- Email: [email protected]
Effects of rate and distance of procurement wheel-running on saccharin-and-sucrose solution drinking by non-deprived rats
Physiology & Behavior, Vol. 36, pp. 539--543.Copyright©Pergamon Press Ltd., 1986. Printed in the U.S.A.
0031-9384/86 $3.00 + .00
Effects of Rate and Distance of Procurement Wheel-Running on Saccharin-and-Sucrose Solution Drinking by Non-Deprived Rats K E N N E T H N. G A N N O N , H O W A R D V. SMITH I A N D K E V I N J. T I E R N E Y D e p a r t m e n t o f P s y c h o l o g y , Trinity College, Dublin 2, Ireland R e c e i v e d 8 M a r c h 1985 GANNON, K. N., H. V. SMITH AND K. J. TIERNEY. Effects of rate and distance of procurement wheel-running on saccharin-and-sucrose solution drinking by non-deprived rats. PHYSIOL BEHAV 36(3) 539--543, 1986.--Four nondeprived rats received daily sessions of 20 rain access to a saccharin-and-sucrose solution following various prior activities, in an attempt to disentangle the normally confounded roles of time spent, amount and rate of procurement responding in causing an increase in consumption once access is gained. After the normal rate of running (approx 35 m/rain) was established, six conditions were run in random order, involving waiting zero and 2.5 min in an immobilised running-wheel, and running, with the wheel rotated by a motor, in four conditions formed by combining two speeds (12 and 30 m/min) with two distances (12 and 30 m), prior to access to the solution. Drinking increased with the speed of prior running, and to a lesser extent with the distance run, but was not related systematically to the time spent running. It is suggested that information from the animal's own behavior in gaining access to a commodity, particularly the rate of energy expenditure, may influence its utilisation of the commodity by affecting the rate of subsequent consummatory responding. Procurement cost
Wheel-running
Non-deprived rats
W H E N there is a procurement cost involved in gaining access to a commodity, such as food or water, animals respond to increases in that cost by reducing the frequency with which they seek access to the commodity and increasing the amount of it they take on each occasion (e.g., [4,11]). The total daily intake of the commodity tends to be defended up to high values of the procurement cost, but this strategy means that the time and energy spent in procurement responding does not have to rise as a linear function of the cost. This strategy appears to be adaptive, but little is known about its immediate causes, although some of the mechanisms involved may have been identified. F o r example, it has been found that when rats are tested at a constant level of deprivation the quantity of food they consume [6], or of water they drink [19], or, in the case of non-deprived rats, the time they spend drinking a saccharin-sucrose solution [7] in a fixed period of access gained by performing a procurement requirement, rises as that requirement increases. This increase in the total consumed appears to occur because the rate of consumption is elevated in the first few minutes of access in proportion to the size of the procurement cost paid immediately before [19]. However, it is not known for certain which aspects of
Foraging
Rate of drinking
Meal size
procurement responding are responsible for these effects. The procurement responses used in experiments have usually been either lever-pressing (e.g., [4,6]), or wheel-running (e.g., [7,8]), and the costs have been increased by requiring the subjects to perform more presses or to run a greater distance. Increasing the cost therefore usually means that the subject will not only have to expend more energy to gain access, but it will also have to expend more time performing the responses, unless it increases the rate at which it responds. Quite possibly more time will be taken even if the rate is also increased. F o r example, in the experiment by Gannon et al. [6], the rate of lever-pressing increased monotonically with the lever-press requirement, but not sufficiently to keep constant the time taken to gain access to the food. If more time is taken and there is consequently a greater delay in gaining access to the commodity the animal's degree of deprivation may be increased, but there is evidence [6,7] which makes it possible to discount extra deprivation as a cause of the effect reported by Gannon et al. [6,7]. It appears likely, therefore, that some aspect of procurement responding itself is responsible for the effect on subsequent consummatory behaviour. The toal amount of responding performed, the time expended in responding (not simply the
1Requests for reprints should be addressed to Howard Smith.
539
54o
GANNON, SMITH AND TIERNEY
delay in gaining access), and the ratio of these (i.e., the rate at which responding took place) could all possibly be involved in this capacity. The available evidence does not make it possible to come to any firm conclusion concerning the way in which animals might use these variables in scaling the procurement cost for the purpose of adjusting their consumption of the commodity once access has been gained. Collier e t a / . [51 performed a factorially designed experiment with cats in which the number of lever-presses and the force required to depress the lever were both varied systematically, but the results did not entirely clarify the issue. The cats lived in the apparatus, and had to pay a procurement cost whenever they wanted to gain access to food. They took fewer and larger meals as both the number of presses and the force were increased, and the two variables interacted, such that each had its greatest effect when the other was small. However, the rate of responding varied as the cost was changed, so that, for example, even when the number of presses was held constant the subjects took longer to perform them as the force required on the bar was increased. Thus the three variables were again confounded. The present study was designed to overcome this confounding by bringing these variables under experimental control. This was done by requiring the subjects to run in a wheel, the rotation of which was controlled by an electric motor, before they were given access to a solution which they could drink for a fixed period of time. Thus the rate and distance for which running took place (and consequently also the time spent running) could be predetermined. In other respects the study was similar to that performed by Gannon ctal. [71, in that non-deprived subjects were employed and they were tested daily, rather than to the experiments of Collier (e.g., [4, 5. 8]) in which the subjects lived in the apparatus permanently and chose when to initiate and terminate their access to whatever commodity was involved. METHOD
Sub.iects The subjects were 4 female albino rats (Wistar strain) which had been bred in the University animal house. They were supplied to the laboratory at approximately 120 days of age, and were initially used in an undergraduate practical involving the shaping of a chain of responses. When the present experiment began they were approximately 180 days old.
