Physiology & Behavior 79 (2003) 183 – 189

Contribution of snacks and meals in the diet of French adults:

a diet-diary study
F. Bellislea,*, A.M. Dalixa, L. Mennenb, P. Galanc, S. Hercbergc, J.M. de Castrod, N. Gausserese
a
INSERM U341, Hôtel-Dieu, 1, Place du Parvis Notre-Dame, 75181 Paris, Cedex 04 75181, France
b
Unité de Surveillance et d’Epidémiologie Nutritionnelle, InVS/CNAM, ISTNA, 5 rue du Vertbois, 75003 Paris, France
c
INSERM U557, INRA U1125, UMR INRA/INSERM/CNAM, ISTNA, 5 rue du Vertbois, 75003, France
d
Department of Psychology, Neuropsychology and Behavioral Neuroscience Program, Georgia State University, Atlanta, GA 30303, USA
e
Danone Vitapole, Route Départementale 128, 91767 Palaiseau, France
Received 16 August 2002; received in revised form 26 November 2002; accepted 5 March 2003

Abstract

To investigate the relative contributions of meals and snacks in the daily intake of free-living humans, 54 French adults maintained food
intake diaries for four 7-day periods. They recorded all food and fluid intakes mentioning whether, in their opinion, each intake event was a
snack or a meal. The weekly food diaries also contained information on the circumstances of each event such as time and place, number of
persons present, and affective states (hunger, satiety, etc.) before and after intake. On average, 2.7 meals and 1.3 snacks were consumed each
day. Very few days included no snacking. Total daily energy and nutrient intake were not different between days with and days without
snacks. Snacks differed from meals in several dimensions. Meals were about twice as large as snacks in energy and weight. Nutrient intake, in
absolute values, was higher in meals. In proportions, however, snacks contained more CHO and less fat and proteins. Most foods were
consumed in larger amounts in the context of meals but a few (sweets, cereal bars, biscuits, and sodas) were mostly consumed as snacks.
Hunger was more intense before but less intense after meals than snacks. The satiety ratio was higher for snacks than meals. Time of day
affected many intake parameters. For example, afternoon snacks exhibited a high satiety ratio for a modest intake. The present study
describes the status of several potential determining factors at the time of snacks in humans, demonstrating a specific role for snacks, as
opposed to meals, in the daily eating pattern of healthy adults.
D 2003 Elsevier Science Inc. All rights reserved.

Keywords: Meals; Snacks; Weekly food diary; Eating patterns; Human subjects

1. Introduction more in general [1], and dieters are commonly advised to

refrain from extraprandial intake.
In humans, the daily food intake is distributed over a In contrast to this negative view of snacking, many
certain number of meals and snacks. Snacking can make an studies have reported that foods ingested as snacks contrib-
important contribution to the total daily energy and nutrient ute significantly to the nutrient quality of the diet [1].
intake, and the nature and quality of this contribution can Whereas early reports of an inverse association between
vary according to numerous factors such as sociocultural eating frequency and body weight have been questioned
context, body weight status, age, gender, etc. Snacking is [2], other beneficial physiological effects, such as reduced
often considered a bad habit because snacks could facilitate blood cholesterol and improved glucose tolerance, seem po-
hyperphagia while providing insignificant quantities of sitively associated with the frequency of daily eating epi-
valuable nutrients due to the choice of low quality snack sodes [3– 5].
foods [1]. Indeed, people who snack frequently tend to eat In the present study, we examined the daily meal and
snack pattern in a group of healthy French adults. The
subjects recorded all foods and fluids ingested in the course
* Corresponding author. Tel.: +33-1-42-34-80-54; fax: +33-1-40-46-
of self-defined snacks and meals over four 7-day periods.
82-48. The weekly food diary method developed by de Castro and
E-mail address: bellisle@[Link] (F. Bellisle). Kreitzman [6] was used. This method allows a precise and

0031-9384/03/$ – see front matter D 2003 Elsevier Science Inc. All rights reserved.
doi:10.1016/S0031-9384(03)00088-X
184 F. Bellisle et al. / Physiology & Behavior 79 (2003) 183–189

