Proximity Measures of Social Play in the Laboratory Rat GRAHAM HOLE Department of Psychology University College London London
Litters of laboratory rats (Rat(u.7 noruegic.rt.7) were filmed between 21 to 60 days of age, and data on 700 social play bouts obtained. These were analyzed for information on how play bouts ended and on the spatial relationship between the players and their physical environment. The results suggest that social play in this species (a) does not appear to be a covert means of securing priority of access to resources such as food, females, or territory; and (b) does not appear to represent practice for behaving in the contexts within which "serious" adult aggression occurs (e.g., practicing being "territorial ," etc. ). Players showed a strong preference for playing close to cover and away from exposed areas; if this is also true for wild rats, it would minimize the potential costs of play in terms of increased conspicuity to predators.
Introduction To date, three themes have dominated research on animal play: the problem of defining play satisfactorily, study of the morphology of the behaviours concerned, and speculation on the functions of play (reviews in Aldis, 1975; Symons, 1978; Fagen, 1981; Bekoff & Byers, 1981; Smith, 1982, 1984; Pellis, 1988). While these are important and interesting issues, our understanding of play would be augmented by research on some of its less well-studied aspects, for example its immediate causation and the behavioural and environmental context within which it takes place. This article looks at a feature of play which, with some exceptions (e.g., Byers, 1977), has received scant attention: the location of players, both in relation to each
Reprint requests should be sent to Graham Hole, School of Social Sciences, Arts Building, University of Sussex, Falmer, Brighton, BNI 9QN, England. Received for publication 23 December 1988 Revised for publication 4 September 1990 Accepted at Wiley 9 November 1990 Developmental Psychobiology 24(2): 117-133 (1991) 0 1991 by John Wiley & Sons, Inc.
CCC 001 2-1630/91/020117-17$O4.00
118
HOLE
other and to their physical environment. Specifically, three aspects of social play will be examined using observational data from 700 social play bouts between laboratory rats aged 21-60 days. First, an examination will be made of the spatial relationships between players before and after play. Does play occur between rats who are already in close proximity or between animals who are initially at some distance from each other? Does play lead to the separation of animals who were previously close to each other? The second issue is who approaches whom in play. Does the initiator of a play bout tend to approach his partner before the play occurs or does he initiate play with an animal who approaches him? Third, the location of play bouts with respect to the players’ physical environment will be investigated. Where does play occur within the animals’ environment? Studying attributes of play may provide information relevant to the problems of defining play and identifying its developmental significance. Data on these three features may throw some light on a hypothesis which has received comparatively little attention in play research: that social play is merely immature, nondamaging aggression (see Fagen, 1981). Paradoxically, while the similarities between social play and adult aggression have repeatedly been noted (e.g., Aldis, 1975; Symons, 1978; Smith, 1982), this has generally been in order to stress the dissimilarities between the two behaviours. It has been assumed that, since play lacks t h e immediate biological goal of “serious” aggression-access to some resource-play is not in itself aggression but may represent direct practice for it. When claims are made that play is “practice” for adult aggressive behaviour, researchers presume that the behaviour patterns of play somehow represent practice of those patterns employed in adult agonistic interactions. There is already good evidence that the behaviour patterns of play cannot represent practice of adult aggressive patterns in any obvious way, at least in rats and hamsters (Pellis, 1988; Pellis & Pellis, 1987, 1988a, b). Pellis and Pellis base their arguments on a consideration of the morphology of the behaviour patterns in play and aggression, and on the fact that the body sites for attack and defense during social play in these species differ from those sites involved in aggressive encounters. Play might, however, represent practice in another, rather different, sense to that normally considered; it might represent a covert means of securing priority of access to resources, such as food, females, or territory. Such preemption of resources would be difficult to recognize, and yet might be a significant factor in adult reproductive success. A second possibility is that play might represent practice in “being territorial,” in “attacking an intruder,” or “defending a resource.” Play might represent practice not of the behaviour patterns involved in “serious” aggression, but rather of behaving in the contexts within which the “serious” adult behaviour patterns occur-in effect, learning that a particular kind of behaviour is the appropriate thing to do in a particular circumstance. Data on the proximity of players can provide some information on this point as follows. If play were a means of securing access to resources, then one might expect the recipient to approach the initiator (so that play marks the defense of a resource by the holding animal). An analysis of where play takes place within the environment may also be relevant here, determining for example, whether play occurs in the vicinity of material resources, such as food, water, or nest sites.
PROXIMITY MEASURES OF SOCIAL PLAY IN THE RAT
119
Information on where play occurs may also provide some insights into the actual survival costs of play-that is, the costs which play has for an animal in practice as opposed to the costs which play has commonly been assumed to possess (Fagen, 1981; Martin & Caro, 1985). For example, it is often claimed that play may increase the conspicuity of players to predators. However, in reality this may not be the ,case. Informal observations of rat play suggest that play occurs around the edges of a cage where the animals are close to protective cover. The present study will investigate this suggestion quantitatively.
Methods Three separate litters were filmed on video. (Some features of these subjects’ play behaviour have already been described in a previous paper, Hole, 1988). The 3 i hr period from 30 min before dusk to 3 hr after dusk was filmed each day, between 21 and 60 days of age. (Previous observations suggested that rats are particularly likely to engage in social play around dawn and dusk). Wrestling bouts were subsequently selected for analysis if they satisfied the criteria outlined below. (Of bouts meeting these criteria, however, only a randomly chosen subset were actually analyzed). This resulted in a total of 700 play bouts being analyzed: 246 for Litter One (LI); 272 for Litter Two (L2); and 182 for Litter Three (L3).
Operational Definition of “Rough-and-Tumble’’ Play A wrestling bout was defined as any interaction occurring between two subjects which began with a pounce, and in which one of the two was held down on its back by the other. (Thus “unsuccessful” play solicitations, in which one animal failed to solicit social play from another, are not considered in this study). The beginning of a bout was considered to be the initial pounce. A bout was considered to have ended when the two animals ceased to maintain ventral-ventral contact for more than three sec. The only interactions selected for analysis were those which consisted principally of wrestling, although some elements of chasing and rearing and/or “boxing” were included by virtue of the definition employed. Only dyadic interactions were analyzed and of these, only those were selected in which the two animals were in complete view throughout the play bout.
Subjects PVG hooded rats were used. Each litter was run independently: Each pregnant mother was placed in the observation cage described below so that her litter was wholly raised under the conditions in which it would be observed. Each litter was culled to six pups (three male, three female) and the mother removed at 21 days, after which the pups were left undisturbed in the test environment until observations ended at 60 days. Subjects were identified on the basis of their natural coat markings.
