Mechanisms Behind the Portion Size Effect: Visibility and Bite Size

Transcription

1 nature publishing group Mechanisms Behind the Portion Size Effect: Visibility and Bite Size Kyle S. Burger 1,2, Jennifer O. Fisher 3 and Susan L. Johnson 1,4 Increases in size lead to increases in energy intake, yet the mechanisms behind this size effect are unclear. This study tested possible mechanisms of the size effect, i.e., bite size and visual cues. A 2 2 repeated measures, within-subject design was used to test the effects of size (410 g vs. 820 g of a pasta dish) and visual cues (blindfolded vs. visible) on energy intake in 30 individuals (15 men, 15 women). At each meal participants were exposed to one of four experimental conditions (small /visible; small /blindfold; large /visible; large /blindfold). Participant characteristics, food intake, number of bites, meal duration, palatability measures and hunger and fullness were assessed. In response to a doubling of the presented, entrée energy intake increased 26% (220 ; P < 0.001) and mean bite size increased 2.4 g/bite (P < 0.05). Overweight (OW) individuals consumed 40% (334 ) more of the entrée in response to the large condition (P < 0.05), while lean individuals intakes did not differ (P < 0.56). A 12% (122 ) decrease in entrée intake was observed in the blindfolded condition (P < 0.01), but no by visual cue interaction was found; indicating that blindfolding did not significantly attenuate the size effect. These data suggest that the size effect is not impacted by removing the visual cue of food and that this effect occurs via changes in bite size in adults. Obesity (2011) 19, doi: /oby Introduction Environmental factors linked to the obesity pandemic have received an increasing amount of attention (1). One of the most studied areas of the eating environment is the effect of varying size on eating behavior. Marketplace food s are consistently larger than in the past and considerably larger than federal standard sizes (2,3). In addition, analyses of household surveys report that individuals are consuming larger sizes at home than they have in the past (4). Numerous studies in the literature have described that increasing size offered increases intake with remarkable consistency in both children and adults (5 16). This size effect has been reported in free-living studies when amounts of food were altered over two and 11 days and up to 1 month without sign of any calorie compensation (8,14,15). Increased intake in response to increase in size has also been demonstrated in acute lab settings when altering the of: amorphous foods, i.e., a pasta and cheese dish (5,6,17), sandwiches (16), liquids (17), and snack foods (13), as well as multiple foods in the same meal (10). Despite this attention, the mechanisms underlying the size effect are poorly understood. It has been reported that, in children, the increase in intake in response to an increase in sizes was a result of larger bite size (6), however, this has not been studied in adults (8,15,18). Fisher et al. suggested that the size served as a visual reference for determining bite size (6). Visual presentation of food is highly influential for meal initiation, amount consumed, and meal termination. When keeping the visual cue of food constant throughout a meal by having participants eat from covert self-refilling soup bowls, individuals unknowingly consumed 73% more soup (19). However, when removing the visual cue of food all together via blindfolding, Rooth et al. reported that consumption decreased 22% in lean and 24% in obese individuals without a change in eating rate or satiety (20,21). These data suggest that individuals may determine the amount of food to consume, at least in part, by visual cues. We propose that the visibly detectable amount of empty plate (i.e., plate space) serves as a cue for meal termination by indirectly providing the eater with information about how much has been consumed, and that this occurs independently of the size presented. Thus, when a large amount of food is offered, more would have to be consumed before the specific amount of plate space/residual food is reached and signals meal termination. The aims of the present study are to examine the effects of removing of the visual cue of plate space on the size 1 Department of Pediatrics, Section of Nutrition, University of Colorado, Denver, Colorado, USA; 2 Oregon Research Institute, Eugene, Oregon, USA; 3 Department of Public Health, Center for Obesity Research and Education, Temple University, Philadelphia, Pennsylvania, USA; 4 Department of Food Science and Human Nutrition, Colorado State University, Fort Collins, Colorado, USA. Correspondence: Susan L. Johnson (Susan.Johnson@ucdenver.edu) Received 9 February 2010; accepted 26 August 2010; published online 14 October doi: /oby VOLUME 19 NUMBER 3 march

