Psychological Reports, 1976, 38, 471-484.

@ Psychological Reports 1976

VARIABLES AFFECTING PREFERENCE OF RATS FOR FREE VS EARNED FOOD KENNETH KOFFER1

GRANT COULSON

HONEY WENT

Pine Ridge

Durham Regional Cm&e

Yorrth Clinical Services

Summary.-Five experiments, using rats as subjects, investigated the variables controlling the choice of reinforcer source in situations in which subjects could obtain reinforcers from a freely available source or by performing an operant. Under a wide variety of conditions, subjects almost always obtained their reinforcers from the freely available source. These results are interpreted to indicate that: ( a ) the conditions affecting choice of reinforcer source are not well understood, ( b ) increasing the reward value of the reinforcer does not necessarily increase the relative amount of earned reinforcer consumed and, ( c ) laboratory rats do not appear to be neophobic in all situations and hence, neophobia cannot be used to explain rats' choice behavior in all siruations involving free and earned reinforcers. Several suggestions are made concerning the design of experiments investigating choice between free and earned reinforcers.

A number of experiments have been reported in which rats or pigeons have continued to emit an established operant (Jensen, 1963; Stolz & Lott, 1964; Neuringer, 1969, 1970; Carder & Berkowitz, 1970; Davidson, 1971; Carder, 1972; Tarte & Snyder, 1972, 1973) or have acquired an operant (Neuringer, 1969) even when the same reinforcer produced by the operant has been simultaneously available. This kind of outcome has fostered the notion that organisms prefer to "work for their living" (Singh, 1970; Carder & Berkowitz, 1970) and that this tendency is widespread in the animal kingdom. There are a number of problems with this conclusion. The first problem is that sometimes the greater portion of the reinforcer is not obtained by emitting the operant, but by consuming the reinforcer from the "free" or respnse-independent source (Koffer & Coulson, 1971; Taylor, 1972; Tarte & Snyder, 1972; Hothersall, Thatcher, & Huey, 1973). Although there are probably several unknown reasons for these discrepancies some of the relevant variables can be specified. Jensen (1963), basing his results on grouped data, showed that 50% or more of a session's reinforcer would be reliably obtained by lever-pressing in rats only when they had at least 1280 responses previously reinforced on a fixed ratio 1 (FR 1) schedule. Tarte and Snyder (1972) showed that there was a direct relationship between deprivation level in rats and percentage of the reinforcer obtained by lever-pressing. Mitchell, Scott, and Williams (1973) showed that rats tended to obtain food pellets from the source with which they had the most recent experience. They also found rats would obtain more pellets by lever-pressing if the pellets were delivered to the same pellet container 'Reprints may be obtained from Kenneth Koffer, Pine Ridge, Box 130, Aurora, Ontario, Canada, L4G 3H3.

47 2

K.KOFFER, I T AL.

used during lever-training sessions. The number of earned pellets decreased when free pellets were presented in a familiar dish and earned pellets were delivered into an unfamiliar dish. Both the Jensen (1963) and the Tarte and Snyder (1972) experiments, however, used only one test session and this leads to problems in interpretation which are discussed below. Another problem associated with the free reinforcer paradigm is evident in some experiments by Singh ( 1970). In three experiments with rats and one experiment with young humans, Singh equalized the frequency of reinforcer delivery in two different locations. Simply being in one location was sufficient to produce intermittent reinforcer delivery, whereas in the other location, presence in the location plus an operant response was needed for reinforcer delivery. When reinforcer density was equal, both species tended to obtain most of their reinforcers at the location requiring an operant. Singh's condicions did not allow a choice between one reinforcer at a time while emitting the operant and many reinforcers simultaneously when choosing the other location. Carder and Berkowitz (1970), using a food-deprived rat, have shown, when no operant was required for either source or reinforcer, the subject preferred the source at which many reinforcers were available simultaneously to that source where reinforcers were delivered singly. In other words, Singh's results may have been different if a large number of free reinforcers had been available in the free location. A number of failures to find a preference for "earned" as opposed to "free" reinforcers have also been reported. Koffer and Coulson (1971), using fooddeprived cats and five or six test sessions, found all the "free" food was consumed in 25 or 31 test sessions, usually before any operants were emitted. Koffer and Coulson tentatively suggested that this preference was related to a species difference in food acquisition and consumption but this conclusion has been brought into question by Taylor (1972) and Hothersall, Huey, and Thatcher (1973). These investigators have reportcd a preference for "free" as opposed to "earned" reinforcers in rats. It appears that any scacement that animals prefer to work for reinforcers rather than freeload is premature, as Taylor (1972) has pointed out. The following five experiments were designed to identify those variables determining the choice between free and earned reinforcers in the rat. Each experiment invescigates one potentially important variable and, in the context of providing an extended number of test sessions, provides some indication of the effects of procedural variables.

