Combining scientific experimentation with conventional housing A pilot study with rhesus monkeys.код для вставкиСкачать
American Journal of Primatology 14:223-234 (1988) Combining Scientific Experimentation With Conventional Housing: A Pilot Study With Rhesus Monkeys BRUNO PREILOWSKI, MICHAEL REGER, AND HANS ENGELE University of Tiibingen, Weissenau Field Station, Rauensburg, Federal Republic of Germany In search of a method to increase the validity of experimental data and simultaneously counteract the negative consequences of restricted laboratory environments, a pilot study was run with three rhesus monkeys (Mucacu mulatta), which were trained and tested with a laboratory-type task in the animal housing facility. The testing apparatus was made available to individual animals 24 hours a day at their living cage and was connected to a computer that controlled the test and the distribution of regular monkey chow as reward. The animals were thus able to work whenever they wanted, for whatever period of time they chose, in their accustomed home environment. Besides furnishing data on the distribution of activity and performance over extended periods of time, the study provided more data than those obtained previously when animals were tested each day for only a limited time in an isolated test chamber. It was also found that the selfinitiated manipulatory activity required by the test considerably reduced the number of motor stereotypies. Thus, testing animals within the housing facility was profitable for the investigators and beneficial for the animals. Key words: Maeaca mulatta, experimentation, animal housing, occupational therapy, activity and performance profiles, behavioral enrichment INTRODUCTION Experiments with non-human primates, especially those that employ training through positive reinforcement, present a dilemma when it comes to satisfying the demands of tight experimental control on one hand and fulfilling the demands of species-adequate housing on the other. The latter would, for example, require housing in social groups and generally enriched environments. Such conditions, however, would render the control of a number of important variables, such as those influencing the level of motivation, nearly impossible. Not only would exact apportionment of food be prevented, but in addition, the attractions of regular housing could severely reduce the relative attractiveness of an isolated test situation. Furthermore, repeatedly removing the animals from their social environment for testing would be both cumbersome and stressful, with subsequent negative consequences Received December 19, 1986;revision accepted August 25, 1987. Address reprint requests to Bruno F’reilowski, University of Tiibingen, Weissenau Field Station, Rasthalde 3, D-7980 Ravensburg, FRG. 0 1988 Alan R. Liss, Inc. 224 1 Preilowski et al. for the experiments. Strict control of the animals’ environment, however, may also endanger the validity of experimental data, as the animals may be behaviorally deficient animals: The negative influence of social deprivation and an impoverished environment on the development of young non-human primates is well established [Markowitz, 1982; Mason, 1978; Mineka & Suomi, 19781. Subadult and adult feral animals that grow up in supposedly “normal” environments also develop abnormal behavioral patterns in the laboratory [Erwin et al, 1973; Rivers et al, 19831. Several hypotheses concerning the causes of these abnormalities have been advanced by those who have described them. However, systematic investigations are rare and are limited primarily to effects of cage size [Draper & Bernstein, 1963; Paulk et al, 19771. Aside from behavioral developmental studies in seminaturalistic environments, there are instances in which animals are tested at their home cages or, in cases of certain long-term studies (eg, metabolic studies), where the test cage becomes the home cage. In most of these cases, however, the integration of experiments into the animal-keeping facilities is forced by the situation and not for the express purpose of improving experimental as well as housing conditions. In zoos, on the other hand, attempts at “occupational therapy,” especially of apes, have been described, which demonstrate that it is possible to combine scientific interests with improvements in conditions of animal housing [Davis & Markowitz, 1978; Markowitz, 1982; Markowitz & Woodworth, 1978; Yanofsky & Markowitz, 19781. The present study was a first attempt to determine whether self-initiated occupation with a manipulandum, made accessible to earn alimentary rewards for 24 hours per day at the home cage of an individual rhesus