valid description of spontaneous eating patterns in human 2.2. Procedure

subjects, and numerous studies using the weekly food diary
method have been published about the particular influence The subjects were given a small pocket-sized intake
of such determining factors as hunger and satiety feelings, diary. The diary was a French translation of the American
circadian rhythms, social facilitation, physiological effects diary that was used in numerous studies over the last 20
of foods and nutrients, palatability, etc. [7 –10]. Recently, years [7,17]. The subjects were instructed to record every
the weekly food diary method has been used in cross- item that they either ate or drank, the time they consumed it,
cultural comparisons of eating patterns in three different the amount, how the food was prepared, and the number of
countries (The Netherlands, France, and the United States) the other people eating with them in as detailed a manner as
[11]. The three-country study revealed that, in contrast to the possible. Self-ratings on seven-point scales were obtained at
American and Dutch patterns, a larger part of daily intake in the beginning and again at the end of each intake event for
France occurred in the afternoon rather than in the evening hunger (from very sated to very hungry), thirst (from very
due to the traditionally large lunch meal and frequent sated to very thirsty), mood (from very depressed to very
afternoon snacking. elated), anxiety (from very relaxed to very anxious), and
In the present population of French adults, the character- overall meal palatability (from very bad to very good). The
istics of subject-defined ‘‘meals’’ and ‘‘snacks’’ were com- subjects had to identify each intake event as either a ‘‘meal’’
pared in terms of size, nutrient content, and circumstances of or a ‘‘snack.’’
intake. The contribution of snacks to the daily energy and In a first training step, the subjects recorded food and
nutrient intake was assessed. fluid intake for a day using a 1-day diary and were
subsequently contacted by the experimenter to review the
information, correct any problems, and answer any question.
2. Methods The data of this first recording day were not used for further
analysis. Then, the subjects recorded their intake for seven
2.1. Subjects consecutive days on four occasions, that is, 1 week during
winter, spring, summer, and fall using the ‘‘weekly food
Data were collected from a total of 54 subjects recruited diary.’’ After each recording period, the subjects were
from two populations: 26 students and 28 middle-aged contacted by the experimenter who reviewed the diaries.
individuals. The students were recruited at the University For a review of the reliability and validity of the diet-diary
of Paris by poster advertisements on campus. They had no procedure, see [17].
declared pathologies and were taking no regular medication
except for oral contraceptives. The Three-Factor Eating 2.3. Data analysis
Questionnaire [12] showed that none had scores exceeding
the normal range of values reported in the French validation The foods reported in the weekly diaries were assigned
study of the questionnaire [13]. The subjects were informed codes from a computer file containing the nutrient compo-
that a study of eating practices was being performed, and sitions of common food items. A computer file of over 3500
that their participation involved recording all their solid and food items was created from the CIQUAL repertory of
fluid intakes for four nonconsecutive weeks, 1 week for French foods [18] and from the local food industry.
every season over 1 year. Students were paid for their The four 7-day records were combined into a single 28-
participation. The food intake data in this population have day record for each subject. Discrete intake ‘‘events’’ were
been described in earlier publications [9,11,14]. The middle- identified according to a definition used in previous works:
aged subjects were participants of the ‘‘Supplementation en They had to contain a minimum of 50 kcal and be separated
Vitamines et Minéraux’’ ([Link]) study, a random- from previous and following ingestive events by a minimum
ized, double-blind, placebo-controlled primary-prevention of 45 min. The compositions of the individual items
trial designed to test the impact of a daily supplementation composing the events were summed. The intake events
with antioxidant vitamins and minerals at nutritional doses were characterized by their total energy content, CHO, fat,
on the incidence of major chronic diseases [15]. The total protein, and alcohol content; duration and rate of intake; and
sample in the [Link] study is over 12,000 mature the amount of time between events; the before-intake and
participants (age>35 years). A subgroup of 100 subjects after-intake intervals; the before-intake and after-intake
were contacted by mail and asked if they would agree to ratings of hunger, thirst, depression, and anxiety; the
participate in a side study focusing on eating practices. They ‘‘deprivation ratio,’’ a measure of the subject’s response to
were informed that their participation would involve record- a period of noneating, which is calculated as the amount of
ing all intakes for four nonconsecutive weeks. Subjects energy ingested on one occasion divided by the duration of
who accepted to participate in this side study followed the the before-intake interval; and the ‘‘satiety ratio,’’ a measure
same procedure as the students. Data on food intake and of the subject’s response to the food energy ingested, which
eating pattern in this group of subjects were published is calculated as the duration of the after-intake interval
earlier [16]. divided by ingested energy at a specific intake occasion.
 F. Bellisle et al. / Physiology & Behavior 79 (2003) 183–189 185

