Appetite 54 (2010) 579–582

Contents lists available at ScienceDirect

Appetite
journal homepage: www.elsevier.com/locate/appet

Short communication

All I saw was the cake. Hunger effects on attentional capture by visual food cues
Richard M. Piech *, Michael T. Pastorino, David H. Zald
Department of Psychology, Vanderbilt University, PMB 407817, Nashville, TN 37240, USA

A R T I C L E I N F O A B S T R A C T

Article history: While effects of hunger on motivation and food reward value are well-established, far less is known
Received 6 August 2009 about the effects of hunger on cognitive processes. Here, we deployed the emotional blink of attention
Received in revised form 5 November 2009 paradigm to investigate the impact of visual food cues on attentional capture under conditions of hunger
Accepted 5 November 2009
and satiety. Participants were asked to detect targets which appeared in a rapid visual stream after
different types of task irrelevant distractors. We observed that food stimuli acquired increased power to
Keywords: capture attention and prevent target detection when participants were hungry. This occurred despite
Food
monetary incentives to perform well. Our findings suggest an attentional mechanism through which
Hunger
Motivation
hunger heightens perception of food cues. As an objective behavioral marker of the attentional
Cognition sensitivity to food cues, the emotional attentional blink paradigm may provide a useful technique for
Attentional blink studying individual differences, and state manipulations in the sensitivity to food cues.
Published by Elsevier Ltd.

Introduction strong incentive to ignore them. This EBA effect has been observed
for threatening, gory, and sexual distractor images (Most et al.,
Food consumption in humans is a highly complex behavior. It is 2005; Most, Smith, Cooter, Levy, & Zald, 2007), as well as for
influenced by a multitude of factors, each of which may act on natively neutral images conditioned to be associated with negative
multiple levels (Berthoud, 2007). Hunger provides a powerful events (Smith, Most, Newsome, & Zald, 2006). However, to date, it
motivational force, which may not only involve physiological and has not been tested whether EBA performance is modulated by the
affective properties, but may modulate aspects of cognition. While state of the participant.
evidence of motivational modulations of the cognitive operations We hypothesized that food cues – images of food – would
has grown in recent years, the cognitive effects specific to being become more powerful distractors when participants completing
hungry have received relatively little attention. Previous studies an EBA task were hungry, than when they were sated. Were that so,
showed that hunger can increase selective attention to food- it would show that hunger can bias perception to involuntarily
related words when enough time is provided (Mogg, Bradley, attend to food cues, even to the extent of disrupting otherwise
Hyare, & Lee, 1998), improve the memory advantage for food items motivated attention to other tasks. In a foraging environment, such
(Morris & Dolan, 2001), and restrict attentional shifting (Piech, a mechanism would seem both plausible and adaptive, as it would
Hampshire, Owen, & Parkinson, 2009). enhance acquisition of food in a physiological state of need. In a
In some cases, emotionally arousing stimuli can capture modern affluent society, it may lead to disruption of goal-directed
attention to such an extent as to eliminate awareness of stimuli behavior and drive attention to food regardless of other goals, such
that appear immediately after the arousing stimulus. In the current as maintaining a diet.
study, we tested if hunger selectively biases attention to food,
creating attentional capture. We utilized the emotional blink of Method
attention (EBA) paradigm (Most, Chun, Widders, & Zald, 2005),
which indexes the ability of stimuli to capture attention. In this Thirty undergraduate students (20 women) participated in the
task, a person attempts to detect a previously defined target in a study for credit in a psychology course. Participants were
stream of rapidly displayed images. Images depicting certain additionally compensated as described below. Participants were
salient objects can prevent target detection if presented shortly informed when signing up for the study that they should not
before the target itself. Crucially, such distractors can prevent participate if they had diabetes, hypoglycemia, or any condition
awareness of subsequent stimuli even if the distractors are which requires regular eating patterns. All participants had normal
completely irrelevant to the task at hand and when there is or corrected-to-normal vision. Each person gave informed written
consent prior to participating. The study was approved by the
Vanderbilt University Institutional Review Board.
* Corresponding author. All stimuli were color images, 9.5 cm wide and 7.5 cm tall, viewed
E-mail address: r.piech@vanderbilt.edu (R.M. Piech). from 50 cm distance. The participants’ task was to detect a rotated

