Conditioned defeat in the Syrian golden hamster (Mesocricetus auratus)

BEHAVIORAL AND NEURAL BIOLOGY 60, 93--102 (1993)

Conditioned Defeat in the Syrian Golden Hamster

(Mesocricetus auratus) M. POTEGAL, K. H U H M A N , 2 T. MOORE, AND J . MEYERHOFF 3 Department of Medical Neurosciences, Division of Neuropsychiatry, Walter Reed Army Institute of Research, Washington, D.C. 20307-5100

pared by treating nonaggressive hamsters with high doses of diazepam: animals so treated locomote more or less continuously around the cage virtually ignoring the subject. An unexpected observation was that subjects in the AF Group tended to closely follow these diazepam-treated, rapidly locomoting NAIs around the cage. Following may be an example of the "risk assessment" activities directed toward a potential threat. The development of a rapid and reliable technique for inducing CD in hamsters sets the stage for further physiological and pharmacological work on this interesting phenomenon. ~ 1993 A c a d e m i c

When singly housed under laboratory conditions, male Syrian golden hamsters routinely attack novel conspecific intruders introduced into their home cages. As we report here, after being repeatedly defeated by a larger, more aggressive intruder, such normal territorial aggression on the part of the resident hamsters is replaced by defensive behavior and flight. We have found that such conditioned defeat (CD) can be reliably induced by a series of 5-min trials with an aggressive intruder whether these trials are spread over 4 days or are all given on the same day. A useful behavioral criterion for the appearance of CD during acquisition is the first occurrence of anticipatory flight (AF), i.e., the first time the resident flees from the next aggressive intruder before being attacked. CD shows generalization: Animals trained to the AF criterion (AF Group) subsequently show defensive behavior toward, and even flee from, intruders which show absolutely no sign of aggressiveness. Animals in the AF Group persisted in such defense behavior for two test sessions; animals given three additional defeat trials beyond the appearance of AF (AF + 3 Group) showed a greater magnitude and persistence of defense and flight. A comparison of CD-trained animals which met a nonaggressive intruder (NAI) every day for 5 days to similarly trained animals which met the intruder only on the fifth day after acquisition suggests that CD diminishes passively as a function of time and not as the consequence of repeated encounters with a nonaggressive stimulus animal. We also found that near ideal NAIs could be pre-

Press, Inc.

Golden h a m s t e r s are t h o u g h t to be solitary animals w h i c h defend h o m e t e r r i t o r i e s in t h e wild (Nowack & Paradiso, 1983; A n d e r s o n & Jones, 1984). W h e n singly h o u s e d u n d e r l a b o r a t o r y conditions, r e s i d e n t h a m s t e r s r o u t i n e l y a t t a c k novel conspecific i n t r u d e r s i n t r o d u c e d into t h e i r h o m e cages. Such aggressive b e h a v i o r is p r e s u m a b l y a n e x p e r i m e n t a l model of n o r m a l t e r r i t o r i a l defense. If it is a r r a n g e d t h a t t h e i n t r u d e r s are l a r g e r a n d / o r m o r e a g g r e s s i v e t h a n t h e t e r r i t o r y resident, however, t h e y will a t t a c k and defeat t h e resident. An e a r l i e r s t u d y in this l a b o r a t o r y found t h a t w h e n h a m s t e r s h a d b e e n exposed to such defeats repeatedly, t h e i r p l a s m a levels of t e s t o s t e r o n e were r e d u c e d and t h e i r levels of adrenocorticotropin, fle n d o r p h i n , a n d cortisol w e r e r a i s e d ( H u h m a n , Moore, Ferris, Mougey, & Meyerhoff, 1991). L e s h n e r (1975) has proposed t h a t one f u n c t i o n of such horm o n a l responses to agonistic e n c o u n t e r s is to modify the f u t u r e b e h a v i o r of the a n i m a l in s i m i l a r situations. One purpose of t h e p r e s e n t studies is to e x a m i n e t h e b e h a v i o r a l consequences of exposing h a m s t e r s to r e p e a t e d defeat. A p a r t i c u l a r l y i n f o r m a t i v e w a y to e v a l u a t e defeat-induced b e h a v i o r a l c h a n g e is to t e s t the responses of chronically d e f e a t e d a n i m a l s to the pres-

1 Research was conducted in compliance with the Animal Welfare Act and other Federal statutes and regulations relating to animals and experiments involving animals, and adheres to principles stated in the Guide for the Care and Use of Laboratory Animals, NIH Publication 85-23. All procedures were reviewed and approved by the WRAIR Animal Use Review Committee. The views of the authors do not purport to reflect the position of the Department of the Army or the Department of Defense (Para 4-3, AR 360-5). 2 Present address: Department of Biology, Georgia State University, P.O. Box 4010, Atlanta, GA 30302-4010. 3 Address correspondence and reprint requests to M. Potegal. 93

0163-1047/93 $5.00 Copyright © 1993 by Academic Press, Inc. All rights of reproduction in any form reserved.

