Volume of food consumed affects satiety in men

Abstract

This study tested the hypothesis that the amount (weight or volume) of food consumed affects the satiating potency of a food, independent of its energy content. Normal-weight young men (n = 20) were tested in a within-subjects design. Subjects were served a milk-based drink or no drink (control), followed 30 min later by a self-selected lunch and > 4 h later by a self-selected dinner. Milk drinks were equal in energy content (2088 kJ, or 499 kcal) and had similar proportions of fat (30.3%), carbohydrate (54.7%), and protein (15%) across three volumes: 300, 450, and 600 mL. Ratings of palatability, sensory properties, and energy content of the drinks and of hunger completed before consumption of the preloads were not significantly different among conditions. The results showed that preload volume affected energy intake at lunch (P < or = 0.009) such that energy intake was less after the 600-mL preload than after the 300-mL preload. This effect was still present when energy intake at dinner was included (P < or = 0.022). At lunch, including energy from the preload, subjects overate relative to the control condition (4323 +/- 322 kJ) after the 300- (5263 +/- 321 kJ) and 450-mL (5011 +/- 300 kJ) preloads but not after the 600-mL (4703 +/- 353 kJ) preload. Thus, the best adjustment for the energy in the preloads was with the largest, least energy-dense drink. Consistent with the effects on intake, the volume of the drinks affected ratings of hunger and fullness. These results indicate that the volume consumed is an important determinant of satiety after milk drinks under these conditions.

Full text

ABSTRACT This study tested the hypothesis that the
amount (weight or volume) of food consumed affects the satiat-
ing potency of a food, independent of its energy content. Nor-
mal-weight young men (n = 20) were tested in a within-subjects
design. Subjects were served a milk-based drink or no drink
(control), followed 30 min later by a self-selected lunch and
>4 h later by a self-selected dinner. Milk drinks were equal in
energy content (2088 kJ, or 499 kcal) and had similar propor-
tions of fat (30.3%), carbohydrate (54.7%), and protein (15%)
across three volumes: 300, 450, and 600 mL. Ratings of palata-
bility, sensory properties, and energy content of the drinks and of
hunger completed before consumption of the preloads were not
significantly different among conditions. The results showed that
preload volume affected energy intake at lunch (P ≤0.009) such
that energy intake was less after the 600-mL preload than after
the 300-mL preload. This effect was still present when energy
intake at dinner was included (P ≤ 0.022). At lunch, including
energy from the preload, subjects overate relative to the control
condition (4323 ± 322 kJ) after the 300- (5263 ± 321 kJ) and
450-mL (5011 ± 300 kJ) preloads but not after the 600-mL (4703
± 353 kJ) preload. Thus, the best adjustment for the energy in the
preloads was with the largest, least energy-dense drink. Consis-
tent with the effects on intake, the volume of the drinks affected
ratings of hunger and fullness. These results indicate that the vol-
ume consumed is an important determinant of satiety after milk
drinks under these conditions. Am J Clin Nutr 1998;
67:1170–77.
KEY WORDS Energy density, humans, food intake, hunger,
satiety, volume, Eating Attitudes Test, Beck Depression Invento-
ry, Zung Self Rating Questionnaire
INTRODUCTION
The high incidence of obesity in affluent societies indicates that
many individuals are consuming more energy than they are
expending. Food intake is affected by a variety of factors, includ-
ing environmental influences and those related to the psychologic
or physiologic characteristics of the individual. In particular,
attributes of the available food, such as palatability (1), energy or
macronutrient content (2), or energy density (3), can affect intake.
We showed previously in a series of studies that some people,
especially lean young men, can adjust their intake in response to
the energy content of a preload that is either ingested (4, 5) or
infused intragastrically (6) before a meal. The type of macronu-
trients, ie, fat or carbohydrate, in the preload did not affect sub-
sequent energy compensation, a measure of satiety, in the lean
men (2, 4–6). In these experiments, the sensory properties and the
weight or volume of the preload were held constant while the
energy or macronutrient content, or both, varied. The purpose of
the present study was to explore further the properties of food that
can affect satiety by keeping the sensory properties and the
energy and macronutrient contents of preloads constant and sys-
tematically varying the volume or weight consumed.
The volume consumed could affect satiety through mechanore-
ceptors or chemoreceptors in the oropharyngeal or gastrointesti-
nal tracts (7). Additionally, the volume of a preload could affect
satiety by affecting the perception of how much has been con-
sumed (3). People may equate portion size with energy content
and adjust subsequent intake accordingly. The present study was
the first to test directly the effects of volume consumed on satiety
in humans when the energy and macronutrient contents and the
sensory properties of the preloads are similar.
SUBJECTS AND METHODS
Design
The experiment used four conditions in a counterbalanced,
within-subjects design. Subjects came to the laboratory on 4 sep-
arate days to eat breakfast, lunch, and dinner. On three of the
days, subjects received a milk-based preload before lunch and on
1 d no preload was served. Test days were separated by ≥1 wk.
Subjects
We recruited participants through advertisements in local and
university newspapers, posters, and mailings. Potential subjects
Volume of food consumed affects satiety in men
1–3
Barbara J Rolls, Victoria H Castellanos, Jason C Halford, Arun Kilara, Dinakar Panyam, Christine L Pelkman,
Gerard P Smith, and Michelle L Thorwart
1
From the Nutrition Department and the Department of Food Science,
Pennsylvania State University, University Park; the Department of Dietetics
and Nutrition, Florida International University, Miami; the Department of
Psychology, University of Central Lancashire, Preston, United Kingdom; and
the Edward W Bourne Behavioral Research Laboratory, New York Hospi-
tal–Cornell Medical Center, White Plains, NY.