monkey, would allow reliable acquisition of performance data (part 1) and, at the same time, have a beneficial effect on general behavior, eg, reduce a number of certain behavioral abnormalities (part ID. EXPERIMENTAL PERFORMANCE The major part of the study concerned the question of whether it is possible to conduct experiments with a somewhat artificial, repetitive task in a regular housing environment for extended periods of time. Of special interest were the rate and reliability of performance under these circumstances. MATERIALS AND METHODS Subjects Two male rhesus monkeys (Macaca rnuZatta) participated in this study: Max (about 8 years old, 8 kg) and Dennis (about 7 years old, 9 kg). These animals had been in the laboratory for 5 and 6 years, respectively, and had been involved in various experiments on sensory-motor learning and intermanual transfer. Housing Subjects were housed in a room together with 12 other male and female rhesus monkeys (Fig. 1).Visual and auditory contact between experimental and nonexperimental animals was possible. The housing room had large, screened windows on two sides, which allowed an average of about 10.4 hours of daylight during the months of testing. In addition, artificial lighting was provided on an automated 12hour on-off cycle (0700 h to 1900 h). Throughout the entire study period, all routines of cleaning, feeding, and general maintenance of the colony were performed normally by the animal caretaker who was continuously present on working days. The nonexperimental animals were fed monkey chow pellets and fruit or vegetables once each day. At irregularly spaced Integrating Experiments Into Housing / 225 F A D Fig. 1. Experimental and animal housing facility. A Control room. B: Maintenance area of housing room separated by wire screen divider from C:, where individual cages were placed (type I cage: 60 x 80 x 120 cm3 depth times width times height; type I1 cage: 52 x 57 x 110 cm3; type I1 cages could be combined to form larger units). D: Outdoor cage with ropes, wooden beams, and tree trunks as a substructure. E: Experimental cage fitted with manipulandum. F: Side view of E. 226 I Preilowski et al. intervals, small groups of compatible animals spent at least half a day in an outdoor cage on the adjacent roof top. The experimental animals received fruit or vegetables once every day but had to earn their monkey chow by working at the manipulandum, as is described in detail below. For the duration of the experiment, they had no access to the outdoor cage. Maintenance of the home cage was kept to a minimum but included regular cleaning of the waste pan; more extensive cleaning occurred only between experimental phases. Apparatus A plexiglass channel (with a 12 x 12 cm2 square cross section, 20 cm in length) was attached in a central position to the solid, opaque side wall of an otherwise conventional monkey cage, with the three remaining side walls and the top made of stainless steel bars. The channel projected horizontally away from the cage and was accessible from inside the cage through a circular opening (10 cm in diameter), which allowed the monkeys to use only one hand at a time. A manipulandum was placed at the outer end of this channel within the animal’s reach (Fig. 1). This manipulandum had been developed for tests involving individual finger movements. Several versions, individually adapted to the hands of the experimental animals, were used. A common feature of all manipulanda were five holes, into which the monkey could insert. five fingers of one hand; the relative position of these holes allowed only the right hand to be used. While operating the apparatus, the monkey could neither see the manipulandum nor his own hand. For Max, a feedback tone of 1,000 Hz was presented whenever he had all five fingers inserted into the manipulandum; for Dennis, this tone was replaced with a signal light above the access opening. Reinforcement consisted of irregularly sized pieces of monkey chow (about 1-5 g) and small (200 mg) sugar-coated pellets, delivered into the plexiglass channel by two different automated dispensers. From an adjacent room the experiment was controlled, and the animal’s responses to the manipulandum were recorded by a computer (PDP 11/34). A video system allowed observations and recordings of the animal’s general behavior. Procedure The two animals were run in succession. Their task was to put all five fingers of the right hand into the holes of the manipulandum (to a specified minimum depth) and to keep them in this position for progressively longer periods of time. (It was planned that this behavior would be used as a basis for training of individual