2 effect and energy intake and determine whether adults bite size relates to the amount of food presented. We hypothesize that: (i) individuals will consume more when offered a large of food compared to when offered a smaller, but the removal of the visual cue of food (and plate space) by blindfolding will attenuate the effect of the size on energy intake; and (ii) individuals bite sizes will significantly increase when given the large. Methods and Procedures Design and participants A 2 2 repeated measures, within-subject design was used to test the interaction between size (small ; SP vs. large ; LP) and visual cues (visible; SEE vs. blindfolded; BLD) on energy intake at a meal. Each participant was exposed to four experimental conditions (SP/BLD, SP/SEE, LP/BLD, LP/SEE). The order of experimental conditions was randomized across the sample. All 24 possible orders were used before repeating any specific order. Demographics, reported eating behaviors, height, weight, food intake, number of bites, meal duration, palatability measures and hunger, and fullness were assessed. It was calculated that 24 participants were needed to achieve power >0.85. This was based on previous literature to determine: an effect of size based on an effect size calculated from Rolls et al. (12) and an effect of blindfolding (21). Thirty individuals were recruited via flyers or from a university distribution list. Inclusion criteria included being: between the ages of years; a willingness to eat the foods offered in the study, and the ability to read and understand the English language at a 6th grade level. Exclusion criteria included, pregnancy, restrictive dietary practices (e.g., vegetarianism or food allergies), taste, or visual impairment that could interfere with data collection. The participants were not told the purpose of the study, but were told that the aim was to investigate the effects of visibility on sensory aspects of food intake (i.e., taste and mouth feel). Participants were compensated for their time and debriefed after the study. All methods and procedures were approved by the Colorado Multiple Institutional Review Board. Pilot testing and experimental menu Two sessions of pilot testing, in separate samples, were completed to determine the flavor of pasta dish and the sizes to be offered. A total of 15 individuals (M = 8, F = 6) tasted a small amount of each of the test four foods and rinsed with water between each taste. Preference was rated (Likert scale, 1 5) anchored by 1 = this food is not pleasant at all to 5 = this food is extremely pleasant and then the participant ranked the four test foods. A Three Cheese Italiano (Kraft, Glenview, IL; energy density 3.1 /g) had the highest preference rating (mean 3.4; range 2 5), was ranked in the top two more frequently than the other flavors (8 of 15), and was chosen as the main dish. Offering altered sizes in food intake studies presents the challenge of floor and ceiling effects. Our aim was to present a small that the majority of individuals would not finish and a large that would not appear overwhelmingly large. We also desired to have a visually detectable difference between the two s. Through pilot testing of different sizes ranging from 400 g to 1,000 g (based on amounts of a similar food presented in a study conducted by Rolls et al. (12)), it was determined that the majority of people would not consume all of a 410 g. The large was then determined by doubling the small (small 410 ± 10 g, 1,255 ; large 820 ± 10 g, 2,509 ). In addition to the entrée, complementary foods were offered with each meal including fixed s of baby carrots (100 ± 5 g, 35 ), sliced apples (130 ± 10 g, 68 ) and water (800 g). The amounts of complementary foods offered (103 ) were fixed across all conditions. Nutritional information was obtained from the manufacturer to determine energy intake. The total energy (SP entrée = 1,255, LP entrée = 2,509 and complementary foods = 103 ) offered in the study sessions was: 1,358 and 2,612 for the small and large conditions, respectively. Pre and postweights to the nearest 0.1 g served as measure of consumption in grams. Study session procedures Participants were instructed to have a typical breakfast on study session days. Each participant came to the Children s Eating Lab at the University of Colorado on four different occasions separated by >4 days. Participants came at the same lunch time for their study sessions. After consent was completed and all questions regarding the study were answered, participants filled out a series of premeal visual analog scales (VAS). Pre and postmeal VAS were used to rate the participants hunger, thirst, and fullness using a mm scale, anchored by not at all and extremely. The participants were then