120
HOLE
Apparatus Subjects were housed in a large purpose-built observation cage, measuring 2 m x 2 m x 50 cm. All four walls and the roof of the cage were made of Plexiglass, except that the upper half of the rear wall was constructed of wire mesh to provide ventilation. Bedding consisted of garden topsoil distributed to an initial depth of approximately 10 cm. Food (SDS rat and mouse maintenance diet number 1 in expanded pellet form) and water were available ad lib from dispensers placed at the right and left corners of the cage, respectively. Lighting was on a 12 h light/dark cycle, with the lights dimmed or brightened over a 2-hr transition period between each phase of the cycle so as to simulate “dawn” and “dusk.” Illumination was provided by six 100-watt tungsten lamps during the light phase, and by two 40-watt red lamps during the dark phase. Filming was carried out using a National Panasonic time-lapse video recorder, (on its continuous rather than time-lapse setting) and a Sony video camera placed at the front of the cage. A zoom lens provided a clear view of animals situated toward the back of the cage. Marks on the walls and floor of the cage enabled fairly accurate judgement of the animals’ distance from the cage walls and from each other. Variables
Initiator or Recipient of Play The identity and sex of the participants in each wrestling bout were recorded. The animal which performed the initial pounce, defined as commencing the interaction, was designated the initiator of the bout (I). The animal who was pounced upon was considered to be the recipient of the wrestling (R).
Bout Endings A record was made of the nature of the behaviour by either of the participants which appeared to terminate the interaction (i.e., behaviour which occurred immediately before the players separated). On the basis of preliminary observations, a number of categories were set up which accounted for the majority of bout endings: Either of the interactants moving away. This was by running or walking away from the partner. Distraction of either interactant. This category included either animal breaking off the current interaction in order to participate in a fresh one with a different animal. For example, a common ending was for the animal on top to break off from his wrestling activities and chase after a different animal in order to play with him. Olfactory investigation of environment. Either animal might suddenly “switch off” from wrestling and engage in prolonged sniffing of the immediate surroundings. Olfactory investigation of other interactant. Either animal might cease wrestling and engage in prolonged sniffing of the wrestling partner. Self-grooming. Either animal might break off to groom himself.
PROXIMITY MEASURES OF SOCIAL PLAY IN THE RAT
121
Sexual behaviours. Sexual activity directed by one of the participants towards the other effectively ended the wrestling interaction. Other. Any behaviours which ended the bout but which did not fall into any of the above categories. For each category, a note was made of which animal was responsible for terminating the bout.
Proximity of Interactants For each play bout, a note was made of the proximity of players to each other in the three sec before and after the bout. If the animals’ bodies were within 20 cm of each other, they were scored as being in “close proximity”; if they were further apart, “no proximity” was recorded.
Who Approached Whom Before Play Commenced This was classified according to whether the initiator of the bout approached the recipient; the recipient approached the initiator; both were already adjacent to each other; or both independently moved towards each other.
Location Within Environment A record was made of the location of each play bout within the cage and the proximity of the interactants to the nearest wall or cover. The cage floor was divided into nine areas of equal size: “far left,” “far middle,” and “far right;” “middle left,” “middle,” and “middle right;” “near left” (which contained the water spouts), “near middle,” and “near right” (which contained the food hopper). The location of each play bout within this grid was recorded. Proximity to the wall was scored as within 10 cm (“close”); between 10-20 cm (“intermediate”); or over 20 cm (“no proximity”). Wrestling interactions within a single bout could often occur at a number of different locations. Therefore, proximity was taken as the distance from cover at which most of the wrestling occurred, ignoring any chases which might take place within a bout. Although chases could result in the interactants moving away from cover for very brief periods, the important factor was where the wrestling itself occurred.
Results Play Bout Endings
Overall Frequency of Bout Endings The “21-60 days” column in Table 1 shows the relative frequencies with which various categories of bout ending were observed to occur, combining the data for both sexes and all age periods. The most common way in which a bout ended (43%) was by one of the interactants moving away from the other. An initiator was as likely to move away from the recipient as was the recipient from the initiator; 23% of bouts were terminated by the initiator moving away, and 20% by the departure of the recipient.
122
HOLE
Table 1 Age Changes in How Play Ended, (“I” = Initiator, “R” = Recipient). Category of bout ending 1 moves away
R moves away I is distracted R is distracted
I sniffs ground R sniffs ground I self-grooms
21-30
3 1-40
21.0% (37) 22.2% (39) 13.1% (23) 7.4% (13) 17.6% (31) 9.7% (17) 0.6% (1)
R self-grooms I sniffs R
R sniffs I
2.3% (4) 3.4% (6) 0.6% (1)
sexual behaviour other
0.0% (0) 2.3% (4)
Age period (days) 41-50
51-60
2 1-60
20.2% (21) 20.2% (21) 12.5% (13) 3.8% (4) 16.4% (17) 4.8% (5) 8.6% (9) 1 .O% (1) 6.7% (7) 1.9% (2) 0.0% (0) 3.8% (4)
* Percent of the total number of play bouts observed in each age period, combining the data for males and females. The second most frequent way in which play ended (26%, 17% terminated by initiator, 8% by recipients) was by one of the players turning from play to olfactory investigation of the surrounding inanimate environment (cage walls or floor). Usually, the initiator stepped off his play partner and began to sniff the ground nearby (or the wail if play was occurring immediately next to it), apparently disregarding totally the animal with which he had just played, becoming unreceptive to any attempts by that animal to renew play. A recipient either ended bouts in the same way (if he was on top at the time) or if he was lying on his back pinned down by the initiator, twisted around to sniff the ground, with the initiator apparently offering no resistance. The third most frequent way in which play ended (18%) was for either the initiator or recipient to be distracted from playing with their present partner. Distraction occurred either by a third animal breaking into the play bout, so that the initiative for terminating the encounter in a sense came from this other pup, or by the player himself breaking off play with his present partner in order to sniff, chase, follow, play, or otherwise interact with a different animal nearby. Small proportions of bouts (under 5% in all cases) were observed to be terminated by self-grooming or by sniffing the play partner. In both cases, initiators performed these behaviours more often than recipients. In the case of self-groom-
PROXIMITY MEASURES OF SOCIAL PLAY IN THE RAT
123
ing, it appeared that the initiator, lying over his partner and in the process of grooming him, transferred his efforts towards grooming himself, allowing the recipient to right himself unimpeded. The termination of play by one of the players in order to sniff the play partner was in practice quite different from the kind of olfactory investigation performed within the play bout itself since it was accompanied by the sniffing animal moving away from ventral-ventral contact with his partner and generally failing to respond to any attempts by the other animal to maintain a playful type of interaction.