2
Supported by NIH grants DK50156 and DK08926.
3
Address reprint requests to BJ Rolls, Nutrition Department, 226 Hender-
son Building, Pennsylvania State University, University Park, PA
16802–6501. E-mail: bjr4@psu.edu.
Received October 30, 1997.
Accepted for publication December 15, 1997.
Am J Clin Nutr 1998;67:1170–77. Printed in USA. © 1998 American Society for Clinical Nutrition
1170
by guest on July 14, 2011www.ajcn.orgDownloaded from

were initially screened through a telephone interview to deter-
mine whether they ate breakfast regularly, were regular milk
drinkers, did not smoke, did not have any food allergies or
restrictions, were not athletes in training, were not dieting to
gain or lose weight, and were not taking any medications or
dietary supplements that could affect appetite. Potential subjects
completed the following questionnaires in our laboratory: the
Eating Attitudes Test (EAT; possible score: 0–140), which
detects symptoms of an eating disorder (8); the Eating Inventory
(EI; 9), which measures dietary restraint (possible score: 0–21),
perceived hunger (possible score: 0–14), and disinhibition (pos-
sible score: 0–16); the Beck Depression Inventory (10; possible
score: 0–63) and the Zung Self Rating Questionnaire (11; possi-
ble score: 20–80), both of which detect depression; a detailed
demographic questionnaire; and a family weight-history ques-
tionnaire. Weight, height, waist, and hip measurements were
taken at this time. Subjects were included in the study only if
their body mass index (BMI; in kg/m
2
) was between 20 and 25.
Subjects were excluded if they scored > 40 on the Zung Self Rat-
ing Questionnaire, > 10 on the Beck Depression Inventory, > 30
on the EAT, or > 10 on the cognitive restraint subscale of the EI.
Finally, subjects were excluded from the study if they disliked
two or more of the foods to be served in any of the test meals.
All aspects of the study were approved by the Institutional
Review Board of Pennsylvania State University.
Procedures
Subjects were asked to keep their evening meals and activity
levels on the day before each test day as similar as possible. Sub-
jects were also asked to refrain from drinking alcohol on the day
before each test day and throughout each test day. Food and activ-
ity diaries were used to monitor compliance. On each test day, sub-
jects consumed breakfast ≥ 3 h before the start of the lunch meal.
Subjects were instructed to not consume any food or beverages,
except water, in the interval between breakfast and 1 h before their
scheduled lunch. During the hour before their lunch, subjects were
instructed to refrain from eating or drinking any foods, including
water.
At the start of the lunch session, subjects tasted and rated a sam-
ple of the preload or were notified that there would be no milk drink
served on that day. On days when a preload was served, subjects
were instructed to consume the preload over a 15-min period. Preset
timers were given to assist subjects in pacing their consumption of
the preload. Subjects were permitted to read magazines before
lunch, except when consuming the preload. Magazines were
screened to exclude any articles pertaining to food, weight loss, or
body image. Lunch was served 30 min after the presentation of the
preload or 30 min after arrival for the no-preload condition. Finally,
subjects returned to the laboratory for dinner ≥ 4 h after lunch
began. Subjects were asked to not consume any foods or energy-
containing beverages outside the laboratory during the experimental
day until after dinner. During all meals, subjects were tested alone
in private cubicles.
Preloads
Three vanilla-flavored, milk-based drinks were developed for
use as preloads. Initial formulation and testing of the drinks was
conducted in the Department of Food Science at Pennsylvania
State University. After these initial tests, we recruited men and
women (n = 18) to taste and rate samples of the final formulations
in our laboratory. These subjects were excluded from participation
in the current preload study. No differences were found for ratings
of perceived energy content of the food, pleasantness of taste,
creaminess, sweetness, thickness, or how much subjects were will-
ing to consume. The preloads were equal in energy and macronutri-
ent contents but varied in volume (300, 450, or 600 mL) (Table 1).
The preloads were chilled (3 °C, 37 °F) and presented to the sub-
jects in a large, clear glass with a straw.
Test meals
Subjects selected their breakfast foods before the study began.
They chose either bagels (served with an assortment of jelly, jam,
and cream cheese) or cereal (there were a variety of choices) and
their beverages (coffee, tea, orange juice, or milk). Subjects
received the same breakfast at the start of each test day.