94

POTEGAL ET AL.

entation of novel conspecifics which have been chosen and/or prepared for a lack of aggressiveness. Chronically defeated rats and mice avoid such nonaggressive stimulus animals; for example, they become immobile (freeze) in the presence of these stimulus animals and assume the upright defense posture upon their approach (Schoeltens & Van de Poll, 1987; Van de Poll, de Jonge, Van Oyen, & Van Pelt, 1982a; Frishknecht, Siegfried, & Waser, 1982). The fact that the defeated animals behave like this without being directly threatened or attacked confirms that a major change in their behavior has occurred: the nonaggressive animals would not elicit these responses in the defeated subjects were it not for their prior defeat experience. The generalization of defensive and fearful behaviors to nonaggressive stimulus animals is an example of conditioned defeat (CD). The current studies explore the phenomenon of CD in hamsters. Experiment 1 of the present report used a CD acquisition procedure spread over several days since such a procedure was effective in producing substantial changes in hormonal expression (Huhman et al., 1991). However, a briefer CD acquisition procedure would be generally more time efficient and would have special advantages in studies of CD persistence, drug interventions, etc. Therefore, Experiment 2 of the present report examined the feasibility of a 1-day CD acquisition procedure. In both of these studies the generalization tests were 5 min in duration. This period of time is long enough to observe an adequate sample of behaviors. It is short enough that the behaviors remain relatively constant throughout, making the mean rate of occurrence a representative measure. Although CD in mice and rats has been demonstrated in single test trials given 3-14 days after initial conditioning (Frishknecht et al., 1982; Corrigan & Flannelly, 1979; Schoeltens and van de Poll, 1987; Van de Poll et al., 1982a), the response extinguishes with continuous or repeated exposure to nonaggressive partners (Frishknecht et al., 1982; Seward, 1946). Experiment 2 examines the persistence of CD in hamsters. Seward's (1946) observations suggest that CD extinction in rats is a function of the extent of previous defeats: in one experiment he found full recovery of aggressive behavior after 1 defeat and 14 extinction trials but little recovery after 3 initial defeats. Experiment 2 examines the effects of the extent of initial CD training on subsequent behavior. Another question of interest is whether CD diminishes as a consequence of repeated exposure to nonaggressive opponents, i.e., because of nonreinforcement, or

whether there is a passive "forgetting" solely as a function of time. This question is also addressed in Experiment 2 by including a group of animals which had a 4-day delay interpolated between defeat acquisition and subsequent testing. Depending on the species and experimental procedures, CD training also has greater or lesser effects upon other behaviors: Chronically defeated rats show reduced weight gain or frank weight loss and also show reduced locomotion and autogrooming (Weiss, Goodman, Losito, Corrigan, Charry, & Bailey, 1981; Van de Poll et al., 1982a; Van de Poll, Smeets, Van Oyen, & Van der Swan, 1982b; Schoeltens and van de Poll, 1987). The effects of CD on weight regulation in hamsters was examined in Experiment 2. Nest building and flank marking, other territorial behaviors characteristic of hamsters, were also examined in Experiment 2. EXPERIMENT 1

Method Subjects (residents and intruders, and nonaggressive stimulus animals). Twelve male hamsters (150-240 g) were individually housed in hanging Plexiglas cages with a woodchip bedding for 3-6 weeks prior to the experiment. The nonaggressive stimulus animals were 140- to 150-g group-housed males. All animals were maintained on a 14:10 daily light:dark cycle. Food and water were available ad lib except during testing.

CD training (acquisition) procedure. All testing took place during the dark phase of the daily light cycle. Subjects were paired by age and weight; one member of each pair was randomly assigned to the resident or intruder condition. Animals were transported from colony room to testing room for CD acquisition and generalization testing. CD acquisition consisted of placing the pair member chosen as the intruder into the cage of the resident for 5 min/day for 4 consecutive days. The sessions were videotaped; the videotapes were later scored using a behavioral inventory modified slightly from Huhman, Bunnell, Mougey, and Meyerhoff (1990). 1. Nonsocial: locomotor/exploratory, self-groom, nesting, feeding, sleeping. 2. Social orienting: attend, approach, investigate, sniff, or touch nose. 3. Defensive behavior: upright and side defense, tail lift, hind limb abduction. 4. Flight: rapid movement away from opponent either across the cage or up.

95

CONDITIONED DEFEAT IN HAMSTERS Generalization Test stimulus animals ~ q ] vs f o r m e r

CD Aquisition (Day 1)

15

T

T

~intruder [ ~ resident

!

~D

dominant residents

8 ~vs

0c~D

m

m

10

former subordinate intruders

6

0

O

>,tD c cD 23 EY

4

2

k.-

i,

ta_ 0 nonsocial

social orienting

def

flight

offense

FIG. 1. Mean frequency and SEM for each of the behavior categories recorded on the first day of CD acquisition in Exp e r i m e n t 1. Filled bars represent intruders; open bars represent residents.