Apparatus Testing took place in an enclosed running-wheel, 31.8 cm in diameter and 7 cm deep. The wheel could be prevented from revolving by means of a solenoid-operated brake, or made to revolve by means of a variable-speed electric motor which was linked to it by a detachable shaft. There was a 2 cm diameter hole in the non-revolving front panel of the wheel, 0.5 cm behing which a drinking-tube could be made available. The latter contained a solution made by dissolving two 17 mg tablets of saccharin sodium (Hermesetas) and 5 g of sucrose in 150 ml of water. The position of the hole was such that a rat in the wheel could drink conveniently by turning its head 3 to 4 cm from the position it was in during running. By means of a solenoid the drinking-tube could be withdrawn sideways to a position in which it could not be reached by a rat in the wheel.
TABLE
1
MEAN TIME SPENT DRINKING, AND MEAN NUMBER OF' LICKS. PER SESSION, IN T H E 8 E X P E R I M E N T A L CONI)ITIONS
Condition Paired Baseline Free Running Waiting Free Access 12 m at 12 m/min 30 m at 12 m/min 12 m at 30 m/min 30 m at 30 m/min
Iime spent drinking Isec)
Number of licks
278.15 324.13 376.25 363.83 ~87.1t1 411.48 434.25 465.00
2013 2847 2811 2735 2927 3099 3254 3492
The experiment was controlled by a C.B.M. microcomputer located in a separate room, which determined the operations of the solenoids and of the motor when it was attached to the wheel, and also recorded the time spent drinking (and the time spent running, when this was free to vary). The latter were recorded by cumulating the number of 0.25 sec intervals containing either at least one lick of the tube, or at least one-sixth of a revolution of the wheel. In addition, electromechanical counters recorded the total number of licks and one-sixth revolutions during each session. Pro( "('dHru
Throughout the experiment, which took place during the summer months, the subjects were maintained under natural daylight conditions and lived individually in standard laboratory cages, in which they had constant free access to food and water. They were removed for a session of testing on each of six days a week, at the same time every day (between 2 and 5 p.m.). Sessions were run in each of eight conditions. No condition was terminated until the subject's performance had been stable for eight sessions. The criteria of stability included the following: (a) Daniel's test for trend [2] indicated there was no statistically significant (p <0.05) increasing or decreasing trend in the time spent drinking (and also running when this was free to vary) over eight successive sessions, and (b) the averages of the five overlapping 4-session blocks within the eight sessions did not differ from their own mean by more than 10%. (The mean number of sessions taken before stability was acheived, across subjects and conditions, was 10.97.) The first condition to which the subjects were exposed was a "paired baseline." The subjects were placed in the wheel for sessions of 20 min duration, and were permitted to run and drink freely, and to alternate between these two types of responding whenever they chose. The brake on the wheel was off, the motor was detached, and the drinkingtube was available throughout the session. The purpose of including this condition was to allow the subjects to become habituated to the apparatus, and also to permit an assessment of their normal running speed. The second condition ("free running") was one in which the subjects were allowed to run freely for the first 150 sec after being placed in the wheel, and were then allowed to drink freely for the following 20 min. The brake on the wheel was switched on at the end of the 150 sec to prevent further
RATE OF P R O C U R E M E N T RESPONDING running, and the drinking tube was made available by the operation of the solenoid. This condition was included mainly to familiarise the subjects with the general sequence of events that was to occur in the subsequent conditions. The six conditions that followed were run in a different random order for each subject. Four of them were ones in which the wheel was driven by the motor. In these four conditions as soon as the subject was placed in the wheel the motor was switched on, and it rotated at one of two speeds (12 or 30 m/min) until the wheel had turned one of two distances (12 or 30 m). This meant that the subject had to run for either 24 sec (12 m at 30 m/min), 60 sec (12 m at 12 m/min, and 30 m at 30 m/min) or 150 sec (30 m at 12 m/min). At the end of this time the wheel was locked, and the drinking-tube became available for 20 min. The distances and runningspeeds used in these four conditions were selected on the basis of the performance of the subjects in the paired baseline condition. The maximum speed at which the wheel was made to rotate by the motor (30 m/min) was comfortably below the average speed (35.5 m/min, SD=5.0) at which the subjects themselves ran in the paired baseline condition and the maximum distance was less than half the average distance (78.4 m, SD=57.8) they ran in that condition. The main difference between the running that took place in these four conditions and that which occurred when the motor was not attached to the wheel was that running in the latter case tended to occur in shorter bursts, but at a faster rate. In one of the other conditions ("free access") the subject was placed in the wheel, with the brake on, and immediately permitted to drink for 20 min, and in the other ("waiting") it was placed in the wheel, with the brake on, and made to wait for 150 sec before the drinking-tube became available for 20 min. RESULTS
The data analysed were the mean time spent drinking, and the mean number of licks, across the sessions when responding had stabilised, by each subject in each of the eight conditions. The conditions affected each of the four subjects in very much the same way. Kendall's coefficient of concordance [18] indicated a significant agreement between the subjects in the rank ordering of the conditions on the basis of both the time spent drinking, W=0.875, p<0.001, and the number of licks, W=0.827, p<0.01. The average across the subjects of the time spent drinking and the number of licks performed in each condition can be seen in Table 1. A one-way repeated-measures analysis of variance of the time spent drinking revealed that the conditions were a signficant source of variance, F(7,21)=9.71, p <0.01. Comparisons of the means by the Newman-Keuls method [20] revealed: (a) that significantly less time was spent drinking in the paired baseline condition than in all of the other conditions except the free running one, p's<0.05; (b) that significantly less time was spent drinking in the free running condition than in any of the conditions in which the wheel was driven by the motor, except for the one in which it was driven for 12 m at 12 m/min, p's<0.05; and (c) that when the wheel was driven for 30 m at 30 m/min there was significantly more time spent drinking than in all of the other conditions except for those in which it was driven for 30 m at 12 m/min and for 12 m at 30 m/rain, p's<0.05. The equivalent analysis of the number of licks performed produced very similar results; the only differences were that there was significantly more licking in the free running con-
541 A
/
/
2'
r" . .