Lunch and dinner were individual, buffet-style, self-selected
meals that allowed participants to choose ad libitum from a variety
of meal-appropriate foods. The foods varied in fat, carbohydrate,
and protein contents to allow subjects to vary their energy intake
and proportions of macronutrients. The lunch consisted of sliced
turkey breast, bologna, American cheese slices, bread, lettuce,
tomato slices, potato chips, pretzels, applesauce, and cookies. Din-
ner included cooked pasta shells, meatless spaghetti sauce,
VOLUME AND SATIETY 1171
TABLE 1
Ingredients and macronutrient and energy contents of preloads
Preload condition
1
300 mL 450 mL 600 mL
Ingredients (g)
Skim milk
2
110.0 110.0 110.0
Instant nonfat dry milk
3
20.0 20.0 20.0
Sweet cream, 40%
2
39.5 39.5 39.5
Total milk protein
4
9.0 9.0 9.0
Corn syrup solids (36 DE)
5
30.0 30.0 30.0
Maltose
6
20.5 14.5 8.3
Polydextrose-K
7
0.0 24.0 49.0
Water 100.0 222.0 352.0
Fat
(g) 16.8 16.8 16.8
(% of energy) 30.3 30.3 30.3
Carbohydrate
(g) 68.3 68.3 68.3
(% of energy) 54.7 54.7 54.7
Protein
(g) 18.7 18.7 18.7
(% of energy) 15.0 15.0 15.0
Energy
(kJ) 2088 2088 2088
(kcal) 499 499 499
Energy density
(kJ/g) 6.3 4.5 3.4
(kcal/g) 1.5 1.1 0.8
1
The preloads were pasteurized in a high-temperature, short-time pas-
teurizer at 80
o
C for 25 s and homogenized in a two-stage homogenizer at
6.9 MPa in the first stage and 3.45 MPa in the second stage. In the 450-mL
and 600-mL drinks, 0.02% carrageenan (SeaKem IC 611; FMC Food Ingre-
dients, Philadelphia) was used. Guar gum (Germantown International,
Broomall, PA) was used in the 450- (0.13%) and 600-mL (0.18%) drinks.
2
The Pennsylvania State University Creamery, University Park, PA.
3
Maryland and Virginia Milk Producers Association Inc, Laurel, MD.
4
Milk Pro800; American Dairy Specialties Ltd, Burlington, NJ.
5
Cerestar USA Inc, Hammond, IN. DE, dextrose equivalents.
6
US Biochemical Corporation, Cleveland
7
Litesse (4.2 kJ/g); Pfizer Chemical Division, New York.
by guest on July 14, 2011www.ajcn.orgDownloaded from

breaded chicken fillets, broccoli, rolls, butter, pound cake, choco-
late bars, and mixed fruit cups. At both meals, a variety of condi-
ments and chilled water were served. To avoid the possibility of
subjects eating to “clean their plates,” we presented more food
than they were likely to consume. More information about the
foods offered at breakfast, lunch, and dinner is available from the
corresponding author on request. During all meals, subjects were
instructed to eat as much or as little of any food item as they
desired and to ask for more if desired. All food items were
weighed before and after consumption to obtain the amount con-
sumed to the nearest 0.1 g. Energy and macronutrient intakes were
calculated by using information from the manufacturers and from
Bowes & Church’s Food Values of Portions Commonly Used (12).
Visual analogue scales
Subjects rated their hunger, thirst, nausea, fullness, and
prospective consumption (ie, how much food they thought they
could eat) on visual analogue scales (VAS; in mm) (13). For
example, hunger was rated on a 100-mm line preceded by the
question, “How hungry are you right now?” and anchored on the
left by “not at all hungry” and on the right by “extremely hun-
gry.” Other anchors consisted of the phrases “not at all...” and
“extremely...” combined with the adjectives “thirsty,” “nause-
ated,” and “full.” Ratings were performed before and after break-
fast, before and after the preload (or at equivalent times in the
no-preload condition), before and after lunch, hourly for 3 h after
lunch, and before and after dinner.
Subjects also rated samples of each preload on 100-mm VAS.
The following attributes were rated: pleasantness of taste, per-
ceived energy content, sweetness, creaminess, thickness, and
prospective consumption. Ratings were performed before con-
sumption of the preload. Subjects were given a 15-mL sample of
the preload to taste and rate. After the ratings were completed,
subjects were served their preload.
Debriefing
Subjects completed a debriefing questionnaire at the end of
the study. They were asked to state the purpose of the study,
whether they noticed any differences between the test days, and
whether there were any factors that affected their responses.
They were also asked whether they noticed any differences in the
milk drinks in terms of color, smell, fat content, energy content,
amount, taste, texture, and other qualities.
Statistical analyses
Data were analyzed by using SAS-PC for Windows (version
6.10; SAS Institute Inc, Cary, NC). One-way analysis of variance
was performed by using the general linear model (GLM) proce-
dure. The percentage of energy from macronutrients was ana-
lyzed by using an equivalent multivariate analysis of variance
procedure with condition entered as a within-subjects factor.
Tukey’s honestly significant difference test was used for post hoc
comparisons of significant effects. Results were considered
significant at the P < 0.05 level. Power analyses were conducted
by using the statistical package nQuery Advisor (version 2.0;
Statistical Solutions, Ltd, Los Angeles) with alpha set to 0.05.
Food intake
Energy intakes (kJ) and amounts consumed (g), with and with-
out water, and the percentage of energy from macronutrients from
lunch and from lunch plus dinner, with and without the preload,
were analyzed. Analyses were also performed excluding the no-
preload (control) condition. The percentage of energy compensa-
tion at lunch was calculated by dividing the energy intake at lunch
in the control condition by the energy intake at lunch (including
energy from the preload) in each of the preload conditions and
multiplying by 100. The percentage energy compensation over
the rest of the day was calculated by dividing energy intake at
lunch and dinner in the control condition by the energy intake at
lunch and dinner in each of the preload conditions and multiply-
ing by 100. For these calculations, values > 100% indicated over-
compensation (undereating) and values < 100% indicated under-
compensation (overeating). Individual t tests were used to test
whether percentage energy compensation scores were different
from 100%. For these tests, results were considered significant at
P < 0.025, based on a modified Bonferroni correction for multi-
ple comparisons.
