Amount of feeding activity and size of meals in free-feeding rats

Amount of feeding activity and size of meals in free-feeding rats

Physiology& Behavior,Vol. 25, pp. 893--899.Pergamon Press and Brain Research Publ., 1980. Printed in the U.S.A. Amount of Feeding Activity and Size o...

589KB Sizes 0 Downloads 31 Views

Physiology& Behavior,Vol. 25, pp. 893--899.Pergamon Press and Brain Research Publ., 1980. Printed in the U.S.A.

Amount of Feeding Activity and Size of Meals in Free-Feeding Rats S. D. MORRISON Laboratory o f Pathophysiology, National Cancer Institute, National Institutes of Health, Bethesda, MD 20205

Received 21 December 1979 MORRISON, S. D. Amount offeeding activity and size of meals infree-feeding rats. PHYSIOL. BEHAV. 25(6) 893-899, 1980.--Instrumentation is described for simultaneous measurement of feeding duration (time actually spent feeding during a nominal meal) and weight of food eaten in individual meals throughout 24-hr periods in free-feeding rats. These variables were measured in two rat strains. Meal weight cannot usefully be estimated directly from meal duration (total time from beginning to end of a nominal meal) or from meal feeding duration. An adjustment procedure is described that removes systematic error and allows a useful estimate to be made from meal feeding duration. The discrepancy remaining between measured and estimated meal size is quite large and is a measure of the independence of control of food intake and control of the behavioral output of feeding. Apparatus

Food intake

Feeding activity

Feeding control

T H E contact " e a t o m e t e r " [3] is an inexpensive and simple way of recording feeding. Although the method is frequently cited [4, 8, 15], it seldom seems to be used, largely because it originated and is thought of as a substitute for direct measurement of food intake although, in common with other indirect measures, it has never been validated for this purpose [6]. Contact with the sensing grid of the " e a t o m e t e r " can be elaborated to operate a delayed drop-out relay with the relay driving an electric clock, so that the total feeding duration (the total amount of time spent feeding) can be accumulated to give an estimate of the behavioral output of feeding that is logically and instrumentally independent of the amount of food ingested [12]. Feeding duration, measured in this way, is quite distinct from "meal duration" which is the total time from the beginning to end of a nominal meal and is heavily influenced by the meal pattern and the meal separation criterion chosen. Meal duration contains a large and unknown component of non-feeding activity that happens to occur in the time vicinity of feeding: feeding duration excludes this component. With constant diet composition and constant environmental conditions and treatment, the ratio of daily food intake to daily feeding duration (feeding efficiency) is substantially constant within individual animals [ 11,12]. Various imposed treatments, however, can markedly shift this ratio [ 11,12], indicating that the behavioral output of feeding (as measured by total duration of feeding activity) and the amount of food ingested are subject to different determinants. Because of the simplicity and small in-cage space demand of the contact eatometer, it is tempting to assume that the ratio of intake to duration for individual meals would be the same as the total ratio for the 24-hr period in which the meal occurs and that, consequently, meal size could be estimated from meal feeding duration multiplied by the 24-hr intake/duration ratio. On the other hand, failure to estimate meal size accurately from a simple function of feeding dura-

tion would be a measure of the extent to which food intake and feeding behavior of individual meals are determined by different sets of conditions. In this study, the weight of food ingested and the feeding duration for individual meals were measured simultaneously. The relationship between the two was evaluated with respect to the utility of meal feeding duration for estimation of meal weight and with respect to the extent to which feeding behavior and food ingestion are separately controlled. METHOD The study covered 65 rat-days from 7 male Fischer rats of BW 148-274 g (a total of 718 meals) and 9 rat-days from 5 male Sprague-Dawley rats of BW 190--260 g (66 meals). A total of 78 rat-days was recorded but 4 were discarded because of instrument malfunction. Each rat was housed individually in a " L u c i t e " cage with stainless steel floor grid (living space 26×26×25 cm), with a casein-based semi-synthetic diet, C21 [9] (metabolizable energy 4.85 kcal/g), and water available ad lib. Each day's record ran for about 23.5 hr with 0.5 hr " d o w n " time when the rat was weighed, food and water vessels were replenished and weighed and recording instrumentation was reset. Environmental temperature of 23-25°C and a constant 12-hr light cycle (lights on at 7:00 a.m. and off at 7:00 p.m.) were maintained. The " d o w n " period occurred between 9:00 a.m. and 10:00 a.m. each day. The recording periods were in sets of 4 consecutive days for the Fischer rats and 2 consecutive days for the Sprague-Dawley rats. All rats were exposed to the experimental diet, housing and instrumentation for at least one week prior to the first record collection.