I-
z
./
U.I a.
00
°
•O •
•
•
•
•
•
•
•
°
°
•
..I-1
•
°
•
[] • : 3Ore
I.U
--
,1
•
°
...... 2"5
:30m/min O e :12m : 12 rnlmin • :Ore (Free Access) 50
7"5
100
1~5
1~o
TIME SPENT RUNNING (Sec) FIG. 1. Time spent drinking as a function of time spent running, in the conditions in which the motor was attached to the wheel.
dition than there was in the paired baseline, p<0.01, and only the 30 m at 30 m/min condition produced significantly more licking than the free running one, p <0.05. [N.B., in the 150 sec available for running in the latter condition the subjects on average ran a distance of 17.4 m (SD=11.3), at a speed of 32.2 m/min (SD=3.6), occupying a time of 31.6 sec (SD= 18.8)]. Although the one-way analyses of variance revealed that there was more drinking in the 30 m at 30 m/rain condition than there was in the 12 m at 12 m/rain one, their results did not permit any more detailed inferences to be drawn concerning the independent or interactive effects of the speed of rotation of the wheel and the distance the subjects were required to run. To elucidate these effects a number of further analyses were performed. The first of these were two 2x2 (speed × distance) analyses of variance, with repeated measures on both factors, of the time spent drinking and the number of licks in the four conditions in which the wheel was driven. The only significant source of variance revealed in both these analyses was that of the speed of rotation of the wheel, F(1,3) = 19.82, p <0.05, and F(1,3) = 16.05, p <0.05, respectively. When the subjects were required to run at the faster speed they subsequently drank more than when they were required to run at the slower speed. The effect of the speed of rotation is easy to discern by inspection of the means in Table 1. However, there is also some evidence in Table 1 of a trend for the amount of drinking to increase as the distance was increased, within each speed of rotation condition. This trend is particularly obvious if the mean of the free access condition (in which the subjects were required to run zero distance) is included in each of the series. The failure to include this condition in the previous analyses may have resulted in the importance of this variable having been overlooked, and therefore a check was made on the significance of these trends by performing Page's L trend test [16] on the two series for each variable. These revealed that there was a significant monotonic trend for both the time spent drinking and the number of licks performed, to increase as the distance increased (across the series 0, 12 and 30 m) when the speed of rotation of the wheel was 30 m/min, L ' s = 5 5 , p's=0.01 in both cases, but when it was 12 m/min the trends just failed to achieve an acceptable level of statistical significance.
542
G A N N O N , SMITH AND TIERNEY
Within each of these series the distance run was confounded with the time spent running, but it is unlikely that the effect is attributable to the latter. There was no consistent relationship between the time spent running and the amount of drinking that followed. This can be seen by reference to Fig. 1, which depicts the time spent drinking as a function of the time spent running in the relevant conditions. For example, 60 sec running at 12 m/min did not produce significantly more drinking than immediate access to the drinking tube, and there was less drinking after 150 sec running at 12 m/rain than there was after 24 sec running at 30 m/min. DISCUSSION
In general terms these results are similar to those of previous experiments in which procurement costs were varied and found to affect the amount of consumption that subsequently occurred in a fixed period of access to a restricted commodity [6, 7, 19]. The results also support the conclusion [6,7] that the effect does not depend on the delay in gaining access to the commodity caused by having to perform the procurement requirement. In addition, the results add considerably to the previous findings, and permit conclusions to be drawn concerning the way in which various features of procurement responding affect subsequent consummatory behaviour. The foremost of these is that when a rat works to gain access to a commodity the major determinant of the amount of consummatory responding that will occur is the rate at which it performs the procurement requirement. The total amount of procurement responding appears to be of lesser importance in this respect, although there was evidence that it does play some part, particularly if the rate of responding is high. On the other hand the duration for which responding occurs does not seem to be systematically related to the amount of subsequent responding. The present results also support the conclusion that the contingency between procurement and consummatory responding is essential in producing the procurement cost effect. The "free running" condition in this experiment produced an amount of drinking that was not significantly different from that which occurred in the "waiting" condition, both of which involved a delay of 150 sec before the drinking tube became available. The "free running" condition also produced significantly less time spent drinking than the condition in which the subjects had to run 12 m at 30 m/min, despite the fact that the subjects on average ran about 17 m at about 32 m/min during the time available. The analysis of the way animals make decisions while foraging is currently provoking much interest within behavioral ecology (e.g., [13,14]). The theoretical basis for much of this research is optimality theory (e.g., [12]). However, the latter does not provide information about the decision processes themselves; it merely specifies the course of