presented with a meal and the blindfold (on blindfolded condition days) and were instructed to eat ad libitum. A standardized tray, bowl, spoon, cup, and containers were used, making the size of the entrée the only varied parameter in the food presentation. Apples were cut into eight pieces and presented in a square plastic container, carrots were presented in round plastic container, water in a plastic cup with a lid and straw and the entrée was offered in a bowl with a spoon. In the blindfolded conditions participants saw the foods prior to eating and then were blindfolded for the duration of their meal. This accomplished three things: (i) it allowed the participant to initially view the different sizes presented; (ii) it eased any reservations the participants had about eating something they could not see and; (iii) it denied the participant the ability to see the amount of food remaining during consumption. Participants were instructed to indicate finishing the meal either by pushing the tray away or removing the blindfold. Participants then completed postmeal VAS which consisted of the same three scales completed premeal and four additional VAS to assess the participants ratings for: overall pleasantness of taste of the entrée, and saltiness, sweetness, and how fatty the entrée tasted (0 100 mm). At the end of the first session, height (to the nearest 0.1 cm) and weight (to the nearest 0.1 kg) were assessed on site with a calibrated scale and stadio meter. No data were collected near any celebration or holiday that involved food or an event that could significantly impact energy balance. On a date prior to the meal sessions, the Three Factor Eating Questionnaire (22) was completed to identify any participants with highly restrictive and/or disinhibited eating patterns. We were primarily interested in dietary restraint (scale range: 0 21) and dietary disinhibition (scale range 0 16) sub-scales. The reliability and validity of the Three Factor Eating Questionnaire and its sub-scales have been established in wide variety of nutrition-based studies (23). Behavioral observations and discharge Meal duration was recorded in (min:s) from the first bite or drink to the time when the participant indicated completion of the meal. One member of the research staff recorded number of bites of the entrée via direct observation behind a two-way mirror at every session. A bite was operationally defined as the point when the spoon transferred food to the mouth. Throughout the study two coders were used. These coders trained together and participated in reliability assessments (r = 0.99) during catch trails. This method of assessing number of bites using direct observation in a one-on-one setting has been used previously (24). Any comments made by the participant were recorded by research staff throughout the study session. An informal discharge interview was performed at the end of the last study session. Participants were queried regarding their thoughts about the purpose of the study, whether they noticed differences in the meal between study sessions and their experiences related to eating blindfolded. Statistical analyses BMI of each participant was calculated, and weight status was dichotomized into lean (BMI <24.99) and overweight (OW; BMI >25). Bite size was calculated by total grams of the entrée consumed obesity VOLUME 19 NUMBER 3 march

3 divided by the total number of bites. Bite frequency was calculated as the sum of the total bites recorded divided by meal duration recorded. Statistical analyses were performed using SAS version 9.1 (2003; SAS Institute, Cary, NC). Descriptive statistics were performed on all data including means, standard deviations and standard error of the means as well as tests of normality of distribution. Exploratory and graphical methods were used to examine data for outliers or other abnormalities. Independent measures t-tests were used to compare participant characteristics (e.g., age, education, dietary disinhibition, and restraint) across sex and weight status. Mixed linear model analyses (PROC MIXED, 2003; SAS Institute, Cary, NC) were conducted to assess main effects ( size, blindfolding, sex, and weight status) and interactions (e.g., size blindfolding; size sex; size weight status; weight status sex; blindfolding size sex). Additionally analyses were performed to test for possible effects of order independent of conditions, no significant effects were observed. Hypothesized covariates and interactions were entered into the models and removed if P > All tests were two-sided with significance levels set at P < In testing the effect of size on intake, consuming all of the entrée (plate cleaning) can skew data, inflating the effect of the increase in. Our study included three steps to account for the effect of