The average of each of the eating event characteristics and Table 2

Characteristics of meals and snacks
the mean daily intakes were calculated for each subject over
the entire 28-day period. These individual means were then Snacks Meals F(1,51)
used to calculate overall group means. Daily number 1.3 ± 0.0 2.7 ± 0.0 398.7* *
Differences between types of eating events were tested Intake per event
Energy (kcal) 302 ± 16 642 ± 20 277.2* *
using ANOVAS. The significance level was set at P=.05.
CHO (kcal) 157 ± 7 248 ± 10 11.7* *
Fat (kcal) 101 ± 8 272 ± 10 277.4* *
Proteins (kcal) 31 ± 2 105 ± 3 336.5* *
3. Results Alcohol (kcal) 14 ± 3 20 ± 3 3.5
Meal size (g) 350 ± 24 693 ± 21 199.0* *
3.1. Population
Ratings of subjective sensations
Before intake event
Table 1 presents the personal characteristics of the men Palatability, bad – good 5.4 ± 0.1 5.4 ± 0.1 0.9
and women recruited in the student and the [Link] Sated – hungry 4.1 ± 0.1 4.9 ± 0.1 55.4* *
populations. In the student subgroups, BMI values were in Sated – thirsty 4.1 ± 0.1 4.5 ± 0.1 15.3* *
Depressed – elated 5.2 ± 0.1 5.0 ± 0.1 6.5*
the normal range (18.5 – 29.9) according to the WHO
Calm – anxious 2.7 ± 0.2 2.7 ± 0.1 0.6
definition [19], except for two men and four women who
had a BMI between 17.1 and 18.5, which is characterized After intake event
as ‘‘mild thinness’’ by the WHO [20]. In the mature Palatability, bad – good 5.7 ± 0.1 5.6 ± 0.1 1
subjects sample, BMIs were in the normal range except Sated – hungry 2.1 ± 0.1 1.7 ± 0.1 13.7* *
Sated – thirsty 2.3 ± 0.1 1.9 ± 0.1 21.9* *
for one woman with mild thinness (BMI = 17.8) and three
Depressed – elated 5.3 ± 0.1 5.2 ± 0.1 3.6
overweight men (BMI between 26.6 and 28.7). The average Calm – anxious 2.6 ± 0.1 2.5 ± 0.1 0.7
daily intake of the two age groups, as reported earlier
[9,16], were close (student males 2450 kcal, student Change from before to after intake
females 1860 kcal; middle-aged males 2440 kcal, middle- Palatability, bad – good 0.3 ± 0.1 0.3 ± 0.1 0
Sated – hungry 1.9 ± 0.1 3.1 ± 0.1 102.4* *
aged females 1794 kcal). Because of the voluntary and
Sated – thirsty 1.9 ± 0.2 2.6 ± 0.2 27.8* *
demanding nature of participation, it was possible to obtain Depressed – elated 0.1 ± 0.0 0.2 ± 0.0 0.8
only five valid sets of data from male [Link] sub- Calm – anxious 0.2 ± 0.1 0.2 ± 0.0 0.7
jects. Given this small n and the comparable daily energy
intake in males and females of both age groups, it was Other characteristics
Number of persons present 1.3 ± 0.0 2.7 ± 0.0 398.7* *
decided to pool the data from all subjects and treat the
Duration of preintake interval (min) 210 ± 7 264 ± 6 173.4* *
population as one group. Duration of postintake interval (min) 177 ± 6 270 ± 6 50.0* *
Satiety ratio 1.0 ± 0.1 0.7 ± 0.0 26.4* *
3.2. Energy intake on days with snacks and days without Deprivation ratio 1.7 ± 0.1 3.8 ± 0.1 157.7* *
snacks Duration of event (min) 27.1 ± 3.9 33.3 ± 1.2 2.6
Ingestion rate (kcal/min) 24.1 ± 2.6 23.9 ± 1.0 0

In all subjects, most recorded days included at least one Means ± S.E.M.
* P < 0.05.
snack. Over a total of 28 days/subject, 20.5 ± 0.9 included at
* * P < 0.01.
least one snack. The total energy intake was 1983 ± 63 kcal
(8296 ± 60 kJ) on days with snacks and 1963 ± 89 kcal
(8213 ± 84 kJ) on days without snacks (N.S.).
3.3. Characteristics of meals versus snacks