action which an animal should choose in order to maximize (or minimize) certain functions. It is frequently assumed (e.g., [1, 10, 15]) that animals utilise simple "rules of thumb'" in making these decisions. The present experiment illustrates one approach to elucidating the nature of these rules in the case of the procurement component of foraging behaviour. Further work will be required to ascertain more precisely how animals scale costs, but the results of this experiment would suggest that rats may utilise information from their own behavior when gaining access to a resource as a basis for deciding how to exploit it. The amount of work performed in gaining access to the resource, and in particular the rate at which that work has been performed, are the important factors which they appear to use in this context. Clearly this analysis will not be applicable to all types of foraging situation. Rate and amount of responding may not be crucial variables for all foragers, either because their prey is distributed in such a fashion that these variables bear little relationship to encounter rate, and hence convey no information about distribution, or because the forager may adopt a strategy which does not include them as variables. They may, however, exert a significant effect on the meal parameters of animals whose foraging strategies do not preclude them. It was somewhat surprising to find that individually neither the amount of procurement repsonding nor the time spent performing the procurement response had particularly important effects on subsequent consummatory behavior. In previous experiments [6,7] the amount of responding was found to have a greater effect than it had in the present one, and temporal variables have been found to be important in other laboratory studies of foraging (e.g., [9]). Optimal foraging theory proposes that animals are sensitive to time constraints, subject to the constraints imposed by satisfying their energy requirements [17]. Of course, the relationship between the size of the procurement cost and the time spent responding, i.e., the rate of responding was found to have a significant effect, and it may be in this way that temporal variables become important in certain cirmcumstances. If the cost of gaining access to a commodity is fixed an animal will have to increase the rate at which it pays that cost if it is to reduce its foraging time. Collier et al. (e.g., [4]) have found that the size and patterning of meals in freely foraging animals depend upon the procurement cost. Smith et al. [19] found that the size of the procurement cost paid immediately before consummatory responding begins affects the rate at which the latter response is performed. Clifton et al. [3] have recently reported that the rate of intake of food influences the size and patterning of meals in a way similar to a procurement cost. These relationships, taken together with the results of the present experiment, offer the possibility that both the size and patterning of meals may ultimately depend upon the rate at which the animal pays the procurement cost.
REFERENCES
I. Barnard, C. Snap decisions for survival. New Scientist 1411: 24-27, 1984. 2. Bradley, J. V. Distribution-Free Statistical Tests. Englewood Cliffs, N J: Prentice-Hail, 1968. 3. Clifton, P. G., D. A. Popplewell and M. J. Burton. Feeding rate and meal patterns in the laboratory rat. Physiol Behav 32: 369374, 1984.
4. Collier, G. H., E. Hirsch and P. Hamlin. The ecological determinants of reinforcement in the rat. Physiol Behav 9: 705-716, 1972.
5. Collier, G. H., L. W. Kaufman, R. Kanarek and J. Fagen. Optimization of time and energy constraints in the feeding behavior of rats: A laboratory simulation. Carnivore 1: 34-41, 1978.
RATE OF PROCUREMENT RESPONDING
6. Gannon, K. N., H. V. Smith and K. J. Tierney. Effects of procurement cost on food consumption in rats. Physiol Behav 31: 331-337, 1983. 7. Gannon, K. N., H. V. Smith and K. J. Tierney. Effect of procurement cost on the drinking of a saccharin-sucrose solution by non-deprived rats. Phyiol Behav 33: 917-921, 1984. 8. Kanarek, R. B. and G. H. Collier. Patterns of eating as a function of the cost of the meal. Physiol Behav 23: 141-145, 1979. 9. Killeen, P. R., J. P. Smith and S. J. Hanson. Central place foraging in Rattus norvegicus. Anim Behav 29: 64-70, 1981. 10. Krebs, J. R. and N. B. Davies. An Introduction to Behavioural Ecology. Oxford: Blackwell, 1981. 11. Marwine, A. and G. H. Collier. The rat at the waterhole. J Comp Physiol Psychol 93: 391--402, 1979. 12. Maynard Smith, J. Optimization theory in evolution. Annu Rev Ecol Syst 9: 31-56, 1978. 13. McFarland, D. J. Decision making in animals. Nature (Lond) 259: 15-21, 1977.
543
14. McFarland, D. and A. Houston. Quantitative Ethology. London: Pitman, 1981. 15. McNamara, J. and A. Houston. The application of statistical decision theory to animal behaviour. J Theor Biol 85: 673-690, 1980. 16. Page, E. B. Ordered hypotheses for multiple treatments: A significance test for linear ranks. J Am Stat Assoc 58: 216-230, 1963. 17. Pyke, G. H., H. R. Pulliam and E. L. Charnov. Optimal foraging: A selective review of theory and tests. Q Rev Biol 52: 137-154, 1977. 18. Siegel, S. Nonparametric Statistics. New York: McGraw-Hill, 1956. 19. Smith, H. V., K. N. Gannon and K. J. Tierney. Effect of procurement wheel running on the rate of drinking in rats. Physiol Behav 33: 927-930, 1984. 20. Winer, B. J. Statistical Principles in Experimental Design. London: McGraw-Hill Kogakusha, 1971.