5. Offense: upright and side offense, chase, attack, and bite. Generalization test. On the fifth day a nonaggressive stimulus animal was placed into the home cage of each subject (former resident or former intruder) for 5 min. The same behaviors were recorded. Results

As shown in Fig. 1, the residents in each of the six pairs produced significantly more offensive be-

40 0

F- Generalization Test (Day 5) ~former subordinate intruders I I f o r m e r dominant residents

30

o

20

L 0 nonsocial

social orienting

def

flight

offense

FIG. 2. Mean frequency and SEM for each of the behavior categories recorded during CD generalization testing (Day 5) of Experiment 1. Filled bars represent intruders; open bars represent residents.

nonsocial

social orienting

def

flight

offense

FIG. 3. Mean frequency and SEM of the behaviors of stimulus animals in Experiment 1 during CD generalization testing (Day 5). Bars with diagonal lines represent behaviors when stimulus animals were confronted with former dominant residents; cross-hatched bars represent behaviors when they were confronted with former subordinate intruders.

haviors [F(1, 10) = 12.8, p < .01] and quickly became dominant on the first training day. The intruders became subordinate, fleeing more often [Mann-Whitney U(6, 6) = 3, p < .01], and engaging in significantly more defensive behaviors [F(1, 10) = 19.3, p < .01]. Once the dominant/subordinate relationship was established, it was maintained over the remaining training days: Residents chased and attacked without fleeing; intruders behaved defensively and fled without attacking. When tested with the small nonaggressive opponent on the Day 5 generalization test, all of the formerly dominant residents (FDRs) attacked the stimulus animal within 2 min. All of the formerly subordinate intruders (FSIs) failed to attack the stimulus animals despite the fact that these animals exhibited no aggressive or offensive behavior (see Fig. 2). This difference in offense between the groups was significant by a Fisher exact test (p < .005). Conversely, as Fig. 2 shows, none of the FDRs fled, while 5 of the 6 FSIs did so (p < .05, Fisher exact test). Furthermore, none of the FDRs behaved defensively, while all of the FSIs did so (p < .005, Fisher exact test). The differences between the behaviors of FDRs and FSIs was not due to differential reactions by the stimulus animals. Except for the time spent responding to subjects' attacks, these animals spent virtually the entire testing period exploring subjects' cages, unaffected by whether the subject was a FDR or a FSI (see nonsocial behavior, Fig. 3).

96

POTEGAL ET AL.

One unexpected observation during the Day 5 generalization testing was that some of the FSIs began to follow the stimulus animal around their cage rather closely. This was surprising because such behavior seemed likely to draw the attention of a potential aggressor and might even have elicited further attack. Following was therefore recorded systematically in Experiment 2.

Discussion This experiment demonstrates that CD can, in fact, be obtained in hamsters. It should be noted that CD was demonstrated in the home cage of the former intruders even though their CD acquisition experience occurred in a cage that was novel to them. This suggests that the CD obtained under these conditions is a robust effect given that Johnston (1975), using other conditions, has found that subordinate hamsters which flee in the home cage of a dominant or even in a "neutral" territory will successfully defend their home cage against intruders. A number of studies show that prior inescapable shock reduces offense and increases defense in intruder rats tested in an intruder/resident paradigm (Williams, 1982; Williams and Lierle, 1986). This raises the possibility that some or all CD effects may be part of a general stress syndrome and this set of chronic defeat effects could be replicated by repeated exposure to any aversive stimulus. We carried out an ancilliary pilot study in which hamsters like those tested in Experiment 1 received 5 min of intermittent exposure to a 1-mA scrambled footshock instead of defeat training on each of 4 days. We found that the repeated shock experience had no net effect on the social behaviors exhibited in a generalization test on the fifth day: the proportions of animals that were dominant, submissive, or neither during the generalization encounters were not different from those of unshocked controls. The contrast between these negative findings and Williams' observations that shock reduces offense and increases defense could easily be due to differences in species and/or methodology. One issue that would require more work is equating the aversiveness of electric shock and conspecific attack. Any attempt to study this question definitively would probably require comparing the effects of a range of shock parameters. It is possible that CD-like effects might be obtained by pairing shock with the presentation of conspecific olfactory cues.