EXPERIMENT1 The present experiment was designed to investigate the effects of offering subjects a choice between water obtained without an operant being necessary-

RATS' PREFERENCE FOR FREE VS E A R N E D FOOD

47 3

this will be called "free" water-and water that could be obtained by emitting an operant on an FR 1 schedule-this will be referred to as "earned" water. Taylor ( 1972 ) and Carder ( 1972 ) both found that water-deprived rats would consume more free water than earned water when earned water could be obtained on an FR 1 schedule. The present experiment differed from these along several procedural dimensions. The main variable investigated here was the effect of requiring some lever-pressing to occur on test sessions before introducing the free water. Carder and Berkowitz (1970) required at least 25 lever-presses on their test sessions before introducing the free food into the experimental chamber. Carder and Taylor both introduced the earned and free water simultaneously. Would requiring 25 levcr-presses before the introduction of free water make subjects more likely to continue lever-pressing in the presence of free water? Method Subjects.-The 4 subjects were male, albino rats of the Sprague-Dawley strain from Canadian Breeding Laboratories, La Prairie, Quebec. The subjects were 90 days of age at the beginning of deprivation and were housed in individual cages with free access to food during all phases of the experiment except during experimental sessions. Apparatw.-The experimental space was a Gerbrands rat station (Model C ) , fitted with a dipper presenting 0.1 ml of water for 3 sec. A Lehigh Valley rat lever (Model 1352) was mounted in one end wall with its centre situated 6 cm from one side and its top 7.8 cm from the grid floor. The lever extended 3 cm into the experimental space and required a downward force of at least 14 gm to operate. On the wall to the left of the lever was a hole, 1.7 cm from the end panel containing the lever and 6.3 cm above the floor. When experimental conditions dictated a water spout projected 1.6 cm into the experimental space. The experimental space was housed inside a sound-attenuating shell with a ventilating fan providing masking noise. Illumination was provided by a 7-w bulb shining through a frosted lens mounted over the dipper aperture. Control and recording equipment were housed in an adjacent room. A closed circuit television system monitored the subjects' behavior in all the experiments. Procedure.-Each subject was allowed access to water for only 15 min. per day in his home cage during the 10 days immediately preceding the first experimental session. On the first experimental day, after 10 min. to explore the experimental space, the dipper was presented once every 30 sec., on the average, for 1 hr. During Session 2, the dipper was presented once every 120 sec., on the average, for 2 hr. After each of the first two sessions, a subject received 10 min. access to water in his home cage. During the next three sessions, which lasted 30 min. each, a subject was al-

474

K. KOFFER,ET AL.

lowed to drink from the drinking tube in the chamber-the free water sourceand received no other water for that day. During the next five sessions the drinking tube was removed, the lever was inserted and a subject was in the experimental chamber for 1 hr. or until he obtained 200 reinforcers on an FR 1 schedule. During the next phase, a subject was placed in the chamber and required to o b tain 25 reinforcers by lever-pressing on the FR 1 schedule before the free water was introduced for the remainder of the 30-min. session. The number of sessions in this phase was 19, 10, 10 and 15 for Subjects A, B, C , and D, respectively. During the next phase, the free water was present from the beginning of the 30-min. session. This phase lasted 7 , 5, 5, and 9 sessions, respectively. Consecutive sessions, in this and succeeding experiments, except where noced, were run on consecutive days.

