Effects of changes in palatability on food intake and the cumulative food intake curve in man

Appetite, 1986, 7, 85-96

Effects of Changes in Palatability on Food Intake and the Cumulative Food Intake Curve in Man ESTHER M. BOBROFF and HARRY R. KISSILEFF Columbia University, College of Physicians and Surgeons, St. Luke's-Roosevelt Hospital Center, New York

This study was undertaken to quantify the relationship between palatability ratings and food consumption and to determine whether the initial rate of eating was affected by changes in food palatability without a change in nutrient content. Both men and women were given small samples offoods at a brief exposure taste test and asked to rate how much they liked or disliked them on a 9-point scale (like extremely to dislike extremely). Those who gave at least a 2-point difference in rating between a banana colada frozen yogurt drink with and without adulteration with cumin were given these foods to eat to satiety on non-consecutive days, and the same 9-point scale was used to rate these foods after they had been eaten as meals. Cumin was used as an adulterant because it is not intrinsically unpalatable, but is not liked by many individuals in yogurt-based foods. Intake was approximately 100 g different for every unit of difference on the scale, and higher for liked than disliked food. Intake was significantly different between the adulterated and unadulterated meals. The percentage of variance explained by the difference in palatability was 34% of the total variance but was 67% of the variance within subjects. Correlation between intake and ratings were poor across subjects for both palatability levels (i.e. adulterated and unadulterated). The initial rate of eating was significantly higher under the better liked than under the less liked food. These results indicate that quantification of effects of hedonic ratings on intake within subjects is possible, but that hedonic ratings may not be good discriminators of intake differences between subjects. The initial rate of eating reflects partly on palatability. INTRODUCTION

It is generally accepted that food consumption is increased and decreased in proportion to the palatability of the food being consumed. This report was stimulated by a conversation reported at a meeting on the chemical senses and food intake. The simplicity of the question and answer are worth quoting: "Maller (Owen): Of what validity are these measures of hedonics in terms of predicting how much of something people will actually consume? Do you have any data on it? Moskowitz (Howard): No I do not know of any good literature on how much one is going to drink v. how palatable the drink is." (Moskowitz, 1977, p. 99). When we began these experiments 4 years later (fall of 1980) the situation had not changed, and it still has not changed. Although there has been some work relating palatability to amount consumed and order of consumption in a meal (Bellisle & Le Magnen, 1980), there have not been any studiesin Supported by Obesity Core Center Grant No. AM-26867. Portions of this work were presented by Esther Bobroff at the Eastern Psychological Association meeting, April, 1981, and were incorporated into her senior thesis at Barnard College, 1981. Requests for reprints should be addressed to: Dr. Harry R. Kissileff, St. Luke's-Roosevelt Hospital, 114th St. and Amsterdam Ave., New York, NY 10025, U.S.A. 0195--{j663/86/010085 + 12 $03'00/0