EXPERIMENT 2 In this experiment we induced CD by introducing large, aggressive intruders into the home cages of resident subjects. Pilot studies revealed substantial differences among subjects in the response to these intruders. At one extreme were a few subjects which fled prior to ever being attacked; at the other extreme were one or two subjects which counterattacked and fought with each of the first few intruders. Using a fixed number of CD trials with such a varied population of animals might be expected to yield a high variability in their subsequent responses. A retrospective, trial-by-trial review of the videotapes of Experiment 1 suggested that for each animal it was possible to detect the first trial on which anticipatory flight (AF) occurred, i.e., the first trial on which they fled at the introduction or approach of the next intruder before it actually attacked them. A correlation analysis showed that the earlier in the acquisition trials AF occurred, the more often the subject fled on the generalization test (r = -.7~1). The first occurrence of AF appears to mark an important behavioral change. We therefore adopted the first AF as a behavioral criterion for the attainment of CD in Experiment 2. The same issue of variability appears in the generalization tests where another animal, the stimulus animal, is being used as a standard stimulus to elicit responses in the subject. In some generalization tests of Experiment 1 both subject and intruder fled from each other. The sight of two hamsters in simultaneous wild flight around the cage was very dramatic but it left something to be desired by way of experimental control. Some pilot observations indicated that the problem of variability in the behavior of the stimulus animals would be greatly exacerbated by the fact that flight by one hamster is capable of eliciting chase and attack in another (cf. Grant & MacKintosh, 1963; Grant, MacKintosh, & Lerwill, 1969). Diazepam treatment was found to provide a very workable solution to the problem of controlling the behavior of the stimulus animals. When placed in the cage of another hamster, stimulus animals treated with 15 mg/kg diazepam show stereotyped locomotion; they circle the periphery of cage almost continuously paying little or no attention to the resident. This serendipitously discovered effect provides a near ideal stimulus situation for eliciting CD. In this experiment the use of the videotape was dropped in favor of direct observation because we knew what behaviors we wished to record.

CONDITIONED DEFEAT IN HAMSTERS

Subjects, aggressive intruders, and nonaggressive stimulus animals. The 28 subjects were individually housed male hamsters (130-170 g). The 20 aggressive intruders (AI) were larger (170-200 g), individually housed males selected for high aggressiveness. Nonaggressive stimulus animals were 2,4 larger group-housed males. All animals were maintained on a 14:10 daily light:dark cycle. Food and water were available ad lib.

Defeat training (acquisition) procedure. All training and testing took place during the dark phase of the daily light cycle. Subjects were matched by weight into sets of four; members of each set were randomly assigned among four groups. Aniraals in three of the groups were given a series of defeat acquisition trials in a single day. Each trial consisted of a 5-min period during which defense and flight was elicited by an AI introduced into the subject's home cage. The defeat acquisition trials alternated with 5-min rest periods. Any given subject met a different AI in each of its trials. The AF group was trained until they reached a criterion of anticipatory flight, i.e., until they fled at the introduction or approach of the next AI before it had attacked them. Two other groups, AF + 3 and AF + 3/D, were also trained to this criterion b u t were then given three additional defeat acquisition trials. The single animal failing to show AF in 10 trials was dropped from the study and replaced. The cages of animals in the fourth, Control, group were simply placed on a table in the same room in which the other groups were being tested. Aggressive behaviors recorded were nip (pulling or tugging at the skin while in a quadrupedal posture), bite, and fight (clinch, locked, or rolling fight). Flight and defensive behaviors were those recorded in Experiment 1. To this list was added tooth chattering, following, and fhll submissive (supine) posture. Nonsocial behaviors recorded included digging and flank marking. CD generalization tests. Subjects in the Control, AF, and AF ÷ 3 groups were given 5 min generalization tests with a nonaggressive intruder on 5 successive days following defeat acquisition. Subjects in the AF + 3/D Group rested for 4 days following acquisition and were tested only on the fifth day. The behaviors of the subjects were evaluated as above. Two to three nonaggressive stimulus animals were used repeatedly on a given day. They were given 15 m g / k g ip diazepam 45 min before use. As an additional safeguard against the initiation of social interaction by the stimulus animals, their snouts were sprayed with an aerosol

97

of Cepacol mouthwash (active ingredients include 0.05% cetylpyridinium chloride and 14% alcohol) to reduce their olfactory sensitivity to odor cues from the subjects. Their snouts were resprayed once or twice between tests.