Age Differences .in Frequency of Bout Endings Table 1 shows how often the different categories of bout endings occurred in each age period, combining the data for males and females. With some fluctuations, the overall pattern described above was shown in all four age periods-that is, the most common endings were first, separation of the players; second, breaking off from play to sniff the environment; and third, distraction by a different animal. This analysis does, however, reveal two age changes in bout endings which may possibly be related. A steady increase with age was seen in the frequency with which play ended by the initiator sniffing the recipient; and sexual activities were seen to terminate play only during the 41-50 day age period (and then only in less than 2% of bouts).
Overall Sex-Differences in Frequency of Bout Endings Table 2 shows the relative frequencies of the different categories of bout endings, analyzed according to the sexes of the animals involved-MI/MR, FI/FR, MI/FR, FI/MR. With minor variations, the overall pattern of bout endings described previously was shown by all classes of interactants. Further analysis also revealed no age-related sex differences in how play bouts were terminated.
Proximity of Interactants Before and After Play
Overall Proximity of Players Table 3 shows the overall frequencies with which players (21-60 days) were observed to be in close proximity or not, both before and after bouts, combining the data for all ages and both sexes. Overall, more play bouts (56%) occurred between individuals who were already in close proximity than between individuals who were not (46%).Similarly, more play bouts (58%) were followed by the interactants remaining close to each other than by separation (42%). When proximity before and after each bout is taken into account, four possible permutations were: proximity before and after; no proximity before or after; proximity before but not after; and no proximity before but proximity after. Table 3 shows that the most frequent of these was proximity both before and after a bout (38%).
HOLE
124
Table 2 Overall Sex Differences in How Play Ended (21-60 days) Category of bout ending
MI-MR
Sexes of interactants FI-FR MI-FR
I moves away
R moves away I is distracted
R is distracted I sniffs ground R sniffs ground
I self-grooms R self-grooms
I sniffs R R sniffs I sexual behaviour other
FI-MR 24.4% (40) 17.1% (28) 12.8% (21) 3. I% (5) 20.1% (33) 10.4% (17) 1.8% (3) 4.3% (7) I .8%
(3) I .2% (2) 0.6% (1) 2.4% (4)
* (Percent of the total number of play bouts involving sexes of interactants concerned, combining data from all age periods. “M” = male, “F” = female, other codes as in Table 1).
Age Changes in Proximity of Players Table 3 shows the data on the proximity ofplayers before and after play, broken down by age but combining the data for both sexes of interactant. Examination of Table 3 suggests that in the first three age periods, the overall pattern described in the previous section was shown, with approximately 60% of play bouts occurring between pups who were already close together. In the final age period, however, there appears to be a reversal of this pattern with 60% of bouts occurring between animals who were not close to each other. (A more detailed analysis for age-related sex differences revealed that this reversal in the final age period was shown by all four permutations of interactant, but most markedly so in bouts initiated by females with males, with 72% of bouts [I3 out of 181 occurring between players who were not already close to each other). The frequencies with which players remained close together after play had ended were very consistent across age periods; in all cases, approximately 60% of play bouts were followed by the participants remaining in close proximity.
Overall Sex Differences in Proximity of Players Table 4 shows the relative frequencies of the proximity measures described above, analyzed according to the participants’ sexes but summing the data over
PROXIMITY MEASURES OF SOCIAL PLAY IN THE RAT
125
Table 3 Overall Age Changes in Frequency of Proximity between Players Before and After Play (Males and Females Combined). Proximity, before and after play
age period 41-50
2 1-30
31-40
P before, P after
36.0% (54)
42.6% (75)
NP before, NP after
22.7% (34)
25.0% (44)
P before, NP after
21.3% (32)
13.6% (24)
NP before, P after
20.0% (30)
18.8%
(33)
17.4% (47)
26.9% (28)
100.0% (150)
100.0% ( 176)
100.0% (270)
100.0% (104)
100.0% (700)
Total P before
57.3% (86)
56.3% (99)
40.4% (42)
56.1% (393)
Total NP before
42.7% (64)
43.7% (77)
59.6% (62)
43.9% (307)
Total P after
56.0% (84)
57.4% (101)
58.6% (61)
Total NP after
44.0% (66)
42.6% (75)
35.4% (43)
Total
21.1% (57)
5 1-60
21-60
32.7% (34)
~~~
* (Percent of the total number of bouts observed in each age period. “P” “NP”
=
= close proximity;
no proximity)
all age periods. In play bouts involving female recipients (i.e., FI/FR and MI/FR play), the interactants were as likely to be apart from each other before or after play as they were to be together. In contrast, when play involved a male recipient, the players were more likely to be in close proximity before and after play than they were to be apart (60%). In other words, male recipients were more likely to be involved in play with an animal near them, and were more likely to remain close to that animal once the play had ended. In all cases, the most commonly observed permutation of proximity scores was proximity both before and after play. However, in play involving a female recipient, the “no proximity before or after” category occurred almost equally as frequently.
Who Approached Whom Before Play Commenced Overall Frequency of Categories of Approach Behaviour Figure 1 gives the overall frequencies with which each of the possible categories of approach behaviour was observed to occur, combining the data for both
126
HOLE
Table 4 Overall Sex Differences in Frequency of Proximity between Players Before and After Play (21-60 Days). MI-MR
P before, P after
41.9% (75) 21.2% (38) 17.9% (32) 19.0% (34) 100.0% ( 179)
31.1% (37) 16.0% (19) 19.3% (23) 100.0% (1 19)
MI-MR
FI-FR
NP before, NP after P before, NP after NP before, P after Total:
Total P before Total NP before Total P after Total NP after
59.8% (107) 40.2% (72) 60.9% (109) 39.1% (70)
FI-FR
Sex of Players MI-FR
Proximity:
FI-MR
33.6% (40)
MI-FR
49.6% (59) 50.4% (60) 52.9% (63) 47.1% (56)
FI-MR 59.8% (98) 40.2% (66) 64.6% (106) 35.4% (58)
* (Percent of the total number of bouts performed by each permutation of players’ sexes)
sexes and all age periods. Nearly half (46%) of all bouts were preceded by I approaching R. The next most common category, accounting for 30% of all bouts, was for both pups to be close to each other already. 18% of bouts occurred after R approached 1. It was relatively uncommon for both partners to approach each other, only 5% of all bouts resulting from this type of encounter. Age-Differences in the Frequency of Categories of Approach Behaviour
Figure 1 shows the relative frequencies of the different categories of approach behaviour in each of the four age periods, combining the data for both sexes. There is some indication of an age-related trend, in that with increasing age, play was more frequently preceded by the initiator approaching the recipient. There was a corresponding decrease with age in the frequency with which play occurred between two animals already in close proximity.