On average, 2.7 meals and 1.3 snacks were recorded each

Table 1 day. Table 2 presents several parameters of meals and
Anthropometric characteristics of the subjects
snacks recorded from the food diaries.
Type of subjects Students [Link] Meals were about twice as large as snacks in terms of
Men Women Men Women energy content and weight and included more CHO, fat, and
n 10 16 5 23 protein, but not alcohol, than snacks. The relative contri-
Age (years) 23.6 ± 2.3 23.3 ± 0.6 51.2 ± 0.9 48.7 ± 0.7 bution of macronutrients (%) in snacks and meals is shown
(range) (17 – 43) (19 – 29) (46 – 57) (36 – 58) in Fig. 1. Snacks were proportionally richer in CHO
Height (cm) 178.9 ± 2.5 164.1 ± 1.2 176.8 ± 0.8 161.1 ± 0.5
Weight (kg) 66.5 ± 2.8 53.9 ± 1.5 81.4 ± 1.6 56.8 ± 0.7
[ F(1,51) = 175.2, P < .001] and contained less fats and
Body mass index 20.7 ± 0.6 20.0 ± 0.5 26.0 ± 0.5 21.8 ± 0.2 proteins than meals [ F(1,51) = 99.2 and 175.9, respectively,
(kg/m2) P < .001].
(range) (18 – 23) (17.1 – 24.6) (23.7 – 28.7) (18.9 – 24.6) Hunger and thirst were more intense before meals than
Means ± S.E.M. snacks. These two sensations were significantly less intense
186 F. Bellisle et al. / Physiology & Behavior 79 (2003) 183–189

Table 3
Food and drink intake in meals and snacks (g)
Snacks Meals F(1,51)
Fruits 27.9 ± 4.1 45.7 ± 5.0 9.6* *
Cheese 2.3 ± 0.6 9.8 ± 2.4 10.6* *
Ice cream 1.7 ± 0.5 1.8 ± 0.6 0
Sweets 2.1 ± 0.5 0.6 ± 0.2 7.5* *
Soup 11.6 ± 2.6 37.4 ± 4.6 27.8* *
Fish 1.1 ± 0.4 13.6 ± 1.3 80.9* *
Beef 2.4 ± 1.4 12.4 ± 1.0 33.6* *
Chicken 0.9 ± 0.3 8.4 ± 0.9 72.4* *
Pork 2.5 ± 0.7 13.4 ± 1.3 68.5* *
Veal 0.2 ± 0.2 1.9 ± 0.3 33.4* *
Lamb 0.5 ± 0.3 2.6 ± 0.4 19.9* *
Turkey 0.0 ± 0.0 1.4 ± 0.3 25.5* *
Rice 1.8 ± 0.8 8.6 ± 0.9 33.2* *
Potatoes 2.3 ± 0.8 18.5 ± 1.6 93.5* *
Bread 5.1 ± 1.3 7.8 ± 1.4 2.8
Fig. 1. Nutrient composition of meals and snacks, expressed as the percent Rusks 1.5 ± 1.0 2.2 ± 0.5 0.8
of total energy derived from each of the four macronutrients. N = 54. Breakfast cereals 0.2 ± 0.1 1.3 ± 0.5 5.0*
Cereal bars 0.7 ± 0.3 0.1 ± 0.0 4.8*
Pastry 13.6 ± 2.9 12.2 ± 2.9 0.1
after meals than snacks. Subjects felt slightly more elated Biscuits 7.6 ± 2.4 1.7 ± 0.3 5.7*
Chocolate 4.0 ± 0.9 2.4 ± 0.5 2.6
before snacks than meals. The change in hunger and thirst
Sugar, honey, and jam 1.0 ± 0.5 4.4 ± 5.0 1.1
sensations was more marked as a result of meal ingestion Dairy products 15.4 ± 3.1 32.4 ± 3.7 16.3* *
than snack ingestion. More other persons were present at Sugar containing sodas 15.7 ± 3.7 7.2 ± 2.2 7.5* *
meals than snacks. ‘‘Light’’ sodas 0.6 ± 0.5 0.4 ± 0.2 0.3
Interprandial intervals were longer before and after Tea and coffee 62.4 ± 9.4 95.5 ± 9.9 10.7* *
Fruit juices 19.3 ± 6.2 17.3 ± 3.8 0.1
meals than snacks. The satiety ratio was higher for snacks
than meals, meaning that energy content being equal, Means ± S.E.M.
* P < 0.05.
snacks preceded a longer satiety than meals. By contrast,
* * P < 0.01.
the deprivation ratio was higher for meals than snacks; in
other words, for an equal duration of preintake fasting,
more energy was ingested in the context of meals than 18 and 20 h, reflecting afternoon snacking in certain
snacks. Although snack duration was shorter than meal subjects and early dinners in others. Some intake was
duration, the difference failed to reach statistical signific- recorded between 10 and 12 h and from 14 to 18 h, clearly
ance. Ingestion rate was not different between meals and outside of traditional meal hours; it was composed of CHO
snacks. for a large part and represented occasional small snacks in
Table 3 shows that certain foods and drinks were some of the participants.
consumed preferentially either in the context of meals or Table 4 presents various parameters of meals and
in the context of snacks. While meats, fish, dairy products, snacks recorded in the morning (until 11:59 h), afternoon
cheese, and fruits were consumed most in the context of
meals, sweets, cereal bars, biscuits, and sodas were
reported mostly in the context of snacks. No difference
was reported in the intake of bread and pastries, among
others.

