children’s reward responses to picture- and odor-cued food stimuli. a developmental analysis...
TRANSCRIPT
Appetite 67 (2013) 88–98
Contents lists available at SciVerse ScienceDirect
Appetite
journal homepage: www.elsevier .com/locate /appet
Research report
Children’s reward responses to picture- and odor-cued food stimuli.A developmental analysis between 6 and 11 years q
0195-6663/$ - see front matter � 2013 Elsevier Ltd. All rights reserved.http://dx.doi.org/10.1016/j.appet.2013.04.003
q Acknowledgments: We thank the children and their parents, and the schoolteachers of primary schools (Saint Joseph, Trinité, Maladière, and Côteaux Suzon) ofDijon city, for their enthusiastic participation in the study. We are grateful to MarieGaillet for her assistance in data collection, and to the Societies Mane et Fils (Bar surLoup, France) and Asquali (Montaigut-Lauragais, France) for graciously providingthe food odorants. This study was supported by the Centre National de la RechercheScientifique (Paris) and by the Conseil Régional de Bourgogne (Dijon).⇑ Corresponding authors.
E-mail addresses: [email protected] (T. Jiang), [email protected] (R. Soussignan).
Tao Jiang a,⇑, Benoist Schaal b, Véronique Boulanger b, Fayez Kontar c, Robert Soussignan b,⇑a Centre de Recherche en Neurosciences de Lyon, UMR 5292 CNRS-INSERM, Université de Lyon 1, Lyon, Franceb Centre des Sciences du Goût et de l’Alimentation, UMR 6265 CNRS, Université de Bourgogne-INRA, Dijon, Francec University of Damascus, Faculty of Education II, Sweida, Syria
a r t i c l e i n f o
Article history:Received 19 October 2012Received in revised form 5 March 2013Accepted 2 April 2013Available online 10 April 2013
Keywords:DevelopmentEating behaviorReward processesLikingWantingOlfactionVisionFood preferenceHungerChildrenGender differences
a b s t r a c t
The reward system is largely involved in the control of food intake. Whether components of this system(i.e., wanting and liking) change during development remains understudied, as well as how proximatefactors (sensory cues, motivational state) modulate reward reactivity across development. We examinedthe developmental pattern of wanting and liking for sensorily-cued food stimuli in 6–11 year old childrenas a function of the child’s motivational state (hunger/satiety), gender, and the nature of foods. Schoolchildren were exposed before or after their lunch on alternative days to visual and odor stimuli represent-ing different categories of familiar foods. Their task was to rate wanting and liking of pictures and odor-ants of pizza, meat, vegetables, fruits, and chocolate. The following results were found: (1) While likingappeared to be stable from age 6 to 11, more particularly for visually-cued foods, wanting decreased, aswell as did subjective hunger perception; (2) there were smaller or absent state-effects in 7-to-9-year-olds; (3) reward ratings were higher in boys than in girls; (4) reward ratings of vegetables were the low-est at all ages. These results suggest that wanting, but not liking, is developmentally variable over child-hood, and that this variation depends on age, gender, motivational state (hunger/satiety), the nature ofthe food and the modality of the sensory cue representing it. Such developmental changes are discussedin relation to biological (adiposity rebound) and cognitive (dietary restraint) factors influencing the moti-vation to eat during middle (6–7 years) and late (9–11 years) childhood.
� 2013 Elsevier Ltd. All rights reserved.
Introduction
In Western and westernizing societies, the availability of foodsthat are consistently improving in palatability exposes people to anincreasing abundance of alluring sensory cues that play an impor-tant role in the psychobiological cycle of food-related arousal, in-take, satiation, and then craving for it again (Birch, 1999; Neary& Batterham, 2010; Smeets, Erkner, & De Graaf, 2010; Sorensen,Moller, Flint, Martens, & Raben, 2003). In this context, it is widelyaccepted that human eating behavior is not only driven by nutri-tional and caloric necessities and by health- and wellbeing-relatedcognitions, but it is also a matter of sensory cues encoded as plea-
surable and rewarding (Lowe & Butryn, 2007; Mela, 2006; Yeo-mans, Blundell, & Leshem, 2004). Reactivity to such reward cuesfrom foods is indeed a good predictor of intake (Bobroff & Kissileff,1986; Rolls, Van Duijvenvoorde, & Rolls, 1984; Zandstra, De Graaf,Mela, & Van Staveren, 2000). Sensitivity to reward cues is also arecognized risk factor for overeating and for the engagement intodevelopmental trajectories of overweight and obesity (Birch,1998; Blass, 2008; Finlayson, King, & Blundell, 2007; Jansenet al., 2003; Mela, 2006; Soussignan, Schaal, Boulanger, Gaillet, &Jiang, 2012).
Thus, a better understanding on whether and how reactivity tofood-related reward cues changes throughout childhood mightshed additional light on the formative processes of food-relatedcognitions and emotions, and of eating habits. Although reactivityto sensory cues from food has been investigated in particular age-groups of participants, including either infants, children or adults(e.g., Birch & Fisher, 1998; Davis et al., 2009; Mennella & Beau-champ, 2002; Soussignan, Schaal, & Marlier, 1999; Soussignan,Schaal, Rigaud, Royet, & Jiang, 2011), ontogenetic changes of re-ward processing have been rarely examined in children followinga developmental perspective. Such a perspective is needed to trackwhether infants or children are more reactive to some food cues
T. Jiang et al. / Appetite 67 (2013) 88–98 89
during some periods of development and whether changes of sen-sitivity to reward cues at given ages may predict future food intakeor attitudes toward eating.
Forerunning studies focused on age-related changes in foodlikes or dislikes relied prevalently on questionnaires completedby parents or children (e.g., Cooke & Wardle, 2005; Nicklaus, Bog-gio, Chabanet, & Issanchou, 2004; Perez-Rodrigo, Ribas, Serra-Maj-em, & Aranceta, 2003; Skinner, Carruth, Bounds, & Ziegler, 2002;Ton Nu, MacLeod, & Barthelemy, 1996), but more rarely on thedevelopmental course of objective food-cue reactivity (e.g., Laing& Clark, 1983) or actual ingestive behavior of children (e.g., Rollins,Loken, & Birch, 2010). Collectively, these studies brought evidencefor both stable and variable likes and dislikes for foods duringchildhood. Some findings reported relatively stable food prefer-ences between 2 and 8 years of age (Nicklaus et al., 2004; Perez-Rodrigo et al., 2003; Rollins et al., 2010; Skinner et al., 2002), withchanges, if any, occurring during puberty, especially in terms ofincreasing preference for vegetables (Nicklaus et al., 2004; TonNu et al., 1996). Other findings indicated a decrease in the numberof liked foods and a stability of disliked foods from 4 to 16 years(Cooke & Wardle, 2005), or a tendency for initially disliked foodsto become more liked from age 5 to 11 (Rollins et al., 2010), or evenan increase of disliked foods from the age of 2 to 8 (Skinner et al.,2002). Thus, extant studies of age-related changes in food prefer-ences do not provide a wholly coherent picture, probably becauseof considerable between-studies differences in terms of age andgender composition of groups, cultural background, design(cross-sectional vs. longitudinal), target respondent (child vs.mother), number, nature, and presentation of test food items, re-sponse format (categorical, ordinal, continuous), and possiblyuncontrolled factors influencing food pleasantness [i.e., motiva-tional state, body-mass index (BMI), earlier experience (breast-feeding)]. These studies being mainly descriptive in nature andbased on questionnaires, they dealt only tangentially with proxi-mate factors influencing age-related changes in food preferencesalong childhood, such as sensory cues, motivational state and nat-ure of reward processes. The present study aimed to further con-tribute to the issue of proximate mechanisms underlyingchildhood variations in food preferences by examining stabilityand change in reactivity to reward elicited by selected food stimuli.
Evidence from neurobiological (Berridge, 2009; Davis et al.,2009), neuroimaging (Born et al., 2011) and behavioral studies (Ep-stein, Truesdale, Wojcik, Paluch, & Raynor, 2003; Finlayson et al.,2007) supports that food reward can be decomposed in at least amotivational component named ‘‘wanting’’ (viz., motivation toeat a food, desire) and an affective component, named ‘‘liking’’(including both the sensory pleasure of food intake and its remem-bered pleasantness) (Berridge & Robinson, 2003). To our knowl-edge, only one recent study (Olsen, Ritz, Hartvig, & Møller, 2011)has simultaneously investigated wanting and liking responses tosensory attributes of foods in comparing 10-year-old childrenand adults. Wanting and liking were assessed on a visual-analoguescale for a target food (sweet pear-banana yoghurt), and for realtest foods conveying sweet-fruity, salty-starchy, sour-fruity, bit-ter-citrus, fatty, and spicy-fatty flavors evaluated (after an actualtaste of them) before/after the target food was eaten to satiation.While some aspects of reward responsiveness were similar acrossage groups (i.e., post-meal drop in wanting for target food largerthan in liking), other aspects differentiated them. Adults decreasedtheir liking and desire for the test foods (sweet, fruity) that resem-bled the target food, while children did not. In contrast, childrenevinced post-meal increased liking for spicy items and wantingfor salty items, while adults did not. Thus, Olsen et al. revealedchild–adult differences in wanting and liking responses to food-stuffs bearing distinct appearances and flavors from a satiatingfood. This finding of age-related shift in the motivational underpin-
nings of eating behavior stimulates a developmental analysis of re-ward responsiveness during childhood. Being focused on a child–adult comparison, Olsen et al.’s study was not developmental interms of characterizing phenomenological and mechanistic pro-cesses underlying food appreciation and appetence along child-hood. Human childhood, viz. the period between weaning andsexual maturity, is indeed highly heterogeneous regarding, amongother biological and psychobiological processes, growth of differ-ent body tissues, energy allocation, patterns of fat deposition,and metabolic efficiency (e.g., Bogin, 1997; Konner, 2010; Rol-land-Cachera et al., 1987). It is thus expectable that sensory, per-ceptual and motivational processes controlling ingestive behaviorare, at least in part, patterned by these changing psychobiologicalconstraints.