Sex Differences in Approach Behaviouv No sex differences were found in approach behaviour; the pattern described above was shown by both sexes at all ages.
PROXIMITY MEASURES OF SOCIAL PLAY IN THE RAT TYPE OF APPROACH BEHAVIOUR
I to R R to I mutual
( A ) 21-30 DAYS:
6.P ..
ltoR Rtol
( 8 ) 31-40 DAYS
(C) 41-50 DAYS:
b
77 28
8 64
-
mutual B.P. ItoR 1
(9) 51-60 DAYS: r!%aIB.P.
I to R
21-60 DAYS:
,",ttal B.P.
(O'JEFtALL)
127
~
17
~~~
62 28
e
18
308
F 1
123
36 204
0 I0 20 30 40 50 60 K TOTAL NO.BOUTS PER AGE-PERIOD
Fig. 1 . Relative frequencies of different types of approach behaviour (data for males and females combined).
Location of Play Within Environment Proximity of Players to Cover During Play Figure 2 shows the overall frequency with which play was observed to occur in each of the three proximity zones used, combining the data from a11 ages and 80
w
TOTAL NO.BOUTS
7
495
60 I I I
40
119
20
OL c10 cm
10-20 crn
,20 cm
PROXIMITY TO CAGE WALL Fig. 2. Relative frequency with which play occurred in proximity to the cage wall.
128
HOLE Z TOTAL NOBOUTS 20
16
10
5
0
FL
FM
FR
ML
MIDDLE
MR
HVITER
NM
FOOD
LOCATION OF PLAY-BOUTS Fig. 3.
Relative frequency of play in different locations within the cage
both sexes. These observed data are shown compared to the frequencies which would be expected if social play was located at random within the environment. In calculating the expected frequencies, account had to be taken of the different areas encompassed by each of the zones: 64% of the cage floor area was contained within the inner zone (i.e., over 20 cm from the wall), 17% by the middle zone, and 19% by the strip nearest the walls, so one might expect the animals to be observed mainly within the central area. In contrast, animals exhibited a very clear preference for playing within 10 cm of the nearest wall. These data underestimate the extent of the preference for cover during play, since many of the bouts scored under “no proximity” were located close to large bundles of nest material or within deep depressions excavated by the animals (not necessarily those that were playing) in the inner regions of the environment. The number of bouts occurring away from any cover was very small, suggesting that rats strongly prefer not to play in exposed areas-that is, that during play, they maintain the marked thigmotaxis that they display during more “serious” activities. No changes in this pattern were found in relation to age or sex of players. Clearly, increased body size of the subjects (so that, relative to the animals, the environment becomes smaller) and increased familiarity with the environment with increasing age did nothing to change the animals’ preference for cover during play.
Locution of Play Bouts in Environment Figure 3 shows the observed frequency of play in each region of the cage. Data were collected only for 95 bouts from L1 (between days 41 and 60), and 92 bouts from L2 (between days 21-30 and 41-60), so a full age and sex analysis was not possible. Fig. 3 suggests that although play was not distributed randomly around the perimeter of the cage, neither was it especially associated with areas of the environment which contained resources such as food and water; play appears to have occurred where the animals were most active.
PROXIMITY MEASURES OF SOCIAL PLAY IN THE RAT
129
Discussion Relationship of Play t o Aggression Various data reported here suggest that social play in this species occurs in a very different context to that of either juvenile or adult aggression, and that it has different immediate consequences, despite the superficially similar appearance of play to aggressive behaviours. The same conclusion has been drawn on the grounds of the morphology of the behaviour patterns in play and aggression (Hole & Einon, 1984), and on the basis of a consideration of the differing body sites of attack and defence during play and aggression (Pellis & Pellis, 1987, 1988a,b; review in Pellis, 1988). In the present study, the most direct evidence of a lack of association between play and agonistic activities comes from the data on how play bouts ended. Playful wrestling always ended in an apparently amicable fashion, and was never seen to escalate into “serious” fighting. Although most play bouts were terminated by physical separation of the interactants, this was not by the animal underneath fleeing; in at least half of the observed bouts, play was ended by the animal on top simply moving away from his partner. There were cases in which the recipient of the play ran away (often into the nest, and into the pile of resting pups, which was apparently an effective way of avoiding a pursuer), but more often than not play was simply “switched off ’: it became apparent that one or both of the pups had simply “lost interest” in the encounter, becoming unreceptive to attempts by the partner to continue the playful interaction. Moving away-in many cases amounting to little more than one of the players turning his head away from the partner-effectively acted as a “play marker” (Chalmers, 1980) that the interaction was finished. Other forms of bout ending were for one of the players to transfer his attentions from playing to sniffing the surrounding environment or his playmate; and for one of the players to be distracted by another animal. Again, both of these were “amicable” forms of bout ending. The data on the proximity of the pups before and after play, and who approached whom before play occurred, also imply that play is essentially nonagonistic, whatever its superficial appearance. Most bouts involved the animals being in close proximity before and after. Just over half of all bouts involved the interactants who were near to each other before play began; and nearly 60% of bouts were followed by the players remaining in close proximity to each other. It appears that play had little effect on the proximity of pups either before or after it; it certainly did not lead to immediate separation of the interactants in the same way as would adult fighting. Similarly, if play were linked to defence of resources (such as the food hopper, water spouts or incipient nesting sites) one might expect it to have occurred more often in those sites than in other parts of the environment, and to be initiated by the animal who was already there rather than by the “intruder” (thus representing incipient defence of resources). However, this was found not to be the case. Although certain areas of the environment appeared to be preferred for play, these were not necessarily areas containing resources; and the single largest category of approach behaviour was for the initiator to approach the recipient, rather than for the initiator to lay in wait for an “intruder.” This is different from what happens in the wrestling interactions of hamsters
130
HOLE
and gerbils; in these species, juvenile wrestling occurs in particular locations (at the nest, food hopper, or water bottle) and is almost always initiated by the animal who is approached (Einon, unpublished). Hole and Einon (1984) considered that although the wrestling interactions of hamsters, gerbils and other rodents superficially resembled those of Raftus species, the formers’ behaviour should not be regarded as being “play” of the same kind as occurs in the latter. Note that this is not to say that other rodents’ wrestling interactions are not some form of “play;” merely that their “play” seems to differ in certain respects from the behaviour that has been described as “play” in rats. Pellis (1988) has shown that the juvenile interactions of Golden hamsters differ in their sites of attack from adult fighting, so that “play” and “aggression” can be clearly differentiated in this species, as it can in the rat. Pellis points out that rat and hamster play differ in the sites chosen for attack; the present data suggest that they may also differ in the context within which they take place. The present data confirm Hole and Einon’s (1984) suggestion that one objective way to substantiate the intuitively-based differentiation between these two types of behaviour may be in terms of where the bouts take place. The overall conclusion from the analyses of how bouts end, where they occur, and how the interactants come together, is that play in this species does not appear to involve practice of aggression-related behaviours, either in terms of practice of the context of adult aggression (for example, “pretend” defense of resources) or in terms of the use of covert damaging tactics against the play partner.
