0 statistical package.
When repeated measures ANOVAs were used, the Green-
house-Geisser epsilon correction was applied. Analysis of
significant interaction effects were performed following
Keppel’s procedure (Keppel, 1991). First, we identified the
levels of the interacting factors explaining the significant
effect (simple effects analysis). Then, if more than two
groups or conditions were involved, multiple pairwise
comparisons were performed using t-tests. The critical
values for these t-tests were derived from Bonferroni to
protect against accumulation of alpha-error. The level of
significance was set at 0.05 for all analysis.
3.1. Self-assessment measures
3.1.1. Food craving questionnaires and eating attitude test
As intended, scores on the FCQ-T showed that chocolate
cravers (M = 150.72; S.D. = 21.21) reported considerably
greater chocolate cravings than noncravers (M = 53.69;
S.D. = 7.75; F[1, 70] = 664.36, p < 0.0001). Although
average eating disorder symptomatology was clearly below
the risk (30) and clinical (50) cut-off scores in both groups,
cravers significantly reported more symptomatology than
noncravers (M = 17.19; S.D. = 12.93 versus M = 10.19;
S.D. = 7.09; F[1, 70] = 8.11, p < 0.01).
There were also clear differences between groups in state
reactivity following the experimental manipulation. The
analysis of variance showed that high chocolate cravers
obtained significantly higher scores in the FCQ-S than low
chocolatecravers(M = 47.91;
M = 22.61; S.D. = 9.84; F[1, 70] = 98.57, p < 0.0001).
S.D. = 11.71versus
3.1.2. Chocolate reactivity questionnaire
We tested whether high and low cravers had different
profiles on the (positive and negative) mood variable set
using a 2 (?2) repeated measures ANOVA, the first between
group factor being group and the second repeated measure
factor being mood. A nonsignificant group by mood
interaction (F[1, 70] = 0.14, p = 0.90), revealed that the
mood profiles of the two groups were parallel. A significant
group effect (F[1, 70] = 19.89, p < 0.0001), indicated that
high chocolate cravers reportedboth more positiveand more
negative affect in response to chocolate images than low
chocolate cravers. A within subjects mood effect (F[1,
70] = 141.77, p < 0.0001), showed that chocolate images
evoked in both groups greater positive than negative affect.
3.1.3. Self-assessment manikin
For each of the SAM measures (valence, arousal, and
dominance), we examined profiles on image-category
variable sets. That is, we conducted three 2 (?4) repeated
measures ANOVAs with Group as the between factor and
Image (pleasant, unpleasant, neutral, and chocolate) as the
within factor. As regards valence, results revealed a
significant Image effect (F[3, 201] = 422.91, p < 0.0001)
and a significant Group by Image interaction effect (F[3,
201] = 12.23, p < 0.001). The Image effect confirmed the
expectation that valence would change as a function of
S. Rodrı ´guez et al./Biological Psychology 70 (2005) 9–18 12
image type. The interaction effect revealed that the valence
profiles across the image types were different between the
two groups. Simple effects analyses testing for valence
differences among groups for each of the image types only
revealed significant between group differences for the
chocolate-image valence (F[1, 67] = 23.70, p < 0.0001).
Valence rankings for pleasant, unpleasant and neutral
images followed the same ordering in both high and low
cravers (see Table 1). However, in high cravers chocolate
images rankedslightly higher than pleasant images, whereas
in low cravers chocolate images ranked considerably lower
than pleasant images.
The results found for arousal and dominance reactivity
in high and low chocolate cravers followed the same pattern
of results described for the variable valence (see Table 1).