3.4. Distribution of intake over 24 h

Fig. 2 presents the distribution of energy and nutrient

intake over the average day (24 h). Two clear peaks
appeared. The higher appeared around noontime and corre-
sponded to lunch, the most abundant meal in the traditional
French diet. The other peak was recorded around 20 –22 h
and reflected dinner intake. A modest elevation of intake in
the morning culminated around 8 –10 h, probably resulting
from the addition of breakfast in certain subjects and
midmorning snacks in others. After lunch, the intake curve Fig. 2. Distribution of the energy and nutrient intake over 24 h, presented as
remained low until 18 h and then increased sharply between averages over 2-h intervals. N = 54.
 F. Bellisle et al. / Physiology & Behavior 79 (2003) 183–189 187

Table 4
Parameters of meals and snacks at three moments of the day
Morning Afternoon Evening Type of event Time of day Interaction
(until 11:59 h) (12:00 – 18:59 h) (after 19:00 h) effect [ F(1,30)] effect [ F(2,60)] [ F(2,60)]
Number of persons present Snacks 1.4 ± 0.3 1.2 ± 0.2 1.5 ± 0.2 1.7 5.9* * 9.1* *
Meals 0.7 ± 0.1 2.2 ± 0.4 2.0 ± 0.3
Size of intake (kcal/event) Snacks 288 ± 30 289 ± 19 370 ± 49 109.6* * 30.8* * 22.7* *
Meals 423 ± 28 760 ± 30 770 ± 34
CHO intake (kcal/event) Snacks 147 ± 13 160 ± 9 141 ± 13 73.7* * 3.3 * 4.0 *
Meals 231 ± 14 262 ± 13 280 ± 13
Fat intake (kcal/event) Snacks 105 ± 15 94 ± 9 138 ± 29 92.0* * 28.9* * 24.9* *
Meals 142 ± 14 345 ± 16 322 ± 16
Proteins intake (kcal/event) Snacks 34 ± 6 27 ± 3 44 ± 7 205.7* * 50.1* * 41.0* *
Meals 52 ± 9 138 ± 6 125 ± 6
Alcohol intake (kcal/event) Snacks 0±0 7.6 ± 2.2 47 ± 11,4 0.4 17.5* * 1.2
Meals 0.09 ± 0.09 17.4 ± 3.7 44 ± 10,6
Size of intake (kcal/event) Snacks 361 ± 43 367 ± 38 420 ± 41 112.2* * 18.6* * 14.5* *
Meals 545 ± 35 798 ± 28 834 ± 37
Intake duration (min) Snacks 28.7 ± 5.1 27.4 ± 6.6 40.6 ± 6.2 0.2 7.6* * 3.1
Meals 22.2 ± 1.2 37.9 ± 2.1 42.7 ± 3.0
Satiety ratio Snacks 1.0 ± 0.1 1.0 ± 0.1 0.6 ± 0.1 55.0* * 17.6* * 0.8
Meals 0.7 ± 0.1 0.5 ± 0.0 0.3 ± 0.0
Deprivation ratio Snacks 2.9 ± 0.9 1.6 ± 0.1 2.6 ± 0.5 10.8* * 0 4.4 *
Meals 3.4 ± 0.4 4.4 ± 0.2 3.6 ± 0.3
Means ± S.E.M.
* P < 0.05.
* * P < 0.01.

