Commentaries
The Cafeteria Diet as a Tool for Studies of
Thermogenesis
NANCY J. ROTHWELL ANDMICHAEL J. STOCK*
Department of Physiological Sciences, University of Manchester, Manchester Ml 3 9PT, England and
"Department of Physiology, St. George's Hospital Medical School, Tooting, London SW17 ORE, England
ABSTRACT The cafeteria diet involves feeding experi
mental animals a choice of palatable human food items to
stimulate energy intake, and has been used extensively to
study diet-induced thermogenesis. In a recent commentary
Moore has argued that this feeding regime is inappropriate
for such studies because the nutrient composition cannot
be controlled, many of the effects seen are due to protein
or nutrient deficiency and accurate measurements of en
ergy intake are difficult to achieve. We argue that all of
these criticisms can be overcome by careful use of the
feeding regime and well-controlled experiments. Gross nu
trient composition of cafeteria diets can be modified over
a wide range, and such studies demonstrate that the effects
of protein deficiency can be clearly dissociated from those
of hyperphagia. There is no experimental evidence for nu
tritional deficiency in cafeteria-fed animals even over very
long periods of time. Furthermore, the alternatives sug
gested by Moore, i.e., presenting sucrose solutions to drink
or high fat diets, suffer the same drawbacks of altered and
often uncontrolled nutrient intake and yet produce little or
no increase in energy intake. Criticism of the cafeteria diet
is not justified simply because of its misuse by nutritionally
naive experimenters. The value and validity of this feeding
regime is further supported by the enormous impact it has
made on our understanding of energy balance regulation
and thermogenesis. J. Nutr. 118: 925-928, 1988.
INDEXING KEY WORDS:
•cafeteria diet •diet-induced thermogenesis
•energy balance
The cafeteria diet is a feeding regime in which ani
mals are offered a choice of several palatable food items
of varied composition, appearance and texture in ad
dition to their normal nonpurified diet. We and others
have used this feeding system extensively to study the
mechanisms of energy balance regulation and diet-in
duced thermogenesis (DIT). However, the use of the
cafeteria diet has been criticized, largely because of the
poor control over nutrient composition, and Moore (1)
recently detailed her criticisms at some length in a
0022-3166/88 $3.00 ©1988 American Institute of Nutrition.
Downloaded from https://academic.oup.com/jn/article-abstract/118/8/925/4738924
by guest
on 25 January 2018
commentary in this journal entitled, "The Cafeteria
Diet—An Inappropriate Tool For Studies of Thermo
genesis." This article has prompted us to address these
criticisms and to discuss their implications. We accept
that there are drawbacks and limitations to the use of
this diet, but believe nevertheless that the cafeteria
feeding system still offers many advantages over con
ventional diets, and has proved invaluable in elucidat
ing some of the mechanisms of energy balance regu
lation. For clarity, we will consider the criticisms raised
by Moore (1) in the same order and categories as in her
commentary.
THE STRUGGLE FOR A CLEAN RESULT
We obviously agree that the ultimate aim of all ex
periments is to obtain a "clean result" that is not open
to variations in interpretation, and that this can only
be achieved when every variable (apart from the one
that is the subject of study) is precisely controlled.
However, because of the very nature of biological sys
tems this aim is rarely possible to achieve, and it could
be argued that when compared with the physical sci
ences, nearly all biological research falls short of this
goal. Nevertheless, because of the complexity and plas
ticity of biological systems, a rigorous and highly re
ductionist approach may actually hinder our under
standing of phenomena. Moreover, if the controls
imposed are too restrictive, one may even lose sight of
the function under investigation. This applies partic
ularly to behavioral studies, of which experiments on
food intake are pertinent to the present discussion. The
use of relatively loosely defined feeding regimes, such
as the cafeteria diet, by nutritionally naive researchers
can lead to some disturbing conclusions. Hence the
highly suspect statement (2)that dietary obesity in caf
eteria-fed dams impairs pup survival and growth rate.
The study referred to (2) involved animals with a lim
ited choice of low-protein foods and a relatively low-
protein nonpurified diet. Thus the experiment was
Received 4 August 1987. Accepted 11 March 1988.
925
 926 ROTHWELL AND STOCK
compromised by an inadvertent limitation on the pro
tein available to support lactation. However, if the op
portunity to abuse or mishandle a technique is suffi
cient to invalidate it, we would be forced to reject nearly
every method available; obviously some compromise
is necessary so that reasonable controls are imposed,
and those variables that cannot be maintained constant
are at least quantified. In addition, more than one ap
proach can be used, with each using controls over dif
ferent parameters.