That is, we found significant Image effects for arousal (F[3,
201] = 202.08,p < 0.0001) and for dominance (F[3,
201] = 76.63, p < 0.0001) and significant Group by Image
interactions for arousal (F[3, 201] = 13.59, p < 0.001) and
dominance (F[3, 201] = 2.67, p < 0.05). Moreover, arousal
(F[1, 67] = 30.85, p < 0.0001) and dominance (F[1,
67] = 11.11, p < 0.01) differences between the two groups
were only found at the chocolate-image level (simple effects
analysis). The only difference between arousal and
dominance was in the direction of the changes: while
arousal ratings for chocolate images—as in the case of
valence—were higher in the high craving group than in the
lowcravinggroup, dominanceratings were lowerinthe high
craving group than in the low craving group.
3.2. Psychophysiological measures
3.2.1. Cardiac defense response
A 2 ? 3 (?10) repeated measures ANOVA with two
between subjects factors, Group (high and low cravers) and
Image (unpleasant, neutral, and chocolate), and a within
subjects factor, Time (the 10 sampling intervals within the
80 s period) was applied. This analysis yielded statistically
significant main effects for Image (F[2, 66] = 3.70,
p < 0.05) and Time (F[9, 594] = 26.58, p < 0.0001), and
significant interaction effects for Group by Image (F[2,
66] = 4.40, p < 0.05) and Group by Image by Time (F[18,
594] = 1.90, p < 0.05).
These interactions effects were examined as follows.
only high cravers displayed CDR responses that were
significantly different between images (F[2, 33] = 5.61,
p < 0.01). Post hoc pairwise comparisons showed signifi-
cant differences in this group only between the unpleasant
and the chocolate images (t  = 3.29, p < 0.01). Simple
effect analyses of the Group by Image by Time interaction
revealed that the significant differences in the high craving
group were concentrated in the time intervals 3, 5, 7, and 8
(all F[2, 33] > 3.18, p < 0.05). As Fig. 1a shows, the
unpleasant image produced potentiation whereas the
chocolate image produced inhibition, with the CDR
corresponding to the neutral image falling in-between the
CDRs of the chocolate and negative images. No significant
simple effects were found in the low craving group (see
3.2.2. Startle reflex
First, we conducted a 2 (?3) repeated measures ANOVA,
with Group (high and low chocolate craving) as the between
subject factor and Image (unpleasant, pleasant, and neutral)
as the within subjects factor, to check whether we had
the results typically obtained by the Lang group (e.g., Lang
et al., 1993). Moreover, we also examined whether high and
low cravers differed on baseline startle reactivity across the
standard unpleasant, pleasant and neutral categories. The
ANOVA results showed a significant effect for Image (F[2,
140] = 41.46,p < 0.0001), but there was neither a
significant effect for Group (F[1, 70] = 1.74, p = 0.19)
nor for the Group by Image interaction (F[2, 140] = 1.18,
p = 0.31). That is, the profiles were not statistically different
in both groups, but they were nonflat with pleasant, neutral,
and unpleasant images, in that order, replicating the
expected ascending ranking of startle-response amplitudes
(cf., Lang et al., 1993; see Fig. 2).
Second, we conducted three simple ANOVAs to contrast
responses to the chocolate images with responses to each of
the three other image categories. In each of these ANOVAs,
the dependent variablewas the difference betweenchocolate
minus other image category. Only the chocolate minus
pleasant difference was statistically different between the
two groups (F[1, 70] = 6.79, p < 0.01). Startle responses to
the chocolate images, compared to pleasant ones, were
(see Fig.2). Responses tochocolate images were,in general,
inhibited with respect tobothneutraland unpleasantimages.
However, neither of these two analyses yielded significant
group differences (see Fig. 2).
S. Rodrı ´guez et al./Biological Psychology 70 (2005) 9–18 13
High LowHigh LowHigh LowHigh Low
Means and (standard deviations) of valence, arousal, and dominance to the images in high and low cravers.