A critical factor that has often been overlooked in previous stud-ies on age-related variations in food preferences (except in Olsenet al., 2011) is the influence of the participant’s actual motivationalstate on self-reported reward assessment. The rating of food pleas-antness varies indeed as a function of hunger/satiety states (viz.,alimentary alliesthesia: Cabanac, 1971). Alimentary alliesthesiahas been observed for odor, taste and sight of foods in adults (Cab-anac & Duclaux, 1973; Stoeckel, Cox, Cook, & Weller, 2007), butalso in infants (Soussignan et al., 1999). In adults, the magnitudeof this variation depends on both the type of food and the sensorymode of its perception (Jiang et al., 2008).
Motivational responses to food are also quite deeply influencedby gender. Gender differences in wanting/liking responses to eitherreal foods or food stimuli have been confirmed in adults (Cornier,Salzberg, Endly, Bessesen, & Tregellas, 2010; Killgore & Yurgelun-Todd, 2010; Nakamura, Shimai, Kikuchi, & Tanaka, 2001) and ado-lescents (Bere, Brug, & Klepp, 2008; Brug, Tak, te Velde, Bere, & deBourdeaudhuij, 2008; Le Bigot Macaux, 2001). But less conclusiveresults derive from research in infants and prepubertal children.Among the studies on food liking in children, some have not re-garded gender as a variable (Olsen et al., 2011), while others foundeither no difference (Perez-Rodrigo et al., 2003) or early variabilityconsistent with that of older participants (Cooke & Wardle, 2005;Nicklaus et al., 2004). However, the onset period of the gender-based differences in both liking and wanting for food remains tobe clarified. As the growth-related metabolic constraints outlinedabove are gender differentiated (e.g., Bogin, 1997), it is expectablethat gender discriminates motivational processes in ingestion. Thepresent study is thus also designed to assess whether and whenduring middle and late childhood gender differences emerge inthe wanting and liking of food-related stimuli.
Finally, available studies on aspects of children’s food rewardassessment (mostly liking) used either large nominal listings offood items in the form of questionnaires, often completed by par-ents (Cooke & Wardle, 2005; Nicklaus et al., 2004; Perez-Rodrigoet al., 2003; Skinner et al., 2002; Ton Nu, MacLeod, & Barthelemy,1996) or a restricted number of real foods (Olsen et al., 2011; Rol-lins et al., 2010). Number of studies have instead opted to presentparticipants with sensory cues – pictures, odors, flavors – derivedfrom foodstuffs as proxies to real foodstuffs, and were able to reli-ably predict food preferences and intake (e.g., Beauchamp & Men-nella, 2009; Born et al., 2011; Cornier, Von Kaenel, Bessesen, &Tregellas, 2007; Jansen, 1998; Jansen et al., 2003; Soussignanet al., 2012; Staiger, Daweb, & McCarthy, 2000). This approach re-lies on the fact that before a food is eaten, vision and olfaction pre-vail in the decision to select it and ingest it (Beauchamp &Mennella, 2011; Fedoroff, Polivy, & Herman, 2003; Ferriday &Brunstrom, 2008; Nederkoorn & Jansen, 2002; Ohla, Toepel, le Cou-tre, & Hudry, 2012). Although offering sensory cues (viz., pictures,odors) extracted from real foods certainly is a reduction in ecolog-ical validity, this approach has experimental advantages, such asthe avoidance of semantic biases that are inherent to lists of food
90 T. Jiang et al. / Appetite 67 (2013) 88–98
names, ease of stimuli management as compared to presenting arange of real foods, and reproducibility (e.g., to assess culturallyor geographically distant populations). Moreover, olfaction playsa major role in the perception of food flavor (Yeomans, 2006),whereas vision is strongly involved in distant food recognitionand selection by providing accurate information about food qual-ity, and often leads to flavor expectations (Verhagen & Engelen,2006). Such a sensory cue-exposure approach is relevant becauseit has been shown to elicit anticipatory food responses (e.g., phys-iological and behavioral reactivity) and to predict food intake (Jan-sen, 1998; Jansen et al., 2003; Soussignan et al., 2012).
The present study was designed to assess changes in reward re-sponses during childhood using olfactory and visual food cues. Byexamining self-reported wanting/liking responses to sensory cuesfor preferred (pizza, animal products, chocolate, fruits) and lesspreferred (vegetables) foods in children aged 6–11 years old testedin contrasting metabolic states, we investigated whether bothcomponents of reward are developmentally aligned. On the onehand, following number of previous studies supporting stabilityof food liking before adolescence (Nicklaus et al., 2004; Rollinset al., 2010) and because liking of foods (or food-related stimuli)is strongly influenced by earlier exposure effects (e.g., Birch & Mar-lin, 1982; Nicklaus et al., 2004; Schaal, 2005; Schaal & Soussignan,2008), we anticipated relative stability of liking between 6 and11 years, within the range of potential variation imparted by gen-der. On the other hand, as a motivational aspect of reward process-ing, wanting is closely related with hunger/satiety states,themselves linked with the current energetic and nutritional needsof the organism. Given that the middle-to-late period of childhood(6–11 years) is certainly not homogeneous in terms of nutritional/energetic requirements in relation with biological (e.g., bodygrowth, adiposity; Drewnowski, 2000; Rolland-Cachera, Deheeger,Maillot, & Bellisle, 2006) and psychological constraints (e.g., die-tary restraint, body self-perception; Birch & Fisher, 1998; Rollins,Loken, & Birch, 2011; Shunk & Birch, 2004), we predicted a highermotivation to eat (wanting) in younger than in older children be-cause middle childhood (5–7 years) appears critical for the onsetof adiposity rebound (AR) which reflects nutritional needs (Rol-land-Cachera et al., 2006), whereas preadolescence (9–11 years)has been associated with dietary restraint in eating behavior (Rol-lins et al., 2011; Shunk & Birch, 2004).
Methods
Participants
The children were recruited from four primary schools of Dijon(Burgundy, France). Parent’s information and authorization formsto let their child participate were sent to the families through theteachers. The study was approved by the bioethics committee ofthe Centre National de la Recherche Scientifique (Paris), the direc-tor of concerned schools, and all parents signed the informed con-sent to participate and let their child participate. The children wereincluded if: (1) their parents authorized them to participate, (2)they accepted to participate, (3) they took lunch in the school can-teen during at least two consecutive days in the week, and (4) theyhad no previous or ongoing adverse conditions affecting vision andsmell. Following these criteria, 156 children participated. The dataof five of them had to be dropped from the analyses because ofincomplete test sessions. The characteristics of the final sampleare given in Table 1. A two-way analysis of variance (ANOVA), withAge (5 levels) and Gender (girls vs. boys) as the between-subjectfactors, was run on BMI. As expected, there was a significant posi-tive linear trend of BMI as a function of age (F1,141 = 11.20,p = 0.001), indicating an increase of BMI across childhood, with sig-
nificant difference between 6–8 and 10–11 year-old children(p < 0.05). However, neither gender (F1,141 = 0.48, p = 0.49), northe interaction between age and gender (F1,141 = 1.31, p = 0.27),reached significance .
Stimuli
Visual stimuliThey were composed of eight color photographs representing
each a foodstuff presented on a white plate. The foodstuffs in-cluded a chocolate bar, a chopped steak, a slice of ham, a portionof pizza, a bowl of leek soup, a portion of cooked cauliflower, an or-ange cut in half, and a cup of strawberries. These pictures were se-lected from a collection of standardized photographs of foodproducts (Su-Vi-Max, 2002). These pictures were then digitalizedat 640 � 480 pixels and presented on a 17-in. LCD color screen atthe homogeneous size of 16 � 12 cm.
Olfactory stimuliThe odorants, graciously donated by Mane (Bar sur Loup,
France) and Asquali (Montaigut-Lauragais, France), were semanti-cally matched with the eight pictures of foodstuffs. Their selectionand matching with the pictures followed the procedure developedin a previous study (Soussignan et al., 2012). Accordingly, odors ofchocolate, beef, bacon, pizza, leek, cooked cauliflower, orange andstrawberry were presented in 60-ml brown glass jars (diame-ter � height: 45 � 76 mm, opening diameter: 28 mm). They werediluted in mineral oil (Aldrich, Saint Quentin-Fallavier, France) ata concentration of 0.5–4% (v/v) depending on the odorants and fol-lowing pretests which procedures were reported previously (Sous-signan et al., 2012). These dilutions were soaked on 6.5 � 11-cmstrips of scentless polypropylene (3 M, Bracknell, UK) to increaseevaporation surface within the test bottles and reduce risks ofspilling during the tests.