(12:00– 18:59 h), and evening (after 19:00 h). The stat- 4. Discussion
istical analysis was performed in a subset of subjects
(n = 31) who had at least one meal and one snack in all In the present sample of French adults, the reported
three periods so that data would be available for all daily meal number (2.7) was close to the traditional three,
subjects in all six eating event  time period cells. Several and one or more extraprandial eating events occurred on
significant main effects and interactions appeared. Time of most days in all subjects regardless of age or gender. The
day and interactions affected the number of persons total daily energy intake was not different whether or
present at the eating occasions and the amount of food not extraprandial intake occurred, suggesting that one or
ingested (in terms of energy and weight). Alcohol intake more snacks do not significantly increase the daily energy
and duration of eating episodes were significantly affec- input.
ted only by time of day (higher in the evening), while Subjects themselves decided whether the eating events
the satiety ratio was affected by both main factors but they were recording in their diaries were meals or snacks.
showed no interaction effect. This analysis shows that The definition of ‘‘meals’’ and ‘‘snacks’’ is the matter of
morning and afternoon snacks produced a high satiety an extensive debate in the scientific literature devoted to
ratio in spite of their small energy content as compared to quantifying the periodicity of eating [21]. In many studies,
meals and evening snacks. In other words, the duration of quantitative definitions were used and meals were discri-
inhibition of eating following these snacks was long minated from snacks on the basis of objective criteria; for
relative to the amount of energy ingested. In addition, example, a ‘‘meal’’ should bring at least so many kilo-
afternoon snacks had a low deprivation ratio, as compared calories, contain so many different types of foods, or be
to other snacks and meals, showing that the energy separated from the preceding and the next eating event by
content of afternoon snacks relative to the time elapsed a minimal time interval. In the present study, the analysis
since the previous eating occasion was lower than for of eating events recorded in the weekly food diaries was
other eating occasions. based on the subjects’ own definitions of meals and
snacks. The criteria used by the participants to discriminate
3.5. Seasonal variations between meals and snacks were not investigated, but it is
highly likely that the French respondents based their
No significant effect of seasons appeared on any of the judgement on the traditional meal definition in the French
food intake variables, although seasons affected waking culture, that is, a social event that occurs on a regular
time that occurred later in the fall and winter than in the basis, day after day, at relatively fixed hours, and includes
spring or summer times [ F(3,276) = 3.2, P < .02]. several dishes presented in succession. Traditionally,
188 F. Bellisle et al. / Physiology & Behavior 79 (2003) 183–189

French adults used to ingest three meals a day. In recent that internal signals associated with hunger and satiety do
years, due to changes in lifestyle, this traditional meal not determine eating patterns in humans in a simple linear
pattern has become less clear cut [22], as is apparent in the causal relationship as was suggested in animals, but that
present data. habits and social constraints, expressed in the present study
According to the participants in the present study, snacks by the subjects’ characterization of certain events as
are clearly different from meals. They differ in size (meals ‘‘snacks’’ or ‘‘meals,’’ also affect how much food may be
are about twice as large as snacks) and nutritional content eaten at a particular time, how much hunger is tolerable
(snacks have a relatively high CHO content) and hunger and before the subject starts eating again, etc.
thirst sensations before and after the event (people reported When laboratory rats are exposed to a humanlike five
more intense hunger and thirst before meals than snacks). meal a day schedule, they shift from postprandial regulation
These observations agree with the results observed using the to preprandial regulation, and meal size then adjusts to the
same method in American subjects [23]. Snacks occur at duration of the preprandial interval [29]. If the scheduled
times when hunger feelings are present. This suggests that feeding pattern is continued, then the animals learn to
an internal, metabolic stimulus plays a role in snacking in adjust meal size to the duration of the postmeal interval.