finger movements in subsequent experiments). Magazine training and shaping. Both animals were magazine trained. During the subsequent shaping, a strict regimen of successive steps toward the correct posture of hand and fingers at the manipulandum was followed for each animal; apart. from usual shaping goals, the aim was to find a series of steps that would allow shaping through automated control. Starting with the shaping phase, the manipulandum was made available to each of the animals for 24 hours a day. Training. Once the animal reliably inserted all five fingers of the right hand into the holes of the manipulandum in the correct manner, it was trained to keep them in this position for progressively longer periods of time. The aim was t o attain a continuous grasp duration of 60 seconds with Max, and of 10 seconds with Dennis. Testing. Max and Dennis were then allowed to work on the task for 34 and 28 days, respectively; activity and performance were recorded by the controlling computer, and a cumulative protocol was stored for each 10-minute interval, yielding a Integrating Experiments Into Housing I 227 total of 144 pairs of scores per animal and day. Whenever an animal brought at least one of his fingers into the manipulandum up to the required minimum depth, this was considered an “attempt.” If the other fingers of the right hand followed within 5 seconds, this was considered a “successful attempt” (and a reward was delivered); otherwise it was scored as a n “unsuccessful attempt” (with no reward). After any such attempt, all fingers had to be retracted for the device to reset and to admit another valid attempt. During the training of the first animal (Max), various kinds of rewards and schedules of reinforcement had been tried. To avoid satiation through the preferred monkey chow reinforcement and also to prevent negative emotional reactions toward the “thinning out” of rewards, we finally settled on a continuous reinforcement schedule with two types of alimentary reward during the training phase of Dennis and during the testing phase of both animals. These two types of reward, pieces of monkey chow and 200-mg sugar-coated pellets, were delivered in a randomized sequence at an average ratio of 1:2, respectively. The pieces of monkey chow were selected in such a way that a minimum total daily amount of reward was assured equivalent to the usual maintenance diet of about 3% of body weight. Irrespective of performance, fruit and vegetables were given daily, at about 0900 h for Dennis and at about 1400 h for Max; water was continuously available. RESULTS With the institution of continuous reinforcement consisting of a random mixture of regular monkey chow and sugar-coated pellets, performance remained relatively stable throughout the training phase, allowing for automated control of progression through successively higher criteria levels. The results of the testing phase are presented in Figure 2. For each animal, scores within the daily 144 10-minute periods were averaged, yielding 24-hour profiles of activity (number of attempts made) and performance (number of successful attempts divided by the number of attempts). Product-moment correlations r between activity and performance profiles are remarkably high: for Max r = 0.93, for Dennis r = 0.92. Both activity and performance profiles of Max exhibit a pronounced bimodality, with the dip at about 1400 h becoming more pronounced as testing proceeded. This dip appears to be caused by the feeding of fruit at approximately this time. The profiles of Dennis, who was given his fruit at 0900 h, did not show this bimodality. A spectral analysis of the average activity and performance profiles revealed no significant periodicity besides the obvious 24-hour day-night cycle. Both animals showed an unexpected level of activity during the night, especially Max, who on the average did not have a single 10-minute period without some activity at the manipulandum. To give a crude estimate of data rates: Max made about 440 attempts per day (280 successful ones), and Dennis made about 930 such attempts (430 successful ones). A comparison of the overall averages reveals more relative variability in the activity than in the performance profiles, the latter remaining fairly stable throughout the testing period. Dividing the entire testing phase into five segments of about equal size, averages over these segments show a decrease in activity with time for both animals, while the level of performance remains almost constant (Fig. 3). DISCUSSION The data demonstrate that with rhesus monkeys it is in fact possible to run an automated test requiring a substantial degree of sensory-motor