Feeding Duration Feeding duration was sensed as described in detail previously [12], but, in this study, the output of the delayed drop-out relay ran a 1 rph clock motor driving a 10-turn

893

894

M()RRISON

Measurement of Food Weight and Feeding Duration Data Acquisition and Analysis

I Wang ii

a



Digital Plotter Graphic Display

Numor,ca,

Analysis

I I

Food Dish

--~ \

I Mag Tape

~ \ /,,Variable ~---~~~TI--'~/Reluctance \ I ' ~ / Transducer

/ Spring Beam

I1 P

~ Filter 'I L

Delayed Drop-Out Relay

Clock Motor

I

I ] A-D Conversion I Scanner

I

Food Weight

10-Turn Potentiometer

-] ~__ , ; D u r a t i o n

Feeding

3 Volt D.C. FIG. 1. (a) Arrangement of contact eatometer and continuous food weigher in animal cage, and accessory equipment. Dimensions in text. (b) Block diagram of system for data acquisition and analysis.

precision potentiometer (Fig. la). The regulated DC supply of 3.0 V across the potentiometer is tapped, as the potentiometer is driven, to yield 0.3 V increment per running hour of feeding activity. On the appropriate range the recording instrumentation (see below) has a resolution of 0.1 mV with a noise level (from the potentiometer power supply) of ___0.2 mV, allowing 3 sec potential resolution of feeding duration in any scanned interval. The actual time resolution of this part of the system is imposed by the time hold of the holding relay which is set at 5 sec.

Food Weight The food dish, with its sensing grid and lead for measure-

ment of feeding duration, is rigidly mounted at the free end of a 15 cm beryllium-copper bronze cantilever beam (section 25x0.4 mm) that gives a deflection of 0.04 mm/g load (Fig. la). The beam, as set up with full food dish, just clears the sensor of a variable impedance transducer (Kaman: Model K-D2300-2S). The sensor is a variable impedance bridge responding to eddy currents induced in adjacent conductive surfaces (the bronze beam). As food is eaten, the clearance of the beam from the sensor increases and increases the output of the system. Over the weight range of food dish from 100 g (full) to 70 g (30 g eaten) the transducer gives a linear output of 21 mV/g. (The DC output can, of course, be amplified if the recording system should require this.) The

F E E D I N G ACTIVITY A N D M E A L SIZE IN F R E E - F E E D I N G RATS TABLE 1 DISTRIBUTION OF FEEDING OVER 24-hr PERIOD (MEAN -+ SE) 24-hr

Light phase (rest)

Dark phase (activity)

Fischer Mean food intake g Mean no. meals Mean meal size

9.60 +_ 0.87

1.65 _+ 0.14

7.95 _ 0.20

11.05 _+ 0.28 0.87 _+ 0.017

2.42 _+ 0.17 0.70 _+ 0.029

8.63 ___0.28 0.92 _ 0.021

g

Total no. rats=7; no. rat-days=65; no. meals=718. Sprague-Dawley Mean food intake

14.27 _+ 1.11

1.60 +_ 0.50

12.65 _+ 0.86

Mean no. meals 7.44 _+ 0.53 1.11 - 0.35 Mean meal size 1.92 _+ 0.12 1.44 _ 0.21 g Total no. rats=5; no. rat-days=9; no. meals=66.