0031-9384/86 $3.00 + .00
Effects of Rate and Distance of Procurement Wheel-Running on Saccharin-and-Sucrose Solution Drinking by Non-Deprived Rats K E N N E T H N. G A N N O N , H O W A R D V. SMITH I A N D K E V I N J. T I E R N E Y D e p a r t m e n t o f P s y c h o l o g y , Trinity College, Dublin 2, Ireland R e c e i v e d 8 M a r c h 1985 GANNON, K. N., H. V. SMITH AND K. J. TIERNEY. Effects of rate and distance of procurement wheel-running on saccharin-and-sucrose solution drinking by non-deprived rats. PHYSIOL BEHAV 36(3) 539--543, 1986.--Four nondeprived rats received daily sessions of 20 rain access to a saccharin-and-sucrose solution following various prior activities, in an attempt to disentangle the normally confounded roles of time spent, amount and rate of procurement responding in causing an increase in consumption once access is gained. After the normal rate of running (approx 35 m/rain) was established, six conditions were run in random order, involving waiting zero and 2.5 min in an immobilised running-wheel, and running, with the wheel rotated by a motor, in four conditions formed by combining two speeds (12 and 30 m/min) with two distances (12 and 30 m), prior to access to the solution. Drinking increased with the speed of prior running, and to a lesser extent with the distance run, but was not related systematically to the time spent running. It is suggested that information from the animal's own behavior in gaining access to a commodity, particularly the rate of energy expenditure, may influence its utilisation of the commodity by affecting the rate of subsequent consummatory responding. Procurement cost
Wheel-running
Non-deprived rats
W H E N there is a procurement cost involved in gaining access to a commodity, such as food or water, animals respond to increases in that cost by reducing the frequency with which they seek access to the commodity and increasing the amount of it they take on each occasion (e.g., [4,11]). The total daily intake of the commodity tends to be defended up to high values of the procurement cost, but this strategy means that the time and energy spent in procurement responding does not have to rise as a linear function of the cost. This strategy appears to be adaptive, but little is known about its immediate causes, although some of the mechanisms involved may have been identified. F o r example, it has been found that when rats are tested at a constant level of deprivation the quantity of food they consume [6], or of water they drink [19], or, in the case of non-deprived rats, the time they spend drinking a saccharin-sucrose solution [7] in a fixed period of access gained by performing a procurement requirement, rises as that requirement increases. This increase in the total consumed appears to occur because the rate of consumption is elevated in the first few minutes of access in proportion to the size of the procurement cost paid immediately before [19]. However, it is not known for certain which aspects of
Foraging
Rate of drinking
Meal size
procurement responding are responsible for these effects. The procurement responses used in experiments have usually been either lever-pressing (e.g., [4,6]), or wheel-running (e.g., [7,8]), and the costs have been increased by requiring the subjects to perform more presses or to run a greater distance. Increasing the cost therefore usually means that the subject will not only have to expend more energy to gain access, but it will also have to expend more time performing the responses, unless it increases the rate at which it responds. Quite possibly more time will be taken even if the rate is also increased. F o r example, in the experiment by Gannon et al. [6], the rate of lever-pressing increased monotonically with the lever-press requirement, but not sufficiently to keep constant the time taken to gain access to the food. If more time is taken and there is consequently a greater delay in gaining access to the commodity the animal's degree of deprivation may be increased, but there is evidence [6,7] which makes it possible to discount extra deprivation as a cause of the effect reported by Gannon et al. [6,7]. It appears likely, therefore, that some aspect of procurement responding itself is responsible for the effect on subsequent consummatory behaviour. The toal amount of responding performed, the time expended in responding (not simply the
1Requests for reprints should be addressed to Howard Smith.
539
54o
GANNON, SMITH AND TIERNEY
delay in gaining access), and the ratio of these (i.e., the rate at which responding took place) could all possibly be involved in this capacity. The available evidence does not make it possible to come to any firm conclusion concerning the way in which animals might use these variables in scaling the procurement cost for the purpose of adjusting their consumption of the commodity once access has been gained. Collier e t a / . [51 performed a factorially designed experiment with cats in which the number of lever-presses and the force required to depress the lever were both varied systematically, but the results did not entirely clarify the issue. The cats lived in the apparatus, and had to pay a procurement cost whenever they wanted to gain access to food. They took fewer and larger meals as both the number of presses and the force were increased, and the two variables interacted, such that each had its greatest effect when the other was small. However, the rate of responding varied as the cost was changed, so that, for example, even when the number of presses was held constant the subjects took longer to perform them as the force required on the bar was increased. Thus the three variables were again confounded. The present study was designed to overcome this confounding by bringing these variables under experimental control. This was done by requiring the subjects to run in a wheel, the rotation of which was controlled by an electric motor, before they were given access to a solution which they could drink for a fixed period of time. Thus the rate and distance for which running took place (and consequently also the time spent running) could be predetermined. In other respects the study was similar to that performed by Gannon ctal. [71, in that non-deprived subjects were employed and they were tested daily, rather than to the experiments of Collier (e.g., [4, 5. 8]) in which the subjects lived in the apparatus permanently and chose when to initiate and terminate their access to whatever commodity was involved. METHOD
Sub.iects The subjects were 4 female albino rats (Wistar strain) which had been bred in the University animal house. They were supplied to the laboratory at approximately 120 days of age, and were initially used in an undergraduate practical involving the shaping of a chain of responses. When the present experiment began they were approximately 180 days old.