Design and procedure
Each participant took part in two sessions during two consecu-tive days in the school, one before (pre-prandial, hunger state) andthe other after the canteen lunch (post-prandial session, satiatedstate). Half of the children were assigned first to the pre-prandial,and then to the post-prandial sessions, while the other half wereexposed to the converse order of sessions. The pre-prandial sessionbegan at 12:00–12:15 and the post-prandial session at 13:00–13:15.
For each session an experimenter welcomed the child in a ded-icated, quiet room and explained that s/he was going to participateto playful tasks with pictures and odors to investigate ‘‘what chil-dren eat’’. The child was comfortably installed in a chair in front ofa table, facing one female experimenter. Several graphic Likert-type scales (see below) and the LCD screen were disposed in frontof the child. On the experimenter’s side was a laptop computer thatcontrolled the visual displays on the LCD screen. At the beginningof the pre-prandial session, the child’s weight (kg) and height (cm)were measured using a digital weight scale (Teraillon, type 9755,precision: 100 g, Valence, France) and a portable height scale(SECA, Leicester, UK).
The children’s hunger/satiety state was measured twice, at thebeginning and at the end of each session, using a 4-point Likert-type scale (ranging from 1: ‘‘I am not at all hungry’’, to 4: ‘‘I am ex-tremely hungry’’). To facilitate the understanding of the rating taskby children, a cartoon figure showing a schematic face and a breadwas associated to each Likert item by a variation in size (from 1:smallest, to 4: largest size) (Soussignan et al., 2012). When theexperimenter was sure that the accurate use of rating scales wasunderstood, the visual and olfactory tests were engaged. The order
Table 1Participants’ sample composition in age, gender, and body-mass index (BMI).
Age class (years) 6–7 7–8 8–9 9–10 10–11Gender Ga Ba G B G B G B G BNumber 16 14 17 13 14 16 13 17 14 17Mean age (SD) 6.43 (0.30) 6.48 (0.34) 7.66 (0.29) 7.57 (0.29) 8.45 (0.39) 8.52 (0.29) 9.56 (0.30) 9.50 (0.29) 10.53 (0.35) 10.74 (0.38)Mean BMI (SD) 16.93 (2.36) 15.68 (0.95) 15.88 (1.87) 17.19 (2.13) 16.65 (2.30) 16.63 (2.36) 17.08 (2.51) 17.76 (2.44) 17.79 (1.73) 18.38 (3.37)
a G: girls; B: boys.
T. Jiang et al. / Appetite 67 (2013) 88–98 91
of both tests was counterbalanced across participants: half partic-ipants (boys and girls of the distinct age groups) began with the vi-sual test, while the other half began with the olfactory test.
For the visual test, the child’s face was about 60 cm apart fromthe LCD screen used to display the food pictures. Each trial beganwith a visual warning signal lasting 2 s (the words ‘Attention tothe picture’) indicating that a picture is going to appear. Then a40-ms tone of 1500 Hz signaled the onset of each picture that re-mained displayed on the screen for 4 s, with an inter-trial intervalof about 15 s. The picture display was timed using the PowerPointsoftware (Microsoft).
For the olfactory test, at an audiovisual onset signal, the childhad to bring an opened odorant-containing glass jar under the nos-trils for 4 s. The interval to the next odor stimulus lasted at least30 s to limit olfactory adaptation. For each picture or odorant,the child was instructed to report at first her/his subjective likingand then her/his subjective wanting of the corresponding food-stuff. Face scales were used in children to facilitate understandingof the instructions designed to rate the stimuli (Chen, Resurreccion,& Paguio, 1996; Stone & Sidel, 2004). For liking, s/he was asked ‘‘Ifyou shall taste this food, how much would you like it?’’ using a 7-point Likert scale depicting a boy or a girl (selected from the Poser7.0 software for male and female participants, respectively) withthree smiles and three disgust expressions varying in intensity,and a neutral face in the middle. The child had to choose the inten-sity of the facial expression depicting better her/his liking (fromneutral: ‘‘Neither I like nor dislike’’, to more intense smiling: ‘‘I likevery much’’) or disliking (from neutral: ‘‘Neither I like nor dislike’’to more intense disgust face: ‘‘I don’t like at all’’) of the foodstuffrepresented by the picture or odor. Children’s verbal responseswere subsequently coded by an experimenter, ranging from �3to +3, with 0 for the neutral choice.
For wanting, each child was asked ‘‘At this moment, how much doyou desire/wish to eat this food’’ and was then required to report his/her wanting using a 4-point Likert scale showing a child who waseating a sandwich with the size of the picture increasing accordingto the intensity of desire to eat (from smallest size: ‘‘I don’t want toeat at all’’ to largest size: ‘‘I want to eat very much’’). The children’sverbal responses were subsequently coded from 1 to 4.
Statistical analysis
Age (five groups: 6–7, 7–8, 8–9, 9–10, 10–11 years) and Gender(girls, boys), taken as between-subjects factors, and State (pre vs.post-prandial) and Stimuli (chocolate, meat, pizza, vegetables,and fruits), as within-subjects factors, were entered in General Lin-ear Models (GLM). The liking/wanting scores for meat were calcu-lated by averaging beefsteak and bacon scores; for vegetables, leekand cauliflower scores were averaged; and for fruits, orange andstrawberry scores were averaged. The effect of Age on wantingand liking scores was analyzed using trend analyses (Howell,2009; Rosenthal & Rosnow, 1985), as for an ordered variable suchas age, classical ANOVA is less powerful and might bias the out-come (Rosenthal & Rosnow, 1985). With five age groups, both lin-ear (i.e., �2, �1, 0, 1, 2) and quadratic (i.e., �2, 1, 2, 1, �2) trendswere tested. To examine the effects of the Gender, Stimuli and
State factors, ANOVAs were performed. Following the significanceof overall F tests (for Stimuli), Tukey’s HSD post hoc tests compareddifferences between means. Contrast tests were used to examinespecific simple effects. Pearson linear correlation coefficients werecomputed to investigate relations between the different variablesregardless of age.
Results
Hunger perception
Hunger perception was calculated by averaging hunger ratingsestimated at the beginning and at the end of each session. A 3-way ANOVA was applied on hunger perception, with State (pre-vs. post-prandial) as within-subjects factor and Gender (Girls vs.Boys) and Age (5 groups) as between-subjects factors. Significantmain effects were found for all three factors: (1) boys expressedsignificantly higher hunger perception than girls (mean ± SE, G:2.37 ± 0.07, B: 2.58 ± 0.06, F1,141 = 5.41, p = 0.02); (2) hunger per-ception was significantly higher in the pre-prandial than in thepost-prandial states (Pre: 3.11 ± 0.06, Post: 1.85 ± 0.06,F1,139 = 276.82, p < 0.001); and (3) an effect of Age (F4,141 = 9.86,p < 0.001), with a significant negative linear trend of hunger per-ception over age (F1,141 = 36.47, p < 0.001). As shown in Fig. 1, thegender difference in reporting hunger perception tended to in-crease over the age period studied (p = 0.056) (Fig. 1a), while theeffect of prandial state remained stable for all age groups(Fig. 1b). No significant interactions between above factors werefound.
Picture-cued reward responses
LikingMain effects on liking reports were detected for: (1) Gender
(F1,141 = 16.23, p < 0.001), with higher liking scores in boys(1.85 ± 0.084) than in girls (1.36 ± 0.09), (2) State (F1,141 = 14.90,p < 0.001), with higher liking scores in pre-prandial (1.69 ± 0.07)than in post-prandial states (1.62 ± 0.06), and (3) Stimuli(F4,564 = 35.83, p < 0.001), with vegetables (0.54 ± 0.12) being sig-nificantly less liked than the other foods (chocolate: 1.85 ± 0.13;meat: 1.76 ± 0.08; pizza: 1.93 ± 0.11; fruits: 1.94 ± 0.08).
There was an interaction between State and Stimuli(F4,564 = 3.89, p = 0.004), showing significantly higher pre-prandialthan post-prandial liking scores for pizza (F1,141 = 14.35,p < 0.001) and vegetables (F1,141 = 5.47, p = 0.02).
Finally, no significant linear or quadratic Age effect was foundon liking. The main effects of Gender, State and Stimuli on likingare shown in Fig. 2 as a function of Age. Figure 2a reveals thatthe gender difference was small in 6–8 year-old children; it thenincreased over age of 8–10 years (F1,141 = 9.49, p = 0.002 for 8–9 years and F1,141 = 9.31, p = 0.003 for 9–10 years), to finally de-crease at age of 10–11 years, following a marginally significant n-type quadratic trend (F1,141 = 3.61, p < 0.06). The effect of State onliking appeared significant only in 6–7 years-old children(F1,141 = 11.59, p < 0.001), but without significant quadratic trend
Fig. 1. Hunger perception in children across age in relation with of: (a) gender, and (b) prandial state. �p 6 0.05; State effects are significant (all ps < 0.001) for all age groups in(b).