6.33 _+ 0.53 2.00 _+ 0.14

g

895

duration of a meal is the amount of time actually spent feeding within that meal period. Unless it can be shown that there is a range of meal separation criterion within which there is no change in meal frequency (or a criterion value at which rate of change of meal frequency passes through a minimum), then any intermeal interval criterion chosen is arbitrary and each analysis of meal patterns is unique to the interval size chosen. This obvious limitation on analysis of meal patterns appears not to have been generally recognized, even in studies primarily devoted to the theoretical basis of analysis of meal patterns [2,13]. Kissileff's study, using an automatic pellet dispensing system, found effectively constant meal frequency with meal separation criterion of 12 to 40 min [4]. I have not found this to hold with free feeding from a pot. In the data used here there is no range of constant meal frequency and meal frequency declines at a uniform 1% per min increase in criterion from l0 to 40 min. However, since no attempt is being made here to analyze meal patterns, arbitrary choice of criterion is valid. The 10 min criterion was chosen here to demonstrate the resolving power of the system and because it is one of the values frequently adopted [l, 2, 4]. RESULTS AND DISCUSSION

beryllium-copper bronze material of the beam minimizes errors due to thermal expansion. The area of the animal floor grid immediately above the beam is filled with a stainless steel sheet to preclude errors from feces and urine dropping onto the beam. The natural frequency of the loaded beam is about 4 Hz. The output of the transducer is fed through a low-pass RC filter to eliminate error from this oscillation. On the appropriate range the recording instrumentation has a resolution of l0/zV and an accuracy of 20/zV, representing a potential weight resolution of 1 mg. In practice, meal increments of <50 mg are rejected.

Recording The outputs from the potentiometer tap and the weighing transducer are scanned every 1 min, the readings are digitized by an automatically ranging digital volt meter (DVM) and, along with time, are recorded on a magnetic tape cassette (Fig. lb).

Analysis The 24-hr data are analyzed on a Wang 2200C programmable calculator and graphics of the data are prepared by the calculator operating through a Tektronix 4662 digital plotter (Fig. lb). The arbitrary definitions of a meal, used for the analysis, are: Feeding duration: meal separation criterion, 10 min; minimum acceptable feeding duration within a l-min scan interval, 5 sec; minimum accumulated duration accepted as a meal, 12 sec. F o r each meal meeting these conditions start time, end time, meal duration and feeding duration are extracted. Food intake: meal separation criterion, 10 min; minimum weight increment accepted as a meal, 0.05 g. F o r each weighed meal start time, end time, meal duration and meal weight are extracted. This means that, for either measurement, bursts of feeding separated by < 10 min are regarded as continuations of a single meal: if separated by /> 10 min they are regarded as distinct meals. Meal duration is the total amount of time from first to last activity of a meal meeting the criteria: feeding

F o r comparison with other feeding data the general characteristics of daily food intake, meal size, meal frequency, and diurnal distribution of meals for both rat strains are given in Table 1. The daily food intake and mean meal size are larger and the meal frequency is smaller in the SpragueDawley rats than in the Fischer rats, and the fraction of the meals and of food intake occurring in the light phase is smaller (Table l). Major differences among rat strains in food intake, metabolism and control systems have been frequently demonstrated [5, 10, 16]. In this instance the differences, especially in food intake, may arise from the fact that male Fischer rats are slow growing and do not continually gain adult weight as males of most strains do, but plateau at a bodyweight of about 400 g [5]. A typical daily record (Fig. 2) shows the high stability of the food weight record. No valid weighed meal occurred that was not precisely associated in time with contact activity (Fig. 2), confirming the basic assumption of the contact eatometer that the rat must make contact with the sensing screen in order to obtain food. Contact activity meeting the acceptance criteria occurred from time to time without any food consumption (Fig. 2). Spurious contact activity meals in Fischer rats raised the apparent meal frequency 12.6% above the true (weighed meal) value. Most of these spurious meals (60%) are of short duration <0.4 min (Fig. 2) and are due to adventitious contact with the sensing screen during exploratory activity. For the Sprague-Dawley rats spurious meals raised the apparent meal frequency by 18%, and 62% of these were of short duration. Contact meal duration and weighed meal duration are determined independently (Methods). The two sets of meal duration had a correlation of +0.887. Mean contact meal duration was 8.7 and 17.6 min, mean weighed meal duration was l l . 3 and 21.3 min for Fischer and Sprague-Dawley rats, respectively. The difference between the means of the two sets of meal duration arises predominantly from the artefact inherent in the meal separation criterion. (Thus, by one method, the actual separation of two successive bursts may be 9 min and so show an elided meal and, by the other, the separation may be l0 min and so show two separate meals.)