© 1986 Academic Press Inc. (London) Limited

86

E. M. BOBROFF AND H. R. KISSILEFF

which palatability has been manipulated independently of nutritent composition for meals consumed as such, in contrast to "taste tests" (Nisbett, 1968) in which the purpose of the test was for the subject simply to sample in order to provide a rating. This study was undertaken to answer three questions: (1) What is the quantitative relationship between palatability on a 9-point scale and change in intake in grams? (2) What percentage of the variability in intake can be attributed to effects of palatability? In other words how important a contribution does palatability make to the control of intake? (3) Does palatability primarily affect the initial rate of eating as predicted by Kissileff and Thornton's (1982) theoretical model for explaining the cumulative food intake curve in man. According to their model, the initial rate of eating, reflected in the linear coefficient of the curve, should increase with increases in 'food palatability, but the rate of deceleration which reflected post-ingestive processes would not be affected by palatability. As Pfaffmann and others (Pfaffmann, Dethier & Hegsted, 1977) have pointed out, definitions of palatability in terms of intake tend to be circular. In animal studies this circularity has been broken by means of operant (Guttman, 1953) conditioning and observation of reflex (Louis-Sylvestre & Le Magnen, 1980), or stereotypical facial, responses (Grill & Nogren, 1978). In man, the verbal response ofthe subject can be used to uncouple palatability and intake. We shall refer to this response to the question "how much do you like or dislike each food" following the ingestion of a small sample (5-7 g) under standardized conditions as the "intrinsic palatability" of the food to underscore the idea that this response reflects a property of the food, rather than a state of the subject. The subject's response to the same question after eating a meal of the food will be referred to as the "post-meal palatability rating". We recognize that the subject's physiological state can contribute to these ratings and that the ratings are likely to be more variable after eating variable amounts. We shall use "palatability" to mean either "intrinsic palatability" or "post-meal palatability rating". Distinction between the two will only be made when the two measures relate differently to intake. The subject's state is considered a random variable from this point of view. The reason for interest in quantifying the relationship between palatability and food intake is to determine whether it might be possible to adjust intakes on standardized eating tests for differences in the intrinsic palatability of the food. If direct manipulation of palatibility with an unchanged nutrient content could be shown to change intake in an orderly way, for a single food, it would justify the possibility of correcting intake for palatability. However, if the relationship between palatability and intake were shown only by correlation, without manipulation, there would be no basis for asserting a causal relationship between palatability and intake, and hence no solid basis for correcting intake for palatability. Although no set of observations alone can establish with certainty that a correlation between intake and rated palatability is causal, the demonstration of such a relationship, by intervention, strengthens the hypothesis that there is a causal basis for similar results obtained by observation alone. Ideally, the intervention should be carried out with a basal food which can be made either more or less palatable by addition of small amounts of agents which alone would not be expected to influence intake. We originally tried to improve the palatability of a yogurt and fruit mixture previously employed in this laboratory, (Kissileff, Klingsberg & Van Italie, 1980) but we could not get a large enough difference between two versions of the test food. We therefore tried the opposite approach of starting with a highly palatable food and adding an aversive adulterant. After some further preliminary testing we decided to use a banana-colada frozen yogurt drink adulterated with cumin.

THE EFFECT OF PALATABILITY ON FOOD INTAKE

87

This adulterant was chosen for three reasons: (1) Small quantities have potent effects on the odor and acceptability of the food. (2) It is not inherently aversive, but is inappropriate in this food for most subjects in our subject pool. (3) The same adulterant could be used in a different food (e.g., Mexican style food) to increase the palatability, so that any effects of its addition could be attributed to psychological, rather than physical, properties of the food. Subjects were therefore given separate meals with the adulterated and unadulterated foods, and their brief exposure taste test ratings and post-meal palatability ratings were compared with their intakes and cumulative intake curves under each condition. METHODS

Overall Procedures

Sixteen people were selected by demographic and behavioral criteria from a population of 114 (see Subject Selection). The first behavioral test required the subjects to rate small portions of the foods they would later eat as meals (see Taste Test). Those who were accepted ate three luncheon test meals in the laboratory, 3 h after a standardized breakfast (see Test Foods), on non-consecutive days as close to one another as the subjects' availability allowed. The first meal (screening) was given to ensure that subjects would eat enough to obtain a reliable cumulative intake curve. Its composition was slightly different than the last two meals. The last two meals differed only in the presence or absence of the adulterant cumin and were given in counterbalanced order for each pair of subjects. After each meal, the subject filled out a rating sheet which included questions about hunger/satiety level and various ratings of the meal (see Daily Procedures). Upon completion of the third meal, subjects were scheduled for a final interview at which additional questions were asked, and they were paid U.S. $15·00. Subject Selection