skill and concentration within the ‘hoisy” environment of these animals’ normal housing facilities. The monkeys showed moderately stable activity and performance, as well as pre- 228 / Preilowski et al. v) k- 4 w I- + a 10 , L 0 a rn W I so 3 z 2¶ W El a U W w a 20 15 10 5 0 2 4 B I 10 12 14 10 10 20 22 24 1 .oo 0.80 v) cn W 0 0 0.w 0.40 3 m 0.20 LL 0 W k 0.00 a a 1.00 W L3 4 0.80 a w > a , I D DENNIS 0.w 0.40 0.20 0.00 0 2 4 0 I 10 12 14 10 10 20 22 24 TIME OF D A Y Fig. 2. Twenty-four-hourprofiles of activity (A, B) and performance (C, D). Each data point represents the number of attempts made (A, B) and the rate of success (C,D) during one of the 144 10-minute intervals, averaged over all days of testing. dictable distributions of these scores over extended periods of time. Taking the average overall data rates as crude indicators of the “usable” data volume that can be expected from such a procedure, the resulting data quantity is excellent; it is greater than data rates that were obtained previously in our laboratory by animals performing roughly comparable tasks in an isolated test chamber. In the test cham- Integrating Experiments Into Housing / 229 1.00 . 0.80 . 0.80 . 0.40 . B _---_---.. -..----_MAX --______----- _/-- *-__ 0.20 DENNIS 1 2 3 4 5 SEGMENT Fig. 3. Number of attempts made (A) and rate of successes (B) during 30-minuteintervals, averaged over consecutive days within five successive segments of the testing phase. ber motivation as well as performance declined seriously after approximately 100 to 150 trials per day. As a result of uncontrollable influences stemming from interactions of the working monkey with other animals and with personnel such as the animal caretaker, or because of satiation and fatigue induced by prolonged testing on a quite stereotyped task, one might expect the data to be of a degraded quality when compared with data obtained in traditional "isolated" test situations using restrictions such as fixed presentations of trials, food deprivation, etc. The present study did not allow a direct comparison between the quality of data obtained in isolated chambers in housing facilities. However, on the basis of a number of experiments run with Max and Dennis in isolated test chambers, we have no reason to suspect major negative influences with regard to either general test compliance or performance under the conditions of the present study. Also, as far as cumulative influences such as fatigue are concerned, there was no detectable decline from average morning to average afternoon activity or performance. Another conceivable source of serial variability in the behavioral data might be short-term, ultradian periodicities, a possibility that cannot be dismissed simply because the spectra revealed no such significant oscillations. It is still possible that the fairly high level of variability in activity and performance reflects an interference between asynchronous, more-or-less random external influences and potentially rhythm-generating internal pacemakers. Of considerable interest is the remarkably high correlation between activity and performance profiles. At present, it cannot be determined whether changes in activity produced changes in attention that in turn influenced performance or whether a common variable, such as motivation, influenced both activity and performance. In any case, the data suggest that a simple model of daily activity and performance, assuming, for example, an almost constant performance level through- 230 I Preilowski et al. out the day, will not describe the data adequately. This daily variation of performance contrasts markedly with the fact that the level of performance averaged over days remained almost constant in both animals throughout the entire phase of testing, while the overall average activity decreased considerably. Most likely this was due to a progressive adaptation to the testing situation: to the requirements of working for food, for example (which had been given in a single daily portion prior to the present study), or to the various distracting influences of the housing situation. The existence of some sort of “constant performance reservoir?’ in rhesus monkeys is a second possible explanation, although this is a less parsimonious hypothesis than the first. With regard to differences between animals, the relatively lower activity rate of Max is probably a result of the requirement to hold the grip for 60 seconds, in contrast to the 10-secondrequirement for Dennis. The higher performance level of Max may be due to more extensive experience during training. However, both activity and performance may also reflect genuine individual differences between these animals. THERAPEUTIC EFFECTS Since an ultimate aim of combining research and housing is to improve the general situation for experimental animals, we wanted to find out whether the selfinitiated occupation with a “typical” laboratory task would have a positive effect on the behavior of an animal, as reflected by a decrease in the frequency or intensity of abnormal behavior. In the present study, we concentrated on the frequency aspect of such behavior. MATERIALS AND METHODS Subject Donald, a six-year-oldmale rhesus monkey (Macaca mulatta) weighing approximately 7 kg, who had been in the laboratory for about 4 years, served as the subject in this part of the study. Donald had not been in any experiment for 1%years at the time the current study was started. At the time Donald entered our laboratory he was underweight and showed several deficits in social behavior reminiscent of the effects of social isolation. These were ameliorated to some extent as he grew up in a group of peers, some of which had a background of social deprivation as well. Apparatus The apparatus used in this part of the study was identical with that used in the first part, except that only monkey chow was given as reinforcement, by means of a newly developed dispenser that has been described in detail elsewhere [Engele et al, 19831. Basically, this dispenser provides access t o food for a specified length of time; in addition, it provides a manipulatory element because it admits only a single attempt to grasp for food per reinforcing event. Procedure The procedure followed in this part of the study was identical with that followed in the previous one, with the exception that the study was limited to magazine training, shaping, and a training phase. Before shaping, during training, and after training, Donald’s behavior was observed. To this end, the animal’s behavior was recorded on video tape for a period of 10 minutes once or twice a day at random between 0800 h and 0900 h andor between 1600 h and 1700 h. Twenty-six observation periods were recorded before shaping, ten during the training phase, and 12 within 24 days after training had been terminated. An observation period was then Integrating Experiments Into Housing / 231 divided into 20 half-minute intervals; each of these intervals was analyzed with respect to the occurrence or non-occurrence of certain types of behavioral patterns (irrespective of intensity or duration), according to 15 nonoverlapping categories that had been established during a 2-day period of observation before the study was started. With respect to the general requirements of training, testing, and colony management, these behavioral patterns were then further classified as “acceptable” (such as occupation with the manipulandum, grooming, exploration) and “nonacceptable” (such as autoaggressive or bizarre behavior, or stereotypic running in circles). RESULTS Frequencies and percentages of behavioral patterns during the three observation periods are presented in Table I. The mean number of behavioral units per observation period increased from 27.5 before shaping to 31.1 during training and decreased to 23.6 after training. The respective proportion of nonacceptable behavior decreased from 41%before shaping to 14%during training and decreased still further to 10%after training. Stereotypic behavior (running in circles, tossing head, and saluting), which had been the most prominent nonacceptable forms of behavior in this animal prior to shaping (91%of nonacceptable behavior, 37% of entire behavior), decreased to 12% of all pattern scores during training. During the period of observation following training, circling and tossing head vanished almost completely, while saluting remained on the previous absolute level, constituting about 8% of total observed behaviors. Some TABLE I. Behavioral Patterns During Observation Periods Before Shaping, During Training, and After Training No. of observation periods Acceptable behaviors Concentrated grooming Manipulating genitals Chewing, biting motions Feeding, drinking Scratching Presenting Manipulating objects except manipulandum Scraping cage bars Busy with manipulandum Nonacceptable behaviors Self-biting,tearing out hair Shaking cage Circling with head tossing Circling with body twirling Saluting (fistbelow ipsilateral eye) Tossing head while at rest Total Before shaping: 26 (%) During training: 10 (%) After training: 12 (%) 102 (14.2) 48 (6.7) 85 (11.9) 26 (3.6) 82 (11.5) 39 (5.4) 28 (3.9) 24 (7.7) 2 (0.6) 121 (38.9) 5 (1.6) 28 (9.0) 3 (1.0) 0 (0.0) 107 (37.8) 18 (6.4) 73 (25.8) 0 (0.0) 42 (14.8) 3 (1.1) 5 (1.8) 12 (1.7) 6 (2.1) 422 (58.9) 8 (2.6) 