0195-6663/$ – see front matter . Published by Elsevier Ltd.

doi:10.1016/j.appet.2009.11.003
580 R.M. Piech et al. / Appetite 54 (2010) 579–582

image among a set of landscape images. The landscape images were told that the target items would always be landscapes, and
consisted of 252 photographs of natural landscapes or pictures of that images of other objects may also appear, but should be
buildings. A subset (168) of these images were rotated by 908, ignored. Participants responded pressing right or left arrow keys to
clockwise or counterclockwise and served as the target images (they indicate target rotation (right: clockwise). To minimize the
remained 9.5 cm wide and 7.5 cm tall). Distractor images consisted number of correct responses due to chance alone, participants
of images of food, images of romantic scenes, and neutral images, first had to indicate whether they saw a target or not, and only
with 56 images coming from each of the distractor categories. trials in which the participant indicated seeing the target and
Neutral distractor stimuli were drawn from the International accurately identified its rotation were counted as correct. A quarter
Affective Picture System (IAPS) database (Lang, Bradley, & of the trials – the catch trials – did not actually have a target.
Cuthbert, 2001), they were selected to have neutral valence and Crucially, every trial included a distractor, belonging to one of
low arousal ratings, and consisted of everyday objects and people. the three categories described above: neutral, romantic, or food.
IAPS-images of food and of romantic scenes were supplemented The distractors preceded targets such that targets occurred either
with images from the internet. Food images depicted a variety of two (lag 2) or eight (lag 8) presentations after the target. In past
dishes and courses, including salads, main courses, and desserts. studies, it has been shown that EBA effects are strong at lag 2, but
Romantic scenes depicted clothed couples (a man and a woman) performance usually recovers by lag 8. The comparison of accurate
holding hands, laughing, dining, walking, or a combination of the target detection performance at lag 2 for the three distractor
above. The three distractor categories were matched on average categories during the hungry and sated sessions is the primary
image luminosity. contrast of interest.
Participants completed the EBA task (Fig. 1A) during two The task consisted of six blocks of 32 trials, totaling 192 trials,
sessions. For one of the sessions (‘hungry’), they were instructed with participants taking short breaks between blocks. Distractors
to refrain from eating, but continue drinking as usual, for 6 hours were positioned as the 4th, 6th, or 8th image within the stream,
prior to the experiment. For the other (‘sated’) session they were to followed by targets two or eight positions later. Distractor
eat as usual. Sated and hungry sessions were counterbalanced across categories and positions were counterbalanced. Participants
the participants. To verify that the procedures modulated hunger, completed 16 practice trials.
participants indicated their hunger level during both sessions, using To ensure that performance impairment associated with
a Likert scale of 0 (not hungry at all) to 7 (extremely hungry). attentional capture by the distractors was involuntary, partici-
The EBA task consisted of rapid presentations of images with pants received a strong incentive to detect targets correctly. They
embedded distractors and targets (see Fig. 1a). Each trial contained knew they would be compensated additionally if they did well,
17 image presentations of 100 ms. The participant’s task was to with 10 USD for an overall performance of 80%, 20 USD for 90% or
identify the target among them – a rotated landscape. Participants more in each session. Additionally, the best participant from each
group of 20 received 50 USD. Thus, ignoring the distractors and
detecting the targets was rewarded with up to 90 USD.
After the second experimental session, participants rated all
distractor images for pleasantness (valence) and arousal using a
labeled magnitude scale (Lishner, Cooter, & Zald, 2008). The images
were presented for 100ms each.

Results

Self-report hunger measure

Prior to commencing the EBA task, participants indicated their

hunger level on a 0 (not hungry at all) to 7 (extremely hungry)
Likert scale. Out of 30 participants, five did not indicate greater
hunger during the hungry session than during the sated session,
and were excluded from further analyses. Two additional
participants were excluded due to accuracy of more than 2
standard deviations below the mean for the respective condition.
This resulted in 23 (7 men) participants with usable data. These
subjects reported a mean hunger level of 5.4 (SD 1.4) during the
hungry and of 2.4 (1.2) during the sated session (t(22) = 10.9,
p < .0005), thus showing a strong manipulation effect.