3.2.3. Skin conductance
The analyses of autonomic arousal as indexed by the SC
changes followed a strategy similar to the one used for the
startle reflexanalyses.That is,we firstexaminedwhetherSC
changes during the 14 half seconds of the startle trials
differed across the neutral, pleasant, and unpleasant images
(e.g., Lang et al., 1997), and whether these changes differed
between high and low cravers.That is, we conducted first a 2
(?3 ?14) repeated measures ANOVAwith Group (high and
low chocolate craving) as the between subject factor, and
Image (unpleasant, pleasant, and neutral) and Time (14 half
seconds of the 7 s interval) as the within subjects factors.
Results showed significant Image (F[2, 140] = 27.91, p <
0.0001) and Time (F[13, 910] = 38.80, p < 0.0001) effects,
and significant Image by Time interaction effects (F[26,
1820] = 28.18, p < 0.0001). No significant main or inter-
action effects were found for Group. These results indicate
that the overall SC levels and the profiles of the two groups
S. Rodrı ´guez et al./Biological Psychology 70 (2005) 9–18 14
Fig.1. CDRpatternto the firstimage in the defense trials(unpleasant, chocolate,and neutral)forthe HighChocolateCraving group(a) and the LowChocolate
Craving group (b).
Fig. 2. Magnitude of the startle reflex to chocolate and pleasant, neutral, and unpleasant images for the High and Low Chocolate Craving groups.
were parallel. However, SC responses differed across image
categories, as indicated by the significant Image effect and
Image by Time interaction effect. That is, the profiles were
nonflat with neutral, pleasant, and unpleasant images, in that
order, replicating the expected ascending ranking of SC
To examine SC in relation to chocolate image exposure,
we repeated the analysis strategy described for the startle
response. That is, we conducted three 2 (?14) repeated
measures ANOVAs to contrast responses to the chocolate
images with responses to each of the three other image
categories. Dependent variable was again the difference
between the response to chocolate images and the response
to each of the other image categories. For the chocolate
minus neutral analysis, there were significant effects for
Group (F[1, 70] = 4.15, p < 0.05) and Group by Time
interaction (F[13, 910] = 3.77, p < 0.05). Fig. 3 shows how
the SC response of high cravers was, in general, higher to
chocolate than neutral images, with this effect being
reversed for low cravers. Simple effects analyses of the
Group by Time interaction showed significant differences
between the high and low cravers at each of the last 5 half-
second points (allF[1, 70] > 4.26,
ANOVAs for the chocolate minus pleasant SC difference
and the chocolate minus unpleasant SC difference did not
reveal any significant Group or Group by Time interaction
effect. However, these two latter ANOVAs revealed a
significant Time effect indicating that the difference
between chocolate minus pleasant (F[13, 910] = 27.54,
p < 0.0001) and chocolate minus unpleasant images (F[13,
910] > 33.60, p < 0.0001) increased as time passed.
p < 0.05). The
To our knowledge, this is the first investigation to
incorporate Lang’s model of emotion and the IAPS
paradigm to study cue-reactivity to chocolate cues. Thus,
S. Rodrı ´guez et al./Biological Psychology 70 (2005) 9–18 15
Fig. 3. Changes in skin conductance to chocolate and neutral images for the High Chocolate Craving group (a) and the Low Chocolate Craving group (b).
it is important to note that the experimental manipulation
succeeded in differentiating reactivity across chocolate and
other types of images, as well as between high and low
chocolate cravers. This finding was true for both self-
reported and physiological measures.
4.1. Self-report measures
group reported higher state cravings in response to the
chocolate cues than the participants selected to be in the low
craving group. Given that research indicates that individuals
who report higher trait cravings do not necessarily report
higher state cravings across situations (Cepeda-Benito et al.,
2003),this finding was important tovalidate the premisethat
chocolate cues would elicit higher cravings in the high
Reports on the self-assessment manikin further corro-
borated the effectiveness of the experimental manipulation.