76 (24.4) 267 (85.9) 254 (89.8) 10 (1.4) 16 (2.2) 146 (20.4) 85 (11.9) 24 (3.4) 13 (1.8) 294 (41.1) l(0.3) 5 (1.6) 7 (2.3) 3 (1.0) 28 (9.0) 0 (0.0) 44 (14.1) 7 (2.5) 0 (0.0) 0 (0.0) 0 (0.0) 21 (7.4) l(0.4) 29 (10.2) 716 (100.0) 311 (100.0) 283 (100.0) _ _ - - 232 I Preilowski et al. forms of aggressive behavior (self-biting and shaking the cage), making up a fairly small percentage of the total behavior to begin with, were less affected by the intervening training. DISCUSSION Comparing observations made before shaping with those made during training, there is considerable reduction in those behaviors that were called “nonacceptable,” primarily because a few specific behavioral abnormalities, including typical cage stereotypies, decreased rapidly. Of course, it should be kept in mind that these stereotypic behaviors were incompatible with the requirements of working at the manipulandum to obtain reward. In a sense then, it might be suspected that unwanted stereotypies were merely replaced by more acceptable, in this case manipulatory, stereotypies. On the other hand, a comparison between observations made before shaping and after training suggests an effect of the intervening training. This supports other observations of generally positive “therapeutic” results even with rather repetitive occupations [Angst & Hess, 1978; Markowitz & Spinelli, 1986; Ridley & Baker, 19821. The fact that some forms of unwanted behavior such as self-biting and saluting showed no change in frequency is disappointing. However, it is to be expected that there may be a hierarchy of abnormal behavioral patterns, of which only a few can be eliminated through a relatively short intervention period [Erwin & Deni, 19791. To what extent different types of abnormal behaviors are responsive to “therapeutic” changes in the experimental environment remains to be investigated. There is some indication that there are differences in the extent to which abnormal behaviors constitute an expression of long-term abnormal developments of the nervous system. If abnormalities are represented by long persistent changes in the central nervous system, one would expect them to be difficult to disrupt or even ameliorate [Ridley & Baker, 19821. CONCLUDING DISCUSSION The results of this study strongly support the idea that laboratory-type tests can be combined with adequate housing facilities with resulting profits for both the experiment and the welfare of the experimental animals: The obtained data rates are excellent, and despite high variability over the day, the overall quality of the data shows no detrimental effects that are due to the “noise” within regular housing environments. Simultaneously, the quality of the data may be positively affected by the absence of stress, which is typically induced in “isolated” test situations by factors such as handling and transport, disruption of social contacts, or the relative strangeness of an experimental compartment. Another beneficial factor is that the animals were allowed to initiate each trial themselves, so that a minimum level of motivation was guaranteed with every attempt. In a sense, these responses can be interpreted as volitional rather than dependent on artificial deprivation schedules. The results also show a positive effect of experimental training on a number of abnormal behavioral patterns often observed in laboratory primates. In our experience, especially with non-isolation-reared, mature rhesus monkeys, it is not cage size that is the critical factor in eliciting abnormal behavior. Instead, the level of active and responsive behavior elicited by the laboratory and housing environment has proven to be a more important variable in determining the amount of stereotypies and maladaptive responses. Our present experimental task, although artificial and repetitive, reduced several forms of cage stereotypies. Even more effective intervention through more complex, variable, and stimulating experimental tasks can be expected. Integrating Experiments Into Housing I 233 So far only one animal has been allowed access to the apparatus at a time. Thus, social contacts of the experimental animal were still limited. However, testing and reliable identification of the working animal in group situations should be feasible and will be attempted in future studies. The approach described above also promises to provide very interesting data with regard to performance changes over the day and over longer periods of time. In addition, it potentially provides a method to study individual differences and social influences on learning and performance, which for highly social animals such as primates appears to be of utmost