plate cleaning: pilot testing of the sizes, operationally defining a plate cleaner and completing an analysis to determine whether a plate cleaner size interaction existed. Based on previous literature (12) a participant was defined as plate cleaner if they left 20 g of the entrée in both of the small conditions (blindfolded and visible). Results Participants A total of 30 individuals (M = 15, F = 15) completed the study, and three men (BMI = 31.3 ± 4.4) were identified as plate cleaners. In addition to consuming all of the small s, one of these men left 20 g of the large entrée in the blindfolded condition. No participant left 20 g of the large entrée in the visible condition. A plate cleaner size interaction was observed (P < 0.001). The plate cleaners had a significantly larger response to the increase in size suggesting that they would have possibly continued to eat in the small condition if there was more food available. Because the plate cleaners were restricted by the amount of food presented in the small conditions and likely were not able to eat until full, their response to size was inflated, thus skewing the data and they were eliminated from further analyses. All subsequent Table 1 Sample characteristics Men (n = 12) Women (n = 15) Total (n = 27) Age (years) 37.7 ± ± ± 11.1 Height (cm) ± 6.1 a ± ± 11.9 Weight (kg) 88.3 ± 12.7 a 67.4 ± ± 15.9 BMI (kg/m 2 ) 26.6 ± ± ± 4.5 Education (years) 16.3 ± ± ± 1.5 Dietary restraint b 7.8 ± 4.3 c 11.4 ± ± 4.9 Dietary disinhibition b 4.1 ± 2.1 c 6.2 ± ± 2.4 Data are presented as means ± s.d. a Significant differences between men and women (P < 0.001). b Three Factor Eating Questionnaire (22). c Trending differences between men and women (P < 0.10). results presented are on the remaining 27 participants (1 Hispanic, 1 Asian, 25 nonhispanic whites). Participants were OW, in their mid thirties and reported high education and low to moderate levels of dietary restraint and disinhibition (Table 1). A total of 13 individuals (five men, eight women) were classified as lean (BMI 24.9) and the remaining 14 individuals (seven men, seven women) were classified as OW (BMI 25). As expected, OW individuals had a significantly higher BMI (29.2 ± 3.7 vs ± 1.7; P < 0.001) and also higher dietary disinhibition scores (6.7 ± 2.8 vs. 3.7 ± 2.7; P < 0.05). OW individuals also had a trend toward higher dietary restraint than had lean individuals (11.4 ± 6.1 vs. 8.1 ± 2.6; P = 0.08). Entrée intake A significant size main effect was observed (Figure 1a; P < 0.001), in which entrée intake increased 26% (220 ; 71.9 g), independent of visibility condition, weight status, and sex. A weight status interaction was also observed (Figure 1b; P < 0.05). OW individuals consumed 40% (334 ; 109 g) more of the entrée in response to the large condition (Figure 2a; P < 0.001) while lean individuals intakes did not differ (Figure 2b; P = 0.56). A 12% (122 ; 31.8 g) decrease in entrée intake was observed in the blindfolded condition (P < 0.01), independent of size, sex, and weight status. No by visual cue interaction was found (P = 0.29); indicating that blindfolding did not attenuate the size effect. An overall sex effect was observed in which men ate 188 (61.6 g) more, independent of size, visibility condition, and weight status (lean and OW; P < 0.05). a 1,300 1, b ** 1,300 * Visible 1,100 Blindfolded 900 Figure 1 Entrée energy intake adjusting for covariates (mean ± s.e.m.). *P < 0.05, **P < (a) resulted in significantly higher energy intake (P < 0.001). (b) Significant weight status by interaction (P < 0.05). No by visibility interaction was observed (P = 0.29). a 1,300 ** 1, OW visible OW blindfolded 1,300 1, OW Lean Lean visible Lean blindfolded Figure 2 Entrée energy intake adjusting for covariates by weight status (mean ± s.e.m.). *P < 0.05, **P < (a) resulted in significantly higher entrée energy intake in overweight (OW) individuals (n = 14; P < 0.001). (b) Lean individuals entrée intake did not differ by (P = 0.56). b 548 VOLUME 19 NUMBER 3 march

4 Complementary foods and total meal intake Significant main effects for size and visibility conditions as well as sex were observed. Specifically, total energy intake at the meal increased 220 in response to the large (P < 0.001; Figure 3), blindfolding resulted in a 102 decrease in total meal intake (P < 0.01) and men consumed significantly more at the meal than women (217 ; P < 0.01). Energy intake of the complementary foods did not significantly differ by weight status (lean 84.0 and OW 78.2 ; P = 0.17). No significant differences of intake in carrots (P = 0.84), apples (P = 0.54) and water (P = 0.89) were observed across the four experimental conditions (SP/BLD, SP/SEE, LP/ BLD, LP/SEE). However, main effects for sex were noted where men consumed more apples (118.2 ± 