Apparatus Testing took place in an enclosed running-wheel, 31.8 cm in diameter and 7 cm deep. The wheel could be prevented from revolving by means of a solenoid-operated brake, or made to revolve by means of a variable-speed electric motor which was linked to it by a detachable shaft. There was a 2 cm diameter hole in the non-revolving front panel of the wheel, 0.5 cm behing which a drinking-tube could be made available. The latter contained a solution made by dissolving two 17 mg tablets of saccharin sodium (Hermesetas) and 5 g of sucrose in 150 ml of water. The position of the hole was such that a rat in the wheel could drink conveniently by turning its head 3 to 4 cm from the position it was in during running. By means of a solenoid the drinking-tube could be withdrawn sideways to a position in which it could not be reached by a rat in the wheel.
TABLE
1
MEAN TIME SPENT DRINKING, AND MEAN NUMBER OF' LICKS. PER SESSION, IN T H E 8 E X P E R I M E N T A L CONI)ITIONS
Condition Paired Baseline Free Running Waiting Free Access 12 m at 12 m/min 30 m at 12 m/min 12 m at 30 m/min 30 m at 30 m/min
Iime spent drinking Isec)
Number of licks
278.15 324.13 376.25 363.83 ~87.1t1 411.48 434.25 465.00
2013 2847 2811 2735 2927 3099 3254 3492
The experiment was controlled by a C.B.M. microcomputer located in a separate room, which determined the operations of the solenoids and of the motor when it was attached to the wheel, and also recorded the time spent drinking (and the time spent running, when this was free to vary). The latter were recorded by cumulating the number of 0.25 sec intervals containing either at least one lick of the tube, or at least one-sixth of a revolution of the wheel. In addition, electromechanical counters recorded the total number of licks and one-sixth revolutions during each session. Pro( "('dHru
Throughout the experiment, which took place during the summer months, the subjects were maintained under natural daylight conditions and lived individually in standard laboratory cages, in which they had constant free access to food and water. They were removed for a session of testing on each of six days a week, at the same time every day (between 2 and 5 p.m.). Sessions were run in each of eight conditions. No condition was terminated until the subject's performance had been stable for eight sessions. The criteria of stability included the following: (a) Daniel's test for trend [2] indicated there was no statistically significant (p <0.05) increasing or decreasing trend in the time spent drinking (and also running when this was free to vary) over eight successive sessions, and (b) the averages of the five overlapping 4-session blocks within the eight sessions did not differ from their own mean by more than 10%. (The mean number of sessions taken before stability was acheived, across subjects and conditions, was 10.97.) The first condition to which the subjects were exposed was a "paired baseline." The subjects were placed in the wheel for sessions of 20 min duration, and were permitted to run and drink freely, and to alternate between these two types of responding whenever they chose. The brake on the wheel was off, the motor was detached, and the drinkingtube was available throughout the session. The purpose of including this condition was to allow the subjects to become habituated to the apparatus, and also to permit an assessment of their normal running speed. The second condition ("free running") was one in which the subjects were allowed to run freely for the first 150 sec after being placed in the wheel, and were then allowed to drink freely for the following 20 min. The brake on the wheel was switched on at the end of the 150 sec to prevent further
RATE OF P R O C U R E M E N T RESPONDING running, and the drinking tube was made available by the operation of the solenoid. This condition was included mainly to familiarise the subjects with the general sequence of events that was to occur in the subsequent conditions. The six conditions that followed were run in a different random order for each subject. Four of them were ones in which the wheel was driven by the motor. In these four conditions as soon as the subject was placed in the wheel the motor was switched on, and it rotated at one of two speeds (12 or 30 m/min) until the wheel had turned one of two distances (12 or 30 m). This meant that the subject had to run for either 24 sec (12 m at 30 m/min), 60 sec (12 m at 12 m/min, and 30 m at 30 m/min) or 150 sec (30 m at 12 m/min). At the end of this time the wheel was locked, and the drinking-tube became available for 20 min. The distances and runningspeeds used in these four conditions were selected on the basis of the performance of the subjects in the paired baseline condition. The maximum speed at which the wheel was made to rotate by the motor (30 m/min) was comfortably below the average speed (35.5 m/min, SD=5.0) at which the subjects themselves ran in the paired baseline condition and the maximum distance was less than half the average distance (78.4 m, SD=57.8) they ran in that condition. The main difference between the running that took place in these four conditions and that which occurred when the motor was not attached to the wheel was that running in the latter case tended to occur in shorter bursts, but at a faster rate. In one of the other conditions ("free access") the subject was placed in the wheel, with the brake on, and immediately permitted to drink for 20 min, and in the other ("waiting") it was placed in the wheel, with the brake on, and made to wait for 150 sec before the drinking-tube became available for 20 min. RESULTS
The data analysed were the mean time spent drinking, and the mean number of licks, across the sessions when responding had stabilised, by each subject in each of the eight conditions. The conditions affected each of the four subjects in very much the same way. Kendall's coefficient of concordance [18] indicated a significant agreement between the subjects in the rank ordering of the conditions on the basis of both the time spent drinking, W=0.875, p<0.001, and the number of licks, W=0.827, p<0.01. The average across the subjects of the time spent drinking and the number of licks performed in each condition can be seen in Table 1. A one-way repeated-measures analysis of variance of the time spent drinking revealed that the conditions were a signficant source of variance, F(7,21)=9.71, p <0.01. Comparisons of the means by the Newman-Keuls method [20] revealed: (a) that significantly less time was spent drinking in the paired baseline condition than in all of the other conditions except the free running one, p's<0.05; (b) that significantly less time was spent drinking in the free running condition than in any of the conditions in which the wheel was driven by the motor, except for the one in which it was driven for 12 m at 12 m/min, p's<0.05; and (c) that when the wheel was driven for 30 m at 30 m/min there was significantly more time spent drinking than in all of the other conditions except for those in which it was driven for 30 m at 12 m/min and for 12 m at 30 m/rain, p's<0.05. The equivalent analysis of the number of licks performed produced very similar results; the only differences were that there was significantly more licking in the free running con-
541 A
/
/
2'
r" . .