896

CO

MORRISON

10 9

W v

8

l-Z IM

7

-1

.<

rx O O ku

r~ W I---I

0 0

6 S

4 3

:t

o z60

ZH H ~"

~\30

W(J WW I,~

0

F

E I

I

NOON

i

i

I

I

l

i

I

l

i

i

7pm TIME

i

I

I

I

MIDN

OF

DAY

-

I

I

I

i

I

I

i

J

7om

HR

FIG. 2. Typical dual 24-hr record of food intake (upper) and contact feeding activity (lower). Lights went off at 7:00 p.m. and on at 7:00 a.m. The height of each spike in a cluster in the feeding activity record represents the number of seconds of feeding recorded in each 1-min scan interval. The thickness of the activity baseline represents electronic noise. Spurious meals can be seen during the first 2 hr and at the end of the record (Fischer rat).

Also, the rat frequently continues to rest its paws on the rim of the food dish after termination of feeding and this appears as continuation of the meal by the weighing criterion but not by the contact criterion. There was no indication that the disturbance of the daily '~down" period stimulated feeding activity. The mean time to first meal after start of daily recording (Fischer series) was 135 min with a mean first meal size of 0.63 g compared with an overall light phase meal of 0.7 g (Table 1). The mean time to first meal of the dark phase was 25 min from time of lights off, with the first dark phase meal size of 0.95 g compared with an overall dark phase meal size of 0.92 g (Table 1). Although the sensing screen for contact activity undoubtedly represents an obstacle to food access, it does not depress the total daily food intake. Average daily food intakes for rats of comparable bodyweight and housing conditions and with identical diet available in identical pots but without sensing screen or instrumentation were 14.5 g (20 SpragueDawley rats) and 9.5 g (10 Fischer rats) (cf Table 1). It was apparent from the first records that the feeding efficiency ratio for individual meals, meal intake/meal feeding duration (I/D), was not constant throughout the day and was not equal to the 24-hr ratio (I*/D*). The simple estimation of meal intake from meal feeding

duration (estimated intake=I'=D.I*/D*) yielded no usable relationship with measured intake (I) (Fig. 3a). Ratios for individual meals started much higher for the earliest meals and ended much lower for the latest meals than the 24-hr ratios (Fig. 3b). This change in feeding efficiency ratio is not related to size of meal: correlations (r) between efficiency and weight of meal are - 0 . 0 5 (Fischer) and -0.11 (Sprague-Dawley). It seemed likely that the declining I/D for successive individual meals was caused by increase in the amount of feeding activity required per unit of food ingested as the level of food in the food dish fell. It was postulated that the ratio I/D declines exponentially with cumulative food consumption. Since the primary object is to attempt to derive meal intake from feeding duration, it was further postulated that I/D declines exponentially with cumulative feeding duration. This is represented by: (I/D).(D*/I*) = A . e x p ( - k . X D )

(1)

where I and D are the measured intake and feeding duration for a given meal, I/D is a numerical differential giving rate of ingestion with respect to feeding duration, ED is the accumulated feeding duration from start of the 24-hr period to the end of that meal, and the normalizing terms I* and D* are the

897

F E E D I N G A C T I V I T Y A N D M E A L SIZE IN F R E E - F E E D I N G RATS ( a )

( b )

8

( c )

2

l,%

hi

Z H

t-%

(.9

(J bJ

r-5 (J Ld

i Q,

O H

tJ

(J v



".