High chocolate cravers rated the chocolate images as more
pleasant, more arousing, but with less sense of control, than
low chocolate cravers. The simultaneous increase in positive
valence and arousal in high cravers is congruent with the
hypothesis that cravings can be conceptualised as emotion-
ally charged experiences (Baker et al., 1987). However,
given that valence and dominance tend to be positively
correlated (Bradley, 2000), the decrease in dominance and
the simultaneous increase in positive valence in response to
chocolate images is a novel finding. This pattern of results
can be accounted for by a multidimensional conceptualiza-
tion of chocolate cravings (e.g., Cepeda-Benito et al.,
2000a,b). For example, reports of low dominance in the
presence of chocolate images should be expected if
individuals experience chocolate cravings as both the
anticipation of positive reinforcement from eating choco-
late, but also as an uncontrollable, guilt-provoking desire to
eat chocolate (see also, Benton, 2001; Hetherington, 2001;
Rogers and Smit, 2000).
The presence of seemingly incompatible affect-related
responses was also captured by the results obtained with
the chocolate reactivity measure (CRQ), which in turn
replicated the finding that chocolate can produce both
positive and negative affect (Rogers and Smit, 2000). That
is, high cravers reported both more positive and more
negative affect in response to chocolate images than low
chocolate cravers. Although, some authors have reported
that procedures that elicit smoking cravings may also
increase negative mood without decreasing positive mood
(e.g., Burton and Tiffany, 1997; Cepeda-Benito and Tiffany,
1996), drug-craving, cue-exposure studies typically find that
stimuli that increase positive mood tends to also decrease
negative mood (e.g., Powell, 1995), and vice versa
(Singleton et al., 2003). Thus, although chocolate cravings
review,Benton,2001),the present results are congruent with
the notion that chocolate ‘‘has a very particular set of
associations’’ (Hetherington, 2001, p. 318) that differenti-
ates chocolate from other addictive substances.
High chocolate cravers obtained higher scores in the
FCQ-S, following the experimental session, and in the EAT.
The association between food craving, as a trait, and some
eating disorders, such as obesity and bulimia has been
reported in several studies (Gendall et al., 1997; Guertin,
1999; Van der Ster Wallin et al., 1994). In general, our
subjective measures confirm previous findings concerning
the presence of both positive and negative affect in response
to food stimuli, as well as the presence of greater eating
disorder symptomatology in people who score high in food
craving questionnaires (Hetherington and Macdiarmid,
1993; Lafay et al., 2001; Rogers and Smit, 2000). However,
levels of symptomatology that were below the risk and
clinical levels for the diagnosis of eating disorders,
according to the EAT. Moreover, the craving differences
were larger than the symptomatology differences between
the two groups. While there were not clinically significant
differences, we found substantial differences in chocolate
craving report between the two groups.
4.2. Physiological measures
The physiological data also validated the effectiveness of
the experimental manipulation. The results replicated
Lang’s findings of blink startle and skin conductance
modulation when the participants viewed pleasant, neutral,
and unpleasant pictures. In line with Lang’s motivational
priming model, blink startle showed the expected valence
effect:potentiation duringunpleasantpicturesand inhibition
during pleasant ones, as compared to neutral pictures. Skin
conductance also showed the expected arousal effect:
greater skin conductance responses to the more arousing
pictures (pleasant and unpleasant) than to the less arousing
pictures (neutral). These effects were observed in both high
and low chocolate cravers, with no significant differences
between the two groups. Moreover, high chocolate cravers
also displayed greater skin conductance responses to the
chocolate pictures relative to the neutral pictures than low
cravers. This latter finding was congruent with the higher
arousal ratings reported by the high chocolate cravers.
Our examination of the startle responses to chocolate
images revealed that high chocolate cravers showed greater
blink startles to chocolate relative to pleasant pictures than
low cravers. These findings suggested a primed effect in the
high craving group due to activation of the aversive
motivational system. With regards to the CDR reflex, high
chocolate cravers showed simultaneously greater aversive
and appetitive motivations induced by the affective images.