Satiating efficiency
The satiating efficiency of the preloads based on Kissileff’s
model (14) was also examined. We calculated the negative of the
slope generated by plotting attributes related to preload size (kJ
or g) against the average amount consumed at lunch (kJ or g). A
satiating efficiency of 1 represents a reduction in lunch intake of
1 unit per unit of preload, a satiating efficiency > 1 represents a
reduction in lunch intake of > 1 unit per unit of preload, and a
satiating efficiency < 1 represents a reduction in lunch intake of
< 1 unit per unit of preload. Two attributes of the preloads were
used in these analyses: weight and hypothetical energy content.
The hypothetical energy content of the preload is the estimated
energy content that would result if subjects assumed that a dif-
ferent volume of the same drink was served on each occasion and
thus volume was an indicator of the energy content (ie, that the
energy density of the drinks remained constant). Using the drink
in the condition in which the best energy compensation was
observed (ie, the 600-mL condition), we calculated the hypo-
thetical energy content of the 300-mL drink to be 50% (1044 kJ)
of that of the 600-mL drink, and the hypothetical energy content
of the 450-mL drink to be 75% (1566 kJ) of that of the 600-mL
drink (2088 kJ).
Visual analogue scale ratings
Subjective ratings of hunger, fullness, thirst, nausea, and
prospective consumption made before the preload was consumed
were analyzed to ensure that systematic differences did not exist
between test days. Changes in ratings due to preload consump-
tion were calculated for each participant by subtracting the rat-
ings taken before preload consumption from the ratings taken
immediately after preload consumption (15 min later). A nega-
tive value indicates a decline in these subjective sensations,
whereas a positive value indicates an increase. Ratings of the
preloads, completed before consumption, were analyzed to con-
firm that subjects could not detect differences between preloads.
RESULTS
Subjects
Twenty-four subjects were selected; however, four were dropped
from the study: one subject failed to report to the laboratory on the
first day of the study, one subject did not follow the experimental
procedures, one subject reported gastrointestinal disturbance (unre-
1172 ROLLS ET AL
by guest on July 14, 2011www.ajcn.orgDownloaded from

lated to the study protocol), and one subject reported increased nau-
sea after consuming the preloads. The final sample consisted of
20 lean men aged 22–42 y (mean age: 28 ± 1.5 y). They weighed
75.4 ± 2.2 kg, were 1.8 ± 0.02 m tall, had a BMI of 23.3 ± 0.3, and
had a waist-to-hip ratio of 0.9 ± 0.01. According to the EI, subjects
scored low in restraint (4.9 ± 0.7), perceived hunger (3.8 ± 0.6), and
disinhibition (3.3 ± 0.4). In addition, the EAT indicated no signs of
an eating disorder (mean score: 6.9 ± 0.6). The two tests used to
assess depression indicated low scores: the Zung Self Rating Ques-
tionnaire (mean score: 27.3 ± 1.0) and the Beck Depression Inven-
tory (meanscore: 08 ± 0.2).
Energy intake
Mean energy intakes at each meal are presented in Table 2.
Subjects consumed significantly less energy at lunch after con-
sumption of the preloads than in the no-preload condition
(F
[3,57]
= 19.86, P < 0.0001). When energy from the preloads was
added to lunch intake, there was still a main effect of condition
(F
[3,57]
= 5.48, P < 0.002). Subjects overate, compared with the no-
preload condition (4323 ± 322 kJ), in the 300-mL (5263 ± 321 kJ)
and 450-mL (5011 ± 300 kJ) conditions but not in the 600-mL
condition (4703 ± 353 kJ) (Figure 1). When the control condition
was excluded from the analysis, the volume of the preload signifi-
cantly affected energy intake at lunch (F
[2,38]
= 5.33, P < 0.009)
such that intake was reduced 17.6% more after consumption of the
600-mL preload than after the 300-mL preload. These results indi-
cate that the volume ingested affected satiety.
When energy intakes at lunch and dinner were combined, subjects
consumed significantly less energy after consumption of the preloads
than in the no-preload condition (F
[3,57]
= 15.83, P < 0.0001). When
energy intake from the preload was added to energy intakes at lunch
and dinner, no differences were detected among the four conditions.
When the no-preload condition was excluded from the analysis we
found that subjects ate significantly more at lunch and dinner com-
bined after the 300-mL preload (10309 ± 537 kJ) than after the 600-
mL preload (9410 ± 588 kJ). Because no significant differences in
energy intake were found for dinner when the no-preload condition
was excluded, these results indicate that subjects did not adjust their
energy intake at dinner to compensate for differences among the
three preload conditions at lunch.
Energy compensation after the preloads
At both lunch and dinner, energy compensation after the
300-mL preload was significantly different from energy com-
pensation after the 600-mL preload. At lunch, energy compensa-
tion after the 300-mL (84%) and 450-mL (88%) preloads was
significantly different from 100% (P < 0.008 and 0.02, respec-
tively). However, energy compensation after the 600-mL preload
(96%) was not significantly different from 100%. At dinner,
compensation scores were not significantly different from 100%
after any of the preloads (95%, 99%, and 104% after the
300-mL, 450-mL, and 600-mL preloads, respectively).