Duration of Play Bouts Hole (1988) found that play bout durations in the rat were randomly determined; the data presented here suggest that this apparent randomness may arise because play bouts cease for a variety of reasons. Some bouts are ended by factors intrinsic to the animals themselves while others are terminated for reasons which are extrinsic to the animals and largely attributable to chance factors (e.g., distraction by other animals). Fagen (1976, 1981) has suggested that the duration of play bouts may be dictated by the function which the play serves-for example, play may have useful exercise and physical training functions, and the length of play bouts may therefore correspond to the optimum length of time for which exercise should be performed. The data reported here, however, suggest that even if there were any patterns in bout duration which resulted directly from play serving some particular function (e.g., exercise), these patterns would in practice be very difficult to detect; bouts whose terminations are dictated by functional considerations may be associated with those terminated for reasons unrelated to any features of the play itself. In short, there are a whole variety of reasons for play to end, many of them probably having little to do with the function of the play itself. In all cases, initiators were responsible for ending more bouts than recipients (60%). The simplest explanation for this is that, since most play bouts involve little reciprocity (Hole, 1988), the animal that initiates the interaction usually remains on top for the duration of the encounter; it is easier for him to terminate the play bout because the animal underneath would have to extricate himself from the pinning before ending the bout.
PROXIMITY MEASURES OF SOCIAL PLAY 1N THE RAT
131
Age and Sex Differences in Play As in previous analyses of the features of play behaviour of these animals (Hole, 1988; Hole, submitted), no major age- or sex-related differences were found. It appears that any ontogenetic changes in play in this particular species are confined largely to the morphology, sequencing, and topography of play (Poole & Fish, 1975, 1976; Meaney & Stewart, 1981; Pellis & Pellis, 1987). However, two age-related changes in the measures described here are worthy of comment despite not being amenable to statistical analysis. With increasing age, there was a steady rise in the frequency with which play ended by the initiator sniffing the recipient (although the frequency of this behaviour compared to other types of bout ending was always small). This probably reflects the maturation of sexual behaviour in this species from 40-50 days of age onwards (Meaney & Stewart, 1981) as does the fact that sexual activities were seen to end play during the 41-50 day age period. Sexual maturation results in a period of a few days when male rats are highly active sexually but also very indiscriminate, mounting males and unreceptive females as well as receptive females (Hole, personal observations). The fact that this flurry of sexual activity is explicitly reflected in the endings of only 2% of play bouts is evidence of the strong dissociation between play and any “serious” social activities, due to play occurring only in a “relaxed field” (Bally, 1945). Finally, between 51-60 days, the pattern of approach behaviour appears to have changed; initiators of bouts were more likely to approach recipients rather than playing with an animal who was already in close proximity. The explanation for this may be that by this age, rats are rapidly-becoming less playful, so that a player has to range more widely around his environment in order to find a partner receptive to play. (Note that, in contrast, there were no apparent ontogenetic changes in the frequency with which players remained close to each other after play).
Survival Costs of Play Young rats were found to show a very marked avoidance of playing in the center of the cage; that is, play occurs close to physical cover and away from exposed areas. Players are also very sensitive to external disturbance and to the introduction of novel objects into their environment; either of these factors will suppress play, at least temporarily (Hole, unpublished observations). If the same features are true of wild animals, then these behaviours presumably minimize the costs of playing by reducing players’ exposure to predators and ensuring that play only occurs in safe conditions. Claims that play has important benefits to young animals have often relied heavily on the assumption that play must have significant advantages in order to compensate for the high survival costs of playing-energy expenditure, risk of injury, and risk of predation due to increased conspicuity (Fagen, 1981). However, this argument has not been supported by detailed examinations of the actual costs of playing as opposed to the potential costs. First, the energetic costs of playing have been shown to be overestimated (Martin, 1984; Miiller-Schwarze, Stagge, & Muller-Schwarze, 1982). Second, various anecdotal and experimental observa-
132
HOLE
tions suggest that the risk of injury to players has been overestimated. For example, rhesus macaques play only in the lower branches of trees, thus minimizing injury if they fall (Symons, 1978);and in a study using separated groups of adults and 1yr-old juvenile monkeys caged outdoors, adult females were found to become more vigilant during periods of spontaneous play among juveniles (Biben, Symmes, & Bernhards, 1989). The present data on play suggest that the supposed risk to players of predation might also have been overstated. A further implication of these results is that it may well be misguided to look at the structure of play as though it were solely dictated by the function or functions that play is considered to fulfil. Much of play might consist of “secondary” adaptations designed to minimize the costs of performing the behaviour, rather than representing “primary” adaptations towards the “main” function of the play itself (whatever that might be).
Conclusion The analyses presented suggest two main conclusions. First, the data on how play bouts end and on the spatial relationships between players lend quantitative support to the intuitive impression that play in this species is essentially an amicable activity, or at least one which is unrelated in any important way to the adult aggressive behaviours which it superficially resembles. They also provide a preliminary demonstration of a technique that has the potential to provide an objective means of differentiating social play from other similar but nonplayful contact behaviours. Second, the data demonstrating that play occurs close to cover suggest that, instead of emphasizing the potential costs of play, one should instead look at ways in which these costs are minimized in the actual performance of the behaviour concerned. The distractibility of playing animals, the diurnal patterning of play bouts and their location within the environment may all be at least partially interpretable as adaptations designed to reduce the costs of playing. This has a number of implications: Many of the observed features of play (including the morphology of its constituent behaviour patterns) may reflect the operation of a number of factors. It may be too simplistic to regard the apperance of play as solely being the reflection of adaptations designed to benefit whatever function it is that play provides. Another consequence, in conjunction with other evidence which suggests that the costs of play have been overemphasized, is that the major argument for supposing that play has important functions (i.e., that play must have benefits to compensate for its high costs) is empirically unsubstantiated. There is no reason why one should not continue to take the stance that play does have functions; but it must be appreciated that this is a working hypothesis for which there is no objective evidence at present. It is, of course, very unwise to extrapolate from captive to wild animals, even of the same species, and anything other than highly tentative considerations of the survival costs of play must await studies performed in more natural habitats. However, these data suggest that systematic and quantitative investigation of aspects of play other than the behaviour patterns concerned may be useful in throwing some new light on this enigmatic behaviour.