Emotional blink of attention task

Fig. 1. Design and results of the emotional blink of attention task. Panel A:
Representation of a single task trial. Seventeen (only six shown) images are As the dependant variable, we calculated the percentage of
presented for 100 ms each. The distractor belongs to one of three image categories: correct trials in each condition. The experimental design resulted
neutral pictures, romantic scenes, or food pictures. The target is a rotated landscape
and appears two (or eight) presentations after the distractor (referred to as Lag 2
in a 4-factor mixed effects analysis of variance (ANOVA). It
and Lag 8). At the end of the trial, participants indicate if they saw the target, and included three repeated-measures factors: Lag (2 or 8), State (Sated
which way it was rotated. Panel B: Accuracy in the task for Lag 2, grouped for or Hungry), and Category (Neutral, Romantic, or Food), and one
neutral, romantic, and food distractors. Lower accuracy indicated greater between-subjects factor, Sequence (of sessions: Hungry first or
attentional blink. The asterisk indicates significant performance decrease after
Sated first). (Sequence was included as a between-subjects factor,
food distractors during the hungry condition (p = .016 (one-tailed)). No significant
decrease was observed for the other distractor categories (ps > .2). Panel C: as we observed practice effects across sessions, i.e. participants
Accuracy in the task for Lag 8 did not differ due to hunger level. Overall performance performed better during the second session, independent of the
at Lag 8 was better after romantic distractors than after either of the other effects of hunger. The accuracy was at 75.5% (SD: 7.3) during the
categories. Error bars indicate one standard error of the mean. first, and at 80.2% (SD: 6.7) during the second session. An ANOVA
 R.M. Piech et al. / Appetite 54 (2010) 579–582 581

revealed the main effect of session number as significant: We also explored participants’ scores on the dietary restraint
F(1,22) = 15.3, p = .001.) scale (Herman, Polivy, Pliner, Threlkeld, & Munic, 1978). The mean
The overall ANOVA revealed a main effect of Lag: accuracy was score for our sample was 12.9 (SD = 4.9), indicating on average
as expected higher at Lag 8 than at Lag 2 (F(1,21) = 130.1, p < .0005). medium restraint (Coelho, Polivy, Herman, & Pliner, 2008). The
This confirmed that our distractors were successful at creating an Pearson correlation between the restraint scores and the perfor-
attentional blink at Lag 2. Planned comparisons showed that this mance impairment in the hungry condition after food distractors at
was the case for all distractor categories (all ts > 4.5, all ps < .0005). Lag 2 was not statistically significant (r = .29, p = .182), suggest-
We therefore focused the analysis on performance at Lag 2, with ing that the relationship between dietary restraint and hunger
which our main hypotheses were concerned. induced changes in attentional capture is at best modest.