Fig. 2. Picture-cued liking rated by children across age in relation with: (a) gender, (b) prandial state, and (c) the type of food stimuli. ��p 6 0.01, ���p 6 0.001.
Fig. 3. Picture-cued wanting rated by children across age in relation with: (a) gender, (b) prandial state, and (c) the type of food stimuli. �p 6 0.05, ��p 6 0.01.
92 T. Jiang et al. / Appetite 67 (2013) 88–98
(Fig. 2b). Finally, liking evoked by the foodstuff pictures did notshow any variation trend in function of age (Fig. 2c).
WantingAn ANOVA indicated the following significant main effects on
wanting for: (1) Age (F4,141 = 8.55, p < 0.001), with a trend analysisfurther revealing that wanting scores decreased linearly from 6 to11 years (F1,141 = 32.24, p < 0.001), with significant difference be-tween ages 6–8 and 9–11 (F1,141 = 33.35, p < 0.001). This negativelinear trend was effective for both boys (F1,141 = 12.01, p < 0.001)and girls (F1,141 = 20.78, p < 0.001) (Fig. 3a), for both pre-prandial(F1,141 = 24.457, p < 0.001) and post-prandial states (F1,141 = 30.49,p < 0.001) (Fig. 3b), and for all five types of foods (all ps < 0.005,Fig. 3c); (2) Gender (F1,141 = 10.58, p = 0.001), with significantlyhigher wanting scores in boys (2.94 ± 0.07) than in girls(2.63 ± 0.07) (cf. Fig. 3a); (3) State (F1,141 = 22.70, p < 0.001) with
significantly higher scores in pre- (2.88 ± 0.05) than in post-pran-dial states (2.69 ± 0.05) (cf. Fig. 3b); and (4) Stimuli(F4,564 = 32.37, p < 0.001) with vegetables’ pictures eliciting lesswanting (2.21 ± 0.06) than all other tested food pictures(F1,141 = 113.36, p < 0.001; chocolate: 3.00 ± 0.08; meat:2.86 ± 0.06; pizza: 2.88 ± 0.08; fruits: 2.97 ± 0.06) (cf. Fig. 3c).
There was also a State by Stimulus interaction (F4,564 = 4.79,p < 0.001), indicating that the prandial state effect was significantlygreater for pizza than for chocolate and vegetables. The wantingscores were significantly lower during the post-prandial than thepre-prandial states for pizza (F1,141 = 25.21, p < 0.001), meats(F1,141 = 13.49, p < 0.001), vegetables (F1,141 = 5.84, p = 0.02), andfruits (F1,141 = 14.35, p = 0.006).
As shown in Fig. 3, the gender effect kept constant across age(Fig. 3a) and significant state effects were noticed for the lowest(F1,141 = 7.85, p = 0.006) and the highest age groups (F1,141 = 7.46,
T. Jiang et al. / Appetite 67 (2013) 88–98 93
p = 0.007 for 9–10 y; F1,141 = 10.90, p = 0.001 for 10–11 y), as indi-cated by a marginally-significant quadratic trend of state(F1,141 = 2.82, p = 0.095) (Fig. 3b).
Odor-cued reward responses
LikingThe ANOVA yielded significant main effects on odor-cued liking
of: (1) Age (F4,141 = 2,61, p = 0.04), with a trend of liking scores todecrease linearly with advancing age (F1,141 = 3,41, p = 0.07); (2)State (F1,141 = 16.55, p < 0.002), with significantly higher likingscores in the pre-prandial (0.77 ± 0.09) than in the post-prandialstates (0.56 ± 0.08); (3) Stimuli (F4,564 = 141.69, p < 0.001), all stim-uli eliciting different liking responses from each other in the fol-lowing order from the most to the least liked: fruits (2.17 ± 0.06),chocolate (1.23 ± 0.13), pizza (0.62 ± 0.12), meats (0.17 ± 0.12)and vegetables (�0.89 ± 0.11).
A significant interaction came out between Stimuli, State, andAge (F4,564 = 1.94, p = 0.02), indicating that the development of lik-ing scores was related to the type of food and the prandial state.Significant negative linear trends were found for fruits in bothpre-prandial (F1,141 = 6.72, p = 0.01) and post-prandial states(F1,141 = 14.65, p < 0.001), while a u-shaped quadratic trend (highliking at lowest and highest ages, and low liking at middle age)was found for chocolate at pre-prandial state (F1,141 = 4.15,p = 0.04), and a n-shaped quadratic trend (low liking at lowestand highest ages, and high liking at middle age) was found for piz-za in post-prandial state (F1,141 = 8.27, p = 0.005). The prandial stateeffect reached significance only in 6–7 year-old children for choco-late (F1,141 = 10.13, p < 0.002) and pizza (F1,141 = 14.10, p < 0.001).Figure 4 depicts the odor-cued liking profiles as a function of age,gender, prandial state, and stimuli. As shown in Fig. 4a, neither sig-nificant linear nor quadratic trends were detected in girls, whileboys showed a significant n-shaped quadratic trend (F1,141 = 3.97,p = 0.05), with highest reports of liking noted at age 8–9 years.The state effect tended to follow an u-shaped quadratic trend(F1,141 = 3.57, p = 0.06) with significant effects in the lowest(F1,141 = 10.40, p = 0.002) and highest age groups (F1,141 = 4.18,p < 0.04) (Fig. 4b). Finally, regarding each food odor, liking de-creased linearly as a function of age for fruits only (F1,141 = 13.00,p < 0.001) (Fig. 4c).
WantingThe ANOVA disclosed significant main effects on wanting re-
ports of: (1) Age (F1,141 = 18.75, p < 0.001); a subsequent trendanalysis indicated a significant linear trend (F1,141 = 69.72,p < 0.001), revealing that wanting scores decreased linearly overthe 6-to-11 year period with significant differences between ages
Fig. 4. Odor-cued liking rated by children across age in relation with: (a) gend
6–9 and 9–11 (F1,141 = 68.91, p < 0.001); this negative linear trendwas noted in both girls and boys (Fig. 5a), in both pre- and post-prandial states (Fig. 5b), and for all odor stimuli (Fig. 5c); (2) Pran-dial State, with significantly higher wanting in the pre-prandial(2.59 ± 0.04) than in the post-prandial states (2.34 ± 0.04)(F1,141 = 47.53, p < 0.001); a significant quadratic trend(F1,141 = 5.18, p = 0.02) pinpointed higher state effects in the lowestand highest age groups (F1,141 = 18.29, 22.17, and 17.02, for 6–7, 7–8, and 10–11 year-olds, respectively; p < 0.001 in all cases), and nostate effect in the 8–10 year-olds (Fig. 5b); (3) Stimuli(F4,564 = 117.26, p < 0.001), with all odorants being rated differentlyfrom each other (except for meat and pizza). The ordering of aver-age wanting scores from the most to the least wanted was: fruits(3.04 ± 0.05), chocolate (2.69 ± 0.07), pizza (2.41 ± 0.06), meats(2.36 ± 0.05), and vegetables (1.81 ± 0.05).
A significant State � Gender � Stimuli � Age interaction(F16,564 = 1.70, p = 0.043) led us to examine Age effects for all com-binations of each level of the remaining variables. Since a signifi-cant linear trend arose for wanting within the age range studied,20 linear trend analyses were performed at each prandial statefor each gender and each stimulus using Bonferroni correction(p < 0.0025). Effect sizes of these trends are shown in Table 2: choc-olate was the only food stimulus for which wanting scores did notchange linearly with age. All other trends were negative in bothpre-prandial and post-prandial states for both girls and boys. Thus,the trend analyses conducted on specific foods confirmed thatwanting was more affected than liking across childhood.
Effects of the sensory modality
The effect of the sensory nature of the food cues on the rewardresponses was analyzed using ANOVAs and trend analyses. For lik-ing, pictures obtained higher scores than odors (1.60 ± 0.06 vs.0.66 ± 0.08; F1,141 = 180.19, p < 0.001). More interestingly, SensoryModality � Stimulus (F4,564 = 29.59, p < .00001), Sensory Modal-ity � Gender (F1,141 = 12.74; p < .0004), and Sensory Modal-ity � Age � State � Stimulus (F16,564 = 2.3, p = .002) interactionsreached significance. These interactions reflected that, on the onehand, the modality effect (picture-cued liking > odor-cued liking)was significantly more important in boys than in girls, and, onthe other hand, that the modality effect depended on the prandialstate, stimulus used, and age group. For example, for chocolate, themodality effect was significant in 6–8-year-olds in the post-pran-dial state, but in 9–10-year old children in pre-prandial state. Formeat, there was a significant modality effect for all ages and inboth pre- and post-prandial states, while for vegetables, it was sig-nificant for all age groups in pre-prandial state.
er, (b) prandial state, and (c) the type of food stimuli. �p 6 0.05, ��p 6 0.01.