Demographic criteria for selection of the eight men and eight women were as foHows: age 17-25, non-smokers, within 10% of desirable body weight for height, and no medical problems. Behavioral criteria were: (1) At a taste test, (see Taste Test) difference of at least 2 on a 9-point scale of like/dislike (Peryam & Pilgrim, 1957) for rating of adulterated and unadulterated versions of the food to be used as test meal. (2) Maintenance of at least this difference in rating ofthe meal after it was consumed for the two test meals. (3) Intake of at least 250 g on the screening test meal. The reason for setting the criteria for selection at a difference of 2 on the 9-point scale, is that this difference exceeds what is necessary for significance in ratings to be achieved, and we believed that a difference of this magnitude would be necessary to obtain a significant difference in intake. The demographic characteristics of the subjects are presented in Table 1 and the subject selection process is diagrammed in Table 2. Taste Test

The details of this procedure have been described elsewhere (Kissileff, Pi-Sunger, Thornton & Smith, 1981; Kissileff, 1985). In brief the subject sampled cups containing 5 to 7 g aliquots of the foods to be tested. Subjects were instructed by a pre-timed tape recording when to sample, report ratings, and rinse. Ratings were reported on the 9-

88

E. M. BOBROFF AND H. R. KISSILEFF TABLE 1 Demographic data (N = 16)

Height (em) Sex Female Female Male Male

Weight (kg)

Age (years)

% ideal weight

Cumin ratio

Mean

SD

Mean

SD

Mean

SD

Mean

SD

1 :250 1:375 1 :250 1:375

165·4 160·7 180-9 178-4

2·9 1·2 1·6 1·9

57·9 55-7 71 -0 73 -3

1·9 1·2 1·9 2-4

20·2 20·7 21·5 20·0

1·8 I-I 0-9 1-7

2·1 4-5 0-0 4-1

2-2 2-2 4-1 1-4

TABLE 2 Selection process

Procedure

N

Screened Rejected because of taste ratings Given meals Dropped from study Rejected for post-meal palatability ratings Rejected for lack of low-palatable food consumption Used in study

114 93 21 5 4 1 16

Note: Number of subjects at each level of indentation adds to level above it.

point like-dislike scale of Peryam and Pilgrim (1957). Descriptors on the scale (from extremes to middle) were: extremely, very much, moderately, slightly, and neither. Test Foods

Subjects were served an English muffin (Thomas) with 1·5 pats of butter and 249 g of a pple juice (Red Cheek) for breakfast. The screening meal consisted of 750 g of vanilla !yogurt (Dannon), 300 g of canned apples (Flagstaff, brand, Fancy York Sliced Imperial Apples) washed to remove sweeteners, and 450 g of canned bananas (Chiquita brand, in heavy syrup, but washed to remove syrup) blended together. The test meals consisted of a banana-colada frozen yogurt drink served plain, with 6 g cumin (McCormick brand crushed cumin seed obtained at a local supermarker) per 1500 g drink (1 : 250 ratio), or with 4·5 g cumin per 1700 g drink (1: 375 ratio)_ For test meal composition, see Table 3. Lunch was served at 50°F. A liquid was selected for two reasons: (1) Cumulative intake curves obtained from liquid food are smoother than curves obtained from solid food, and (2) the difference in palatability could be obtained by addition of a small quantity of an adulterant, that would mix uniformly and would not substantially change the caloric value. Two levels of adulterant were used because after the first group was tested, it appeared that the level of cumin was too high. As a result intakes were too low and meal durations too short to model the cumulative intake curve, in two of the subjects (who were replaced). No problems were encountered with the 1 : 375 adulteration amount.

89

THE EFFECT OF PALATABILITY ON FOOD INTAKE TABLE 3 Composition of banana-colada fro zen yogurt drink

Ingredient Vanilla frozen yogurt" Banana-colada drink b Coconut milk Shredded coconut Banana White grape juice Pear Lime juice Drink subtotal Total Total calories (approx.)