All of the subjects magazine-trained quickly, discovered the free water as soon as they had access to it and began lever-pressing almost as soon as the lever was first introduced. During the third phase, when the lever was present but the free water was absent, the 200 reinforcers were obtained in from 24 to 31 rnin. with the exception of one session for % in which he obtained 186 reinforcers and stopped pressing for the resc of the hour. The percentage of free water consumed during each session in Phases 4 and 5 are presented for individual subjects in Fig. 1. In Phase 4, when each subject was required to press 25 times before the free water was introduced, the percentage of free water obtained was derived from the amounts consumed after its introduction. The calculations were based on the assumption that all of the water obtained by pressing was consumed. Observation of the subjects showed this to be the case with the exception of the last few response-produced reinforcers obtained during Phases 4 and 5. Except for two sessions for D, more than 50% of the water consumed during Phases 4 and 5 was free wacer. During the last five sessions of Phases 4 and 5, the average percentage of free water consumed by all four subjects was 84 and 91, respectively. During Phase 4, subjects would drink considerable amounts of free water immediately after its introduction before returning to lever-pressing. During Phase 5 , when the free water was available from the beginning of the session, almost all the free water consumed in the session was taken before lever-pressing began. The present resulcs are similar to those of Taylor (1972) and Carder (1972) in that majority of wacer consumed during test sessions was from the free wacer source. This held true whether or not 25 lever-pressing responses were required before the introduction of the free water. Although there was a

RATS' PREFERENCE FOR FREE VS EARNED FOOD

FIG.1. Percentage of free water consumed for Subjects A through D as a function of session number in Phases 4 and 5. The phases are separated by a broken vertical line.

SESSIONS

fair amount of between- and within-subject variability, the general trend, in keeping with Taylor's finding, is that relatively more free water was consumed as the number of sessions increased. EXPERIMENT2 The importance of qualitative parameters in the choice of free or earned reinforcers has been studied by Carder (1972) who found that rats obtained a much greater percentage of the reinforcer from the earned source when the reinforcer was a 10% sucrose solution than when it was water. Carder reasoned chat a 10% sucrose solution, since it is like food, is ". . . an incentive of higher quality . . ." (p. 25) than water and this produced the greater preference for earned sucrose than for earned water. The present experiment used a mixture of condensed milk and water. This mixture is in effect a food, which without additional food supplements, will sustain rats for extended periods of time. If Carder's reasoning is correct, the milk solution will produce more consumption of earned reinforcer than did the water in Exp. 1.

Method Sicbjects.-The 4 subjects had the same characteristics as those in the first experiment. Apparutxr.-The apparatus was the same as that used in the first experiment except that water was replaced by an equal mixture, by volume, of water and Nestle's Sweetened Condensed Milk. Procedure.-During the seven days preceding the experiment a subject was food deprived to 80% of his free-feeding weight and maintained at that weight, with corrections for normal growth, throughout the experiment. Magazine-training and exposure to free food were the same as Exp. 1. On the sixth experimental day each subject was placed in the experimental chamber with the lever present and an FR 1 schedule in effect. Initial lever-pressing was inconsistent and each of the first three sessions lasted for 45 min. or until 150 reinforcers had been obtained. The next two sessions lasted for 60 min. or until 200 reinforcers had been obtained. Because not all the subjects obtained 200 re-

476

K.KOFFER,ET AL.

inforcers within 60 min., the duration of the last 3 sessions without free food returned to 45 min. or 150 reinforcers. During the Phase 4 sessions, the subject was placed in the experimental chamber and required to obtain 25 reinforcers before the free-food bottle was introduced. Sessions during Phases 4 and 5 lasted for 30 min. The number of sessions in this phase were, for E, F, G and H, 12, 6, 9 and 6, respectively. During Phase 5 sessions the free-food bottle was present from the beginning of the session. The number of sessions in this phase were 12, 6, 9 and 6, respectively. Successive sessions were run on successive days except for occasional days off for weight stabilization.