That is, only high cravers displayed the expected CDR
potentiation and inhibition whenviewing the unpleasant and
chocolate pictures, respectively. Thus, whereas the startle
response suggested that chocolate images activated the
aversive motivational system, the CDR findings indicated
S. Rodrı ´guez et al./Biological Psychology 70 (2005) 9–1816
that reactivity to the chocolate pictures activated the
approach motivational system. This pattern of results runs
counter to Lang’s theory of motivation and Baker’s theory of
affective craving, both of which predict that exposure to
affect laden images should activate either appetitive or
avoidance states but not both simultaneously.
In general, our results are consistent with those reported
by Drobes et al. (2001) and Mauler et al. (1997). Like us,
these authors found that food deprived participants rated
food pictures as pleasant and attractive but displayed blink
potentiation instead of blink inhibition. This seemingly
conflicting pattern of results has two plausible explanations.
First, in our design the assessment of the CDR preceded the
measurement of the startle reflex. It is possible that initial
exposure to the chocolate pictures activated appetitive
motivation but subsequent exposures resulted in aversive
(frustration), cues that may be initially associated with
reward can lead to unpleasant affect and a reversal of the
pattern of reflex modulation (Drobes et al., 2001).
Alternatively, the physiological data may reflect the
activation of an ambivalent (Bradley, 2000; Cacioppo, 2001;
Elash et al., 1995; Miller, 1959; Tiffany, 1990) rather than a
(Baker etal., 1987).According toCacioppo’snewversionof
Miller’s approach-avoidance conflict (Cacioppo, 2001),
appetitive and aversive mechanisms are not mutually
exclusive. They may show reciprocity or coactivation.
When reciprocity is working, one mechanism is activated
and the other inhibited. When coactivation is working, both
mechanisms are activated simultaneously, provoking a
motivational conflict. Food craving, in general, and
chocolate craving, in particular, may represent an instance
of this type of conflict. In the present study, both self-
reported and autonomic physiological responses appeared to
capture this ambivalence in response to chocolate cues.
4.3. Relevance of the findings
The theoretical and practical importance of the results
should be evaluated taking into consideration that our
findings are specific to chocolate cues and may or may not
generalize to other foods. The implications of the results are
also moderated by the extent to which food cues might
promote binge eating and overeating (Jansen, 1998). Based
on models of extinction and habituation, exposure and
responsepreventionmethods havebeen effectivemethods to
treat avoidant and other anxiety-related behaviors (e.g.,
Tiffany, 2002) and bulimia nervosa (Toro et al., 2003).
However, although exposure/response–prevention methods
have been effective in reducing both drug-craving and food-
craving reactivity, the data have not consistently shown that
these techniques can reduce drug use (Conklin and Tiffany,
2002) or binge eating (Jansen, 2001), respectively.
more consistently is that the therapeutic models that have
been used may have failed to take into consideration the
multidimensionality of craving phenomena. That is, in the
present study, we found that chocolate cues can simulta-
neously activate both appetitive and aversive motivational
states. Perhaps, for cue-exposure response–prevention
treatments to be effective the procedures need to provoke
and extinguish both the positive and negative affective
treatments may depend on the extent to which they
incorporate the theoretical and empirical findings reported
in the scientific literature (Conklin and Tiffany, 2002).
The present research was supported by two joint grants
from the Spanish Ministry of Science and Technology and
the European Union (projects BSO 2001-3015 and
BSO2001-3211), and the Junta de Andalucı ´a (research
Baker, T.B., Morse, E., Sherman, J.E., 1987. The motivation to use drugs:
psychobiological analysis of urges. In: Rivers, C. (Ed.), The Nebraska
Symposium on Motivation: Alcohol Use and Abuse. University of
Nebraska Press, Lincoln, pp. 257–323.