Weight (g)

Energy (kcal)

Protein (g)

Fat (g)

Carbohydrate (g)

1125·00

1243·00

45·20

11·30

237-30

93-75 37·50 112'50 93·75 30·00 7·50 375·00 1500·00

236·25 129·75 95·68 49·63 75·60 8·85 595·77 1838·80

lOO

23-34 13-24 0·22 0·00 0·51 0·04 37·35 48·65 11·99

4·88 3·52 24·98 12·47 18·96 3·06 67·87 305·17 75·22

(%)

1·31 1·23 0-19 0·87 0·10 6·70 51·90 12·79

Notes: Caloric density = l·Zkcal/g. a Source, Dannon Yogurt Co. b Source, Lakewood Co., Miami, Florida. The yogurt was in a 3 : 1 ratio by weight with the drink.

Daily Procedures Each day of testing, a subject was served breakfast, which lasted 15 min, between 0800 and 1000 h. After the first breakfast session, the subject's height and weight was recorded. The subject returned after a 3-4 h deprivation period for lunch. The subject ate lunch (for meal contents see Test Foods) after a tape recording gave instructions. The universal eating monitor (Kissileff, et aI., 1980) recorded intake every 3 sec and showed this information in the adjacent room on a computer terminal. When the subject had not eaten for 5 min the computer terminal signaled the experimenter. Then the experimenter returned and handed the subject a rating sheet to fill out. The rating sheet contained questions about the palatability (same question and scale as on taste testHow much do you like or dislike the food you just ate?) taste, texture, and satiating effect of the food (Kissileff et ai., 1980). It also contained an extensive series of objective questions about the subjects' physical and psychological state Mather (Note 1) and a visual analogue scale of hunger and satiety (Silverstone, 1975). When the subject completed the questionnaire he/she was free to leave.

Design of Study and Analysis of Data Two types of analysis were done, each emphasizing a different aspect of the data. First univariate analyses of variance (2 x 2 x 2 with repeated measures on the last factor) were performed to determined the effects of sex, the two adulteration levels (l : 250 and 1 : 375) and the presence of adulterant (present or absent) on the following dependent variables: intake, duration, coefficients of cumulative intake curves (Kissileff, Thornton & Becker, 1982) hunger, satiety, taste test ratings and post-meal palatability ratings. The latter two ratings were made using the 9-point scale (Peryam & Pilgrim, 1957) at the taste test and on the sheet given for evaluation 5 min after the subject had finished the meal. Second, ratios of intake differences to palatability

90

E. M. BOBROFF AND H. R. KlSSILEFF

differences were computed and expressed directly and as percentages of the mean intake under both palatability levels in order to evaluate the quantitative relationship between intake and palatability.

RESULTS

Intake, meal duration, linear coefficient, post-meal satiety rating, post-meal palatability rating and taste test rating were all significantly less under the cumin adulteration level than under the unadulterated banana-colada, for both sexes and under both adulteration amounts. Hunger rating was significantly greater after the adulterated than after the unadulterated food. There were no significant effects of sex, adulteration level or sex x adulteration interaction. The post-meal palatability ratings averaged approximately one unit lower than the taste test ratings. The quadratic coefficient was reduced in absolute value (i.e. it became less negative) but not significantly so by the cumin. (See Table 4 and Appendix for details on all variables). Intake of the high (815'9 g) and low (768·8 g) palatable foods was not significantly different for the four subjects who were dropped because they did not maintain their taste test differences when given the post-meal palatability questionnaire (mean ± SD 47·1 ±68·4, t=O·689, dj=3, p
1400 1200 ~

'"

1000

c:

800

.><

~

"'" E

600

cv

400

-.;

f-

G

200

o 1•8=--=2:-'-'4-:-=-::3.L.0~3.L.6~4"':'2-4.L8---="5.L4-6-:-"-'-0-6=-"-:-6---:7::'-.2::--7...1...8=--8.L.4-9 L1.

.LJ.

0

Taste test rating

FIGURE 1. Intake plotted against palatability rating, for the brief exposure taste test given when subject was screened and before any other food was eaten in the morning. Each subject is indicated by a separate letter. Lines connecting the same letters indicate the slope of intake to palatability rating for each subject. The letters at the left of each line indicate intakes for the adulterated version and the letters at the right indicate intakes for the unadulterated version of the test meal.