R esdts As in Exp. 1, all the subjects magazine-trained quickly, discovered the free food almost immediately upon its introduction, and began lever-pressing directly following insertion of the lever. During the last 3 sesslons of Phase 3, when the lever was present but the free food was not, the 150 reinforcers were obtained in an average of 25 min. with session length ranging from 13 to 45 min. The percentage of free food consumed during each session in Phases -4 and 5 is presented for individual subjects in Fig. 2. The percentages in Fig. 2 were

FIG. 2. Percentage of free food consumed for Subjects E through H as a function of session number in Phases 4 and 5. The phases are separated by a broken vertical line. W U

a W 40

5

10

15

20

25

5

I0

15

20

SESSIONS

derived in the same way as those in Fig. 1. In all but two of the sessions, one for

E and one for G, the percentage of free food consumed was greater than the amount of response-produced food. Again, there was no indication that introduction of the freely available reinforcer at the beginning of the session rather than after 25 reinforcers had been obtained had any effect on the amount of free food consumed. During Phase 4, the average percentage of free food consumed for the last five sessions was 94. During Phase 5, the average amount consumed during the last five sessions was 89. These data resemble those of Exp. 1 in two respects. Firstly, most of the freely available reinforcer was consumed before the subject resumed lever-pressing in Phase 4 or began lever-pressing in Phase 5. Secondly, a few of the rein-

RATS' PREFERENCE FOR FREE VS EARNED FOOD

477

forcers obtained during the last part of the sessions during Phases 4 and 5 were not consumed. In both of the first two experiments there was considerable "overshoot" in lever-pressing when the only source of the reinforcer was from the FR 1 schedule. That is, subjects would press two or three times before consuming the reinforcer. The expectation that a liquid food reinforcer would enhance rats' preference for the earned reinforcer was not confirmed. All subjects exhibited a substantial preference for free food in all but two sessions. The reasons for the discrepancy between the present results and those of Carder ( 1972) are not clear. Subjects in the present experiment seemed "excited" about milk delivery and often spent considerable time licking the remnants of the reinforcer from the hole through which the dipper delivered the reinforcer. In addition, the reinforcer had enough caloric content to be highly reinforcing since additional food supplements were not needed once experimental sessions started. EXPERIMENT 3 One possibility that the reinforcer delivered from the FR 1 schedule in Exps. 1 and 2 was not as reinforcing as the freely available reinforcer is chat the reinforcer produced by the lever-press was only available for 3 sec. If the subject did not entirely consume the reinforcer in the 3 sec., the remainder was lost to him. The withdrawal of reinforcer during consumption may have been aversive. Conditions in the present experiment were arranged so that the earned reinforcer remained available until it was consumed. Method Szbjects.-The subjects were similar to those in the first 2 experiments. Appmatw.-The experimental chamber was that used in Exp. 1 except that a small dish was inserted in the hole through which the dipper had delivered the reinforcer. A solenoid valve, which replaced the dipper, delivered 0.1 ml of the same mixture of milk and water used in Exp. 2 whenever experimental conditions dictated. Procedure.-Deprivation conditions, magazine-training and initial exposure to free food were the same as those in Exp. 2. All subjects received 8 FR 1 training sessions before the test phases. The FR 1 craining sessions lasted 60 min. or until 150 reinforcers were delivered. The next two phases were the same as in Exp. 2. In Phase 4 each 30-min. session began with 25 reinforcers being obtained from the FR 1 schedule before the free food was introduced. The Subjects I, J, K and L had 10, 10, 22, and 10 sessions, respectively, during Phase 4. During Phase 5, the free food was present from the beginning of each session. The subjects had 11, 18, 15, and 13 sessions, respectively, in this phase. Succes-

478

K. KOFFER, l T AL.

sive sessions were run on successive days except for occasional days off for weight stabilization.