Benton, D., 2001. Psychological and pharmacological explanations of
chocolate craving. In: Hetherington, M. (Ed.), Food Cravings and
Addiction. Leatherhead, Surrey, UK, pp. 265–293.
Bradley, M.M., 2000. Emotion and motivation. In: Cacioppo, J.T., Tas-
sinary, L.G., Berntson, G.G. (Eds.), Handbook of Psychophysiology.
University Press, Cambridge, pp. 602–642.
Bradley, M.M., Lang, P.J., 2000. Measuring emotion, behavior, feeling, and
physiology. In: Lane, R.D., Nadel, L. (Eds.), Cognitive Neuroscience of
Emotion. Oxford University Press, New York, pp. 242–276.
Burton, S.M., Tiffany, S.T., 1997. The effect of alcohol consumption on
craving to smoke. Addiction 92 (1), 15–26.
Cacioppo, J.T., 2001. Affective space. In: Abstracts of the II International
Workshop on Emotion and the Brain, Universidad de las Islas Baleares.
Castro, J., Toro, J., Salamero, M., Guimera, E., 1991. The eating attitude
test: validation of the spanish version. Evaluacio ´n Psicolo ´gica 7 (2),
Castro, J., Toro, J., Lazaro, L., Pons, F., Halperin, I., 2002. Bone mineral
density in male adolescents with anorexia nervosa. Journal of the
American Academy of Child and Adolescent Psychiatry 41, 613–618.
Cepeda-Benito, A., Ferna ´ndez, M.C., Moreno, S., 2003. Relationship of
gender and eating disorder symptoms to reported cravings for food:
construct validation of state and trait craving questionnaires in Spanish.
Appetite 40, 47–54.
Cepeda-Benito, A., Gleaves, D.H., Fernandez, M.C., Vila, J., Reynoso, J.,
2000a. The development and validation of the state and trait food
cravings questionnaires. Behaviour Research and Therapy 38, 1125–
Cepeda-Benito, A., Gleaves, D.H., Williams, T.L., Erath, S.T., 2000b. The
development and validation of the state and trait food cravings ques-
tionnaires. Behaviour Therapy 31, 151–173.
S. Rodrı ´guez et al./Biological Psychology 70 (2005) 9–18 17
Cepeda-Benito, A., Tiffany, S.T., 1996. The use of a dual-task procedurefor Download full-text
the assessment of cognitive effort associated with cigarette craving.
Psychopharmacology 127 (2), 155–163.
Coffey, S.F., Saladin, M.E., Drobes, D.J., Brady, K.T., Dansky, B.S.,
Kilpatrick, D.G., 2002. Trauma and substance cue reactivity in indivi-
duals with comorbid posttraumatic stress disorder and cocaine or
alcohol dependence. Drug and Alcohol Dependence 65, 115–127.
Conklin, C.A., Tiffany, S.T., 2002. Applying extinction research and theory
to cue-exposure addiction treatments. Addiction 97, 155–167.
Cook III, E.W., 1994. VPM Reference Manual. Author, Birmingham,
Drobes, D.J., Miller, E.J., Hillman, C.H., Bradley, M.M., Cuthbert, B.N.,
Lang,P.J., 2001. Food deprivation and emotionalreactions to food cues:
implications for eating disorders. Biological Psychology 57, 153–177.
Elash, C.A., Tiffany, S.T., Vrana, S.R., 1995. Manipulation of smoking
urges and affect through a brief-imagery procedure: self-report, psy-
chophysiological, and startle probe responses. Experimental and Clin-
ical Psychopharmacology 3 (2), 156–162.
Garner, D.M., Garfinkel, P.E., 1979. The eating attitude test: an index of the
symptoms ofanorexia nervosa.PsychologicalMedicine9 (22), 273–279.
Gendall, K.A., Sullivan, P.F., Joyce, P.R., Bulik, C.M., 1997. Food cravings
in women with a history of anorexia nervosa. International Journal of
Eating Disorders 22, 403–409.