Intake (g) Duration (min) Linear coefficient Quadratic coefficient Hunger Satiety Taste test ratings Post-meal ratings

Variable

Female 1:375

784·7 4·9 217·0 -14·2 5·1 7·1 7·5 6·7

232·3 1·8 104·4 -17-7 9·0 3·1 3·2 0·,

652-1 4·7 291-8 -31·5 4·6 8·9 7·2 \-S:

354·0 5·2 129·2 -12·8 9·2 6·0 3·9 3·2

829·3 7·3 159·7 -9·2 2·3 11·6 7·6 8·0

284·2 2·9 62·3 40·9 g·3 4·6 2·9 2·5

429·4 2·5 106·4 -15·7 -4·3 404 4·5 404

1 :250

628·8 306·8 5·8 2·7 211·5 158·2 -20·3 ' -22·9 2·5 5·2 10·0 6·2 g·3 2·7 7·5 3·2

1:375

Mean

Male

82·5 0·7 31·6 11·6 1·0 0·9 0·2 004

SED

Difference (Banana colada- Cumin)

Average

Banana Banana Banana Banana colada Cumin colada Cumin colada Cumin colada Cumin

1 :250

4 Results

TABLE

.....,

\0

tTl

;0:::

>

Z .....,

80

~

z

><: 0

:;

t'"'

E:::

>

> t'"' > .....,

."

~

0

n .....,

tTl

~

~

tTl

tTl

::t

92

E. M. BOBROFF AND H. R. KISSILEFF

1400 1200 ~ ~

1000

0

~

SOO

.,

"0 E

SOO

~

t!

400 200

Post-meal rating

FIGURE 2. Intake plotted against palatability rating for post meal rating given 5 min after meal was finished. Rest of information is same as Figure 1.

Another important relationship was that between the average intakes of each subject and the average palatability rating. There was a surprise here in that intake was positively correlated for the post-meal ratings but was negatively correlated for the taste test ratings (see Figures 3 and 4). One contributor to this anomalous finding is that individuals with the highest intakes tended to show higher post-meal than taste test ratings and individuals with lower intakes tended to show lower post-meal ratings than taste test ·ratings. Finally, the percentages of variance in intake explained by the various sources in this study were as follows: The between-subject variance accounted for 47·4% and the within-subject variance accounted for 52·6% of total variance. Of the between subject variance 0·9% (0·4% of the total) was attributable to adulteration levels and only 0·02% of between (0·008% of the total) was attributable to sex. The between-subject error variance accounted for 46·6% of the total variance. Of the within-subject variance

1000 900

K B

SOO 700 o

II>

E

M

A

p

SOO 500 400

c: o

II>

300

:::<

200 100

F

L C

0 D J

G

N

o'Lt1-.4:L.2::-:-4.L.4:-:-4.'-::S-4-="'."=S-c5::'-.-=-O--=5"-.2::-::5.L.4:-::-5.'-::S-c5='.-=-S-cS::'-.-=-0-::6:L·2::-::S.L.4:-S~·S Mean taste test rating

FIGURE 3. Mean intake plotted against mean palatability rating, for the brief exposure taste test rating given in the morning before any food was eaten. Each subject is indicated by a separate letter. The regression line for intake (/) as a function of rating (R) is 1= 2163 - 305R; r = - 0 516,

p=O·0406.

THE EFFECT OF PALATAHILITY ON FOOD I:-.ITAKL

-

..'"

-"

E

!OOO

c

700 600

;;

500

..

400

~

c 0

:<

B

800

.,

E

K

900

"0

300

A

M

:~

~"

200

D

100

'4,,1

o

93

:3·0

I

3·5

I

4·0

C

G

J,N

I

I

4·5

5·0

I

5·5

6·0 6·5

7·0

7·5

Mean post -meal rat,ng

FIGURl 4. Mean intake plotted against mean palatability rating, for the post-meal rating given 5 min after meals were eaten. Each subject is indicated by a separate letter. The regression line for intake (/) as a function of rating (R) is 1= - 105 + 120K r = O· 502, p = 0·04 7.