Fig. 5. Odor-cued wanting rated by children across age in relation with: (a) gender, (b) prandial state, and (c) the type of food stimuli. �p 6 0.05, ��p 6 0.01, ���p 6 0.001.
Table 2Effect sizes (r) between the wanting score and age (n = 151) for the different food odorants.
Gender Girls Boys
State Pre-prandial Post-prandial Pre-prandial Post-prandial
Stimuli Correlations Wanting Wanting Wanting WantingChocolate Age �0.23 (ns) �0.11 (ns) �0.30 (ns) �0.26 (ns)Meat Age �0.42 (medium)a �0.39 (medium) �0.58 (large) �0.55 (large)Pizza Age �0.32 (ns) �0.39 (medium) �0.48 (medium) �0.24 (ns)Vegetable Age �0.23 (ns) �0.46 (medium) �0.62 (large) �0.37 (medium)Fruit Age �0.46 (medium) �0.56 (large) �0.49 (medium) �0.40 (medium)
a Significant effect size (r) classified as small (|r| = 0.10–0.29), medium (|r| = 0.30–0.49), and large (|r| P 0.5) (Cohen, 1992); ns: non-significant.
94 T. Jiang et al. / Appetite 67 (2013) 88–98
For wanting, the pictures also obtained higher scores than theodors, (2.78 ± 0.05 vs. 2.46 ± 0.04; F1,141 = 72.27, p < 0.001). Signifi-cant effects were detected for the following interactions: SensoryModality � Age � Gender (F4,141 = 6.20, p < 0.001), Sensory Modal-ity � Age � State (F4,141 = 3.52, p = 0.009), Sensory Modal-ity � State � Gender (F1,141 = 8.41, p = 0.004) and SensoryModality � State � Stimulus � Gender (F4,564 = 3.18, p = .01). Theseinteractions indicate that boys of all ages rated pictures higherthan odors, whereas girls rated pictures higher than odors onlyat 7–8 and 9–10 years. When they were satiated, girls rated fruitodors higher than fruit pictures, while no modality differenceswere detected for fruits in boys and girls who were in hunger state.
Relationships between the variables
Picture- and odor-cued wanting/liking scoresPearson correlation coefficients were calculated to examine
whether picture-cued reward responses were concordant withodor-cued reward responses. Correlations between picture- andodor-cued liking for the considered food stimuli ranged from�0.09 to 0.43. Significant correlations (Bonferroni correction withp < 0.005) were found for meats (r = 0.31) and fruits (r = 0.43) inthe pre-prandial state. Correlations between picture- and odor-cued wanting scores ranged from 0.18 to 0.56, with significant cor-relations (Bonferroni correction with p < 0.005) found in pre-pran-dial states (for meats: r = 0.42, fruits: r = 0.56, chocolate: r = 0.33,and pizza: r = 0.28) and in post-prandial states (for meats:r = 0.43, and fruits: r = 0.51). These correlations indicate first thatmore frequent olfaction–vision associations appeared for wantingthan for liking reports, suggesting that both notions of rewardevoked dissociable responses in 6-to-11 year-old children. Sec-ondly, picture- and odor-cued wanting or liking scores for a givenfood tended to elicit more agreement when the children were hun-gry than when they were sated.
Reward scores and hunger perceptionThe liking scores (for both picture- and odor-cued food objects)
were poorly correlated with hunger scores. Among the 20 correla-tions, only picture-cued meats (r = 0.37) and odor-cued fruits(r = 0.33) reached statistical significance in the pre-prandial state(in both cases, p < 0.001). Effect sizes between liking and hungerscores ranged from �0.06 to 0.37. No difference was found be-tween picture- and odor-cued correlations of liking and hunger.In contrast, 13 among 20 correlations between wanting and hungerscores reached statistical significance (all ps < 0.001, effect sizerange: 0.10–0.45). Picture-cued wanting scores produced signifi-cantly higher correlations with hunger scores than did odor-cuedwanting (F1,8 = 5.37, p = 0.05). Thus, picture- and odor-mediatedreward rating by children differed for liking and wanting in rela-tion with subjective hunger. In the present experimental condi-tions, the children’s wanting responses appeared indeed morestrongly associated with hunger than did liking responses. Further,the children’s wanting–hunger link may be stronger for picturecues than for odor cues.
Liking and wanting scoresLiking and wanting scores were significantly correlated in both
pre-prandial and post-prandial states for all foodstuffs and sensorymodalities studied here (all ps < 0.001, n = 151; effect sizes forvisually-cued and odor-cued responses ranging from 0.63 to 0.77and 0.37 to 0.59, respectively). Picture-cued correlations betweenliking and wanting scores were significantly higher than odor-cuedcorrelations (F1,8 = 44.24, p < 0.001). Figure 6 shows the level of lik-ing/wanting correlations as a function of age. A trend analysis indi-cated significant linear trends for both picture- (F1,45 = 13.21,p = 0.001) and odor-cued responses (F1,45 = 7.58, p = 0.008), butwith opposite, negative and positive trends for picture- and odor-cued responses, respectively. As can be seen in Fig. 6, the food pic-tures elicited more convergent reward scores than the food odor-ants in younger children (6–9 years), whereas both odor and
Fig. 6. Liking–wanting correlations (Pearson’s r) in function of age.
T. Jiang et al. / Appetite 67 (2013) 88–98 95
visual cues elicited concordant responses in older children (9–11 years).
Discussion
In the present study, we used a sensory cue-reactivity approachand manipulated proximate factors of the organism (motivationalstate) and the stimuli (sensory modalities) to begin delineatingthe developmental patterning of affective (liking, sensory pleasure)and motivational (wanting, desire) components of reward pro-cesses in school-age children exposed to both visual and olfactoryfood stimuli.
Differential developmental trajectories for wanting and liking?
Our findings indicate that both dimensions of reward followeddistinct developmental trajectories. While food liking remainedrelatively stable along late childhood, in particular for the presentset of picture-cued foodstuffs, food wanting showed a negative lin-ear trend over the age range studied, with a strong decrease notedfor both picture- and odor-cued foods in late childhood (9–11 years) as compared to middle childhood (6–7 years). It remainsto be clarified why liking reports were higher and more develop-mentally stable for pictures than for odors of the same foodstuffs.Vision certainly predominates in the distal selection of foodstuffsor beverages in humans (e.g., Morrot, Brochet, & Dubourdieu,2001; Verhagen & Engelen, 2006; Zampini, Wantling, Phillips, &Spence, 2008; Zellner, Bartoli, & Eckard, 1991). By allowing anaccurate identification of objects, specifically foodstuffs, visionprovides a fuller and more immediate representation of the aspect,presentation, and hedonic nature of specific foods than olfaction.Olfaction plays a more important role in the proximal examinationof foodstuffs, for testing its familiarity/novelty, level of freshness/ripeness, and ultimately confirming its comestibility. The domi-nance of vision in the present experimental conditions was furthersupported by the fact that liking and wanting scores were bettercorrelated for pictures than for odorants, especially in the groupof younger children, suggesting that visual cues engender higheragreement than odor cues in the perception of food reward in 6–9-year-olds (as compared to 9–11-year-olds). A previous studyrun on participants aged 2–7, 8–9, 10–11, 12–18 and >18 years alsoindicated that before 8–9 years flavor perception of drinks manip-ulated to be visually (in)congruent with their flavor is dominatedby vision. It is only after 8–9 years that children became able to fo-cus their attention on the flavor properties of stimuli, and this fla-vor-based response then steadily increased with age (Oram et al.,1995). This is also in line with previous studies in adults and chil-dren showing that pictures of food and non-food items differenti-
ate the hedonic value of objects more clearly than odors, usingboth self-report and facial electromyographic measurement (Jianget al., 2008; Soussignan, Ehrlé, Henry, Schaal, & Bakchine, 2005;Soussignan et al., 2011).
The relative stability of liking across development was previ-ously reported for a wide range of conventional food stimuli in2-to-8 year-old children (Nicklaus et al., 2004; Skinner et al.,2002), for snack foods in girls followed longitudinally from age 5to 11 (Rollins et al., 2010), or for a fruity yoghurt between 10-year-olds and adults (Olsen et al., 2011). Although some changesin liking have also been reported during childhood (Cooke & War-dle, 2005; Olsen et al., 2011), it is unclear whether they resultedfrom differences in methods or food items or even from uncon-trolled factors (e.g., motivational state, familiarity) that can affectthe decision to select the most preferred foods or the rating ofliked/disliked foods. For example, although food preference mea-sured by questionnaires has been conceptualized as food liking/disliking, it is unclear whether motivational differences (in hun-ger/satiety levels) may impact liking and wanting responses whenpreferences are assessed in this way.