Satiating efficiency
The satiating efficiency of preload weight was 0.30 for food
intake at lunch and 0.35 for water intake at lunch. The satiating effi-
ciency for intake of food and water combined was 0.64. Thus,
adjustments in lunch intake for preload weight were not made on a
strictly one-to-one basis. For each gram of preload consumed, sub-
jects reduced their intake of food and water at lunch by
0.64 g. As shown in Figure 2, a satiating efficiency of 1.0 would
result if subjects assumed that changes in energy content of the pre-
loads occurred in direct proportion with changes in volume (ie, vol-
ume cues overrode energy cues) and adjusted their energy intake
VOLUME AND SATIETY 1173
TABLE 2
Intakes of food and water
1
Preload condition
No preload 300 mL 450 mL 600 mL
Breakfast
Energy intake (kJ) 2658 ± 194 2761 ± 181 2827 ± 180 2676 ± 187
Preload
Energy intake (kJ) NA 2088 2088 2088
Amount consumed (g) NA 329 469 618
Lunch
Energy intake (kJ) 4323 ± 322
a
3175 ± 321
b
2923 ± 300
b
2615 ± 353
b,2
Amount consumed (g)
3
588 ± 35
a
452 ± 34
b 4
423 ± 32
b,c,4
366 ± 37
c
Water consumed (g) 552 ± 70
a
496 ± 53
a,b
469 ± 55
a,b
396 ± 57
b,2
Dinner
Energy intake (kJ) 5405 ± 362
a
5046 ± 338
a,b
4752 ± 261
b
4708 ± 300
b
Amount consumed (g)
3
814 ± 48 782 ± 46 744 ± 42 726 ± 45
Water consumed (g) 578 ± 55 578 ± 68 538 ± 52 495 ± 70
Total (preload + lunch + dinner)
Energy intake (kJ) 9728 ± 627 10309 ± 537 9763 ± 518 9410 ± 588
2
Amount consumed (g)
3
1402 ± 76
a
1563 ± 65
b
1635 ± 64
b,c
1711 ± 73
c,2
Water consumed (g) 1131 ± 120
a
1073 ± 118
a
1007 ± 100
a,b
891 ± 123
b,2
Fat (% of energy) 22 ± 1.2 22 ± 0.8 22 ± 0.9 23 ± 1.0
Carbohydrate (% of energy) 65 ± 1.4 64 ± 0.9 65 ± 1.1 65 ± 1.1
Protein (% of energy) 13 ± 0.5 14 ± 0.4 14 ± 0.5 13 ± 0.4
1
Data were analyzed with and without data from the no-preload condition. Values within a row with different superscript letters are significantly
different when models included the no-preload condition, P < 0.05. NA, not applicable.
2
Significantly different from 300 mL when the no-preload condition was excluded, P < 0.05.
3
Amount consumed excluding water.
4
Significantly different from 600 mL when the no-preload condition was excluded, P < 0.05.
by guest on July 14, 2011www.ajcn.orgDownloaded from

accordingly. A satiating efficiency of 0.0 would result if subjects
relied on energy cues only, ie, all preloads would reduce intake at
lunch by the same amount (kJ). A satiating efficiency of 0.54 was
found when the hypothetical energy content of the preload was
regressed against actual energy intake at lunch. Thus, subjects
reduced their energy intake at lunch by 0.54 kJ for every kJ of
hypothetical preload. These results indicate that volume cues par-
tially overrode energy cues.
Macronutrient energy intake and weight of food consumed
There were no significant differences in the macronutrient
composition of lunch and dinner. Consumption of the preload
reduced both food and water intakes at lunch (Table 2). Subjects
consumed less food (g) in the three preload conditions than in
the control condition (F
[3,57]
= 23.5, P < 0.0001). Subjects con-
sumed less food and water at lunch after the 600-mL preload
than after the 300-mL preload. Effects of volume were also noted
when the weight of the preload was added to the weight of food
and water intakes at lunch. Subjects consumed a greater total
weight of food and water in the 450- and 600-mL conditions than
in the control condition, and consumed more food and water (g)
in the 600-mL condition than in the 300-mL condition.
VAS ratings
Subjective sensations
No significant differences were found for subjective ratings of
hunger, fullness, thirst, nausea, or prospective consumption reported
before the preload was served. Thus, subjects did not differ system-
atically across conditions. Differential changes in ratings due to con-
dition were detected in the interval between when the preload and
lunch were served (Table 3). Subjects reported greater reductions in
hunger and prospective consumption and greater increases in full-
ness after consumption of the 600-mL preload than after consump-
tion of the 300-mL preload. These results show a clear effect of vol-
ume consumed on subjective ratings of satiety. No effect of condition
was noted for changes in ratings of nausea. Ratings were low before
preloads were served (between 3.2 and 2.2 mm) and changed little
after the preloads were consumed (Table 3).
Preload ratings
Ratings of the preloads, completed on each test day before
subjects were aware of the volume to be consumed, were not
1174 ROLLS ET AL
FIGURE 1. Mean (± SEM) energy intake at lunch in each of the four
conditions (n = 20). Means with different letters are significantly differ-
ent, P < 0.05.
FIGURE 2. Satiating efficiency of volume consumed. For each
regression line, intake at lunch (kJ) was regressed against the hypotheti-
cal energy content of the preloads, which was calculated on the basis of
the energy content estimates that would result if subjects equated
changes in volume with changes in energy content. Using the 600-mL
drink as a reference, we calculated the hypothetical energy content of the
300 mL drink to be 50% (1044 kJ) of and the 450-mL drink to be 75%
(1566 kJ) of the energy content in the 600 mL drink (2088 kJ). Theoret-
ically, a satiating efficiency of 1.0 would result if lunch intake was
adjusted in direct proportion to changes in volume, and a satiating effi-
ciency of 0.0 would result if no adjustments were made.