Lag 2 analysis Distractor valence and arousal

The ANOVA addressing performance only at Lag 2 consisted of
two repeated-measures factors: Category (Food, Neutral, Roman- Participants’ valence rating scores for neutral, romantic, and
tic) and State (Hungry, Sated), and one between-subjects factor, food distractors were: 0.1 (SD 10.7), 26.1 (14.3), 34.0 (22.6),
Sequence (Fig. 1). This ANOVA revealed main effects for Category respectively. An ANOVA showed a significant effect of Category
(F(2,21) = 11.4, p = .001) and State (F(1,21) = 8.3, p = .009; see (F(1,22) = 29.6, p < .0005), and planned comparisons indicated
Fig. 1B). The main effects reflected a lower performance after that both romantic and food distractors received higher valence
romantic distractors and an overall lower performance during the scores than neutral ones (ps < .0005), but were not different from
hungry session. Planned comparisons showed romantic distractors one another (p = .153).
produced worse performance than either neutral or food Arousal scores showed a similar pattern. The scores for neutral,
distractors (t(22) = 4.5, p < .0005; t(22) = 3.9, p = .001, respective- romantic, and food distractors were: 16.4 (SD 12.1), 32.7 (18.0),
ly). The ANOVA also revealed an interaction between Category and 40.8 (27.2), respectively, with a significant effect of Category
State (F(2,21) = 3.8, p = .030). To understand this interaction, we (F(1,22) = 13.0, p = .001). Both romantic and food distractors
compared the performance after each distractor category during received higher arousal scores than neutral ones (ps < .0005),
the Sated and Hungry sessions using paired-samples T-tests. These but were not different from one another (p = .186).
showed that only the food distractors led to significantly worse Half of the participants rated the stimuli while hungry, and half
performance during the hungry session (t(22) = 2.3, p = .016 (one- while sated. Only food stimuli were rated differently based on State:
tailed)). For the other distractor categories, no significant they were rated as more arousing and more positive by the hungry
impairment after fasting was observed (t(22) = .1, p = .874 group. Valence ratings showed an interaction effect with State
(neutral); t(22) = 1.3, p = .212 (romantic). The results confirmed (F(2,42) = 9.0, p = .001). Independent T-tests showed a higher food
the main hypothesis of the study: food pictures create an cue valence rating for the hungry group (t(21) = 4.2, p < .0005), but
attentional blink that is enhanced after fasting. not for neutral or romantic distractors (ps > .4). Similarly, arousal
The ANOVA at Lag 2 also showed interactions of the ratings showed an interaction with State (F(2,42) = 11.9, p = .001).
experimental Sequence with the State factor (F(1,21) = 31.2, Independent T-tests showed a higher food cue arousal rating for the
p < .0005) and with the State–Category interaction F(2,21) = 5.0, hungry group (t(21) = 3.8, p = .001), but not for neutral or romantic
p = .011. The first interaction reflects performance being overall distractors (ps > .7). In summary, neutral distractors were less
somewhat better during session two (consistent with a practice positive and less arousing than either of the affective distractor
effect), but the presence and extent of this improvement depends categories. There was no significant difference between the food and
upon the sequence of conditions, since half the subjects performed romantic distractors on either scale, although numerically food
the sated condition first, and the other half performed the hungry distractors were rated higher on both scales.
session first. The second interaction reflects that the degree of
improvement from session one to two was greatest for romantic Discussion
distractors, regardless of state, whereas the presence and extent of
improvement for food distractors from the first to second session Hunger influences food intake through mechanisms beyond the
was dependent upon the sequence of conditions. These interac- mere desire to eat. Here, we have demonstrated a cognitive effect
tions validate the inclusion of sequence as a factor in the statistical of hunger, namely the increased capture of attention by food cues.
model and were not further investigated. This process appears involuntary, in that it occurs even when
participants are rewarded for tasks that require them to ignore the
Lag 8 analysis foods. The modulation of attentional capture by hunger may prove
An ANOVA addressing performance at Lag 8 resembled the one particularly problematic for individuals who are dieting, as the
for Lag 2. It revealed a main effect of Category (F(2,21) = 34.7, hungrier they get, the more likely food stimuli are to capture their
p < .0005), but planned comparisons showed better performance attention and interrupt goal-directed behavior. This relationship
after romantic distractors, compared to both neutral and food resembles that of drug addiction: biasing cognitive processing
distractors (both ps < .0005), which were at an equivalent level. toward reinforcers (Garavan & Hester, 2007).
Thus, romantic distractors showed a different pattern of effects for Visual attention processes are frequently considered to be
Lag 2 and Lag 8. While they were the most distracting category at driven by two principal kinds of mechanisms, bottom-up (stimulus
Lag 2 (see analysis above), leading to impaired performance in the driven) and top-down (endogenous control) (Connor, Egeth, &
task, at Lag 8, performance in romantic distractor trials was Yantis, 2004). Bottom-up mechanisms of attentional capture are
elevated compared to the other categories. This was confirmed by thought to be automatic and due to salient properties of a given
an interaction of Lag and Category in the overall 4-factor ANOVA stimulus, for example bright colors or movement. Top-down
described above (F(2,21) = 38.1, p < .0005). control mechanisms are endogenous to the individual performing
No effects of State or interaction of Category with State were a task. In the attention literature, top-down effects are typically
observed in the ANOVA of Lag 8 (see Fig. 1C). Like for Lag 2, the Lag exercised deliberately and result from awareness and knowledge
8 ANOVA showed an interaction between experimental Sequence about the current task demands. The ability of certain high arousal
and State (F(1,21) = 5.7, p = .026), reflecting improved performance stimuli to capture attention even when they are irrelevant to the
during the second session. task has the hallmark characteristics of a bottom-up mechanism.
582 R.M. Piech et al. / Appetite 54 (2010) 579–582