In the present study, the age-dependent decrease in the motiva-tion to eat foods evoked by sensory cues was concurrent of areduction in hunger perception across childhood, with wanting,but not liking, being correlated with hunger scores during bothpre- and post-prandial states. Thus, momentary interoceptive cueslinked to hunger may affect wanting more than liking (Jiang et al.,2008; Soussignan, Jiang, Rigaud, Royet, & Schaal, 2010). Two fac-tors may be invoked to account for the developmental patterningin the desire to eat foods as they were modeled in the presentstudy. On the one hand, the higher score of wanting in 6–7 yearold children might be partly explained by findings showing thatmiddle childhood (5–7 years) is critical for the onset of the adipos-ity rebound (AR) (Rolland-Cachera et al., 1984, 1987, 2006; Whi-taker, Pepe, Wright, Seidel, & Dietz, 1998). A rise in body fatnessoccurring between ages 5 and 7 years might indeed be causal orconsequential of a higher motivation to eat palatable foods. Onthe other hand, the reduction in eating motivation in preadoles-cents may be explained in terms of concerns about weight, self-body image and dieting, which increase between middle (5–7 years) and late childhood (9–11 years), specifically in girls (Ber-nier, Kozyrskyj, Benoit, Becker, & Marchessault, 2010; Hill, Draper,& Stack, 1994; Packard & Krogstrand, 2002). These psychologicalchanges are associated with dietary restraint abilities and self-con-trol in preadolescence (Raffaelli, Crockett, & Shen, 2005; Rollinset al., 2011; Shunk & Birch, 2004). This view is supported in thepresent findings by the fact that girls reported lower wantingscores than boys, which is consistent with the fact that girls aremore concerned by body image and dieting issues, and expressmore self-control or food intake inhibition (Hill et al., 1994; Pack-ard & Krogstrand, 2002). Although it is more difficult to explain thedecrease in hunger perception noted in the present study acrosschildhood, and more in girls than in boys, one may speculate thatan increase in dietary restraint during preadolescence might biasthe perception of hunger state and lead children to missense or ne-glect interoceptive cues. Clearly, additional investigations are re-quired to evaluate this hypothesis.
Otherwise, the relative developmental stability found for liking inthe present study probably reflects the influence of early exposureand familiarity. Indeed, food preference has been shown to increaseas a function of exposure frequency (Birch & Marlin, 1982) and learn-ing is a strong determinant of the developmental stability of foodpreference and acceptance (Birch & Fisher, 1998; Nicklaus et al.,2004). The fact that picture-cued liking appeared more stable overage than odor-cued liking also supports the interpretation of learn-ing/exposure predominance in food liking, as visual stimuli do more
96 T. Jiang et al. / Appetite 67 (2013) 88–98
fully foster the recall of hedonics associated with food and eatingcontexts (Davis, 1977; Schifferstein, Smeets, & Postma, 2010).
Effect of the prandial state on wanting and liking
As expected, self-reported desire and pleasure evoked by ourfood-related pictures and odors decreased after the meal. However,this general prandial effect was not homogenous as a function ofage and of reward component. It was more pronounced for want-ing than for liking in younger and older children (6–7 and 10–11 years), and was absent or very shallow in 8–10-year-olds. Asthis result is, to the best of our knowledge, unprecedented in theliterature, the causes of such reduced or absent prandial effect onliking and wanting of food cues in 8–10-year-olds are unclear.Although alliesthesia to food cues has been shown in newborns(Soussignan et al., 1999) and adults (Cabanac & Duclaux, 1973;Jiang et al., 2008; Stoeckel et al., 2007; Uher, Treasure, Heining,Brammer, & Campbell, 2006), it is reasonable to conceive that itfluctuates as a function of variations in energetic/nutritional needsproper to infancy and childhood. Middle and late childhood consti-tutes indeed a period of transition where the prandial effect on re-ward perception may be transitionally stabilized from 8 years tobefore the pubertal growth spurt. Longitudinal findings have re-vealed that American children from 5 to 9 years of age increasedtheir intake of highly palatable snack foods after a standard meal,suggesting that a higher responsiveness to palatable foods acrossmiddle childhood may resist prandial state effects (Birch, Fisher,& Davison, 2003). In addition, 6–8 year-old children adjusted moreeffectively to a caloric preload than 8–10 year-old children andwere more able to discriminate between low-energy and high-en-ergy preload snacks (Cecil et al., 2005). Further studies are neededat different ontogenetic periods to confirm this developmentaltrend not only with food cues in the lab, but also with real foodsin the lab and in real meals.
Effect of gender on liking and wanting
Boys expressed higher wanting and liking scores than girls. Butthis effect operated in response to food pictures only, no differencebeing noted with the food odors. As already mentioned, visual cuesprovide a more cognitively-elaborated representation of objectsthan odors, suggesting that this difference may be mainly medi-ated by food cognition and gender-based food attitudes (Burton,Smit, & Lightowler, 2007; Cornier et al., 2010; Gravener, Haedt,Heatherton, & Keel, 2008). In contrast with a number of prior stud-ies, we did not find that girls display higher liking for (sensory cuesof) vegetables and fruits (as noted by Bere et al., 2008; Brug et al.,2008; Cooke & Wardle, 2005; Rasmussen et al., 2006) and that boysliked more (the sensory cues of) animal products (as noted byCooke & Wardle, 2005; Nicklaus et al., 2004). However, as somestudies did not find food-related gender differences in liking duringchildhood (Hill, Saxton, Webber, Blundell, & Wardle, 2009; Russell& Worsley, 2008), it is unclear whether between-studies differ-ences on this point reflect methodological disparities (e.g. sensorycues, lists of food words, real foods), or other factors such as foodavailability (due to social/economic status of parents; e.g., for lik-ing of fruits) (Hilsen, van Stralen, Klepp, & Bere, 2011). Interest-ingly, the lower wanting/liking scores of girls for food pictureswere paralleled with lower hunger perception in girls than in boys.This result is reminiscent of adult women declaring significantlyhigher post-meal satiety and fullness and lower hunger ratingsthan did men (Cornier et al., 2010; Gregersen et al., 2011).
Effect of the nature of food
From middle to late childhood the assessment of changes in re-ward responses to the five categories of food pictures revealed thatvegetables were rated as bearing low reward value compared tothe other foods considered here. However, by using food odors, aclearer differentiation emerged among the represented foodstuffs,with fruits being the most liked and desired, followed by chocolateor pizza and then meats, and finally vegetables. This pattern of foodpreference was globally concordant with that reported by Cookeand Wardle (2005), and was stable over middle and late childhood.
Limitations
Several limitations of our study must be acknowledged. Oneobvious concern in interpreting any age-related difference iswhether it resulted from task comprehension difficulties (e.g., dif-ferentiation between food liking and food wanting), rather than, aswe argue above, from causes that are specific to motivational and/or perceptual fluctuations. Several lines of evidence suggest thatthe children fully understood the task that we used here. First,we used facial scales that are appropriate for young children (Chenet al., 1996). In our study, these scales provided distinctive cues tochildren allowing a clear discrimination between food liking (tast-ing being related to positive and negative facial expressions) andfood wanting (avidity for a food being related to the size of a pic-ture). Second, although liking and wanting scores were signifi-cantly correlated in both pre-prandial and post-prandial statesfor all foodstuffs and sensory modalities, the participants differen-tiated them as the prandial state had more effect on the assess-ment of wanting than on the assessment of liking. Anotherpossibility is that the present results could derive from some spe-cific attributes of the present set of stimuli. Thus, to be generaliz-able, future investigations tracking developmental changes inreward processing of food-related stimuli might use different setsof familiar stimuli in different populations of children participants.Moreover, as the study was conducted in the school environment,no control could be exerted on the foods that children ate duringthe lunches on the day they were tested. Although this might con-stitute an additional source of within age-groups variability duringthe postprandial session, it can be minimized as the menu rota-tions in the school canteens extended over 1 month (i.e., the samemeal composition being repeated every 4 weeks). Finally, the pat-tern of results reported in the present study should be interpretedwith caution considering the small sample of children and consid-ered as providing a new hypothesis about the effect of develop-mental organismic needs on the magnitude of alliesthesia thatneeds to be further explored in replication studies including youn-ger and older participants.
In conclusion, the present study highlighted that the motiva-tional and affective components of food reward might follow dif-ferent developmental trajectories from middle through latechildhood. Since sensory sensitivity and reactivity, and food re-ward perception are known to influence food intake and overeating(Finlayson et al., 2007; Jansen et al., 2003; Naish & Harris, 2012;Soussignan et al., 2012), such data must be confirmed to verifywhether children are more vulnerable to food-related sensory cuesduring particular stages of growth, and whether the developmentof food intake across childhood may be differentially predictedby wanting or by liking. Finally, these findings should be extendedto a larger age range and confirmed using both subjective–declar-ative and objective measures [e.g., overt and covert (electromyog-raphy) facial responsiveness, progressive reinforcement schedules,reaction times, intake, metabolic and somatic variables].
T. Jiang et al. / Appetite 67 (2013) 88–98 97
References
Beauchamp, G. K., & Mennella, J. A. (2009). Early flavor learning and its impact onlater feeding behavior. Journal of Pediatric Gastroenterology and Nutrition, 48,S25–S30.
Beauchamp, G. K., & Mennella, J. A. (2011). Flavor perception in human infants.Development and functional significance. Digestion, 83, 1–6.
Bere, E., Brug, J., & Klepp, K. I. (2008). Why do boys eat less fruit and vegetables thangirls? Public Health Nutrition, 11, 321–325.