Guertin, T.L., 1999. Eating behaviour of bulimics, self-identified binge
eaters and non-eating disordered individuals: what diferentiates these
population? Clinical Psychology Reviewdkjdot 19 (1), 1–25.
Hetherington, M.M., Macdiarmid, J.I., 1993. ‘‘Chocolate addiction’’: a
preliminary study of its description and its relationship to problem
eating. Appetite 21, 233–246.
Hetherington, M.M., 2001. Food Cravings and Addiction. Leatherhead
Hill, A.J., Weaver, C.F., Blundell, J.E., 1991. Food craving, dietary restraint
and mood. Appetite 17 (3), 187–197.
Hodes, R.L., Cook, E., Lang, P., 1985. Individual differences in autonomic
response: conditioned association or conditioned fear? Psychophysiol-
ogy 22 (5), 545–560.
Jansen, A., 1998. A learning model of binge eating: cue reactivity and cue
exposure. Behaviour, Research and Therapy 36, 257–272.
Jansen, A., 2001. Craving and binge eating. In: Hetherington, M.M.
(Ed.), Food Cravings and Addiction. Leatherhead Publishing, UK,
Keppel, G., 1991. Designand Analysis: A Researcher’s Handbook. Prentice
Lafay, L., Thomas, F., Mennen, L., Charles, M.A., Eschwege, E., Borys,
J.M., Basdevant, A., 2001. Gender differences in the relation between
food cravings and mood in a adult community: Results from the
fleurbaix laventie ville sante ´ study. The International Journal of Eating
Disorders 29 (2), 195–204.
Lang, P.J., 1980. Behavioral treatment and bio-behavioral assessment:
computer applications. In: Sidowski, J.B., Johnson, J.H., Williams,
T.A. (Eds.), Technology in Mental Health Care Delivery Systems.
Ablex, Norwood, NJ, pp. 119–137.
Lang, P.J., 1995. The emotion probe: studies of motivation and attention.
American Psychology 50, 372–385.
Lang, P.J., Bradley, M., Cuthbert, B.N., 1995. The International Affective
Picture System (IAPS): Technical Manual and Affective Ratings. The
Center for Research in Psychophysiology, University of Florida, Gains-
Lang, P.J., Bradley, M.M., Cuthbert, B.N., 1997. Motivated attention:
affect, activation and action. In: Lang, P.J., Simons, R.F., Balaban,
M. (Eds.), Attention and Orienting: Sensory and Motivational Pro-
cesses. Erlbaum, Hillsdale, NJ.
Lang, P.J., Greenwald, M.K., Bradley, M.M., Hamm, A.O., 1993. Looking
at pictures: affective, facial, visceral, and behavioral reactions. Psycho-
physiology 30, 261–273.
Marks, I.M., 1987. Fears, Phobias and Rituals: Panic, Anxiety and their
Disorders. University Press, Oxford.
Mauler, B., Tuschen, B., Hamm, A., 1997. Fear or craving: startle reflex
modulation during viewing of food. Psychophysiology 34, S62.
Miller, N.E., 1959. Liberation of basic S–R concepts: Extensions to conflict
behavior, motivational and social learning. In: Koch, S. (Ed.), Psy-
chology: A Study of a Science, Study 1. McGraw-Hill, New York, pp.
Mintz, L.B., O’Halloran, M.S., 2000. The eating attitudes test: validation
Molto ´, J., Montan ˜e ´s, S., Poy, R., Segarra, P., Pastor, M.C., Tormo, M.P.,
Ramı ´rez, I., Herna ´ndez, M.A., Sa ´nchez, M., Ferna ´ndez, M.C., Vila, J.,
1999. Un nuevo me ´todo para el estudio experimental de las emociones:
ElInternacionalAffectivePictureSystem(IAPS).Adaptacio ´nespan ˜ola.