importance. One might question the costs involved in combining scientific experimentation and adequate housing. Since for most scientific experiments the use of computers or microprocessor control is quite common, the additional costs for the proposed approach are tolerable. Except for normal precautionary measures in animal management, for example, allowing the animals both visual contacts and the possibility of hiding from the view of others, no special arrangements are necessary as far as housing is concerned. Of course, special care has to be taken to protect the equipment from destructive manipulations of the animals. The present study was a first attempt to combine research and environmental enrichment in our laboratory. In future research we may find some deleterious effects of this approach for scientific experimentation. In any case, our knowledge about the detrimental effects of traditional nonresponsive animal housing on the observable behavior of primates [Markowitz, 19861 and the increasing knowledge about external influences on almost every physiological parameter leave very little choice but to do our best to improve living conditions for the primates we study. CONCLUSIONS 1. Experiments employing self-initiated operant activity of monkeys can be conducted in conventional housing facilities with resulting profits for the experiment and for the welfare of the experimental animal. 2. Automated equipment allowing the experiment to run around the clock can result in high data rates at an overall high quality, despite considerable fluctuations of activity over the day. 3. Even relatively artificial and repetitive experimental tasks can reduce cage stereotypies, which are suspected to be due to lack of active responsive behavior in monkeys housed under controlled laboratory conditions. ACKNOWLEDGMENTS The authors would like to thank Franz Gruber and Eugen Weber for their help in maintaining the monkey colony and Hal Markowitz for his valuable critique of the manuscript. We thank the Universities of Tubingen and Konstanz as well as the Deutsche Forschungsgemeinschaft (Pr 117/7-1)for support. REFERENCES Angst, W.; Hess, J.P. M e n zwischen Freile- Draper, W.A.; Bernstein I.S. Stereotyped beben und Totalisolation. Eine ethologische havior and cage size. PERCEPTUAL AND Betrachtung der zunehmenden Stimulus- MOTOR SKILLS 16~231-234,1963. verarmung. Pp. 91-114 in DAS TIER IM Engele, H.; Preilowski, B.; Reger, M. A verEXPERIMENT. W.H. Weihe, ed. Bern, satile dispenser for commercial monkey chow reinforcement. BEHAVIORAL REHuber, 1978. Davis, R.R.; Markowitz, H. Orangutan per- SEARCH METHODS AND INSTRUMENformance on a light-dark reversal discrimi- TATION 15:360-362.1983. nation in the zoo. PRIMATES 19:755-759, Erwin, J.; Deni, R. Strangers in a strange land: Abnormal behaviors or abnormal en1978. 234 I Preilowski et al. vironments? Pp. 1-28 in CAPTIVITY AND BEHAVIOR. J. Erwin, T.L. Maple, G. Mitchell, eds. New York, Van Nostrand Reinhold, 1979. Erwin, J.; Mitchell, G.; Maple, T. Abnormal behavior in non-isolate reared rhesus monkeys. PSYCHOLOGICAL REPORTS 33: 515-523,1973. Markowitz, H. BEHAVIORAL ENRICHMENT IN THE ZOO. New York, Van Nostrand Reinhold, 1982. Markowitz, H. Environmental engineering projects for captive primates. Paper presented at the XIth Congress of the International Primatological Society, Goettingen (FRG), July 20-25, 1986. Markowitz, H.; Spinelli, J.S. The interface between research and environment. Paper presented at a Symposium on improving the quality of life of laboratory primates. San Diego, Aug. 25, 1986. Markowitz, H.; Woodworth, G. Experimental analysis and control of group behavior. Pp. 107-131 in BEHAVIOR OF CAPTIVE WILD ANIMALS. H. Markowitz, V.J. Stevens, eds. Chicago, Nelson-Hall, 1978. Mason, W.A. Social experience and primate cognitive development. Pp.233-251 in THE DEVELOPMENT OF BEHAVIOR COMPARATIVE AND EVOLUTIONARY ASPECTS. G.M. Burghardt; M. Bekoff, eds. New York, Garland Press, 1978. Mineka, S.; Suomi, S.J. Social separation in monkeys. PSYCHOLOGICAL BULLETIN 851376-1400,1978. Paulk, H.H.; Dienske, H.; Ribbens, L.G. Abnormal behavior in relation to cage size in rhesus monkeys. JOURNAL OF ABNORMAL PSYCHOLOGY 86~87-92,1977. Ridley, R.M.; Baker, H.F. Stereotypy in monkeys and humans. PSYCHOLOGICAL MEDICINE 12:61-72,1982. Rivers, A.; Bartecki, U.; Brown, J.V.; Ettlinger, G. An unexpected “epidemic” of a rare stereotypy: Unidentified stress or imitation? LABORATORY PRIMATE NEWSLETTER 22~5-7,1983. Yanofsky, R.; Markowitz, H. Changes in general behavior of two mandrills (Papio sphinx) concomitant with behavioral testing in the zoo. THE PSYCHOLOGICAL RECORD 28~369-373,1978.