addition to the triggering action of external stimuli present The assumption behind this ‘‘provisional appetite’’ being
in the environment. Meals bring more energy and nutrient that they learn to eat enough at each meal in order not to
than snacks; consequently, they reduce hunger more. They become hungry until food will become available again [24].
contain more liquid and consequently produce a larger thirst The behavior of the human participants in the present study
quenching effect than snacks. Another observation associ- suggests that they ate enough in meals and snacks to reduce
ated with the larger size of meals, as compared to snacks, is the level of subjective hunger after eating, but not enough
that they are followed by longer periods of no eating, an to prevent the return of hunger before the next eating
effect that is analogous to observations first reported in occasion. The small amount of food ingested as snacks
laboratory rats in the 1950s [24]. The animals exhibited was not sufficient to keep hunger sensations at minimal
what was called the ‘‘postprandial correlation’’ between levels until the next meal occasion. By contrast, it allowed
meal size and the duration of the postprandial interval until the return of hunger feelings of relatively high intensity
the onset of the next spontaneous meal. The hypothesized before the next socially predictable occasion to ingest a
mechanism explaining the postprandial correlation was that large meal. Then, human consumers do not eat in sufficient
the animals completely used up the metabolic fuel contained amounts to prevent the experience of hunger, in contrast to
in the preceding meal before resuming consumption. the hypothesized strategy of laboratory rats, rather they
In the animal literature, there has been controversy over seem to use snacks as modulators of hunger between
the strength and reliability of prandial correlations [25 – 28] socially determined meal occasions, allowing hunger to
suggesting that other factors than metabolic signals actually return and perhaps appetite to increase before the next
determine the feeding patterns even in a solitary rat cage. meal. To paraphrase Le Magnen [24], humans seems to
This conclusion is in line with the present human data eat just enough (at snacks) in order to be hungry at the next
analyzed according to a subject-specific cognitive category meal.
(meals vs. snacks). Our task was to examine how behavioral The traditional French three meals a day pattern should
parameters traditionally hypothesized to reflect internal give three clear peaks on Fig. 2, with almost nonexistent
signals (meal size, meal number, meal to meal intervals, intake between these peaks. This is clearly not what is
etc.) would vary in the context of ingestive events discrimi- observed since intake occurs at all times over the waking
nated on the basis of cultural, traditional learning. Although hours, with two marked peaks corresponding to the two
laboratory animals with free access to food and water could main meals of the day, lunch, and dinner. It is not clear
be hypothesized to eat exclusively or essentially in response whether this pattern is the result of recent changes in the
to physiological, regulatory signals, such a hypothesis lifestyle of French adults, or the representation of a situation
cannot be made in a human society context. The cognitive that has existed for a long time but was never adequately
discrimination between snacks and meals, that all subjects assessed.
understood and used without difficulty, allows us to objec- In the present population, the data presented in Table 2
tively demonstrate the differential influence of factors indicate that energy from snacks represented 18.5% of total
reflecting the nutritional status on cognitively distinct eating daily energy intake, on average. This is less than what was
events, in the same person, at different moment of a single reported (25%) in adults living in the UK [30,31] but close
recording day. to the value (17%) obtained in a group of middle-aged
It is interesting that both the deprivation ratio and the sedentary American men [32]. Several earlier reports indi-
satiety ratio were significantly different between events cated that snacks contain proportionally more CHO than
called ‘‘snacks’’ and events called ‘‘meals.’’ Energy load meals and less fat and protein than meals [30,31,33]. This
being equal, snacks appeared more satiating than meals, and was confirmed in the present study. Different foods are
previous deprivation time being equal, more energy was selected for snacks and for meals. In consequence, snacking
consumed in meals than in snacks. These observations show appears to be a potentially useful strategy to increase food
 F. Bellisle et al. / Physiology & Behavior 79 (2003) 183–189 189