PROXIMITY MEASURES OF SOCIAL PLAY IN THE RAT
133
Notes I would like to thank Dr. H. C. Plotkin and my anonymous reviewers for their helpful comments on a previous version of this manuscript.
References Aldis, 0. (1975). Play $fighting. New York: Academic Press. Bally, G. (1945). Vom ursprung and von den grenzen der freiheit, eine deutung des spieles bei tier und mensch. Basel: Schwabe. Bekoff, M., & Byers, J. A. (1981). A critical reanalysis of the ontogeny and phylogeny of mammalian social and locomotor play: An ethological hornet’s nest. In K. Immelmann, G. W. Barlow, L. Petrinovich, & M. Main (Eds.), Behuuiorul development: The Bielefeld interdisciplinary project Cambridge: Cambridge University Press, pp. 296-337. Biben, M., Symmes, D., & Bernhards, D. (1989). Vigilance during play in squirrel monkeys. American Journal of Primatology, 17, 41-49. Byers, J. A. (1977). Terrain preferences in the play behavior of Siberian ibex kids (Capra ibex sibirica). Zeitschrift fur Tierpsychologie, 45, 199-209. Chalmers, N. R. (1980). The ontogeny of play in feral olive baboons (Papio anubis). AnimalBehaviour, 28, 570-585. Fagen, R. (1976). Exercise, play and physical training in animals. In P. P. G. Bateson & P. H. Klopfer (Eds.), Perspectives in ethology, Vol. 2 New York: Plenum Press, pp. 189-219. Fagen, R. (1981). Animal play behavior. London: Oxford University Press. Hole, G. (1988). Temporal features of social play in the laboratory rat. Ethology, 78, 1-20. Hole, G. J., & Einon, D. F. (1984). Play in rodents, In P. K. Smith (Ed.), Play in animals and humans (pp. 95-1 17). Oxford: Basil Blackwell. Martin, P. (1984). The (four) whys and wherefores of play in cats: A review of functional, evolutionary, developmental and causal issues. In P. K. Smith (Ed.), Play in animals and humans (pp. 71-94). Oxford: Basil Blackwell. Martin, P., & Caro, T. M. (1985). On the functions of play and its role in behavioral development. Advances in the Study of Behavior, 15, 59-103. Meaney, M. J., & Stewart, J. (1981). A descriptive study of social behaviour in the rat Rattus norvegicus. Animal Behaviour 29, 34-45. Miiller-Schwarze, D., Stagge, B., & Miiller-Schwarze. C. (1982). Play behavior: Persistence, decrease and energetic compensations during food shortage in deer fawns. Science, 215, 85-87. Pellis, S . M. (1988). Agonistic versus amicable targets of attack and defense: Consequences for the origin, function, and descriptive classification of play-fighting. Aggressive Behavior 14, 85-104. Pellis, S. M., & Pellis, V. C. (1987). Play-fighting differs from serious fighting in both target of attack and tactics of fighting in the laboratory rat Ratfus norvegicus. Aggressive Behavior, 13, 227-242. Pellis, S . M., & Pellis, V. C. (1988a). Identification of the possible origin of the body target that differentiates play fighting from serious fighting in Syrian golden hamsters (Mesocricetus auratus). Aggressive Behavior, 14, 437-449. Pellis, S . M., & Pellis, V. C. (1988b). Play-fighting in the Syrian golden hamster Mesocricetus auratus waterhouse, and its relationship to serious fighting during post-weaning development. Deuelopmental Psychobiology, 21, 323-337. Poole, T. B., & Fish, J. (1975). An investigation of playful behaviour in Rattus norvegicus and Mus musculus (Mammalia). Journal of the Zoological Society of London, 175, 61-71. Poole, T. B., & Fish, J. (1976). An investigation of individual, age and sexual differences in the play of Rattus norvegicus (Mammalia: Rodentia). Journal of the Zoological Society of London 179, 249-260. Smith, P. K. (1982). Does play matter? Functional and evolutionary aspects of animal and human play. Behavioral & Brain Sciences, 5 , 139-184. Smith, P. K. (1984). Play in animals and humans. Oxford: Basil Blackwell. Symons, D. (1978). Play and aggression: A study ofrhesus monkeys. New York: Columbia University Press.
Litters of laboratory rats (Rat(u.7 noruegic.rt.7) were filmed between 21 to 60 days of age, and data on 700 social play bouts obtained. These were analyzed for information on how play bouts ended and on the spatial relationship between the players and their physical environment. The results suggest that social play in this species (a) does not appear to be a covert means of securing priority of access to resources such as food, females, or territory; and (b) does not appear to represent practice for behaving in the contexts within which "serious" adult aggression occurs (e.g., practicing being "territorial ," etc. ). Players showed a strong preference for playing close to cover and away from exposed areas; if this is also true for wild rats, it would minimize the potential costs of play in terms of increased conspicuity to predators.
Introduction To date, three themes have dominated research on animal play: the problem of defining play satisfactorily, study of the morphology of the behaviours concerned, and speculation on the functions of play (reviews in Aldis, 1975; Symons, 1978; Fagen, 1981; Bekoff & Byers, 1981; Smith, 1982, 1984; Pellis, 1988). While these are important and interesting issues, our understanding of play would be augmented by research on some of its less well-studied aspects, for example its immediate causation and the behavioural and environmental context within which it takes place. This article looks at a feature of play which, with some exceptions (e.g., Byers, 1977), has received scant attention: the location of players, both in relation to each
Reprint requests should be sent to Graham Hole, School of Social Sciences, Arts Building, University of Sussex, Falmer, Brighton, BNI 9QN, England. Received for publication 23 December 1988 Revised for publication 4 September 1990 Accepted at Wiley 9 November 1990 Developmental Psychobiology 24(2): 117-133 (1991) 0 1991 by John Wiley & Sons, Inc.