Bernier, C. D., Kozyrskyj, A. L., Benoit, C., Becker, A. B., & Marchessault, G. (2010).Body image and dieting attitudes among preadolescents. Canadian Journal ofDietetic Practice and Research, 71, e34–e40.
Berridge, K. C. (2009). ‘‘Liking’’ and ‘‘wanting’’ food rewards. Brain substrates androles in eating disorders. Physiology and Behavior, 97, 537–550.
Berridge, K. C., & Robinson, T. E. (2003). Parsing reward. Trends in Neurosciences, 26,507–513.
Birch, L. L. (1998). Development of food acceptance patterns in the first years of life.Proceedings of the Nutrition Society, 57, 617–624.
Birch, L. L. (1999). Development of food preferences. Annual Review of Nutrition, 19,41–62.
Birch, L. L., & Fisher, J. O. (1998). Development of eating behaviors among childrenand adolescents. Pediatrics, 101, 539–549.
Birch, L. L., Fisher, J. O., & Davison, K. K. (2003). Learning to overeat. Maternal use ofrestrictive feeding practices promotes girls’ eating in the absence of hunger. TheAmerican Journal of Clinical Nutrition, 78, 215–220.
Birch, L. L., & Marlin, D. W. (1982). I don’t like it; I never tried it. Effects of exposureon two-year-old children’s food preferences. Appetite, 3, 353–360.
Blass, E. M. (2008). Obesity. Causes, mechanisms, prevention, and treatment.Sunderland, MA: Sinauer.
Bobroff, E. M., & Kissileff, H. R. (1986). Effects of changes in palatability on foodintake and the cumulative food intake curve in man. Appetite, 7, 85–96.
Bogin, B. (1997). Evolutionary hypotheses for human childhood. Yearbook of PhysicalAnthropology, 40, 63–89.
Born, J. M., Lemmens, S. G. T., Martens, M. J. I., Formisano, E., Goebel, R., &Westerterp-Plantenga, M. S. (2011). Differences between liking and wantingsignals in the human brain and relations with cognitive dietary restraint andbody mass index. The American Journal of Clinical Nutrition, 94, 392–403.
Brug, J., Tak, N. I., te Velde, S. J., Bere, E., & de Bourdeaudhuij, I. (2008). Tastepreferences, liking and other factors related to fruit and vegetable intakesamong schoolchildren. Results from observational studies. The British Journal ofNutrition, 99, S7–S14.
Burton, P., Smit, H. J., & Lightowler, H. J. (2007). The influence of restrained andexternal eating patterns on overeating. Appetite, 49, 191–197.
Cabanac, M. (1971). Physiological role of pleasure. Science, 173, 1103–1107.Cabanac, M., & Duclaux, R. (1973). Alliesthésie olfacto-gustative et prise alimentaire
chez l’homme. Journal de Physiologie (Paris), 66, 113–135.Cecil, J. E., Palmer, C. N., Wrieden, W., Murrie, I., Bolton-Smith, C., Watt, P., Wallis, D.
J., & Hetherington, M. M. (2005). Energy intakes of children after preloads.Adjustment, not compensation. The American Journal of Clinical Nutrition, 82,302–308.
Chen, A., Resurreccion, A. V. A., & Paguio, L. P. (1996). Age appropriate hedonic scalesto measure food preferences of young children. Journal of Sensory Studies, 11,141–163.
Cohen, J. (1992). A power primer. Psychological Bulletin, 112, 155–159.Cooke, L. J., & Wardle, J. (2005). Age and gender differences in children’s food
preferences. British Journal of Nutrition, 93, 741–746.Cornier, M.-A., Salzberg, A. K., Endly, D. C., Bessesen, D. H., & Tregellas, J. R. (2010).
Sex-based differences in the behavioral and neuronal responses to food.Physiology & Behavior, 99, 538–543.
Cornier, M.-A., Von Kaenel, S. S., Bessesen, D. H., & Tregellas, J. R. (2007). Effects ofoverfeeding on the neuronal response to visual food cues. American Journal ofClinical Nutrition, 86, 965–971.
Davis, R. G. (1977). Acquisition and retention of verbal associations to olfactory andabstract visual stimuli of varying similarity. Journal of Experimental Psychology.Human Learning and Memory, 3, 37–51.
Davis, C. A., Levitan, R. D., Reid, C., Carter, J. C., Kaplan, A. S., Patte, K. A., King, N.,Curtis, C., & Kennedy, J. L. (2009). Dopamine for ‘‘wanting’’ and opioids for‘‘liking’’. A comparison of obese adults with and without binge eating. Obesity,17, 1220–1225.
Drewnowski, A. (2000). Sensory control of energy density at different life stages.Proceedings of the Nutrition Society, 59, 239–244.
Epstein, L. H., Truesdale, R., Wojcik, A., Paluch, R. A., & Raynor, H. A. (2003). Effects ofdeprivation on hedonics and reinforcing value of food. Physiology & Behavior, 78,221–227.
Fedoroff, I., Polivy, J., & Herman, P. (2003). The specificity of restrained versusunrestrained eaters’ responses to food cues. General desire to eat, or craving forthe cued food? Appetite, 41, 7–13.
Ferriday, D., & Brunstrom, J. M. (2008). How does food-cue exposure lead to largermeal sizes? British Journal of Nutrition, 100, 1325–1332.
Finlayson, G., King, N., & Blundell, J. E. (2007). Is it possible to dissociate ‘liking’ and‘wanting’ for foods in humans? A novel experimental procedure. Physiology &Behavior, 90, 36–42.
Gravener, J., Haedt, A. A., Heatherton, T. F., & Keel, P. K. (2008). Gender and agedifferences in associations between peer dieting and drive for thinness.International Journal of Eating Disorders, 41, 57–63.
Gregersen, N. T., Mrller, B. K., Raben, A., Kristensen, S. T., Holm, L., Flint, A., & Astrup,A. (2011). Determinants of appetite ratings. The role of age, gender, BMI,physical activity, smoking habits, and diet/weight concern. Food NutritionResearch, 55. http://dx.doi.org/10.3402/fnr.v55i0.7028.
Hill, A. J., Draper, E., & Stack, J. (1994). A weight on children’s minds. Body shapedissatisfactions at 9-years old. International Journal of Obesity, 18, 383–389.
Hill, C., Saxton, J., Webber, L., Blundell, J., & Wardle, J. (2009). The relativereinforcing value of food predicts weight gain in a longitudinal study of 7–10-y-old children. The American Journal of Clinical Nutrition, 90, 276–281.
Hilsen, M., van Stralen, M. M., Klepp, K.-I., & Bere, E. (2011). Changes in 10–12 yearold’s fruit and vegetable intake in Norway from 2001 to 2008 in relation togender and socioeconomic status. A comparison of two cross-sectional groups.International Journal of Behavioral Nutrition and Physical Activity, 8, 108. http://dx.doi.org/10.1186/1479-5868-8-108.
Howell, D. C. (2009). Statistical methods for psychology. Belmont, USA: Wadsworth,Cengage Learning.
Jansen, A. (1998). A learning model of binge eating. Cue reactivity and cue exposure.Behaviour Research and Therapy, 36, 257–272.
Jansen, A., Theunissen, N., Slechten, K., Nederkoorn, C., Boon, B., Mulkens, S., & Roefs,A. (2003). Overweight children overeat after exposure to food cues. EatingBehaviors, 4, 197–209.
Jiang, T., Soussignan, R., Rigaud, D., Martin, S., Royet, J.-P., & Schaal, B. (2008).Alliesthesia to food cues. Heterogeneity across stimuli and sensory modalities.Physiology & Behavior, 95, 464–470.
Killgore, W. D. S., & Yurgelun-Todd, D. A. (2010). Sex differences in cerebralresponses to images of high versus low-calorie food. NeuroReport, 21, 354–358.
Konner, M. (2010). The evolution of childhood. Relationships, emotion, mind.Cambridge, MA: Belknap Press–Harvard University Press.
Laing, D. G., & Clark, P. J. (1983). Puberty and olfactory preferences of males.Physiology & Behavior, 30, 591–597.
Le Bigot Macaux, A. (2001). Eat to live or live to eat? Do parents and children agree?Public Health Nutrition, 4, 141–146.
Lowe, M. R., & Butryn, M. L. (2007). Hedonic hunger. A new dimension of appetite?Physiology & Behavior, 91, 432–439.
Mela, D. J. (2006). Eating for pleasure or just wanting to eat? Reconsidering sensoryhedonic responses as a driver of obesity. Appetite, 47, 10–17.
Mennella, J. A., & Beauchamp, G. K. (2002). Flavor experiences during formulafeeding are related to preferences during childhood. Early Human Development,68, 71–82.
Morrot, G., Brochet, F., & Dubourdieu, D. (2001). The color of odors. Brain andLanguage, 79, 309–320.
Naish, K. R., & Harris, G. (2012). Food intake is influenced by sensory sensitivity.PLoS One, 7, e43622.
Nakamura, K., Shimai, S., Kikuchi, S., & Tanaka, M. (2001). Correlation between aliking for fat-rich foods and body fatness in adult Japanese. A gender difference.Appetite, 36, 1–7.