Other measures. J u s t prior to each generalization test the subject's nest was evaluated: A score of 0 was given if no nest had been constructed, a score of 2 was awarded if woodchips were piled up in one corner of the cage and a well-defined nest cup had been scooped out. A score of 1 was given in intermediate cases, e.g., if there was a woodchip pile but it contained no nest cup. The m a x i m u m height of the woodchip pile was also measured. Subjects were weighed immediately after each test. Results CD acquisition. Hamsters in the AF, AF ÷ 3, and AF + 3/D groups achieved AF criterion in a mean (_+ standard deviation) of 3.0 ___ 1.4, 4.5 _+ 2.2, and 3.4 __ 1.9 defeat acquisition trials, respectively. There were no differences among the groups. Both the AF and AF + 3 groups included four animals which fought with the first AI. Two animals in the AF + 3/D group also fought. Generalization tests. Figure 4 depicts group differences and behavioral trends over days. In general, animals in the AF ÷ 3 group showed more extreme and persistent flight and defensive behavior than those in the AF group. As shown in the top panel of Fig. 4, 6 of 7 animals in the AF ÷ 3 fled on the first day, whereas only 3 animals in the AF group did so. The median number of flights in the AF group was 6 which was close to the median of 4 observed in the subordinate animals of Experiment 1. No control animals fled. The resulting distribution of scores across groups permitted only nonparametric analyses of variance. A Freidman test of the total numbers of flights over all 5 test days for the Control, AF, and AF + 3 groups yielded an overall X~(2) = 10.3, p < .006. Post hoc Wilcoxon tests showed that the AF ÷ 3 group fled significantly more often than the AF group (Z= 2.1, p < .04, corrected for ties) or the control group (Z = 2.4, p < .02, corrected for ties). Note that Fig. 4 shows that the response o f A F + 3/D animals which experienced a 5-day delay between CD acquisition and generalization testing was most similar to that of AF ÷ 3 animals on their fifth day of generalization testing. The same combined measure of defense used in

98

POTEGAL ET AL.

8

• ~7 V []

l-

Control AF AF + 3 AF" + 3/Deloy

ii

iI

E "E <

"6 .O

E z

v

v

v

v

v

30

2,5 3

0 Q.

T

20

O) > 15

c ID

a

10

÷

5

0

m

6

<2

E

G)

3

E

z~

v

omz 1

2

3

4

5

AF + 3 group on their first and fifth days. The responses of AF + 3/D group most closely resembled those of AF + 3 animals on their fifth day: AF + 3/D scores were significantly less t h a n AF + 3 scores on Test Day 1 [t(12) = 3.02, p < .03)] but were not different from the AF + 3 scores on Test Day 5 [t(12) = 0.5)]. Because offense behaviors during generalization testing were relatively rare, nip, offense posture, and bite were collapsed into a single score. The bottom panel of Fig. 4 shows the cumulative number of animals which had engaged in any one or combination of these behaviors as a function of generalization trial. Note t h a t the pattern shown by this composite measure of offense is the converse of t h a t shown by flight and defensive behavior in terms of both group differences and trends over days. A Freidman ANOVA of the total numbers of offensive behaviors over all 5 test days yielded a X2(2) = 6.95, p < .04. A post hoc Wilcoxon test showed t h a t the AF + 3 group was significantly different from the control group (Z = 2.2, p < .03, corrected for ties). Becoming supine represents an extreme in defensive postures. Figure 5 (left) shows the significant difference in the median number of supine postures among groups over all 5 test days (Freidman X2(2) = 6.4, p < .05). A post hoc Wilcoxon test showed t h a t the AF + 3 group was significantly different from the control group (Z = 2.0, p < .05, corrected for ties). As shown in Fig. 5 (right), following was greatest in the AF group. A repeated measures ANOVA of the data of matched animals collapsed over trials showed t h a t the difference among the groups was

Generalization Trials

FIG. 4. Behaviors of subjects in Experiment 2 over the 5 days of CD generalization testing. Error bars for defensive postures are SEM.

n~ 4

1 1

Experiment 1 was used in this experiment. As shown in the middle panel of Fig. 4, AF + 3 animals also showed maximal scores in these behaviors. The distributions of these scores permitted parametric analyses. A repeated measures ANOVA revealed a significant difference among Control, AF, and AF + 3 groups [F(2, 18) = 20.6, p < .0001], a significant effect over trials [F(4, 72) = 4.9, p < .002], and a significant Group × Trials interaction IF(8, 72) = 2.3, p < .03]. The defensive behavior of the AF + 3/D group which was tested only on the fifth day after CD acquisition was compared to t h a t of the

0 a.

03

LU Z

.J

U3 Z

Z

,<,,

<_ a

CONTROL

AF

AF+3

CONTROL

AF"

AF+3

FIG. 5. (Left) Median n u m b e r of supine postures. (Right) Means and SEM of the total n u m b e r of times each subject followed the i n t r u d e r in Experiment 2.

99

CONDITIONED DEFEAT IN HAMSTERS

TABLE 1 Measures Unaffected by CD Training Weight (g) Postacquisition Group Control AF AF + 3

Flank marks

Digs

Preacquisition

Day 1

Day 5

Nest height (cm)

2.9 ~ 2.6 3.7 _+ 3.7 1.3 +- 2.1

7.4 ± 7.8 4.4 _+ 4.0 5.2 ± 8.0

152 ± 10 148 ± 11 149 _ 14

151 ± 10 151 ± 12 149 ± 13

156 ± 9 155 ± 14 151 _ 13

6.6 ± .6 6.6 -+ .3 6.4 + .4

Note. For each group the means and standard deviations of flank marks and digs were calculated from the total numbers of these behaviors performed over all five generalization tests by each group member. Nest heights were measured daily and the mean for each subject was calculated. The overall means and standard deviations shown here were calculated from these individual means.