Revista de Psicologı ´a General y Aplicada 52, 55–87.
Penas-Lliedo, E., Vaz-Leal, F.J., Wallen, G., 2002. Excessive exercise in
anorexia nervosa and bulimia nervosa:Relation to eating characteristics
and general psychopathology. International Journal of Eating Disorders
31 (4), 370–375.
Pe ´rez, N., Vila, J., Ferna ´ndez, M.C., Garcı ´a, A., Turpin, G., 1998. Difer-
encias individuales asociadas a la respuesta cardiaca de defensa: vari-
ables psicofisiolo ´gicas y de personalidad. Psicothema 10, 609–621.
Powell, J., 1995. Conditioned responses to drug-related stimuli: is context
crucial? Addiction 90 (8), 1089–1096.
Reyes del Paso, G.A., Vila, J., 1998. The continuing problem of incorrect
heartrateestimationin psychophysiologicalstudies:anoff-line solution
for cardiotachometer users. Biological Psychology 48, 269–279.
Rivas, T., Bersabe, R., Castro, S., 2001. Psychometric properties of the
questionnaire for eating disorder diagnosis (Q-EDD). Psicologı ´a Con-
ductual 9 (2), 255–266.
review of the evidence from a biopsychosocial perspective. Pharmacol-
ogy Biochemistry and Behaviour 66 (1), 3–14.
Sa ´nchez, M.B., 2000. Modulacio ´n emocional de los reflejos de sobresalto y
defensa. Tesis Doctoral. Universidad de Granada, unpublished thesis.
Sa ´nchez,M.B.,Ruiz-Padial,E.,Pe ´rez, N.,Ferna ´ndez,M.C.,Cobos,P.,Vila,
J., 2002. Modulacio ´n emocional de los reflejos defensivos mediante
visualizacio ´n de ima ´genes afectivas. Psicothema 14 (4), 702–707.
Singleton, E.G., Anderson, L.M., Heishman, S.J., 2003. Reliability and
validity of the Tobacco Craving Questionnaire and validation of a
craving-induction procedure using multiple measures of craving and
mood. Addiction 98 (11), 1537–1546.
Tiffany, S.T., 1990. A cognitive model of drug urges an drug use behaviour:
role of automatic and nonautomatic processes. Psychological Review
Toro, J., Cervera, M., Feliu, M.H., Garriga, N., Jou, M., Martı ´nez, E., Toro,
E., 2003. Cue exposure in the treatment of resistant bulimia nervosa.
International Journal of Eating Disorders 34, 227–234.
problem of differentiating orienting and defense reflexes. Psychophy-
siology 23, 1–14.
Van der Ster Wallin, G., Norring, C., Holmgren, S., 1994. Binge eating
versus nonpurged eating in bulimics: is there a carbohydrate craving
after all? Acta Psychiatrica Scandinavica 89, 376–381.
Vila, J., 1998. Psicologı ´a y salud: entre el corazo ´n y el cerebro. Anuario de
Psicologı ´a 29 (2), 109–138.
Vila, J., Ferna ´ndez, M.C., Godoy, J., 1992. The cardiac defense response in
humans: Effects of stimulus modalityand gender differences. Journal of
Psychophysiology 6, 140–154.
Vila, J., Sa ´nchez, M., Ramı ´rez, I., Ferna ´ndez, M.C., Cobos, P., Rodrı ´guez,
S., Mun ˜oz, M.A., Tormo, M.P., Herrero, M., Segarra, P., Pastor, M.C.,
Montan ˜e ´s, S., Poy, R., Molto ´, J., 2001. El Sistema Internacional de
Ima ´genes Afectivas (IAPS): Adaptacio ´n espan ˜ola. Segunda parte.
Revista de Psicologı ´a General y Aplicada 54 (4), 635–657.
S. Rodrı ´guez et al./Biological Psychology 70 (2005) 9–18 18