Results All the subjects except one discovered the free food quickly. Subject I did not consume any free food during the first session in which he had the o p p r tunity but drank an amount equal to that of other subjects during the next two sessions. During magazine-training it was occasionally necessary to stop reinforcer presentation until the diluted milk already delivered had been consumed. All subjects except J, who needed three additional hours for his first lever-training session, began lever-pressing quickly when the lever was first introduced. Observation indicated that the behavior seen during magazine and lever-training resulted from the great length of time subjects spent "cleaning up" the reinforcer and this competed with any movement away from the reinforcer dish. During lever-training, sessions lasted from 22 to 60 min. resulting in greater variability of session duration than during the first two experiments. The percentage of free food obtained during each session in Phases 4 and 5 is presented for individual subjects in Fig. 3. There war no systematic change

SESSIONS

FIG. 3. Percentage of free food consumed for Subjects I through K as a function

of session number in Phases 4 and 5. The phases are separated by a broken vertical line.

in the percentage of free food consumed as a function of procedure. The average percentage of free food consumed during the last five sessions of Phase 4 was 98. The average percentage for the last five sessions of Phase 5 was 94. In only one session did the percentage of free food fall below 50. Most free food was consumed before the subject went back to lever-pressing in Phase 4 or began lever-pressing in Phase 5. Note that Subject I who did not drink any free food in the first session when it was the only available source, subsequently obtained almost all of his reinforcer from the free-food source during choice sessions al-

RATS' PREFERENCE FOR FREE VS EARNED FOOD

479

though he had the least total exposure to free-food consumption before the choice sessions began. In Exps. 1 and 2, as indicated above, some responses occurred while the dipper was coming up or while it was up. In the present experiment, two or more responses in close temporal order were rarely emitted except during the first one or two sessions of lever-pressing. Changing the reinforcer-delivery system so that the reinforcer remained available until it was consumed did not change the overwhelming preference for free food, although it did seem to have some effect on magazine-training and perhaps on lever-training. In agreement with the results of Exps. 1 and 2, the requirement that the delivery of 25 reinforcers had to precede the introduction of the free food did not affect the relative amount of free food consumed. The results of the first three experiments greatly restrict the generality of the contrafreeloading phenomenon and suggest that it occurs only under very special sets of experimental conditions. Procedures in the present experiment were intended to closely resemble those of Carder and Berkowitz (1970) who showed that rats, given a choice between a free and an earned source of food pellets, tended to take the majority of their food from the earned source if the schedule associated with the earned food was FR 1 or FR 2. There were several differences between the procedures used here and those of Carder and Berkowitz. They did not give the extensive magazine-training as provided in the present experiment. Subjects in the present experiment were held at 80% of their free-feeding weight while those in Carder and Berkowitz's experiment were under a 23-hr. food deprivation schedule. It did not appear that either of these differences should decrease the probability of lever-pressing during choice sessions. In addition, initial test sessions in the present experiment were 30 min. long, while those of Carder and Berkowitz lasted 1 hr. Method

Subjects.-The 4 subjects had the same properties as those in the first experiment. Apparata~.-The experimental chamber for the last two experiments measured 29.5 cm )( 27 cm )( 25 cm. A Lehigh Valley (Model 1352) lever was mounted with its centre 10 cm from the door and its top 7.2 cm from the grid floor. The lever projected 3 cm into chamber and required a weight of 14 gm to operate. Standard formula, 45-mgm Noyes food pellets produced by leverpressing were delivered to a food dish set into the wall beside the lever. Illumination was provided by a 7-w bulb shining through a frosted lens above the food cup. The experimental space was housed inside a sound-insulating chamber with a ventilating fan that also provided masking noise. Electromechanical control and recording equipment were housed in an adjacent room.