21.4 g vs ± 37.9 g; P < 0.01), carrots (79.7 ± 25.7 g vs ± 34.3 g; P < 0.001), and water (393.6 ± vs ± 202.3; P < 0.05) than women. In addition, lean individuals consumed more apples than their OW counterparts (115.9 ± 26.4 vs ± 36.6; P < 0.05). Bite size and meal duration No differences in number of bites taken across the four experimental conditions were observed (mean range: 22.4 ± ± 14.4; P = 0.89), by weight status (P = 0.51) or by sex (P = 0.98). Mean bite size increased 2.4 g/bite in the large condition (P < 0.05; Figure 4) and blindfolding resulted in a 2.3 g/bite decrease in size (P < 0.05; Figure 4). Men took significantly larger bites than women (16.6 ± 7.3 compared with 13.2 ± 7.6 g/bite; P < 0.01). Portion size did not significantly interact with blindfolding for either number of bites (P = 0.78) or bite size (P = 0.88). In addition, number of bites was not associated with sex (P = 0.98) or weight status (P = 0.52). Mean meal duration for all participants was 15:30 ± 4:06; (min:s). No differences were observed in meal duration across any of the experimental conditions or by sex or weight status (P = ). Visual analog scales and participant comments Pre and postmeal hunger, thirst, and fullness, as well as palatability measures (pleasantness of taste, saltiness, sweetness, and how fatty the entrée tasted) did not vary across any of the four experimental conditions (P = ; Table 2). Men 1,400 1,200 ** 1, Complimentary foods Entrée Figure 3 Mean (±s.e.m.) entrée and complimentary energy intake during the large and small meals. **Significantly higher total intake (P < 0.001). reported higher premeal hunger (69.4 ± 12.0 vs ± 24.2; P < 0.01) and OW individuals reported being less full postmeal (77.4 ± 13.8) compared to lean individuals (83.0 ± 8.3; P < 0.05). The majority of the participants noticed the difference in size, yet no participant was able to deduce the purpose of the study. When asked about eating blindfolded, participants reported that it felt a little unusual at first, but that they found it easier than they had anticipated; many participants reported it as being fun. No participant felt that blindfolding significantly impacted their intake, nor did any participant spill. Discussion The primary aim of this study was to investigate the mechanisms underlying size effects on food intake. In concert with previous research, we demonstrated that an increase in size results in increased intake (12,14,16). Our data suggests that the removal of the visual cue of the plate space during the meal failed to attenuate the participant s response to the increase in size. We propose that plate space might serve as a proxy of how much one has eaten, but the amount of plate space visible does not appear to prompt meal termination. This is consistent with data reported that plate size (and thus amount of plate space visible) did not impact energy intake (25). The retention of the size effect despite blindfolding does, on the other hand, suggest that the initial visual cue of the amount of food presented may be powerful enough to influence how much is eaten. Data presented in the current study suggest that there is a possible interaction between the visual cue and bite size which ultimately influences food intake. Our data are similar to those reported by Fisher et al. where increases in size have been shown to increase children s bite size (6) and a large bite size increased intake in adults (24). Increases in bite size in response to increases in size suggest a different explanation as to why the size effect remained despite participants being blindfolded. One premise relates to physical property of having more food available relative to the spoon. The spoon may have greater probability of capturing more food simply because of the volume of food it is Bite size (g/bites) * * Blindfolded Visible Figure 4 Mean (±s.e.m.) bite size by and visibility conditions. *Bite size significantly in the large and visible conditions (P < 0.05). obesity VOLUME 19 NUMBER 3 march

5 Table 2 Ratings of hunger, satiety, and palatability by experimental condition SP/SEE LP/SEE SP/BLD LP/BLD Overall mean Premeal hunger 61.3 ± ± ± ± ± 20.7 Premeal thirst 61.6 ± ± ± ± ± 18.4 Premeal fullness 22.8 ± ± ± ± ± 15.9 Postmeal hunger 9.8 ± ± ± ± ± 8.6 Postmeal thirst 16.5 ± ± ± ± ± 15.4 Postmeal fullness 79.6 ± ± ± ± ± 11.9 How pleasant did How fatty did How salty did How sweet did 56.9 ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± 23.1 Data are presented as mean ± s.d. There were no significant differences in means among conditions. BLD, blindfolded; LP, large ; SEE, visible; SP, small. scooping from is greater. It has been previously reported that when self-serving, an increase in serving spoon size leads to an increase in food taken (26), suggesting that spoon