I-
z
./
U.I a.
00
°
•O •
•
•
•
•
•
•
•
°
°
•
..I-1
•
°
•
[] • : 3Ore
I.U
--
,1
•
°
...... 2"5
:30m/min O e :12m : 12 rnlmin • :Ore (Free Access) 50
7"5
100
1~5
1~o
TIME SPENT RUNNING (Sec) FIG. 1. Time spent drinking as a function of time spent running, in the conditions in which the motor was attached to the wheel.
dition than there was in the paired baseline, p<0.01, and only the 30 m at 30 m/min condition produced significantly more licking than the free running one, p <0.05. [N.B., in the 150 sec available for running in the latter condition the subjects on average ran a distance of 17.4 m (SD=11.3), at a speed of 32.2 m/min (SD=3.6), occupying a time of 31.6 sec (SD= 18.8)]. Although the one-way analyses of variance revealed that there was more drinking in the 30 m at 30 m/rain condition than there was in the 12 m at 12 m/rain one, their results did not permit any more detailed inferences to be drawn concerning the independent or interactive effects of the speed of rotation of the wheel and the distance the subjects were required to run. To elucidate these effects a number of further analyses were performed. The first of these were two 2x2 (speed × distance) analyses of variance, with repeated measures on both factors, of the time spent drinking and the number of licks in the four conditions in which the wheel was driven. The only significant source of variance revealed in both these analyses was that of the speed of rotation of the wheel, F(1,3) = 19.82, p <0.05, and F(1,3) = 16.05, p <0.05, respectively. When the subjects were required to run at the faster speed they subsequently drank more than when they were required to run at the slower speed. The effect of the speed of rotation is easy to discern by inspection of the means in Table 1. However, there is also some evidence in Table 1 of a trend for the amount of drinking to increase as the distance was increased, within each speed of rotation condition. This trend is particularly obvious if the mean of the free access condition (in which the subjects were required to run zero distance) is included in each of the series. The failure to include this condition in the previous analyses may have resulted in the importance of this variable having been overlooked, and therefore a check was made on the significance of these trends by performing Page's L trend test [16] on the two series for each variable. These revealed that there was a significant monotonic trend for both the time spent drinking and the number of licks performed, to increase as the distance increased (across the series 0, 12 and 30 m) when the speed of rotation of the wheel was 30 m/min, L ' s = 5 5 , p's=0.01 in both cases, but when it was 12 m/min the trends just failed to achieve an acceptable level of statistical significance.
542
G A N N O N , SMITH AND TIERNEY
Within each of these series the distance run was confounded with the time spent running, but it is unlikely that the effect is attributable to the latter. There was no consistent relationship between the time spent running and the amount of drinking that followed. This can be seen by reference to Fig. 1, which depicts the time spent drinking as a function of the time spent running in the relevant conditions. For example, 60 sec running at 12 m/min did not produce significantly more drinking than immediate access to the drinking tube, and there was less drinking after 150 sec running at 12 m/rain than there was after 24 sec running at 30 m/min. DISCUSSION
In general terms these results are similar to those of previous experiments in which procurement costs were varied and found to affect the amount of consumption that subsequently occurred in a fixed period of access to a restricted commodity [6, 7, 19]. The results also support the conclusion [6,7] that the effect does not depend on the delay in gaining access to the commodity caused by having to perform the procurement requirement. In addition, the results add considerably to the previous findings, and permit conclusions to be drawn concerning the way in which various features of procurement responding affect subsequent consummatory behaviour. The foremost of these is that when a rat works to gain access to a commodity the major determinant of the amount of consummatory responding that will occur is the rate at which it performs the procurement requirement. The total amount of procurement responding appears to be of lesser importance in this respect, although there was evidence that it does play some part, particularly if the rate of responding is high. On the other hand the duration for which responding occurs does not seem to be systematically related to the amount of subsequent responding. The present results also support the conclusion that the contingency between procurement and consummatory responding is essential in producing the procurement cost effect. The "free running" condition in this experiment produced an amount of drinking that was not significantly different from that which occurred in the "waiting" condition, both of which involved a delay of 150 sec before the drinking tube became available. The "free running" condition also produced significantly less time spent drinking than the condition in which the subjects had to run 12 m at 30 m/min, despite the fact that the subjects on average ran about 17 m at about 32 m/min during the time available. The analysis of the way animals make decisions while foraging is currently provoking much interest within behavioral ecology (e.g., [13,14]). The theoretical basis for much of this research is optimality theory (e.g., [12]). However, the latter does not provide information about the decision processes themselves; it merely specifies the course of