CCC 001 2-1630/91/020117-17$O4.00
118
HOLE
other and to their physical environment. Specifically, three aspects of social play will be examined using observational data from 700 social play bouts between laboratory rats aged 21-60 days. First, an examination will be made of the spatial relationships between players before and after play. Does play occur between rats who are already in close proximity or between animals who are initially at some distance from each other? Does play lead to the separation of animals who were previously close to each other? The second issue is who approaches whom in play. Does the initiator of a play bout tend to approach his partner before the play occurs or does he initiate play with an animal who approaches him? Third, the location of play bouts with respect to the players’ physical environment will be investigated. Where does play occur within the animals’ environment? Studying attributes of play may provide information relevant to the problems of defining play and identifying its developmental significance. Data on these three features may throw some light on a hypothesis which has received comparatively little attention in play research: that social play is merely immature, nondamaging aggression (see Fagen, 1981). Paradoxically, while the similarities between social play and adult aggression have repeatedly been noted (e.g., Aldis, 1975; Symons, 1978; Smith, 1982), this has generally been in order to stress the dissimilarities between the two behaviours. It has been assumed that, since play lacks t h e immediate biological goal of “serious” aggression-access to some resource-play is not in itself aggression but may represent direct practice for it. When claims are made that play is “practice” for adult aggressive behaviour, researchers presume that the behaviour patterns of play somehow represent practice of those patterns employed in adult agonistic interactions. There is already good evidence that the behaviour patterns of play cannot represent practice of adult aggressive patterns in any obvious way, at least in rats and hamsters (Pellis, 1988; Pellis & Pellis, 1987, 1988a, b). Pellis and Pellis base their arguments on a consideration of the morphology of the behaviour patterns in play and aggression, and on the fact that the body sites for attack and defense during social play in these species differ from those sites involved in aggressive encounters. Play might, however, represent practice in another, rather different, sense to that normally considered; it might represent a covert means of securing priority of access to resources, such as food, females, or territory. Such preemption of resources would be difficult to recognize, and yet might be a significant factor in adult reproductive success. A second possibility is that play might represent practice in “being territorial,” in “attacking an intruder,” or “defending a resource.” Play might represent practice not of the behaviour patterns involved in “serious” aggression, but rather of behaving in the contexts within which the “serious” adult behaviour patterns occur-in effect, learning that a particular kind of behaviour is the appropriate thing to do in a particular circumstance. Data on the proximity of players can provide some information on this point as follows. If play were a means of securing access to resources, then one might expect the recipient to approach the initiator (so that play marks the defense of a resource by the holding animal). An analysis of where play takes place within the environment may also be relevant here, determining for example, whether play occurs in the vicinity of material resources, such as food, water, or nest sites.
PROXIMITY MEASURES OF SOCIAL PLAY IN THE RAT
119
Information on where play occurs may also provide some insights into the actual survival costs of play-that is, the costs which play has for an animal in practice as opposed to the costs which play has commonly been assumed to possess (Fagen, 1981; Martin & Caro, 1985). For example, it is often claimed that play may increase the conspicuity of players to predators. However, in reality this may not be the ,case. Informal observations of rat play suggest that play occurs around the edges of a cage where the animals are close to protective cover. The present study will investigate this suggestion quantitatively.
Methods Three separate litters were filmed on video. (Some features of these subjects’ play behaviour have already been described in a previous paper, Hole, 1988). The 3 i hr period from 30 min before dusk to 3 hr after dusk was filmed each day, between 21 and 60 days of age. (Previous observations suggested that rats are particularly likely to engage in social play around dawn and dusk). Wrestling bouts were subsequently selected for analysis if they satisfied the criteria outlined below. (Of bouts meeting these criteria, however, only a randomly chosen subset were actually analyzed). This resulted in a total of 700 play bouts being analyzed: 246 for Litter One (LI); 272 for Litter Two (L2); and 182 for Litter Three (L3).
Operational Definition of “Rough-and-Tumble’’ Play A wrestling bout was defined as any interaction occurring between two subjects which began with a pounce, and in which one of the two was held down on its back by the other. (Thus “unsuccessful” play solicitations, in which one animal failed to solicit social play from another, are not considered in this study). The beginning of a bout was considered to be the initial pounce. A bout was considered to have ended when the two animals ceased to maintain ventral-ventral contact for more than three sec. The only interactions selected for analysis were those which consisted principally of wrestling, although some elements of chasing and rearing and/or “boxing” were included by virtue of the definition employed. Only dyadic interactions were analyzed and of these, only those were selected in which the two animals were in complete view throughout the play bout.
Subjects PVG hooded rats were used. Each litter was run independently: Each pregnant mother was placed in the observation cage described below so that her litter was wholly raised under the conditions in which it would be observed. Each litter was culled to six pups (three male, three female) and the mother removed at 21 days, after which the pups were left undisturbed in the test environment until observations ended at 60 days. Subjects were identified on the basis of their natural coat markings.
120
HOLE
Apparatus Subjects were housed in a large purpose-built observation cage, measuring 2 m x 2 m x 50 cm. All four walls and the roof of the cage were made of Plexiglass, except that the upper half of the rear wall was constructed of wire mesh to provide ventilation. Bedding consisted of garden topsoil distributed to an initial depth of approximately 10 cm. Food (SDS rat and mouse maintenance diet number 1 in expanded pellet form) and water were available ad lib from dispensers placed at the right and left corners of the cage, respectively. Lighting was on a 12 h light/dark cycle, with the lights dimmed or brightened over a 2-hr transition period between each phase of the cycle so as to simulate “dawn” and “dusk.” Illumination was provided by six 100-watt tungsten lamps during the light phase, and by two 40-watt red lamps during the dark phase. Filming was carried out using a National Panasonic time-lapse video recorder, (on its continuous rather than time-lapse setting) and a Sony video camera placed at the front of the cage. A zoom lens provided a clear view of animals situated toward the back of the cage. Marks on the walls and floor of the cage enabled fairly accurate judgement of the animals’ distance from the cage walls and from each other. Variables
Initiator or Recipient of Play The identity and sex of the participants in each wrestling bout were recorded. The animal which performed the initial pounce, defined as commencing the interaction, was designated the initiator of the bout (I). The animal who was pounced upon was considered to be the recipient of the wrestling (R).