size can influence the amount scooped. Whereas our data suggest that the volume of food presented affects the amount scooped and independent of visibility. It is also possible that individuals, when blindfolded, used the spoon to feel how much food was left and estimate how much they had eaten. However, it was observed that individuals in the blindfolded conditions generally ate from the same area of the bowl creating a one large divot in the entrée, suggesting that they did not feel around the bowl to determine the amount of residual food. The present investigation is a first step in systematically uncovering the mechanisms that are responsible for the effects of size on eating behavior. Based on the data presented here, the authors posit two means by which size effect may occur: (i) an intuitive cognitive process where the initial visual cue of the size influences the amount ingested; and (ii) a passive mechanical process where the physical amount of food available impacts the bite size. While these are two very different mechanisms they are both testable hypotheses. The former of the two, could be tested using a neuroimaging paradigm to assess regional brain activity in response to pictures of foods varied by. We have previously found, when compared to lean individuals, OW individuals rate large s of food (relative to small s of the same food) higher for wanting to eat (K.S. Burger, M.A. Cornier, J.P. Ingebrigtsen and S.L. Johnson, unpublished data). Given this information, we would hypothesize that OW individuals, when exposed to pictures of large s of food, would have increased brain activity in somatosensory, reward, and attention related regions of the brain. This type of study would allow for the assessment of brain reactivity to a wide variety of visual food stimuli without the confounding effects of taste and satiety. The passive mechanical process influencing bite size is also a readily testable hypothesis. Loading studies could be designed to assess amounts taken on utensils in response to different s or, in contrast, studies with predetermined bite size could be conducted. The former paradigm would directly address the mechanical aspects of bite size and without the confounding effects of taste and satiety. Given the effect of size has on intake has been observed in foods that do not involve utensils (7,13 16), it is unlikely that the mechanical process is solely responsible for the size effect. It is more reasonable to suggest that these two processes are additive in influencing intake with additional factors that cannot be addressed using results from the present study design. In addition to the processes and studies proposed above which assess how visual and mechanical aspects of eating interact with size, additional information could be gathered by studies that directly assess individuals opinions and attention to their eating process (e.g., what is an individual thinking when they are given a large or small? How does that opinion change during and after the meal?). Other factors likely to interact with size and leading to an influence on intake are: the physical state of the food (e.g., large solid, smaller solid units, amorphous, liquid), presentation style (e.g., plate, bowl, self-serve), method of delivery (e.g., hands, spoon, fork, drinking), perception of cost (monetary or calorie), and social environment, among others. The majority of these factors are largely unstudied under the context of the influence of size and intake. Investigations could be designed to systematically study each of these factors to better uncover the mechanisms underlying the size effect. With a better understanding of why size influences intake, an accurate conceptual model could be formulated then serve as a foundation for providing evidenced-based obesity prevention programs, weight loss interventions, and interventions to increase consumption of fruits and vegetables. Contrary to the present study s findings, previous similarly designed studies have not reported a relationship between weight status and response to size. The foremost 550 VOLUME 19 NUMBER 3 march