65'8'>0 (34'6~'n of total) was explained by palatability and 29·1 ~;; (15'3~~ of total) by error. The palatability by sex and adulteration level interactions made negligible contributions to intake variance. DISCUSSION

These results show a quantitative relationship between differences in intake and differences in palatability rating within the same person and suggest its magnitude using a commonly employed scale: A change of one unit on the scale results in a mean change of 98·56 g in intake for post-meal palatability rating and 97·55 for taste test rating. Obviously it would be important to know whether this relationship would hold under other dietary conditions. A more generalizable expression for the relationship of intake to palatability could be derived by expressing the intake difference for different palatability levels as a percentage of the mean across all palatability levels. When this was done in the present study, the average intake changed by 19'3'):" of the mean per unit of post-meal palatability rating and by 18'7~~ of the mean per unit of taste test rating. When a similar calculation was carried out on a previously reported study (;'IJisbett, 1968), which also used a homogeneous food adulterated with a single adulterant. quinine. at approximately the same estimated concentration as our cumin adulteration (2·5 g of quinine sulphate per quart of ice cream = 1 g/363 g), intake changed by 20·4"0' !\ similar computation for another study (Hill & McCutcheon, 1975) in which subjects ate a variety of mixed meals resulted in a calculated ratio of only 8·3. In order to carry out these comparisons. the 6-point scale of Nisbett and the 11point scale of Hill and McCutcheon were converted to a 9-point scale. The following formula was used for the conversion:

9-Point scale difference =

(N-Point scale difference) (Maximum difference M' d·~··· x ( aXlmum I erence on 9-point scale) on N -point scale

where N = number of points in scale.

94

E. M. BOBROFF AND H. R. KISSILEFF

These considerations make it possible to predict the amount of change in intake associated with a difference in palatability without change in nutrient content for any given level of intake, including, for example, large intakes of main courses or small intakes of desserts. Although differences in intake within a subject were obtained when there were large differences in palatability rating within the same subject, absolute intakes correlated poorly with both the taste test and post-meal ratings, and the average intake correlated negatively with taste test ratings. These results indicate that taste test ratings are poor predictors of the actual amount a person will consume in comparison to other people, but that they can be reliable guides to consumption differences within a person. Therefore it would not be practical to adjust intake scores between people using palatability ratings as covariates, but it might be practical to adjust differences within subjects by using such ratings. More work will be necessary to determine whether this relation is linear or non-linear. It is also important to note that the four subjects who were dropped due to a lack of difference in the post-meal palatability ratings illustrate that a brief exposure taste test only predicts intakes accounting for about 75% of the subjects. We have no explanation for the inconsistency between taste test and postmeal ratings in the other 25%. These results also support the hypothesis advanced by Kissileff and Thornton (1982) that the linear coefficient of the cumulative intake curve is influenced by food palatability. This idea is consistent with data from the animal literature originally reported by Shulford (1959). He showed that palatability measured in a brief exposure preference test predicted the magnitude of the early rise in the cumulative consumption curve while post-ingestive effects influenced deceleration of the curve. He suggested that intake was determined by a force that facilitates ingestion and by an opposing force which inhibits it. Kissileff and Thornton (1982) quantified the expression of these forces and Kissileff et al. (1982) showed that a quadratic equation is sufficient to describe them in man. These results are also consistent with the findings of Bellisle and Le Magnen (1980) who showed that the rates of chewing and swallowing are increased with increasing palatability and that these parameters were as sensitive at the end of the meal as at the beginning. That is, chewing rates were lower at the end of the meal than at the beginning of the meal and were lower for less palatable than for more palatable foods. But there was no interaction between time of consumption (beginning or end of meal) and palatability. It remains to be determined what the relationship is between the slowing of intake, the hedonic value of the food, and the development of satiety. The model of Kissileff and Thornton (1982) proposes that the decline in intake is exclusively the result of the onset of satiety. Bellisle and Le Magnen (1980) also found that "both chewing time per food unit and the pause between two food units increased from the beginning of the meal to the end, independently of palatability. In other words, the eating rate slows down as satiation grows. Satiation thus seems to act in the same direction as a decrease in palatability." In contrast, the results of Rolls, Rolls, Rowe and Sweeney (1981) suggest that there are small changes in palatability during the course of a meal. They found that, from the beginning to the end of the meal, palatability ratings decline for foods eaten but remain the same for foods not eaten in the meal. Although it is not clear whether satiation causes decreases in palatability or whether decreases in palatability lead to satiation or whether these two variables are simply independent factors resulting from food consumption, additional observations in the present experiment support the latter view. First, subjects voluntarily terminated