"

"-':':

I

_~

I

~

..j

• •



.

•.°

:'.

.•-2..L:. •

II

~L •

I

,-, l

r

. ,.

'

tU



H

". • : - "'.. .~'~,:~': " . . ~ " . ..," " '"



• "

"'

~.I~,,.

.

• ."

Q'I~

8

I

WEIGHED



~

,

,

2

3

MEAL

-

G



.

,

"" ""~"

8

.~:.

•,

" ~

"''""

58 ACC.FEED.DUR.-

8

8

/t t-.~..

T.-.:.~.. • ~S

~It

"~

.4~

$"o



,L ,'L "

,I

! e8 MIN



I

8 I WEIGHED

I

2 MEAL

I

:3 G

-

FIG. 3. (a) Relationship of meal size estimated as meal feeding duration multiplied by total daily intake/total daily duration (uncorrected estimate) against measured meal size; (b) Relationship of individual meal ratio, intake/duration (I/D), to accumulated feeding duration ED. Mean 24-hr ratio (I*/D*)=0.27 -+ SD 0.09; (c) Relationship of corrected estimated meal size (estimated as described in text) to measured meal size. All data for Fischer rat series.

total intake (measured directly) and the total feeding duration for the 24-hr period in which the meal occurs. The constants A and k were evaluated from the logarithmic form of eq. (1). The removal of differences among rat-days (Table 2) yielded, for the joint regression, k=0.025 -+ 0.001 and In A=0.536 -+ 0.021 (A=l.71). There is a family of curves, one for each rat-day, which can be evaluated since both size of meal and feeding duration are available. However, these individual curves cannot be utilized where the objective is to predict meal size from feeding duration when no direct measurement of meal size would be available. The meal size (weight) I', estimated from feeding duration by integration of eq. (1) from beginning to end of each meal, is: I',~ = (I*/D*)-A-(exp(-k'Y--,Dn) - exp(-k'7--,Dn.0)

Source of Variance

D.F.

Regr. coeff, (k)

Regr. variance

Error variance

Variance ratio

Among rat-days

64

-0.0016

0.136

0.09

Within rat-days

653

+0.0253

98.822

0.117

844.4

Total

717

+0.0187

71.426

0.153

466.5

1.5

(2)

where I'n is the estimated size of the nth meal of the day, and XDn and XD,-I are the accumulated feeding durations from the start of the 24-hr period to the end of the nth and n-lth meals. F o r optimum estimation, the summed estimates of meal weight should equal the total measured food intake for the 24-hr period. Because of the variability of k from day to day this is not generally true when I ' , is estimated directly from eq. (2). Accordingly, after estimation of I'n from eq. (2), the individual meal estimates are multiplied by I*/~I'n, so that the final meal weight estimates, I"n, are such that: XI"n = I*

TABLE 2 ANALYSISOF CO-VARIANCEFOR DERIVATIONOF EXPONENT (k) FOR CALCULATIONOF MEAL SIZE FROM MEAL FEEDING DURATION (FISCHER SERIES)

(3)

The plot of meal intakes estimated in this way against measured meal intake is linear, with a correlation coefficient (r) of +0.92, a slope of +0.93 _+ 0.013 and SE of estimate (SD of deviation from regression) of -+0.22 g (Fig. 3c). These values and the values given above for A and k are all for Fischer data. F o r the Sprague-Dawley rats, k=0.015, r = + 0 . 8 8 and SE of estimate is -+0.58 g. However, use of a standard k (0.025) for the Sprague-Dawley data does not importantly alter the estimate (r=0.84, SE of estimate= _-.0.66 g). The deviation from 1.0 of the slope of the relationship arises from the redistribution of food intake onto spurious meals and away from true meals.