and nutrient diversity in the diet, at least in healthy adult [10] de Castro JM, Bellisle F, Dalix AM, Pearcey S. Palatability and intake
relationships in free-living humans: characterization and independence
populations, without compromising the energy balance.
of influence in North Americans. Physiol Behav 2000;70:343 – 50.
Our analysis of three successive moments over the [11] de Castro JM, Bellisle F, Feunekes GU, Dalix AM, De Graaf C.
waking hours (morning, afternoon, and evening) revealed Culture and meal patterns: a comparison of the food intake of free-
numerous main effects and interactions between the ‘‘type living American, Dutch, and French subjects. Nutr Res 1997;17:
of event’’ factor and time of day. Among other effects, 807 – 29.
afternoon snacks have a high satiety ratio and a low [12] Stunkard AJ, Messick S. The Three-Factor Eating Questionnaire to
measure dietary restraint, disinhibition and hunger. J Psychosom Res
deprivation ratio. In the traditionally long time interval that 1985;29:71 – 83.
separates lunch from dinner in the French culture, small [13] Lluch A. Identification des conduites alimentaires par approches nu-
snacks produce long lasting satiety. This observation con- tritionnelles et psychométriques: implications thérapeutiques et pré-
firms that ingested amounts and satiety are not a fixed ventives dans l’obésité humaine. Doctoral dissertation. Nancy I,
function of previous energy intake. More factors, such as France: University Henri Poincaré; 1995.
[14] Bellisle F, Dalix AM, de Castro JM. Eating patterns in French sub-
time of day, number of people present, learned habits, and jects studied by the ‘‘weekly food diary’’ method. Appetite 1999;32:
cognitive set shape the daily eating pattern in order to adapt 46 – 52.
it to energy needs, at least in healthy individuals. [15] Hercberg S, Preziosi P, Briançon S, Galan P, Triol I, Malvy D, et al. A
primary prevention trial using nutritional doses of antioxidant vita-
mins and minerals in cardiovascular disease and cancers in a general
5. Conclusion population: the [Link] study—design, methods, and participants
characteristics. Control Clin Trials 1998;19:336 – 51.
[16] de Castro JM, Bellisle F, Dalix AM, Slama G. The effect of type I
In the present study, snacking was shown to occur when diabetes on the eating patterns of free living French: a diet diary study.
a certain level of hunger was present. The presence of one or Physiol Behav 2002;75:583 – 92.
more snacks during a day did not significantly increase [17] de Castro JM. Methodology, correlational analysis, and interpretation
energy intake while it shifted nutrient content somewhat of diet diary records of the food and fluid intake of free-living humans.
Appetite 1994;23:179 – 92.
towards less fat and more CHO. Whether or not snacking is [18] CIQUAL. Répertoire général des aliments. Paris: Lavoisier; 1991.
considered a troublesome behavior that should be avoided [19] World Health Organization. Obesity. Preventing and managing
or a useful way to diversify nutrient intake, the data the global epidemic. Report of a WHO Consultation on Obesity.
presented here show, for the first time, the status of several Geneva, 1997.
potentially determining factors on the intake of snacks in [20] World Health Organization. Physical status: the use and inter-
pretation of anthropometry. Report of a WHO Expert Committee.
free-living subjects. More studies of such factors are needed, Geneva, 1995.
in different populations and different societies, so that the [21] Gibney MJ, Wolever TMS. Periodicity of eating and human health:
very prevalent behavior of snacking can be better under- present perspective and future directions. Br J Nutr 1997;77:S3 – 5.
stood and evaluated in terms of consequences on the human [22] Poulain JP. The contemporary diet in France. Appetite 2002;28:1 – 13.
nutritional status. [23] de Castro JM, Brewer EM, Elmore DK, Orozco S. Social facilitation
of the spontaneous meal size of humans occurs regardless of time,
place, alcohol or snacks. Appetite 1990;15:89 – 101.
[24] Le Magnen J. Neurobiology of feeding and nutrition. San Diego:
References Academic Press; 1992.
[25] Collier G, Johnson DF, Mitchell C. The relation between meal size
[1] Gatenby SJ. Eating frequency: methodological and dietary aspects. Br and the time between meals: effects of cage complexity and food cost.
J Nutr 1997;77(Suppl 1):S7 – 20. Physiol Behav 1999;67:339 – 46.
[2] Bellisle F, McDevitt R, Prentice A. Meal frequency and energy bal- [26] Castonguay TW, Kaiser L, Stern JS. Meal patterns analysis: artifacts,
ance. Br J Nutr 1997;77(Suppl 1):S57 – 70. assumptions and implications. Brain Res Bull 1986;17:439 – 43.
[3] Debry G. Newer knowledge on the treatment of obesity. Bibl Nutr [27] de Castro JM. Meal pattern correlations: facts and artifacts. Physiol
Dieta 1978;26:44 – 59. Behav 1975;15:13 – 5.
[4] Jenkins DJA. Carbohydrate tolerance and food frequency. Br J Nutr [28] Panksepp J. Reanalysis of feeding patterns in rats. J Comp Physiol
1997;77(Suppl 1):S71 – 81. Psychol 1973;82:78 – 94.
[5] Arnold L, Ball MJ, Duncan AW, Mann J. Effect of isoenergetic intake [29] de Castro JM. The meal pattern of rats shifts from postprandial reg-
of three or nine meals on plasma lipoproteins and glucose metabolism. ulation to preprandial regulation when only five meals per day are
Am J Clin Nutr 1993;46:611 – 20. scheduled. Physiol Behav 1988;43:739 – 46.
[6] de Castro JM, Kreitzman SN. A microregulatory analysis of sponta- [30] Drummond S, Crombie N, Kirk TR. A critique of the effects of
neous human feeding patterns. Physiol Behav 1985;35:329 – 35. snacking on body weight status. Eur J Clin Nutr 1996;50:779 – 83.
[7] de Castro JM. Circadian rhythms of the spontaneous meal pattern, [31] Summerbell CD, Moody RC, Shanks J, Stock MJ, Geissler C. Rela-
macronutrient intake, and mood of humans. Physiol Behav 1987; tionship between feeding pattern and body mass index in 220 free-
40:437 – 46. living people in four age groups. Eur J Clin Nutr 1996;50:513 – 9.
[8] de Castro JM, Elmore DK. Subjective hunger relationships with meal [32] Dreon DM, Frey-Hewitt B, Ellsworth N, Williams PT, Terry RB,
patterns in the spontaneous feeding behavior of humans: evidence for Wood PD. Dietary fat: carbohydrate ratio and obesity in middle-aged
a causal connection. Physiol Behav 1988;43:159 – 65. men. Am J Clin Nutr 1988;47:995 – 1000.
[9] de Castro JM, Bellisle F, Dalix AM. Palatability and intake relation- [33] Ballard-Barbash R, Thompson FE, Graubard BI, Krebs-Smith SM.
ships in free-living humans: measurement and characterization in the Variability in percent energy from fat throughout the day: implications
French. Physiol Behav 2000;68:271 – 7. for application of total diet goals. J Nutr Educ 1994;26:278 – 83.