6 possibility for this difference is that the present study recruited a sample with a nearly equal number of OW and lean individuals, who had considerably different BMIs. Lean individuals did respond to the increase in size similarly to previous literature (12), although in the current study this was not statistically significant. In a previous study performed by Rolls et al. the s presented were differed by sex (14). While different energy requirement exist between men in women, food s presented in the eating environment are fixed i.e., not varied by sex. The present study is an acute assessment of eating in a lab setting, which is more similar to Rolls et al. (2002), where the sizes were not varied by sex (12). In conclusion, this study has provided evidence that OW adults appear to be more susceptible to the effects of increases in size than their lean counterparts. The present investigation has provided insight regarding the mechanisms underlying the size effect in adults, i.e., increases in bite size. Why the phenomena of increased intake and increased bite size occur is less clear. We propose that these results can be interpreted to suggest that: (i) the initial visual cue of the amount presented may influence intake independent of plate space, and/or (ii) the physical amount of food effects how much is transferred to the spoon for a bite. Solely based on the data here, one could suggest that limiting the initial size and/or using smaller utensils could lead to a decrease in energy intake. Additional research is needed to study these concepts and determine how this can affect long term intake and weight regulation. Acknowledgments This work was supported by the Helen F. McHugh Graduate Research Fellowship, Colorado State University and the National Research Initiative of the USDA Cooperative State Research, Education and Extension Service, grant number # We thank Gary Grunwald his input in data analysis and interpreting results and Sarah Zellmer, Adrienn Hollonds, Richard Boles and the other members of the Children s Eating Lab for their assistance in the data collection and manuscript preparation. Disclosure The authors declared no conflict of interest The Obesity Society REFERENCES 1. Hill JO, Peters JC. Environmental contributions to the obesity epidemic. Science 1998;280: Young LR, Nestle M. The contribution of expanding sizes to the US obesity epidemic. Am J Public Health 2002;92: Nielsen AJ, Popkin BM. Patterns and trends in food sizes, JAMA 2003;4: Smiciklas-Wright H, Mitchell D, Mickle S, Goldman J, Cook A. Foods commonly eaten in the U.S., and : Are sizes changing? J Am Diet Assoc 2003;1: Fisher JO, Liu Y, Birch LL, Rolls BJ. Effects of size and energy density on young children s intake at a meal. Am J Clin Nutr 2007;86: Fisher JO, Rolls BJ, Birch LL. Children s bite size and intake of an entree are greater with large s than with age-appropriate or self-selected s. Am J Clin Nutr 2003;5: Flood JE, Roe LS, Rolls BJ. The effect of increased beverage size on energy intake at a meal. J Am Diet Assoc 2006;106: ; discussion Jeffery RW, Rydell S, Dunn CL et al. Effects of size on chronic energy intake. Int J Behav Nutr Phys Act 2007;4: Kral TV, Roe LS, Rolls BJ. Combined effects of energy density and size on energy intake in women. Am J Clin Nutr 2004;79: Levitsky DA, Youn T. The more food young adults are served, the more they overeat. J Nutr 2004;134: Raynor HA, Wing RR. Package unit size and amount of food: do both influence intake? Obesity (Silver Spring) 2007;15: Rolls BJ, Morris EL, Roe LS. Portion size of food affects energy intake in normal-weight and overweight men and women. Am J Clin Nutr 2002;76: Rolls BJ, Roe LS, Kral TV, Meengs JS, Wall DE. Increasing the size of a packaged snack increases energy intake in men and women. Appetite 2004;42: Rolls BJ, Roe LS, Meengs JS. r sizes lead to a sustained increase in energy intake over 2 days. J Am Diet Assoc 2006;106: Rolls BJ, Roe LS, Meengs JS. The effect of large sizes on energy intake is sustained for 11 days. Obesity (Silver Spring) 2007;15: Rolls BJ, Roe LS, Meengs JS, Wall DE. Increasing the size of a sandwich increases energy intake. J Am Diet Assoc 2004;104: Diliberti N, Bordi PL, Conklin MT, Roe LS, Rolls BJ. Increased size leads to increased energy intake in a restaurant meal. Obes Res 2004;12: Kelly MT, Wallace JMW, Robson PJ et al. Increased size leads to a sustained increase in energy intake over 4d in normal-weight and overweight men and women. Br J Nutr 2009;3: Wansink B, Painter JE, North J. Bottomless bowls: why visual cues of size may influence intake. Obes Res 2005;13: Linné Y, Barkeling B, Rössner S, Rooth P. Vision and eating behavior. Obes Res 2002;10: Barkeling B, Linné Y, Melin E, Rooth P. Vision and eating behavior in obese subjects. Obes Res 2003;11: Stunkard A, Messick S. The three factor eating questionnaire to measure dietary restraint, disinhibition, and hunger. J Psychosom Res 1985;1: Williamson DA, Martin CK, York-Crowe E et al. Measurement of dietary restraint: validity tests of four questionnaires. Appetite 2007;48: Spiegel TA, Kaplan JM, Tomassini A, Stellar E. Bite size, ingestion rate, and meal size in lean and obese women. Appetite 1993;21: Rolls BJ, Roe LS, Halverson KH, Meengs JS. Using a smaller plate did not reduce energy intake at meals. Appetite 2007;49: Wansink B, van Ittersum K, Painter JE. Ice cream illusions - Bowls, spoons, and self-served sizes. Am J Prev Med 2006;3: obesity VOLUME 19 NUMBER 3 march