~9~

M()RRIS()N

1-1 • •





,

.

%

.

-o

* •

• o

• ••~,



0

.

t

T

l

H

0



""

b_l



".



:

~"

"

".

• ""

"

~...: ".

,~,~,:~.'.

" ":~l""~



,."

~.'~-y., ":.

.-.:

"," . ~ ' " . . ~

~',,~, ..:.,._ , , . ' ~ , ) , . ' : . . . ". , . . " , ' .

~_'.-.,e.....,~., , "

/'.'.

""

"."





..':

..

o'1O

I

c;-'I

LIGHTS OFF

W n

-2

|

I

NOON

I

|

|

i

i

I

t i

LIGHTS ON i

i

|

I

7pm

TZME

|

|

H'I'DN

OF

DAY

-

!

I

i

|

t 7Qm

Hi~

FIG. 4. Deviation (true meal size -estimated meal size) plotted against start time of meal in the 24-hr period (Fischer rat series). There is no evidence of variation in tightness of coupling of meal size and feeding behavior over the 24-hr.

Although the particular forms of curves (Fig. 3) and the parameters found here are dependent on the particular method used to assess feeding activity, the general approach is valid for assessment of the predictive value of any indirect measure of meal size that is not complicated by the meal separation criterion. The artefact introduced into meal duration by the meal separation criterion is such that no useful estimate of meal size can be made from meal duration. The linearity and slope of the relationship between measured and estimated meal size (Fig. 3c) indicates that the proposed exponential relationship accounts for almost all of the artefact. The error of estimate is small enough to allow useful estimate of meal size from feeding duration in situations where it is technically difficult or impossible to accommodate a direct food weighing device (e.g. in an animal calorimeter). However, the scatter about the ideal line of estimate is still very large (Fig. 3c) and, where direct weighing is feasible, that should certainly be used in preference to estimation from feeding duration. The precision of the measurements of feeding duration and meal weight is high and errors in these measurements, along with errors from redistribution of food among meals because of spurious meals, could not account for an error of estimate of more than +_0.05 g. (There was no detectable food scatter in any day.) Use of the individual values of k calculated for individual rat-days, to derive the durationestimated meal size does not appreciably alter the scatter. The residual scatter, then, must represent biological slackness between control of behavioral output of feeding (measured as feeding duration) and the control of food intake. It seemed possible that the stiffness of coupling might be higher during periods when the intensity of feeding might be expected to be higher, particularly the early dark period when the animal is restoring the negative energy balance incurred during the light period [7]. The deviation of estimated from measured meal intake has been plotted against meal start time in Fig. 4, and no systematic diurnal variation

in tightness of coupling is apparent over the 24-hr period. It has not, so far, been possible to find any source of systematic variation in tightness of coupling. This suggests, as has been found with total daily feeding duration and food intake when food intake is altered experimentally [11,12], that feeding activity and food ingested are determined by different sets of conditions, and that it is not justifiable to assume that food intake and feeding activity are accurate measures of one another. The shortcomings of meal duration as a measure of meal size have already been shown [2], but these are largely the result of error introduced by the meal separation criterion. The present results show that even when this error is excluded, by considering only the activity immediately involved in feeding, there is still substantial slack between feeding activity and food intake. Looseness of coupling between an external behavior and its primary consequence is not unusual, physiologically. when the behavior subserves several functions. Respiratory activity, which subserves oxygen demand, acid-base balance and heat elimination, is not always closely coupled to oxygen demand, as a glance at a panting dog will illustrate. It has been proposed that the sometimes conflicting demands are resolved by operating external respiratory activity in the way that minimizes its energy cost [14]. In a generalized feeding situation there may be several varying environmental constraints on feeding which the animal meets by varying its food-gathering strategies [1] possibly, again, to minimize energy costs of feeding [17]. It is understandable, in such a situation, that feeding activity may be only loosely coupled with its consequence, food ingestion. In the situation existing here, food is continuously and freely available and the need for strategy changes sufficient to alter feeding activity is not obvious. It is possible, however, that continued nutrient demand as sensed by, for example, blood nutrient levels sustains feeding activity, while immediately conflicting conditions, such as change in gastric wall compliance or in