TABLE 3
Change in visual analogue scale (VAS) scores after preload consumption
1
Preload condition
No preload 300 mL 450 mL 600 mL
mm
“How hungry are you right now?” 5.1 ± 3.3
a
29.5 ± 4.1
b
217.9 ± 3.4
b,c
228.5 ± 3.4
c
“How thirsty are you right now?” 3.8 ± 3.5
a
211.6 ± 5.2
a,b
212.8 ± 5.0
b
217.8 ± 5.9
b
“How much food do you think you could eat right now?” 2.1 ± 3.4
a
28.6 ± 2.5
a,b
212.5 ± 3.2
b
226.0 ± 3.5
c
“How nauseated are you right now?” 0.2 ± 0.5 1.5 ± 0.8 1.2 ± 1.0 3.5 ± 1.9
“How full are you right now?” 20.4 ± 2.4
a
11.9 ± 4.6
a,b
23.2 ± 3.8
b
39.7 ± 3.8
c
1
Values within a row with different superscript letters are significantly different, P < 0.05. x
–
± SEM. Change determined by subtracting the value before
the preload from the value 15 min after the preload.
by guest on July 14, 2011www.ajcn.orgDownloaded from

significantly different among conditions (Table 4). This indi-
cates that the sensory properties of the milk drinks were well-
matched and confirmed the results of the sensory testing per-
formed during formulation of the drinks.
Power analyses
We conducted analyses based on current data to determine
whether sufficient power was obtained to detect meaningful dif-
ferences in energy intake at lunch. We estimated the power to
detect differences in lunch intake in a situation in which subjects
adjusted their intake in response to a hypothetical preload
amount (Figure 2). The power to detect a difference of 522 kJ
(the hypothetical difference in energy content between preloads)
was 0.74 for comparisons between the 600- and 450-mL condi-
tions and between the 450- and 300-mL conditions. The esti-
mated power to detect a difference of 1044 kJ between the 600-
and 300-mL conditions was 0.99.
We also conducted analyses to ensure that the sample size was
sufficient to detect meaningful differences in ratings of the pre-
loads. We found that differences of 7 mm between the 300- and
450-mL preloads and between the 450- and 600-mL preloads
would result in power estimates ≥ 0.80 for ratings of pleasant-
ness of taste, energy content, creaminess, sweetness, thickness,
and prospective consumption. These results indicate that failure
to detect differences in ratings of the milk drinks was not due to
a lack of statistical power.
Debriefing
Forty percent of the subjects correctly stated that the purpose
of the study was to examine intake after consumption of drinks
differing in volume. Two subjects failed to notice a difference in
the amount that was served. Most subjects (70–90%) reported no
differences in the milk drinks with respect to color, fat content,
temperature, energy content, smell, sweetness, texture, or taste,
confirming the results obtained with VAS ratings.
DISCUSSION
The results show clearly that the weight or volume of a liquid
food affects satiety independent of its sensory properties or its
energy or macronutrient content. Consistent with the effects on
food intake, the volume consumed in the preload affected ratings
of hunger, fullness, and the amount subjects wanted to eat. In this
study we tested only lean, nondieting young men, a group that
was shown in previous studies to be sensitive to the energy con-
tent of preloads (4, 5). However, the visual, cognitive, and physi-
ologic cues from the volume consumed were sufficient to modu-
late the effect of energy content on satiety in these young men.
The three formulations of milk drinks were matched for total
energy content and macronutrient composition (percentage of
energy). The main difference in the drinks was in water content
(Table 1). Because the energy content of the drinks was held con-
stant and the volume was varied by adding water, energy density
(kJ/g) also varied. Thus, we could not separate the effects of vol-
ume from the effects of energy density. This result was obviously
unavoidable because total energy content and energy density
cannot be controlled for simultaneously while varying volume.
Small amounts of carrageenan (0.02% in the 450-mL and 600-
mL drinks) and guar gum (0.13% in the
450-mL drink and 0.18% in the 600-mL drink) were added to
maintain thickness in the more dilute drinks. These small
amounts of thickener were unlikely to have affected satiety (15).
Differences in sensory properties and palatability were not
confounding variables in this study because no significant dif-
ferences in pleasantness of taste, sensory qualities, and prospec-
tive consumption were detected among the three versions of the
drink. Thus, we specifically tested the effects of the volume of
the drinks because the sensory properties, palatability, and
energy and macronutrient contents were controlled for. Further-
more, we showed that the number of subjects tested in this study
was adequate to detect meaningful differences between drinks;
therefore, the failure to detect differences between ratings was
not due to a lack of statistical power.
Several previous studies examined the role of volume or
weight of food consumed in the control of food intake (16–18).
Holt et al (16) compared preloads of isoenergetic servings of
38 different foods and found that serving weight was the factor
most positively correlated with ratings of satiety and subse-
quent food intake in a test meal. However, in another study, De
Graaf and Hulshof (17) investigated the relative effects of
weight and energy content of preloads of dairy products on
intake 2 h after they were consumed and found that the effect
of energy content was greater than that of the weight of the pre-
loads. The results of these studies, although provocative, do not
directly indicate an effect of weight or volume of food on sati-
ety because other qualities of the preloads varied. For example,
in these previous studies the preloads were not matched for
palatability and macronutrient content (16, 17). Also, the test
meal was not given for several hours after the preload, although
one would expect the effects of volume on satiety to peak some
time during the first hour after consumption, when cognitive
cues, oropharyngeal stimulation, and gastrointestinal disten-
sion are greatest (19).