THE CAFETERIA DIET AND DIT
We have reviewed the evidence for DIT in cafeteria-
fed rats elsewhere (3-7) and have shown that the hy-
perphagia associated with cafeteria feeding, rather than
the altered dietary composition (as Moore suggests), is
the primary stimulus to DIT. We described 18 experi
ments performed by ourselves and other groups that
demonstrate increases in DIT when energy intake is
increased by cafeteria feeding (6).Interestingly, the two
groups (8, 9) that Moore reported as having failed to
observe significant DIT obtained only small increases
in energy intake in spite of an altered dietary compo
sition, thereby illustrating the importance of hyper-
phagia. Moore misquoted our data (10, 11) by stating
that DIT was not present in older rats or in some strains.
In fact DIT was always present, but varied in magnitude
and was usually inversely related to adiposity. We have
also shown that rats fed a cafeteria diet at the same
level of energy intake as control rats fed a purified diet
fail to show an increase in DIT (12).
MACRONÃœTRIENTCOMPOSITION
Protein deficiency is a potent stimulus to DIT (13),
but the effects on thermogenesis may be complicated
or masked by the accompanying hypophagia and growth
retardation. These problems can be partly overcome by
feeding of a palatable protein-deficient (< 10%)cafeteria
diet to maintain normal levels of energy intake (13).
The effects of variations in nutrient intake and hyper-
phagia on DIT have been distinguished in two ways.
First, DIT is still present when adequate or high (20-
30% ) protein diets are fed, or when the carbohydrate
and fat content are varied dramatically (13, 14).Second,
when a purified diet of the same composition as the
cafeteria diet was fed at a normal level of intake, en
ergetic efficiency was not affected (15). In all our ex
periments, rats have been allowed access to high pro
tein purified or nonpurified diets, and protein usually
accounts for 17-23% of energy intake in cafeteria-fed
rats. However, because of their hyperphagia, total pro
tein intake was greater than that of controls.
Moore (1) claimed that our finding (15) of a greater
acute thermic response to fat than carbohydrate in caf
eteria-fed rats contradicts the observations of others
Downloaded from https://academic.oup.com/jn/article-abstract/118/8/925/4738924
by guest
on 25 January 2018
(16), and questioned the validity of data obtained from
cafeteria-fed rats. Because the rats in this other study
(16) were fed a purified diet and were neither hyper-
phagic nor thermogenic, the contradiction is not ob
vious and the comparison not justified. Moore also stated
that "some data in rats fed carefully controlled diets"
indicate that high carbohydrate diets stimulate ther
mogenesis to a greater extent than high fat diets, but
she ignored many other data (see refs. 3-7 for reviews)
indicating the reverse. These apparent discrepancies can
be ascribed to many factors, particularly differences in
the energy cost of depositing fat, which is lower when
high fat rather than high carbohydrate diets are con
sumed. However, these differences will only affect the
obligatory DIT (or specific dynamic action), which Moore
herself has clearly distinguished from regulatory DIT.
MICRONÃœTRIENTCOMPOSITION
We have seen no evidence to support Moore's state
ments that "cafeteria foods are low in vitamin and min
eral content" and that "it is likely that cafeteria-fed
rats consume diets of suboptimal vitamin and mineral
content," and no reference to these claims was given.
Moreover, even if the proportion of micronutrients is
reduced in cafeteria-fed rats, total intake may be nor
mal or elevated as a result of hyperphagia. It is highly
unlikely that the cafeteria diets that we have used are
nutritionally deficient because they do not adversely
affect litter size, pup weight, growth rate or longevity
when fed for long periods of time (réf. 17; and N. J.
Rothwell and M. J. Stock, unpublished data).
THE CAFETERIA DIET IS SELF-SELECTED AND
UNCONTROLLED
Because of the choice of foods available to cafeteria-
fed rats, it is impossible to control the composition of
the diet consumed by each individual rat or by the
group as a whole, although this has been deliberately
and extensively varied (protein 7-33%, fat or carbo
hydrate 7-73%) by presenting foods of different com
positions (12-15). The composition of the diet con
sumed can be measured, either by direct analysis or
from manufacturer's values, and variations within groups
are relatively small. For example, in a recent experi
ment (to be published) in which the intake of each
nutrient was calculated we observed a 74% increase in
energy intake of cafeteria-fed rats and 20% of total in
take was derived from the nonpurified diet. The coef
ficient of variation in intake (energy and nutrients) was
similar for control and cafeteria-fed rats (energy 10.4%,
protein 9.6%, fat 11.7%, carbohydrate 11.1%). Total
protein intake did not differ significantly between the
two groups and was adequate for growth. We routinely
measure intake of nonpurified diet in cafeteria-fed rats
and find that this remains relatively constant at 17—
 CAFETERIA DIET AND THERMOGENESIS
25% of metabolizable energy intake. Statements, such
as that quoted by Moore, that "animals tended to ig
nore" the nonpurified diet, contribute little to the de
bate.