480

K . KOFFER, ET AL.

The free food was provided as 45-mgm standard-formula Noyes pellets contained in a round metal dish, 8 cm in diameter and 3 cm high, which could be fastened to the wall opposite the lever. The dish held 28 gm of pellets which was more than any subject consumed during a single session. Procedare.-Magazine-training and free-food exposure were the same as those in Exp. 3. Subjects were given five sessions of lever-training on FR 1 which lasted for 60 min. or until 200 pellets were delivered. After lever-training, sessions began with 25 lever-presses reinforced on an FR 1 schedule and then the free food was introduced. Sessions during the fourth phase lasted 30 min. During the fifth phase, conditions were identical except sessions lasted 60 min. During the sixth phase, sessions lasted 30 min. Because Subject 2 pressed the lever only a few times during choice sessions, he was not run during the sixth phase.

Resulrs All subjects magazine-trained quickly, began to eat the free food quickly on the first occasion it was available and started to press the lever almost as soon as it was made available. The percentage of free food consumed for each session is presented for individual subjects in Fig. 4. Again, there was a clear preference for free food during Phase 4, when sessions were 30 min. long. In Phase 5, when sessions were 60 min. long, all subjects except Subject 2 showed a decline in preference for free food. This was most marked for Subject 1. In Phase 6, when sessions were again 30 min. long, preference for free food increased markedly for Subject 4, remained stationary for Subject 3 and continued to decline for Subject 1. The decline in preference for free food when session length was increased which occurred in three or four subjects took place by way of the following process. During the 30-min. session a subject would begin eating the free food as soon as it was introduced and continue to eat it for 10 to 20 min. Then he would

SESSIONS

FIG. 4. Percentage of free food consumed for Subjects 1 through 4 as a function o f session number in Phases 4, 5 and 6. The phases are separated by a broken vertical line and session lengths are as indicated in each phase.

RATS' PREFERENCE FOR FREE VS EARNED FOOD

48 1

lever-press a few times and the session would be over. During the longer sessions, more lever-presses would occur after the free Food was consumed. The periods of eating free food got progressively shorter for Subject 1 during the 60-min. sessions and continued at a low level even when session length was decreased. The present experiment represents more exactly the procedure used by Carder and Berkowitz ( 1970) than did Exp. 4. It differs mainly in the addition of a feature that tends to increase the probability of selecting earned food on choice sessions. Mitchell, Scott, and Williams (1973) showed that the more recent the experience with a food source, the more likely food would be obtained from that source and that food was more likely to be obtained from a familiar source than from an unfamiliar one. Mitchell, et al. conclude that laboratory rats are "neophobic" and that experiments in which earned food has been preferred indicate subjects were avoiding the relatively novel free-food container. In addition, Tarte and Snyder (1973) found that equalizing the amount of exposure to free food and earned food prior to choice sessions results in a large preference for free food. Tarte and Snyder also found that giving more exposure to lever-pressing resulted in more pellets being obtained by lever-pressing, a result consonant with that of Mitchell, et al. Taken together, these results lead to the expectation that if much experience with lever-pressing and no experience with free food was given before choice sessions, earned food would be preferred. Tarte and Snyder ( 1972) have shown that the greater the deprivation the greater the percentage of earned food consumed. A secondary purpose of the present experiment was to determine if this relationship could be demonstrated in a within-subject design. Method Subjects and apparatus.-The 4 subjects had the same characteristics as those in the first experiment, and the apparatus was that described in Exp. 4. Procedwe.-During the first experimental session, subjects were trained to press the lever on an FR 1 schedule. An FR 1 schedule was in effect during the first six sessions which lasted 1.25 hr. each. Subjects were deprived of food for the first time 23 hr. before the first session. During the first eight sessions the only food received was that obtained during the experimenttal sessions. All sessions after the first six were 1-hr. test sessions which began with 25 reinforcers being obtained before free food was introduced. During the 11 days following the first two test sessions, subjects' weights were reduced to 80% of their free-feeding weights and each subject received at least eight more test sessions. Enough food was consumed during the six lever-training sessions and the first two test sessions to maintain the subjects' weights above 90% of their free-

482

K.