F E E D I N G A C T I V I T Y A N D M E A L SIZE IN F R E E - F E E D I N G RATS gastro-intestinal transfer rate, require reduction in ingestion rate. This could be resolved by reducing the effectiveness of the feeding activity which would slacken the coupling between food ingestion and feeding activity. Increase of feeding duration with falling level of food in the food dish will also affect whole day feeding efficiencies since, as food intake increases, feeding duration will increase more rapidly and so will generate a spuriously reduced feeding efficiency. However, the changes in whole day feeding efficiency that have been found [11,12] are in the opposite

899

direction (i.e. food intake constant or increasing with a reduced feeding duration). The 24-hr effects previously reported, then, are underestimated by the artefact rather than due to it. ACKNOWLEDGEMENTS I wish to thank Dr. S. Goldstein and Messrs. M. Curtis and H. Withers, all of Biological Engineering and Instrumentation Branch, N.I.H., for design and construction of the continuous food weighing device, and Mr. E. McDuffie for technical assistance.

REFERENCES 1. Collier, G., E. Hirsch and P. H. Hamlin. The ecological determinants of reinforcement in the rat. Physiol. Behav. 9: 705-716, 1972. 2. De Castro, J. M. Meal pattern correlations: facts and artefacts. Physiol. Behav. 15: 13-15, 1975. 3. Fallon, D. Eatometer: a device for continuous recording of free-feeding behavior. Science 148: 977-978, 1965. 4. Kissileff, H. R. Free feeding in normal and "recovered lateral" rats monitored by a pellet-detecting eatometer. Physiol. Behav. 5: 163-173, 1970. 5. Kritchevsky, D. Diet, lipid metabolism, and aging. Fedn Proc. 38: 2001-2006, 1979. 6. Kumer, R., I. P. Stoberman and H. Steinberg. Psychopharmacology. A. Rev. Psychol. 21: 595-628, 1970. 7. Le Magnen, J. and M. Devos. Metabolic correlates of the meal onset in the free food intake of rats. Physiol. Behav. 5:805-814, 1970. 8. Levitsky, D. A. Feeding condition and internal relationships. Physiol. Behav. 12: 77%787, 1974. 9. Millar, F. K., J. White, R. H. Brooks and G. B. Mider. Walker carcinosarcoma tissue as a dietary constituent. I. Stimulation of appetite and growth in the tumor-bearing rat. J. natn. Cancer Inst. 19: 957-967, 1957.

10. Morrison, S. D. Differences between rat strains in metabolic activity and in control systems. Am. J. Physiol. 224: 1305-1308, 1973. 11. Morrison, S. D. Generation and compensation of the cancer cachectic process by spontaneous modification of feeding behavior. Cancer Res. 36: 228-233, 1976. 12. Morrison, S. D. and N. F. Coffey. Feeding activity and feeding efficiency as distinct modes of change in food intake. J. appl. Physiol. 34: 268-270, 1973. 13. Panksepp, J. Reanalysis of feeding patterns in the rat. J. comp. physiol. Psychol. 82: 78-94, 1973. 14. Priban, I. P. and W. F. Fincham. Self-adaptive control and the respiratory system. Nature 208: 33%343, 1965. 15. Robins, M. W. Circadian pattern of the use of incisor teeth by laboratory rats. Arch. Oral Biol. 18: 641-645, 1973. 16. Schemmel, R., O. Mickelsen and K. Motawi. Conversion of dietary to body energy in rats as affected by strain, sex and ration. J. Nutr. 102: 1187-1198, 1972. 17. Westerterp, K. How rats economize--energy loss in starvation. Physiol. Zool. 50: 331-362, 1977.