ADDITIONAL LIMITATIONS
Undoubtedly the composition of human foods used
for cafeteria diets will vary, but this effect can be min
imized by selection of standard food items. We have
found that repetitive analysis of the same foods pur
chased over several years yields reproducible values (less
than 5% variation). Furthermore, it is not always rea
sonable to assume that commercially available non-
purified diets are of constant composition (18).We have
seen variations of up to 12% in metabolizable energy
density and protein content of commercial nonpurified
diets, and variations in the source of the nutrient, and
hence taste, smell and quality may vary according to
the market prices of feedstuffs.
ALTERNATIVE DIETS
We (3-6) and others (19-22) have reviewed and tested
the methods available for producing reversible in
creases in energy intake (e.g., forced feeding, high fat
diets, sucrose solutions, pharmacological interven
tions). All have some drawbacks or limitations, the
most obvious being the alterations in nutrient com
position. The statement by Moore (1) that "one can
effectively stimulate hyperphagia in rats without com
promising control over dietary composition if one sup
plements the diet with sucrose solutions.. ." contra
dicts her earlier arguments. Moore does not quote, nor
are we aware of, any studies that have been able to
"control" the amount of sucrose solution consumed by
rats. Thus, the total and proportional intake of nu
trients will differ from controls and vary between an
imals, as it does for the cafeteria diet. Because the ad
ditional food (in this case sucrose) contains no essential
nutrients and the intake of nonpurified diet is de
pressed, the likelihood of protein and/or micronutrient
deficiency is probably greater than for the cafeteria diet.
Moore (1) stated that alternative palatable diets to
the cafeteria system (e.g., high fat, sucrose) stimulate
hyperphagia by 50-60%, but gave no reference to this
claim. We have found (21) that such diets stimulate
energy intake by up to 30%, whereas others (22, 23)
report hyperphagia of 0-25% in animals presented with
high carbohydrate or high fat diets or with sucrose so
lutions.
CONCLUSIONS
We acknowledge that the major drawbacks of the
cafeteria diet are the variations in nutrient composition
Downloaded from https://academic.oup.com/jn/article-abstract/118/8/925/4738924
by guest
on 25 January 2018
927
and the poor control over this factor. Nevertheless, it
is possible to measure micronutrient intake and to vary
this substantially by selection of the food items pre
sented.
There is no evidence for nutritional deficiencies in
cafeteria-fed rats and, for a given level of hyperphagia,
the effects on DIT and energy balance are consistent.
Both hyperphagia and alterations in nutrient consump
tion stimulate DIT independently and synergistically.
Cafeteria experiments have been performed that clearly
dissociate the effects of hyperphagia from nutrient
composition and/or deficiency and that demonstrate
that increases in intake per se can stimulate thermo-
genesis.
Finally, although biologists must strive to design and
perform controlled experiments that yield readily in
terpretable data, they must not lose sight of the rele
vance of this work to natural biological conditions. In
this respect the cafeteria diet may have some advan
tages over purified powdered or pelleted nonpurified
diets. We believe that the value of the cafeteria diet
can be best appreciated by considering its contribution
to our understanding of energy balance regulation. This
argument is based on that well-known nutritional theo
rem that "the proof of the pudding is in the eating."
As Moore (1)clearly stated, "the concept of DIT is not
new" and first appeared (as luxuskonsumption) over a
century ago. However, in spite of seemingly adequate
methodology, little progress was made in this area until
the late seventies. It can hardly be considered coinci
dental that since the development of the cafeteria feed
ing system there has been an explosive increase in stud
ies on DIT that, hopefully, has considerably enhanced
our understanding of energy balance regulation.
A Mediine search for papers describing animal ex
periments using the cafeteria diet revealed that over
400 have appeared in the past 10 years. The cafeteria
diet has obviously proved useful to many workers, and
Moore's criticisms should not dissuade other research
ers from adopting it. Provided that the limitations of
cafeteria diets are realized and appropriate caution is
exercised over interpretation of the data, this system
can be an appropriate and useful tool for studies of
thermogenesis and has already proved invaluable.
LITERATURE CITED
1. MOORE,B. J. (1987| The cafeteria diet—an inappropriate tool
for studies of thermogenesis. /. ÑutÃ-.117: 227-231.
2. ROLLS,B. J., ROWE,E. A., FAHRBACK, S. E., AGUIS,L. & WILLIAMSON,
D. H. (1980] Obesity and high energy diets reduce survival
and growth rates of rat pups. Proc. Nutr. Soc. 39: 51A (abs.).
3. ROTHWELL, N. J. & STOCK,M. J. |1981) Regulation of energy
balance. Annu. Rev. Nutr. 1: 235-256.