KOFFER,

AL.

feeding weights so that reducing their weights to 80% of free-feeding weights represented a substantial increase in deprivation level. All subjects were lever-trained within 10 to 20 min. of the start of the first session. The subjects had an average of 1843 total reinforcers in the first six sessions when no free food was available. The results for individual subjects are presented in Fig. 5. Only Subject 8 displayed a preference for earned food in any of the tesc sessions. The percentage of free food consumed did not change systematically for any subject after the increase in deprivation level. Even when subjects' most recent experience was with lever-pressing and when no prior experience with free food was allowed, only one subject chose earned food. This brings into question Mitchell, Scott, and Williams' (1973) contention that the "neophobia" of laboratory rats accounts for situations in which earned food is preferred. The subjects in the present experiments had no prior experience with the free-food container and still took the majority of their food from it, even on the first exposure. The present results are not due to a limited number of reinforced responses preceding the tesc sessions since the smallest number of responses before the test sessions was 1644, far greater than the number reported by Jensen (1963) to produce a preference for earned food. The absence of any relationshsip between body weight and amount of free food consumed may mean that the direct relationship found by Tarte and Snyder ( 1972) between amount of deprivation and percentage of pellets earned obtains only in a between-subjects design. It could be argued that the 3 hr. of magazine-training given to subjects in the first four experiments provided experience in obtaining reinforcers without

FIG. 5 . Percentage of free food consumed for Subjects 5 through 8 as a function of session number. The two sessions before the broken vertical line were given under 23-hr. food deprivation and those after were given when each subject was at 80% of his freefecding weight.

5 1 0

5 1 0 1 5

SESSIONS

RATS' PREFERENCE FOR FREE

vs EARNED FOOD

483

the operant and that this tendency influenced behavior during choice sessions. The results of the present experiment, which contained close to an irreducible minimum of magazine-training, indicate that extended magazine-training was probably not the decisive factor in the results of the first four experiments. The present results, however, are consistent with those of Hothersall, Huey, and Thatcher (1973) who replicated Carder and Berkowitz's procedure but found that most subjects took most of their food from the free source.

GENERAL DISCUSSION The present series of experiments ~rovidedthree types of reinforcers (water, diluted condensed milk, or food pellets), two testing procedures (0 or 25 reinforcers had to be obtained before the free reinforcer was introduced), two different durations of earned reinforcer availability (the earned reinforcer was available for 3 sec. or until it was consumed) and two different training/deprivation procedures. In only one set of these conditions did one of 20 subjects consistently prefer the earned reinforcer during initial sessions and only one subject developed a preference for the earned reinforcer during an extended number of choice sessions. These results strongly question any assumption that animals prefer earned to free reinforcers (see also Taylor, 1972; Hothersall, Huey, & Thatcher, 1973 ). It seems unlikely that container-neophobia was a significant factor in any of the present experiments. In none of the experiments did any of the subjects have any exposure to the free reinforcer source before encounteribg it in the experimental chamber. Only two subjects exhibited behavior that could be interpreted as neophobic. Subject 1 in Exp. 3 did not take any reinforcer from the free source on the first occasion it was available and in Exp. 5 , when subjects had no experience with the free source before choice sessions, only Subject 8 took the majority of his food from the familiar source. Since the results of experiments which give subjects a choice between free and earned reinforcers are contradictory and failures to replicate are frequent, it seems that further work needs to be done to determine the variables controlling choice of reinforcer in such situations. W e suggest the following guidelines for procedures, descriptions of procedures and data presentation. ( 1 ) Subjects should not have an unspecified experimental history. Since Jensen (1963), Tarte and Snyder (1973) and Mitchell, Scott, and Williams (1973) have shown that experience with the reinforced response is important, such experience should be specifiable and specified. In Koffer and Coulson's (1971) experiment, for example, the 2 of 6 subjects that had long and unspecified histories of intermiaent reinforcement were responsible for all the sessions in which all the free food was not eaten. ( 2 ) Since Tarte and Snyder (1972) have shown that deprivation can be a relevant variable, deprivation history should be controlled and specified. ( 3 ) Magazine-training and method of