Yet, the demonstrated influence of hunger on capture by food Dolan, 2001). Our data appear congruent with this conclusion.
stimuli also bears top-down characteristics. The effect of hunger Participants’ ratings of arousal for food images were increased in
cannot be viewed as purely stimulus driven given that the stimuli the hungry sessions, reflecting the increased attentional blink
remained unchanged for the two sessions. Rather, what alters the caused by food stimuli. Consistent with an arousal interpretation,
distractors’ potential to disrupt task performance is a parameter ratings of arousal for the nonfood stimuli in this study did not
which is crucially internal to the participants, namely their change, and EBA performance for such stimuli did not demonstrate
motivational state. As such, the effect can be argued to be a form of a statistically significant effect of the hunger manipulation.
top-down control. However, as is often the case in studies with emotional stimuli,
The described hunger effect departs from standard top-down the valence profile of our food conditions mirrored the one for
mechanisms in a critical respect. It is unlikely to reflect a conscious arousal, with participants also rating the food as more pleasant
intentional process of cognitive control, since there was no when hungry. Thus, while consistent with an arousal explanation
advantage gained from paying more attention to the food stimuli, of the attentional blink, the study does not specifically distinguish
and indeed directing attention to these irrelevant stimuli actually between arousal and valence.
impairs performance. Given the monetary incentive to perform well, Research on the importance of social, cognitive and environ-
participants are on the contrary likely to exert top-down effort to mental factors for human eating behavior have recently led some
counter attentional capture by food cues. Thus, to the extent that to favor these factors over the ‘traditional’ hunger when trying to
hunger provides a top-down bias, it is likely involuntary in nature, explain healthy eating patterns, as well as obesity or eating
which stands in sharp contrast to the deliberate and conscious disorders (Herman & Polivy, 2005). Our experiment shows that
effects that typically characterize top-down control of visual hunger can very well influence these non-homeostatic factors, and
attention. Existing reports of motivational effects on visual attention that the influence is likely to be exercised despite a person’s
typically show a performance gain when participants expect higher deliberate attempts to the contrary.
rewards from doing well on an attentional task (Engelmann,
Damaraju, Padmala, & Pessoa, 2009). However, it is important to References
distinguish between paradigms in which the motivationally
relevant stimuli are targets vs. distractors, as the role of the stimuli Anderson, A. (2005). Affective influences on the attentional dynamics supporting
awareness. Journal of Experimental Psychology-General, 134(2), 258–280.
in the task is critical in determining whether they will have positive Berthoud, H. (2007). Interactions between the ‘cognitive’ and ‘metabolic’ brain in the
or negative effects on task performance. control of food intake. Physiology & Behavior, 91(5), 486–498.
The present study was conducted as a proof of concept for the Brignell, C., Griffiths, T., Bradley, B., & Mogg, K. (2009). Attentional and approach biases
for pictorial food cues. Influence of external eating. Appetite, 52(2), 299–306.
potential utility of the EBA paradigm for research on eating
Castellanos, E., Charboneau, E., Dietrich, M., Park, S., Bradley, B., Mogg, K., et al. (2009).
behavior. We tested the effects in an undergraduate student Obese adults have visual attention bias for food cue images: Evidence for altered
population without specific weight requirements. Because of this reward system function. International Journal of Obesity, 1476–5497.
and the small sample size, it would be wrong to consider the size of Coelho, J., Polivy, J., Herman, C., & Pliner, P. (2008). Effects of food-cue exposure on
dieting-related goals: A limitation to counteractive-control theory. Appetite, 51(2),
the hunger effect as normative in nature. Future testing of clinical 347–349.
populations (and well-screened healthy controls) will be crucial Connor, C., Egeth, H., & Yantis, S. (2004). Visual attention: bottom-up versus top-down.
for determining the clinical relevance of our finding. Current Biology, 14(19), 850–852.
Engelmann, J., Damaraju, E., Padmala, S., & Pessoa, L. (2009). Combined effects of