4. ROTHWELL,N. J. & STOCK,M. J. (1983) Luxuskonsumption,
diet-induced thermogenesis and brown fat: the case in favour.
Clin. Sci. 64: 19-23.
 928 ROTHWELL
5. ROTHWELL, N. J. & STOCK,M. I. |1984) Diet-induced thermo-
genesis and brown fat. In: Endocrinology (Labrie, F. & Proulx,
L., eds.), pp. 419-422, Elsevier, Amsterdam.
6. ROTHWELL, N. ]. & STOCK,M. J. (1986) Brown adipose tissue
and diet-induced thermogenesis. In: Brown Adipose Tissue
(Trayhurn, P. & Nicholls, D., eds.), pp. 269-298, Edward Arnold,
London.
7. STOCK,M. }. & ROTHWELL,N. J. (1987) Criteria and experi
mental evidence for luxuskonsumption. In: Recent Advances in
Obesity Research (Blondheim, S. H. &. Berry, E. M., eds.), pp.
124-130, Libbey, London.
8. ARMITAGE, G., HERVEY, G. R., ROLLS,B. J., ROWE,E. A. & TOBIN,G.
(1983) The effects of supplementation of the diet with highly
palatable foods upon energy balance in the rat. /. PhysioL (Lon
don) 342: 229-251.
9. BARR,H. G. & MCCRACKEN,K. J. (1984) High efficiency of
energy utilization in "cafeteria" and force-fed rats kept at 29°C.
Bi. J. Nutr. 51:379-387.
10. ROTHWELL, N. J., SAVILLE,
M. E. & STOCK,M. ]. (1982) Effects
of feeding a "cafeteria" diet on energy balance and diet-induced
thermogenesis in four strains of rat. /. Nutr. 112: 1515-1524.
11. ROTHWELL, N. J. & STOCK,M. J. (1983) Effects of age on diet-
induced thermogenesis and brown adipose tissue metabolism in
the rat. Int. ¡.Obes. 7: 538-589.
12. ROTHWELL, N. I-, STOCK,M. ]. & WARWICK,B. P. (1985) Energy
balance and brown fat activity in rats fed cafeteria diets or high-
fat semisynthetic diets at several levels of intake. Metabolism
34: 474-480.
13. ROTHWELL, N. J., STOCK,M. J. & TYZBIR,R. S. (1982) Energy
balance and mitochondrial function in liver and brown fat of rats
Downloaded from https://academic.oup.com/jn/article-abstract/118/8/925/4738924
by guest
on 25 January 2018
AND STOCK
fed "cafeteria" diets of varying protein content. /. Nutr. 112:
1663-1672.
14. ROTHWELL, N. J. & STOCK,M. }. (1987) Influence of carbohy
drate and fat intake on diet-induced thermogenesis and brown
fat activity in rats fed low-protein diets. /.Nutr 117: 1721 -1726.
15. ROTHWELL,N. J., STOCK,M. J. & WARWICK,B. P. (1983) The
effects of high fat and high carbohydrate cafeteria diets on diet-
induced thermogenesis in the rat. Int. /. Obes. 7: 263-270.
16. CLICK,Z., WICKLER,S. J., STERN,J. S. & HORWITZ,B. A. (1984)
Regional blood flow in rats after a single low protein high car
bohydrate test meal. Am. f. PhysioL 247: R160-R166.
17. ROTHWELL, N. J. & STOCK,M. ]. (1986) Body weight and brown
fat activity in hyperphagic cafeteria-fed female rats and their
offspring. Biol. Neonate 49: 284-291.
18. WISE, A. (1980) Nutritional differences between stock diets
for the breeding and maintenance of rats. Proc. Nutr. Soc. 39:
46A (abs.).
19. KANAREK,R. B. & HIRSCH,E. (1977) Dietary-induced overeat
ing in experimental animals. Fed. Proc. 36: 154-158.
20. SCLAFANI,A. (1978) Dietary obesity. In: Recent Advances in
Obesity Research, vol. 2 (Bray, G. A., éd.),pp. 123-132, New
man, London.
21. ROTHWELL, N. J., SAVILLE, M. E. & STOCK,M. J. (1982) Acute
and chronic effects of sucrose and glucose on thermogenic re
sponses and weight gain in the rat. Proc. Nutr. Soc. 41: 8A (abs.).
22. SCLAFANI,A. & XENAKIS,S. (1984) Sucrose and polysacchride
induced obesity in the rat. PhysioL Behav. 32: 169-174.
23. BARTNESS,T. J., RUBY,N. F. &. WADE, G. N. (1984) Dietary
obesity in exercising or cold-exposed Syrian hamsters. PhysioL
Behav. 32: 85-90.