One way in which the volume consumed could affect satiety
is via receptors in the gastrointestinal tract. Several studies in
animals that examined the influences of gastric fullness or dis-
VOLUME AND SATIETY 1175
TABLE 4
Sensory and prospective consumption visual analogue scale (VAS) scores before preload consumption
1
300 mL 450 mL 600 mL
mm
“How pleasant is the taste of this food right now?” 67.5 ± 2.9 66.1 ± 4.3 67.6 ± 4.6
“How many calories do you think this food has?” 59.6 ± 4.1 59.5 ± 4.1 55.7 ± 4.2
“How sweet does this food taste?” 71.6 ± 3.4 70.4 ± 3.7 70.8 ± 4.8
“How creamy does this food taste?” 66.7 ± 3.6 68.5 ± 4.2 68.4 ± 4.5
“How thick is this food?” 57.6 ± 5.2 67.4 ± 3.7 54.4 ± 4.3
“How much of this food could you eat right now?” 49.8 ± 4.1 54.1 ± 5.2 50.6 ± 4.7
1
x
–
± SEM. There were no significant differences.
by guest on July 14, 2011www.ajcn.orgDownloaded from

tension and the modulatory effects of gastric emptying suggest
that mechanical and nutrient stimulation of gastric and postgas-
tric compartments can play an important role in the regulation of
food intake (20, 21). In humans, this hypothesis is supported by
the results of studies in which the level of gastric distension was
manipulated, usually through the use of gastric balloons, and
subsequent intake was examined. The findings showed that
expansion of a gastric balloon to a volume of 400 mL reduced
subsequent food intake (22). It is not clear in humans how sensi-
tive the stomach is to volume cues, although a recent study in
rats showed that the stomach is sensitive to its contents when it
contains only a small proportion of its capacity (23). Future stud-
ies should examine further the role of gastrointestinal factors in
the effects of volume on satiety by intubating directly into the
stomach drinks similar to those consumed in the present study.
It is of interest that in the present study the best compensation
for the energy in the drinks was seen after consumption of the
drink with the largest volume and lowest energy density, ie, the
600-mL preload. There are many possible explanations for this.
One is that the 600-mL drink may have been the one to best con-
form to the participants’ expectations of the energy content.
Another is that the more concentrated drinks of lower volume
(300 and 450 mL) may have been below a threshold required to
stimulate volume receptors. It could be that there is an optimal
energy density of foods for the maintenance of energy balance
and that consumption of foods with higher energy density will
contribute to the development of obesity, whereas those of lower
energy density will aid in weight loss.
Kissileff (14) proposed calculating satiating efficiency as a
means for comparing the satiating potency of different foods.
Numerous attributes of foods that are postulated to affect satiety,
can be evaluated by comparing the magnitude of the chosen
attribute of a preload with intake of a subsequent test meal. We
used this strategy in a previous study to compare the satiating
effects of preloads varying in macronutrient content and found
that fat in yogurt was slightly less satiating than was carbohy-
drate in some individuals (5). However, lean, nondieting young
men compensated remarkably well for the energy content of pre-
loads regardless of the macronutrient composition: satiating effi-
ciency scores were close to 1.0 for both high-fat and high-carbo-
hydrate preloads (5).
Although satiating efficiency is often used to relate energy
content in a preload to energy consumed at lunch (24), it is also
possible to relate other attributes of preloads, such as weight or
the perceived energy content, to lunch intake (14). We calculated
how much energy the subjects should have consumed if the
energy density of the drinks had remained constant and the vol-
ume actually reflected the energy content of the drinks, ie, the
hypothetical energy content. The resulting satiating efficiency
(0.54) indicated that subjects were responsive to volume cues but
not on a one-to-one basis. That is, the reduction in intake was
<50% less than what may have been expected if the subjects
equated the doubling in volume to a doubling in energy content.
These results suggest that for lean, young men, cues related to
volume only partially override cues related to energy content in
the short term. Further research is warranted to investigate how
volume may affect intake in other populations and over longer
time periods. For example, we showed that obese men and
women and individuals highly concerned with diet and body
weight (ie, restrained eaters) are less responsive to the energy
content of preloads than are lean, young men (5). We also
showed that women are highly responsive to cognitive cues
about food (25). Thus, it is conceivable that the satiating effi-
ciency of volume may be greater in these groups because they
have been shown to be less responsive to energy cues than the
young men tested in this study.
The present study was designed to maximize the effects of
volume on satiety rather than to distinguish any separate effects
of visual, cognitive, and physiologic cues related to the volume
of the drinks. Several studies showed that beliefs about the
energy or fat content of a preload can affect satiety (25, 26).
Because we emphasized the visual differences in the drinks by
serving them in clear glasses, both visual and cognitive cues
could have affected the results. We did not ask subjects how
much energy they thought they had consumed in the drinks
because we did not want to draw their attention to the energy
content of the preloads. However, it is likely that they thought
that the volume of the drinks was related to total energy content
because the sensory properties such as thickness and creaminess
were perceived to be similar. Also, when the subjects were asked
to rate the energy content of the drinks based on the 15-mL sam-
ple tasted at the start of lunch, they indicated no significant dif-
ferences between the drinks.