significant IF(2, 12) = 4.2 p < .05]. Post hoc Scheff~ tests indicated a significant difference between AF and AF + 3 groups (p < .05). Following was the only behavior to depart from the pattern of greatest increase in the most extreme CD group. Teeth chattering in rodents reflects anxiety or fear. Four animals in the AF + 3 group chattered their teeth during at least one generalization test. Two of the AF animals and none of the controls did so. Flank marking and digging were reduced in the experimental groups but these effects were not significant (see Table 1). There were no differences in nest quality: The median and modal nest score for all groups was 2. There was also no difference in mean nest height (see Table 1). The majority of subjects in every group gained weight from the CD acquisition day to the first generalization test and over the course of the 5-day testing period. There were no differences in the magnitude of the weight gain (see Table 1). No within-group differences on the generalization test could be found between subjects which fought with the AI on the first defense acquisition trial and those that did not. Furthermore, there were no significant correlations between the number of trials or flights during acquisition and any measures of flight and defense on the generalization tests.

An ancillary observation. Hamsters engage in at least one species typical grooming behavior not observed in mice or rats: they will sit on their hindlimbs, ventroflex their heads, and remove fecal boli from their anus with their teeth. The bolus is usually flung away with a toss of the head. We had originally recorded this behavior along with the more common mouth and paw movement as "grooming" until midway through Experiment 2 when we noted that only CD trained animals performed this behavior during generalization tests. Because the grooming recorded up to that point represents dif-

ferent sorts of behavior for controls and experimentals these data have not been reported here. The fact that only CD-trained animals performed this behavior suggests that it may be related to the "emotional" defecation occurring in other rodent species. Future studies of CD in hamsters will distinguish and separately record grooming of the body surface and oral feces removal.

Discussion The phenomenon of conditioned defeat was first reported in rats and mice, two closely related species within the family Muridae. The current report extends the demonstration of CD to the golden hamster, a more distantly related rodent of the family Cricetidae (Honacki, Kinman, & Koeppl, 1982), indicating the cross-species generality of this phenomenon. CD in hamsters appears to be robust since it has been obtained in two experiments carried out several years apart under the supervision of two different authors (K.H. and M.P.) using different conditions of acquisition. It is also robust in the sense that a number of different behaviors were affected. Chronic defeat enhances learning of a shock avoidance task in mice (Hudgens & MacNeil, 1970) and impairs shuttlebox performance in rats (Williams & Lierle, 1988). These findings are difficult to interpret because the relationship between the commonly used shock-motivated experimental learning paradigms and the real life challenges to free living animals is unclear. The present studies of CD involve a model of aversive social learning which takes place under more naturalistic conditions. The importance of these results is suggested by the observation that male mice made submissive by CD fail to copulate with, and may even avoid, receptive females (Kahn, 1951; Yoshimura & Kimura, 1991). Behavioral data of this sort is lacking in rats and hamsters (cf. Schoeltens & van de Poll,

100

POTEGAL ET AL.

1987) but defeat-induced reductions in plasma testosterone and other physiological indices of reproductive capacity have been demonstrated in these and other species (Brain & Benton, 1983; Huhman et al., 1991). It is therefore clear that chronic defeat is a highly significant event in the life of a rodent which could directly and dramatically affect its inclusive fitness. Experiment 2 demonstrates that a 1-day CD acquisition procedure is feasible in hamsters; this is comparable to the 1-day acquisition procedure for mice developed by Frishknecht et al, (1982). When generalization trials are restricted to 5 min in length, the persistence of CD over days is sufficient to allow repeated postacquisition testing. It would be of interest to determine the time course of the effect with longer generalization trials. In Experiment 2 the total duration of exposure to nonaggressive animals was 25 min. Would the same changes in behavior occur during a single, 25-min generalization test? Although not explicitly tested, it seems reasonable to suppose that the strength and persistence of the CD found in Experiment 2 may be a function of the use of several different opponents during acquisition. That is, subjects could not associate their defeat solely with confrontation with some particular opponent which is the case in the development of the dominance/submission relationship in standard paradigms. The flight and defense behavior of animals in the AF ÷ 3 group of Experiment 2 was more frequent and extreme than that of the AF group. The three additional CD trials given to the AF + 3 group produced avoidance and escape behaviors of a higher energetic cost, an effect presumably mediated by their greater fearfulness. Remaining supine (assuming the full submissive posture) is an alternative strategy with a lower energetic cost. To date we have encountered 2-3% of animals who show no flight but show more prolonged maintenance of the supine posture with increasing number of CD trials. Adams(1979) has argued that the same stimuli that provoke defense, e.g., sudden noise, pain, restraint, will also cause submission if familiar conspecific odors are present. According to Adams, the switch in behavioral mode of responding is controlled by an hypothetical neural mechanism located in the ventromedial hypothalamus. If so, the individual differences may have to do with the sensitivity and/or "set point" of this mechanism. The following of the stimulus animal by the subjects departed from the pattern of other behaviors in that it was maximal in the AF group. Following may be an example of the "risk assessment" ac-