484

K. KOFFER, ET AL.

operant training may be important contributors to experimental history and should be specified. ( 4 ) Some experiments do not specify what percentage of reinforcer is obtained from the earned or the free source. This should be shown as a dependent variable as a matter of course. ( 5 ) Amount and duration of availability must be specified for earned reinforcers. A dipper presentation of 3 sec. would presumably be more reinforcing than one of 0.5 sec. ( 6 ) Because there are wide individual differences in percentage of free reinforcer chosen, individual data should be presented. ( 7 ) Since the present results, as well as those of Taylor (1972) and Koffer and Coulson ( 1971), indicate a change in choice L.ehavior over test sessions, ac least 5 to 10 sessions should be given under each experimental condition. This number should, of course, be increased if a systematic change in preference is observed during the first sessions. REFERENCES

CARDER, B. Rats' preference for earned in comparison with free liquid reinforcers. Psychonomic Science, 1972, 26, 25-26. CARDER, B., & BERKOWITZ,K. Rats' preference for earned in comparison with free food. Science, 1970, 167, 1273-1274.

DAVIDSON, A. B. Factors affecting keypress responding by rats in the presence of free food. Psychonomic Science, 197 1, 24, 135-137. HOTHERSALL,D., HUEY, D.. & THATCHER.K The preference of rats for free or response-produced food. Animal Learning and Behavior, 1973, 1 , 241-243. JENSEN, G. D. Preference for bar pressing over "freeloading" as a function of number of rewarded presses. lournal o f Experimental Psychology, 1963, 65, 451-454. K O F F ~K.,, & COULSON,G. Feline indolence: cats prefer free to response-produced food. Psychonomic Science, 197 1, 24, 41-42. MITCHELL,D., SCOTT,D. W., & WILLIAMS, K. D. Container neophobia and the rat's preference for carned food. Behavioral Biology, 1973, 9, 613-624. NEURINGER,A. J. Animals respond for food in the presence of free food. Science, 1969, 166, 399-401. NEURINGER,A. J. Many responses per food reward with free food present. Science. 1970, 169, 503-504. SINGH,D. Preference for bar pressing to obtain reward over freeloading in rats and children. Journal o f Comparative and Physiological Psychology, 1970, 73, 320-327. STOLZ, S. B., & L o n , D. F. Establishment in rats of a persistent res onse producing a net 1055 of reinforcement, Journal o f Comparative and ~ h y s i o i ~ i c aP~ychology. l 1964, 57, 147-149. TARTE,R. D., & SNYDER,R. L. Barpressing in the presence of free food as a function of food deprivation. Psychonomic Science, 1972, 26, 169-170. TARTE, R. D., & SNYDER,R. L. Some sources of variation in the bar-pressing versus freeloading phenomenon in rats. Journal o f Comparati?ie and Physiological Psychology, 1973,84, 128-133. TAYLOR, G . T. A limitation of the contrafreeloading phenomenon. Psychonomic Science, 1972, 29, 173-174.

Accepted December 24, 1975.

Variables affecting preference of rats for free vs earned food.

Psychological Reports, 1976, 38, 471-484. @ Psychological Reports 1976 VARIABLES AFFECTING PREFERENCE OF RATS FOR FREE VS EARNED FOOD KENNETH KOFFER...
515KB Sizes 0 Downloads 0 Views