THE EFFECT OF PALATABILITY ON FOOD INTAKE

95

meals of different palatabilities without equal hunger and satiety ratings. Subjects were less satiated when they stopped eating the less palatable food than when they stopped eating the more palatable food. These results indicate that when food palatability is substantially decreased by experimental manipulation, subjects stop eating before they are normally satiated. If satiety were induced when palatability reached a critical value we would have obtained similar satiety ratings when subjects stopped, regardless of initial or terminal palatability. Second, our data on palatability ratings following the meal are not consistent with the idea that a drop in palatability terminates themeal. Ratings after the meal were at most one unit below ratings at the brief exposure test. Furthermore, termination of a meal without satiety when the unpalatable food was eaten are also not consistent with the hypothesis that a change in palatability constitutes a component of satiety, sensory-specific or otherwise. These facts are most parsimoniously explained by the two-component hypothesis. According to it, a less palatable food will produce a lower initial rate of eating and a smaller level of inhibition from post-ingestinal factors will be sufficient to offset it, thereby leading to a lower intake. The fact that palatability remains relatively high following the meal, and yet eating stops, is explained by the rise of a post-ingestive inhibitory factor rather than a decline in acceptability or palatability of the food. This view is also consistent with animal data showing that when food is drained from the stomach shortly after eating, consumption resumes (Snowdon, 1970) and that, even when large amounts-sufficient to lower palatability substantially by fatigue alone-pass through the mouth but do not remain in the stomach, eating persists (Young, Gibbs, Antin, Holt & Smith, 1974). REFERENCE NOTE I. Mather, P. Covert nutrient supplementation and normal feeding in man and rat: Experiment-

ation and simulation. Master's Thesis, Birmingham University, 1977. REFERENCES

Bellisle, F. & Le Magnen, J. The analysis of huma n feeding patterns: the Edogram. Appetite, 1980, 1, 141-150. Grill, H. & Norgren, R. Taste reactivity test. I. Mimetic responses to gustatory stimuli in neurologically normal rats. Brain Research, 1978, 143, 263-279. Guttman, N. Operant conditioning, extinction, and periodic reinforcement in relation to concentration of sucrose used as reinforcing agent. Journal of Comparative and Physiological Psychology, 1953, 46, 213-224. Hill, S. W. & McCutcheon, N. B. Eating responses of obese and non obese humans during dinner meals. Psychosomatic Medicine, 1975,37, 395-40L Kissileff, H. R. Quantitative relationship between palatability and food intake in man. Tn M. Kare & 1. Brand (Eds.), The chemical senses and nutrition (Vol. 3). New York: Academic Press (in press). Kissileff, H. R. & Thornton, J. Facilitation and inhibition in the cumulative food intake curve in man. In A. 1. Morrison & P . Strick (Eds.), Changing concepts of the nervous system. Pp. 585607. New York: Academic Press, 1982. Kissileff, H. R., Klingsberg, G . & Van Itallie, T. B. Universal eating monitor for continuous recording of solid or liquid consumption in man. American Journal of Physiology, 1980, 238, RI4- R22. Kissileff, H. R., Pi-Sunyer, F. X., Thornton, J. & Smith, G. P. C-terminal octapeptide of cholecystokinin decreases food intake in man. American Journal of Clinical Nutrition, 1981, 34, 154- 160. Kissileff, H. R., Thornton, 1. & Becker, E. A quadratic equation adequately describes the cumulative food intake curve in man. Appetite, 1982, 3, 255-272. Louis-Sylvestre, J. & Le Magnen, J. Palatability and prea bsorptive insulin release. Neuroscience and Biobehavioral Reviews, 1980, 4 (Suppl. 1), 43-46.