action which an animal should choose in order to maximize (or minimize) certain functions. It is frequently assumed (e.g., [1, 10, 15]) that animals utilise simple "rules of thumb'" in making these decisions. The present experiment illustrates one approach to elucidating the nature of these rules in the case of the procurement component of foraging behaviour. Further work will be required to ascertain more precisely how animals scale costs, but the results of this experiment would suggest that rats may utilise information from their own behavior when gaining access to a resource as a basis for deciding how to exploit it. The amount of work performed in gaining access to the resource, and in particular the rate at which that work has been performed, are the important factors which they appear to use in this context. Clearly this analysis will not be applicable to all types of foraging situation. Rate and amount of responding may not be crucial variables for all foragers, either because their prey is distributed in such a fashion that these variables bear little relationship to encounter rate, and hence convey no information about distribution, or because the forager may adopt a strategy which does not include them as variables. They may, however, exert a significant effect on the meal parameters of animals whose foraging strategies do not preclude them. It was somewhat surprising to find that individually neither the amount of procurement repsonding nor the time spent performing the procurement response had particularly important effects on subsequent consummatory behavior. In previous experiments [6,7] the amount of responding was found to have a greater effect than it had in the present one, and temporal variables have been found to be important in other laboratory studies of foraging (e.g., [9]). Optimal foraging theory proposes that animals are sensitive to time constraints, subject to the constraints imposed by satisfying their energy requirements [17]. Of course, the relationship between the size of the procurement cost and the time spent responding, i.e., the rate of responding was found to have a significant effect, and it may be in this way that temporal variables become important in certain cirmcumstances. If the cost of gaining access to a commodity is fixed an animal will have to increase the rate at which it pays that cost if it is to reduce its foraging time. Collier et al. (e.g., [4]) have found that the size and patterning of meals in freely foraging animals depend upon the procurement cost. Smith et al. [19] found that the size of the procurement cost paid immediately before consummatory responding begins affects the rate at which the latter response is performed. Clifton et al. [3] have recently reported that the rate of intake of food influences the size and patterning of meals in a way similar to a procurement cost. These relationships, taken together with the results of the present experiment, offer the possibility that both the size and patterning of meals may ultimately depend upon the rate at which the animal pays the procurement cost.
REFERENCES
I. Barnard, C. Snap decisions for survival. New Scientist 1411: 24-27, 1984. 2. Bradley, J. V. Distribution-Free Statistical Tests. Englewood Cliffs, N J: Prentice-Hail, 1968. 3. Clifton, P. G., D. A. Popplewell and M. J. Burton. Feeding rate and meal patterns in the laboratory rat. Physiol Behav 32: 369374, 1984.
4. Collier, G. H., E. Hirsch and P. Hamlin. The ecological determinants of reinforcement in the rat. Physiol Behav 9: 705-716, 1972.
5. Collier, G. H., L. W. Kaufman, R. Kanarek and J. Fagen. Optimization of time and energy constraints in the feeding behavior of rats: A laboratory simulation. Carnivore 1: 34-41, 1978.
RATE OF PROCUREMENT RESPONDING
6. Gannon, K. N., H. V. Smith and K. J. Tierney. Effects of procurement cost on food consumption in rats. Physiol Behav 31: 331-337, 1983. 7. Gannon, K. N., H. V. Smith and K. J. Tierney. Effect of procurement cost on the drinking of a saccharin-sucrose solution by non-deprived rats. Phyiol Behav 33: 917-921, 1984. 8. Kanarek, R. B. and G. H. Collier. Patterns of eating as a function of the cost of the meal. Physiol Behav 23: 141-145, 1979. 9. Killeen, P. R., J. P. Smith and S. J. Hanson. Central place foraging in Rattus norvegicus. Anim Behav 29: 64-70, 1981. 10. Krebs, J. R. and N. B. Davies. An Introduction to Behavioural Ecology. Oxford: Blackwell, 1981. 11. Marwine, A. and G. H. Collier. The rat at the waterhole. J Comp Physiol Psychol 93: 391--402, 1979. 12. Maynard Smith, J. Optimization theory in evolution. Annu Rev Ecol Syst 9: 31-56, 1978. 13. McFarland, D. J. Decision making in animals. Nature (Lond) 259: 15-21, 1977.
543
14. McFarland, D. and A. Houston. Quantitative Ethology. London: Pitman, 1981. 15. McNamara, J. and A. Houston. The application of statistical decision theory to animal behaviour. J Theor Biol 85: 673-690, 1980. 16. Page, E. B. Ordered hypotheses for multiple treatments: A significance test for linear ranks. J Am Stat Assoc 58: 216-230, 1963. 17. Pyke, G. H., H. R. Pulliam and E. L. Charnov. Optimal foraging: A selective review of theory and tests. Q Rev Biol 52: 137-154, 1977. 18. Siegel, S. Nonparametric Statistics. New York: McGraw-Hill, 1956. 19. Smith, H. V., K. N. Gannon and K. J. Tierney. Effect of procurement wheel running on the rate of drinking in rats. Physiol Behav 33: 927-930, 1984. 20. Winer, B. J. Statistical Principles in Experimental Design. London: McGraw-Hill Kogakusha, 1971.