tivities directed toward a potential threat; these activities are an integral part of defensive behavior according to the Blanchards (e.g., Blanchard & Blanchard, 1989). Following also occurs in defeated rats where it is usually associated with ultrasonic emissions directed at the aggressive rat. A curious parallel to following in these rodents has been reported in vervet monkeys. In this species a subordinate animal may follow, and even appear to chase, a dominant just after having been harrassed by that animal. The dominant animal remains in control, however, since the "chase" is terminated as soon as the dominant stops moving (Kaplan, 1987). The fact that it was the AF group in Experiment 2 that showed maximum following suggests that this behavior emerges at intermediate levels of fearfulness. There were no within-group differences on the generalization tests between subjects which fought with the intruder on the first CD acquisition trial and those that did not. Furthermore, there was a lack of correlation between the number of trials or flights to the AF criterion and measures of flight and defense on the generalization tests. This set of findings, taken together, suggests that training animals to AF criterion minimizes or abolishes the effects of preexisting individual differences. From a methodological viewpoint this technique is clearly to be preferred because it reduces variance. From a theoretical viewpoint we may have identified a behavioral indicator reflecting a critical internal event, threshold, or turning point at which the neural circuitry controlling fearful behavior becomes dominant over the circuitry promoting aggressive or nonsocial behavior. The flight and defensive behaviors of animals which experienced a 5-day delay between CD acquisition and generalization testing was most similar to that of animals in the AF and AF + 3 Groups on their fifth day of generalization testing. This observation suggests that CD in the repeatedly tested groups diminished as a function of time and not as the consequence of repeated exposure to a nonaggressive conspecific. A passive decay model of CD diminution appears to be a more appropriate representation of the change than does an extinction model. Note that the disappearance of overt CD behavior does not imply that all traces of CD training have vanished; it would not be surprising if animals which had once acquired CD were later susceptible to a more rapid relearning of CD. The differences in flank marking among the groups did not reach conventional levels of statistical significance. As indicated in Table 1, however,

CONDITIONED DEFEAT IN HAMSTERS

the rate of flank marking in the AF + 3 group was less than 50% that of the control group. This difference was the same order of magnitude as that between dominant and submissive hamsters in the study of Ferris, Axelson, Shinto, and Albers (1987). The fact that CD had no effect on nest construction and maintenance indicates that the CD-induced changes in behavior are directly related to interactions with conspecifics and not to nonsocial territorial behaviors. It also shows that the animals' ability to carry out species-typical behaviors was unaffected by the stress of CD. Hamsters are an important addition to the group of rodent species in which CD may be studied; their brains are larger than those of mice making neurochemical and neuropharmacological experiments easier to carry out. Furthermore, CD appears to be achieved more quickly in hamsters than in mice. Compared to rats, on the other hand, hamsters are more reliably aggressive. Hamsters show high frequencies of aggressive behavior after short periods of individual housing, unlike rats which must be group housed in colonies for extended periods before they reliably exhibit aggression. These characteristics make the hamster a very appropriate species in which to carry out the interesting and important studies which will further our understanding of CD. On a technical note, a number of techniques for the preparation of standard stimulus animals to be used in studies of social behavior have appeared in the literature. These include anosmia, olfactory bulbectomy, and group housing (e.g., Benton, 1981). Published techniques for hamsters include the use of a muzzle and the analgesic/sedative methotrimaprazine (Potegal, Blau, Black, & Glusman, 1980). The range of manipulations for preparing standard stimulus hamsters has been extended by the discovery that high doses of diazepam produces animals which locomote almost continuously around the cage while paying little attention to the other animal. This effect may be an instance of the exaggerated exploration induced by benzodiazepines (see Treit, 1985, for a review of these effects). The utility of this approach for preparing standard stimulus animals in other rodent species needs to be explored. REFERENCES Adams, D. B. (1979). Brain mechanisms for offense, defense, and submission. Behavioral and Brain Sciences, 2, 201-241. Anderson, S., & Jones, J., Jr. (1984). Order and families of recent mammals of the world. New York: Wiley. Benton, D. (1981). The measurement of aggression in the lab-