Neary, M. T., & Batterham, R. L. (2010). Gaining new insights into food reward withfunctional neuroimaging. In W. Langhans & N. Geary (Eds.), Frontiers in eatingand weight regulation (pp. 152–163). Basel: Karger.
Nederkoorn, C., & Jansen, A. (2002). Cue reactivity and regulation of food intake.Eating Behaviors, 3, 61–72.
Nicklaus, S., Boggio, V., Chabanet, C., & Issanchou, S. (2004). A prospective study offood preferences in childhood. Fifth Rose Marie Pangborn Sensory ScienceSymposium. Food Quality and Preference, 15, 805–818.
Ohla, K., Toepel, U., le Coutre, J., & Hudry, J. (2012). Visual–gustatory interaction.Orbitofrontal and insular cortices mediate the effect of high-calorie visual foodcues on taste pleasantness. PLoS One, 7, e32434.
Olsen, A., Ritz, C., Hartvig, D. L., & Møller, P. (2011). Comparison of sensory specificsatiety and sensory specific desires to eat in children and adults. Appetite, 57,6–13.
Oram, N., Laing, D. G., Hutchinson, I., Owen, J., Rose, G., Freeman, M., & Newell, G.(1995). The influence of flavor and color on drink identification by children andadults. Developmental Psychobiology, 28, 239–246.
Packard, P., & Krogstrand, K. S. (2002). Half of rural girls aged 8 to 17 years reportweight concerns and dietary changes, with both more prevalent with increasedage. Journal of the American Dietetic Association, 102, 672–677.
Perez-Rodrigo, C., Ribas, L., Serra-Majem, L., & Aranceta, J. (2003). Food preferencesof Spanish children and young people. The enKid study. European Journal ofClinical Nutrition, 57, S45–S48.
Raffaelli, M., Crockett, L. J., & Shen, Y.-L. (2005). Developmental stability and changein self-regulation from childhood to adolescence. Journal of Genetic Psychology,166, 54–75.
Rasmussen, M., Krrlner, R., Klepp, K.-I., Lytle, L., Brug, J., Bere, E., & Due, P. (2006).Determinants of fruit and vegetable consumption among children andadolescents. A review of the literature. Part I. Quantitative studies.International Journal of Behavioral Nutrition and Physical Activity, 3, 22. http://dx.doi.org/10.1186/1479-5868-3-22.
Rolland-Cachera, M. F., Deheeger, M., Bellisle, F., Semp, M., Guilloud-Bataille, M., &Patois, E. (1984). Adiposity rebound in children. A simple indicator forpredicting obesity. The American Journal of Clinical Nutrition, 39, 129–135.
Rolland-Cachera, M. F., Deheeger, M., Guilloud-Bataille, M., Avons, P., Patois, E., &Sempé, M. (1987). Tracking the development of adiposity from one month ofage to adulthood. Annals of Human Biology, 14, 219–229.
Rolland-Cachera, M. F., Deheeger, M., Maillot, M., & Bellisle, F. (2006). Earlyadiposity rebound. Causes and consequences for obesity in children and adults.International Journal of Obesity, 30, S11–S17.
98 T. Jiang et al. / Appetite 67 (2013) 88–98
Rollins, B. Y., Loken, E., & Birch, L. L. (2010). Stability and change in snack food likesand dislikes from 5 to 11 years. Appetite, 55, 371–373.
Rollins, B. Y., Loken, E., & Birch, L. L. (2011). Preferences predict food intake from 5 to11 years, but not in girls with higher weight concerns, dietary restraint, and%body fat. Obesity, 19, 2190–2197.
Rolls, B. J., Van Duijvenvoorde, P. M., & Rolls, E. T. (1984). Pleasantness changes andfood intake in a varied four-course meal. Appetite, 5, 337–348.
Rosenthal, R., & Rosnow, R. L. (1985). Contrast analysis, focused comparison in theanalysis of variance. London: Cambridge University Press.
Russell, C. G., & Worsley, A. (2008). A population-based study of preschoolers’ foodneophobia and its associations with food preferences. Journal of NutritionEducation and Behavior, 40, 11–19.
Schaal, B. (2005). From amnion to colostrum to milk. Odour bridging in earlydevelopmental transitions. In B. Hopkins & S. Johnson (Eds.), Prenataldevelopment of postnatal functions (pp. 52–102). Westport, CT: Praeger.
Schaal, B., & Soussignan, R. (Eds.). (2008). L’enfant face aux aliments [Children facingfoods]. Paris: Presses Universitaires de France (Enfance).
Schifferstein, H. N., Smeets, M. A., & Postma, A. (2010). Comparing location memoryfor 4 sensory modalities. Chemical Senses, 35, 135–145.
Shunk, J. A., & Birch, L. L. (2004). Girls at risk for overweight at age 5 are at risk fordietary restraint, disinhibited overeating, weight concerns, and greater weightgain from 5 to 9 years. Journal of the American Dietetic Association, 104,1120–1126.
Skinner, J. D., Carruth, B. R., Bounds, W., & Ziegler, P. J. (2002). Children’s foodpreferences. A longitudinal analysis. Journal of the American Dietetic Association,102, 1638–1647.
Smeets, P. A., Erkner, A., & De Graaf, C. (2010). Cephalic phase responses andappetite. Nutrition Review, 68, 643–655.
Sorensen, L. B., Moller, P., Flint, A., Martens, M., & Raben, A. (2003). Effect of sensoryperception of foods on appetite and food intake. A review of studies on humans.International Journal of Obesity, 27, 1152–1166.
Soussignan, R., Ehrlé, N., Henry, A., Schaal, B., & Bakchine, S. (2005). Dissociation ofemotional processes in response to visual and olfactory stimuli followingfrontotemporal damages. Neurocase, 11, 114–128.
Soussignan, R., Jiang, T., Rigaud, D., Royet, J. P., & Schaal, B. (2010). Subliminal fearpriming potentiates negative facial reactions to food pictures in women withanorexia nervosa. Psychological Medicine, 40, 503–514.
Soussignan, R., Schaal, B., Boulanger, V., Gaillet, M., & Jiang, T. (2012). Orofacialreactivity to the sight and smell of food stimuli. Evidence for anticipatory likingrelated to food reward cues in overweight children. Appetite, 58, 508–516.
Soussignan, R., Schaal, B., & Marlier, L. (1999). Olfactory alliesthesia in humanneonates. Prandial state and stimulus familiarity modulate facial andautonomic responses to milk odors. Developmental Psychobiology, 35, 3–14.
Soussignan, R., Schaal, B., Rigaud, D., Royet, J. P., & Jiang, T. (2011). Hedonicreactivity to visual and olfactory cues. Rapid facial electromyographic reactionsare altered in anorexia nervosa. Biological Psychology, 86, 265–272.
Staiger, P., Daweb, S., & McCarthy, R. (2000). Responsivity to food cues in bulimicwomen and controls. Appetite, 35, 27–33.
Stoeckel, L. E., Cox, J. E., Cook, E. W., III, & Weller, R. E. (2007). Motivational statemodulates the hedonic value of food images differently in men and women.Appetite, 48, 139–144.
Stone, H., & Sidel, J. L. (2004). Sensory evaluation practices (3rd ed.). Boston:Academic Press.
SU-VI-MAX (2002). Portions alimentaires. Manuel photos pour l’estimation desquantités. Paris: Economia (Polytechnica).
Ton Nu, C., MacLeod, P., & Barthelemy, J. (1996). Effects of age and gender onadolescents’ food habits and preferences. Second Rose Marie PangbornMemorial Symposium. Food Quality and Preference, 7, 251–262.
Uher, R., Treasure, J., Heining, M., Brammer, M. J., & Campbell, I. C. (2006). Cerebralprocessing of food-related stimuli. Effects of fasting and gender. BehaviouralBrain Research, 169, 111–119.
Verhagen, J. V., & Engelen, L. (2006). The neurocognitive bases of humanmultimodal food perception. Sensory integration. Neuroscience andBiobehavioral Reviews, 30, 613–650.
Whitaker, R. C., Pepe, M. S., Wright, J. A., Seidel, K. D., & Dietz, W. H. (1998). Earlyadiposity rebound and the risk of adult obesity. Pediatrics, 101, E5.
Yeomans, M. R. (2006). Olfactory influences on appetite and satiety in humans.Physiology & Behavior, 89, 10–14.
Yeomans, M. R., Blundell, J. E., & Leshem, M. (2004). Palatability. Response tonutritional need or need-free stimulation of appetite? British Journal ofNutrition, 92, S3–S14.
Zampini, M., Wantling, E., Phillips, N., & Spence, C. (2008). Multisensory flavorperception. Assessing the influence of fruit acids and color cues on theperception of fruit-flavored beverage. Food Quality and Preference, 19, 335–343.
Zandstra, E. H., De Graaf, C., Mela, D. J., & Van Staveren, W. A. (2000). Short- andlong-term effects of changes in pleasantness on food intake. Appetite, 34,253–260.
Zellner, D. A., Bartoli, A. M., & Eckard, R. (1991). Influence of color on odoridentification and liking ratings. The American Journal of Psychology, 104,547–561.