The current paradigm may be of particular utility for testing attention and motivation on visual task performance: Transient and sustained
models of overeating that emphasize individual differences in motivational effects. Frontiers in Human Neuroscience 3.
sensitivity to external food cues (Schachter & Rodin, 1974). Our Garavan, H., & Hester, R. (2007). The role of cognitive control in cocaine dependence.
Neuropsychology Review, 17(3), 337–345.
study did not measure such sensitivity, and future investigations of Herman, C., & Polivy, J. (2005). Normative influences on food intake. Physiology &
this relationship will be valuable. We did, however, include a behavior, 86(5), 762–772.
measure of dietary restraint, which notably did not show a Herman, C., Polivy, J., Pliner, P., Threlkeld, J., & Munic, D. (1978). Distractibility in dieters
and nondieters: An alternative view of ‘‘externality’’. Journal of Personality and
significant relationship of hunger induced changes in attentional Social Psychology, 36(5), 536.
capture by food. But, sensitivity to external food cues may reflect a Lang, P., Bradley, M., & Cuthbert, B. (2001). International affective picture system (IAPS):
distinct process from what is measured by dietary restraint. The Instruction manual and affective ratings. The Center for Research in Psychophysiol-
ogy, University of Florida.
described mechanism of attentional capture has the potential to
Lishner, D., Cooter, A., & Zald, D. (2008). Addressing measurement limitations in
influence eating habits, reflecting the extent to which food stimuli affective rating scales: Development of an empirical valence scale. Cognition &
draw an individual’s attention even when the person attempts to Emotion, 22(1), 12.
ignore them. Differential levels of this tendency may predict the Mogg, K., Bradley, B. P., Hyare, H., & Lee, S. (1998). Selective attention to food-related
stimuli in hunger: Are attentional biases specific to emotional and psychopatho-
risk of obesity or the success at maintaining a diet, akin to the logical states, or are they also found in normal drive states? Behaviour Research and
concept of external food sensitivity (Passamonti et al., 2009). Therapy, 36(2), 227–237.
Interestingly, Brignell, Griffiths, Bradley, & Mogg (2009) recently Morris, J. S., & Dolan, R. J. (2001). Involvement of human amygdala and orbitofrontal
cortex in hunger-enhanced memory for food stimuli. Journal of Neuroscience,
reported an association between reaction times in a food-related 21(14), 5304.
dot probe paradigm and high external eating, such that the high Most, S., Chun, M., Widders, D., & Zald, D. (2005). Attentional rubbernecking: Cognitive
external eaters showed greater allocation of attention to food control and personality in emotion-induced blindness. Psychonomic Bulletin and
Review, 12(4), 654.
words. Obese individuals show a similar pattern, with greater Most, S., Smith, S., Cooter, A., Levy, B., & Zald, D. (2007). The naked truth: Positive,
orienting to food cues in general, and greater time looking at food arousing distractors impair rapid target perception. Cognition & Emotion, 21(5),
cues than normal weight individuals in situations in which they are 964–981.
Passamonti, L., Rowe, J., Schwarzbauer, C., Ewbank, M., von dem Hagen, E., & Calder, A.
not hungry (Castellanos et al., 2009).
(2009). Personality predicts the brain’s response to viewing appetizing foods: The
Attentional blink effects akin to the ones reported here have neural basis of a risk factor for overeating. Journal of Neuroscience, 29(1), 43.
been shown for a variety of motivationally significant stimuli, Piech, R., Hampshire, A., Owen, A., & Parkinson, J. (2009). Modulation of cognitive
flexibility by hunger and desire. Cognition & Emotion, 23(3), 528–540.
including gore, erotic scenes, and negatively conditioned stimuli
Schachter, S., & Rodin, J. (1974). Obese humans and rats. Washnigton, DC: Erlbaum/
(Most et al., 2005; Most et al., 2007; Smith et al., 2006). These Halstead.
stimulus classes share a high level of perceived arousal. Such Smith, S., Most, S., Newsome, L., & Zald, D. (2006). An emotion-induced attentional
observations lead to the conclusion that the arousal value of blink elicited by aversively conditioned stimuli. Emotion, 6(3), 523–527.
Vuilleumier, P., Armony, J., Driver, J., & Dolan, R. (2001). Effects of attention and
stimuli drives their potential to capture attention and create an emotion on face processing in the human brain an event-related fMRI study.
attentional blink (Anderson, 2005; Vuilleumier, Armony, Driver, & Neuron, 30(3), 829–841.