101

oratory. In P. F. Brain and D. Benton, (Eds.). The biology of aggression. Alphen aan den Rijn, The Netherlands: Sijthoff and Noordhoff. Blanchard, R. J., & Blanchard, D. C. (1989). Attack and defense in rodents as ethoexperimental models for the study of motion. Progress in Neuro-Psychopharmacology and Biological Psychiatry, 13(Supplement), $3-$14. Brain, P. F., & Benton, D. (1983). Conditions of housing, hormones, and aggressive behavior. In B. Svare (Ed.). Hormones and aggressive behavior. New York: Plenum. Corrigan, J. G., & Flannelly, K. J. (1979). Ultrasonic vocalizations of defeated male rats. Journal of Comparative and Physiological Psychology, 93, 105-115. Ferris, C. F., Axelson, J. F., Shinto, L. H., & Albers, H. E. (1987). Scent marking and the maintenance of dominant/ subordinate status in male golden hamsters. Physiology and Behavior, 40, 661-664. Frishknecht, H-R., Seigfreid, B., & Waser, P. G. (1982). Learning of submissive behavior in mice: A new model. Behavioral Processes, 7, 235-245. Grant, E. C., & MacKintosh, J. H. (1963). A comparison of the social postures of some common laboratory rodents. Behaviour, 21, 246-259. Grant, E. C., MacKintosh, J. H., & Lerwill, C. J. (1969). The effect of a visual stimulus on the agonistic behaviour of the golden hamster. Behaviour, 73-77. Honacki, J., Kinman, K., & Koeppl, J. (1982). Mammal species of the world. Lawrence, Kansas: Allen Press & The Association of Systematics Collections. Hudgens, G. A., & MacNeil, D. A. (1970). Aggressiveness and learning ability: Effect of histories of wins or defeats on avoidance learning in mice. Psychonomic Science, 20, 5153. Huhman, K. L., Bunnell, B. N., Mougey, E. H., & Meyerhoff, J. L. (1990). Effects of social conflict on POMC-derived peptides and glucocorticoids in male golden hamsters. Physiology and Behavior, 47, 949-956. Huhman, K. L., Moore, T., Ferris, C., Mougey, E. H., & Meyerhoff, J. L. (1991). Acute and repeated exposure to social conflict in male golden hamsters: Increases in plasma POMC-peptides and cortisol and decreases in plasma testosterone. Hormones and Behavior, 25, 200-216. Johnston, R. E. (1975). Scent marking by male golden hamsters (Mesocricetus auratus). III. Behavior in a seminatural environment. Zietschrift fur Tierpsychologie, 37, 213-221. Kahn, M. W. (1951). The effect of severe defeat at various age levels on the aggressive behavior of mice. Journal of Genetic Psychology, 79, 117. Kaplan, J. (1987). Comparative behavior of African monkeys. New York: A. R. Liss. Nowack, R. M., & Paradiso, J. L. (1983). Walker's mammals of the world, 4th ed. Baltimore, MD: Johns Hopkins University Press. Potegal, M., Blau, A., Black, M., & Glusman, M. (1980). A technique for the study of intraspecific aggression in the golden hamster under conditions of reduced target variability. The Psychological Record, 30, 191-200. Schoeltens, J., & Van de Poll, N. E. (1987). Behavioral consequences of agonistic experiences in the male $3 (Tryon maze dull) rat. Aggressive Behavior, 13, 213-226. Seigfreid, B., Frischknecht, H.-R., & Waser, P. G. (1982). A new

102

POTEGAL ET AL.

learning model for submissive behavior in mice: Effects of nalozone. Aggressive Behavior, 8, 112-115. Seward, J. P. (1946). Aggressive behavior in the rat. IV. Submission as determined by conditioning, extinction and disuse. Journal of Comparative Psychology, 39, 51-76. Treit, D. (1985). Animal models for the study of anti-anxiety agents: A review. Neuroscience and Biobehavioral Reviews, 9, 203-222. Van de Poll, N. E., de Jonge, F., Van Oyen, H. G., & Van Pelt, J. (1982a). Aggressive behavior in rats: Effects of winning or losing on subsequent aggressive interactions. Behavioral Processes, 7, 143-155. Van de Poll, N. E., Smeets, J., Van Oyen, H. G., & Van der Swan, S. M. (1982b). Behavioral consequences of agonistic experience in rats: Sex differences and the effects of testosterone. Journal of Comparative and Physiological Psychology, 96, 893-903. Weiss, J., Goodman, P., Losito, B., Corrigan, S. M., Charry, J.,

& Bailey, W. (1981). Behavioral depression produced by an uncontrollable stressor: Relationship to norepinephrine, dopamine, and serotonin levels in various aspects of rat brain. Brain Research Review, 3, 167-205. Williams, J. (1982). Influence of shock controllability by dominant rats on subsequent attack and defensive behaviors toward colony intruders. Animal Learning and Behavior, 10, 305-313. Williams, J., & Lierle, D. M. (1986). Effects of stress controllability, immunization, and therapy of the subsequent defeat of colony intruders. Animal Learning and Behavior, 14, 305314. Williams, J., & Lierle, D. M. (1988). Effects of repeated defeat by a dominant conspecific on subsequent pain sensitivity, open-field activity, and escape learning. Animal Learning and Behavior, 16, 477-485. Yoshimura, H., & Kimura, N. (1991). Ethopharmacology of copulatory disorder induced by chronic social conflict in male mice. Neuroscience and Biobehavioral Review, 15, 497-500.