In the present study, we showed that the effect of a given level
of energy in a preload on satiety can be affected by adding water
and thickeners to increase the volume while maintaining the sen-
sory characteristics. The finding that the volume of food ingested
can affect satiety has important health implications. Because of the
high incidence of obesity in the United States, strategies are needed
to help individuals reduce their energy intake. The results from this
study showed that the amount eaten after consumption of isoener-
getic preloads varied with the volume of the preload. In addition,
eating a food of high volume helped to suppress hunger and to
increase fullness. The reduction in energy intake at lunch after the
high-volume preload was not compensated for by an increased
intake at dinner. Clearly, it is of interest to conduct a longer-term
study in which a high-volume, low-energy preload precedes every
meal over a longer time period to determine whether this would be
an effective strategy for weight loss. Increasing the volume con-
sumed by adding water or foods with a high water content, such as
fruit or vegetables, to the diet can be a safe and easy approach to
controlling hunger and reducing food intake.
REFERENCES
1. Rolls BJ. Effects of food quality, quantity, and variety on intake. In:
Marriott BM, ed. Not eating enough. Washington, DC: National
Academy Press, 1995:203–15.
2. Rolls BJ. Carbohydrates, fats, and satiety. Am J Clin Nutr
1995;61(suppl):960S–7S.
3. Poppitt SD, Prentice AM. Energy density and its role in the control
of food intake: evidence from metabolic and community studies.
Appetite 1996;26:153–74.
4. Rolls BJ, Kim S, McNelis AL, Fischman MW, Foltin RW, Moran T.
Time course of effects of preloads high in fat or carbohydrate on
food intake and hunger ratings in humans. Am J Physiol
1991;260:R756–63.
5. Rolls BJ, Kim-Harris S, Fischman MW, Foltin RW, Moran TH, Stoner
SA. Satiety after preloads with different amounts of fat and carbohy-
drate: implications for obesity. Am J Clin Nutr 1994;60:476–87.
6. Shide DJ, Caballero B, Reidelberger R, Rolls BJ. Accurate energy
compensation for intragastric and oral nutrients in lean males. Am J
Clin Nutr 1995;61:754–64.
1176 ROLLS ET AL
by guest on July 14, 2011www.ajcn.orgDownloaded from

VOLUME AND SATIETY 1177
7. Read N, French S, Cunningham K. The role of the gut in regulating
food intake in man. Nutr Rev 1994;52:1–10.
8. Garner DM, Garfinkel PE. The Eating Attitudes Test: an index of the
symptoms of anorexia nervosa. Psychol Med 1979;9:273–80.
9. Stunkard AJ, Messick S. The three-factor eating questionnaire to
measure dietary restraint, disinhibition, and hunger. J Psychosom
Res 1985;29:71–83.
10. Beck AT, Beamesdorfer A. Assessment of depression: the Depres-
sion Inventory. In: Pinchot P, ed. Psychological measurements in
psychopharmacology, modern problems in pharmacopsychiatry. Vol
7. Basel, Switzerland: Karger, 1974:151–69.
11. Zung WWK. A self-rating depression scale. Arch Gen Psychiatry
1970;12:63–70.
12. Pennington JAT. Bowes & Church’s food values of portions com-
monly used. Philadelphia: JB Lippincott Company, 1994.
13. Hetherington M, Rolls BJ. Methods of investigating human eating
behavior. In: Toates F, Rowland N, eds. Feeding and drinking.
Amsterdam: Elsevier Science Publishers BV, 1987:77–109.14.
Kissileff HR. Satiating efficiency and a strategy for conducting food
loading experiments. Neurosci Biobehav Rev 1984;8:129–35.
15. French SJ, Read NW. Effect of guar gum on hunger and satiety after
meals of differing fat content: relationship with gastric emptying.
Am J Clin Nutr 1994;59:87–91.
16. Holt SHA, Miller JCB, Petocz P, Farmakalidis E. A satiety index of
common foods. Eur J Clin Nutr 1995;49:675–90.
17. De Graaf C, Hulshof T. Effects of weight and energy content of pre-
loads on subsequent appetite and food intake. Appetite 1996;26:
139–51.
18. Porrini M, Santangelo A, Crovetti R, Riso P, Testolin G, Blundell
JE. Weight, protein, fat, and timing of preloads affect food intake.
Physiol Behav 1997;62:563–70.
19. Spiegel TA, Kaplan JM, Alavi A, Kim PSY, Tse KKM. Effects of
soup preloads on gastric emptying and fullness ratings following an
egg sandwich meal. Physiol Behav 1994;56:571–5.
20. Rayner DV. Gastrointestinal satiety in animals other than man. Proc
Nutr Soc 1992;51:1–6.
21. Phillips RJ, Powley TL. Gastric volume rather than nutrient content
inhibits food intake. Am J Physiol 1996;271:R766–79.
22. Geliebter A. Gastric distension and gastric capacity in relation to
food intake in humans. Physiol Behav 1988;44:665–8.
23. Davis JD, Smith GP, Sayler JL. Reduction of intake in the rat due to
gastric filling. Am J Physiol 1997;272:R1599–605.
24. Kissileff HR, Gruss LP, Thornton J, Jordan HA. The satiating effi-
ciency of foods. Physiol Behav 1984;32:319–32.
25. Shide DJ, Rolls BJ. Information about the fat content of preloads
influences energy intake in healthy women. J Am Diet Assoc
1995;95:993–8.
26. Caputo FA, Mattes RD. Human dietary responses to perceived
manipulation of fat content in a midday meal. Int J Obes 1993;
17:237–40.
by guest on July 14, 2011www.ajcn.orgDownloaded from