96

E. M. BOBROFF AND H. R. KISSILEFF

Moskowitz, H. R., Intensity and hedonic functions for chemosensory stimuli. In M. R. K are & O. Maller (Eds.), The chemical senses and nutrition. Pp. 71-99. New York: Academic Press, 1977. Nisbett, R. E. Taste, deprivation, and weight determinants of eating behavior. journal of Personality and Social Psychology, 1968, 10, 107-116. Peryam, D. R. & Pilgrim, F. J. Hedonic scale method of measuring food preferences. Food Technology Supplement, 1957, 11, 9-14. Pfaffmann, c., Dethier, V. G. & Hegsted, D. M. Concluding comments. In M. R. Kare & O. Maller (Eds.), The chemical senses and nutrition. Pp. 463-475. New York: Academic Press, 1977. Rolls, B. J., Rolls, E. T., Rowe, E. A., & Sweeney, K. Sensory specific satiety in man. Physiology and Behavior, 1981,27, 137- 142. Shuford, E. H. Jr. Palatability and osmotic pressure of glucose and sucrose solutions as determinants of intake. Journal of Comparative and Physiological Psychology, 1959, 52, 150--153. Silverstone, T. Anorectic drugs. In T. Silverstone (Ed.), Obesity, its pathogenesis and management. Pp. 193-227. Acton: Publishing Sciences Group, 1975. Snowdon, C. T. Gastrointestinal sensory and motor controls of food intake. journal of Comparative and Physiological Psychology, 1970, 71, 68-75: Young, R. C., Gibbs, 1., Antin, J., Holt, J., & Smith, G. P. Absence of satiety during sham feeding in the rat. Journal of Comparative and Physiological Psychology, 1974, 87, 795-800.

Received 13 August 1984, revision 11 February 1985 ApPENDIX

Analysis of variance Intake Source

df

MS

AdAmt" Sex AxS Error Pal b PxA P xS PxAxS Error

1 1

17813 336 13936 165621 1475160 113931 140 485 54394

1 12 1 1 1 1 12

Hunger Source

MS

4·81 AdAmt 46·08 Sex AxS 3-25 15·48 Error 147·06 Pal 3·12 PxA 0·02 PxS PxAxS 8·20 8·66 Error

linear ' coefficient

Duration

F

MS

1·38 2·28 11·82 6·95 27 ' 12~ 50·63 2·09 12·44 0'00 11·94 0·01 2·50 0'11 0·00 0·08

J88

- - -- - MS

0·20 0·33 1·70

30"596 11347 1155 14517 90627 17 7766 4430 7987

l3-04§ 3·20 3·08 0·64

Taste test

Satiety

F

MS

F

MS

0·31 2·98 0·21

11·05 26;10 11·05 12·00 155'76 8·92 7'80 2·26 7'12

0·92 2·17 0·92

0·46 0·06 0·00 0'44 162·68 0·01 3·54 1·69 0·38

16·97§ 0·36 0·00 0·95

F

F

21'86~

1·25

HO

0·32

F

Quadratic coefficient

3815 2094 1949 1104 ll '35t 1961 0·00 464 0·97 526 0·55 2809 1072 2·11 0·78 0·08

3-45 1·90 1·76 1·83 0-43 0·49 2·62

Post-meal ratings MS

1·06 0'13 0·00

0·50 1·12 0·13 2-81 425-44 11 153-12 0·02 1·12 2·00 9'27t 4'42* 0·50

*p <0'06; **p<0'05; tp<0'OO5; fp
F

MS

HO

F 0·18 0-40 0·04 138·6811 1·02 1·81 0-45