Bout Endings A record was made of the nature of the behaviour by either of the participants which appeared to terminate the interaction (i.e., behaviour which occurred immediately before the players separated). On the basis of preliminary observations, a number of categories were set up which accounted for the majority of bout endings: Either of the interactants moving away. This was by running or walking away from the partner. Distraction of either interactant. This category included either animal breaking off the current interaction in order to participate in a fresh one with a different animal. For example, a common ending was for the animal on top to break off from his wrestling activities and chase after a different animal in order to play with him. Olfactory investigation of environment. Either animal might suddenly “switch off” from wrestling and engage in prolonged sniffing of the immediate surroundings. Olfactory investigation of other interactant. Either animal might cease wrestling and engage in prolonged sniffing of the wrestling partner. Self-grooming. Either animal might break off to groom himself.
PROXIMITY MEASURES OF SOCIAL PLAY IN THE RAT
121
Sexual behaviours. Sexual activity directed by one of the participants towards the other effectively ended the wrestling interaction. Other. Any behaviours which ended the bout but which did not fall into any of the above categories. For each category, a note was made of which animal was responsible for terminating the bout.
Proximity of Interactants For each play bout, a note was made of the proximity of players to each other in the three sec before and after the bout. If the animals’ bodies were within 20 cm of each other, they were scored as being in “close proximity”; if they were further apart, “no proximity” was recorded.
Who Approached Whom Before Play Commenced This was classified according to whether the initiator of the bout approached the recipient; the recipient approached the initiator; both were already adjacent to each other; or both independently moved towards each other.
Location Within Environment A record was made of the location of each play bout within the cage and the proximity of the interactants to the nearest wall or cover. The cage floor was divided into nine areas of equal size: “far left,” “far middle,” and “far right;” “middle left,” “middle,” and “middle right;” “near left” (which contained the water spouts), “near middle,” and “near right” (which contained the food hopper). The location of each play bout within this grid was recorded. Proximity to the wall was scored as within 10 cm (“close”); between 10-20 cm (“intermediate”); or over 20 cm (“no proximity”). Wrestling interactions within a single bout could often occur at a number of different locations. Therefore, proximity was taken as the distance from cover at which most of the wrestling occurred, ignoring any chases which might take place within a bout. Although chases could result in the interactants moving away from cover for very brief periods, the important factor was where the wrestling itself occurred.
Results Play Bout Endings
Overall Frequency of Bout Endings The “21-60 days” column in Table 1 shows the relative frequencies with which various categories of bout ending were observed to occur, combining the data for both sexes and all age periods. The most common way in which a bout ended (43%) was by one of the interactants moving away from the other. An initiator was as likely to move away from the recipient as was the recipient from the initiator; 23% of bouts were terminated by the initiator moving away, and 20% by the departure of the recipient.
122
HOLE
Table 1 Age Changes in How Play Ended, (“I” = Initiator, “R” = Recipient). Category of bout ending 1 moves away
R moves away I is distracted R is distracted
I sniffs ground R sniffs ground I self-grooms
21-30
3 1-40
21.0% (37) 22.2% (39) 13.1% (23) 7.4% (13) 17.6% (31) 9.7% (17) 0.6% (1)
R self-grooms I sniffs R
R sniffs I
2.3% (4) 3.4% (6) 0.6% (1)
sexual behaviour other
0.0% (0) 2.3% (4)
Age period (days) 41-50
51-60
2 1-60
20.2% (21) 20.2% (21) 12.5% (13) 3.8% (4) 16.4% (17) 4.8% (5) 8.6% (9) 1 .O% (1) 6.7% (7) 1.9% (2) 0.0% (0) 3.8% (4)
* Percent of the total number of play bouts observed in each age period, combining the data for males and females. The second most frequent way in which play ended (26%, 17% terminated by initiator, 8% by recipients) was by one of the players turning from play to olfactory investigation of the surrounding inanimate environment (cage walls or floor). Usually, the initiator stepped off his play partner and began to sniff the ground nearby (or the wail if play was occurring immediately next to it), apparently disregarding totally the animal with which he had just played, becoming unreceptive to any attempts by that animal to renew play. A recipient either ended bouts in the same way (if he was on top at the time) or if he was lying on his back pinned down by the initiator, twisted around to sniff the ground, with the initiator apparently offering no resistance. The third most frequent way in which play ended (18%) was for either the initiator or recipient to be distracted from playing with their present partner. Distraction occurred either by a third animal breaking into the play bout, so that the initiative for terminating the encounter in a sense came from this other pup, or by the player himself breaking off play with his present partner in order to sniff, chase, follow, play, or otherwise interact with a different animal nearby. Small proportions of bouts (under 5% in all cases) were observed to be terminated by self-grooming or by sniffing the play partner. In both cases, initiators performed these behaviours more often than recipients. In the case of self-groom-
PROXIMITY MEASURES OF SOCIAL PLAY IN THE RAT
123
ing, it appeared that the initiator, lying over his partner and in the process of grooming him, transferred his efforts towards grooming himself, allowing the recipient to right himself unimpeded. The termination of play by one of the players in order to sniff the play partner was in practice quite different from the kind of olfactory investigation performed within the play bout itself since it was accompanied by the sniffing animal moving away from ventral-ventral contact with his partner and generally failing to respond to any attempts by the other animal to maintain a playful type of interaction.
Age Differences .in Frequency of Bout Endings Table 1 shows how often the different categories of bout endings occurred in each age period, combining the data for males and females. With some fluctuations, the overall pattern described above was shown in all four age periods-that is, the most common endings were first, separation of the players; second, breaking off from play to sniff the environment; and third, distraction by a different animal. This analysis does, however, reveal two age changes in bout endings which may possibly be related. A steady increase with age was seen in the frequency with which play ended by the initiator sniffing the recipient; and sexual activities were seen to terminate play only during the 41-50 day age period (and then only in less than 2% of bouts).
Overall Sex-Differences in Frequency of Bout Endings Table 2 shows the relative frequencies of the different categories of bout endings, analyzed according to the sexes of the animals involved-MI/MR, FI/FR, MI/FR, FI/MR. With minor variations, the overall pattern of bout endings described previously was shown by all classes of interactants. Further analysis also revealed no age-related sex differences in how play bouts were terminated.
Proximity of Interactants Before and After Play
Overall Proximity of Players Table 3 shows the overall frequencies with which players (21-60 days) were observed to be in close proximity or not, both before and after bouts, combining the data for all ages and both sexes. Overall, more play bouts (56%) occurred between individuals who were already in close proximity than between individuals who were not (46%).Similarly, more play bouts (58%) were followed by the interactants remaining close to each other than by separation (42%). When proximity before and after each bout is taken into account, four possible permutations were: proximity before and after; no proximity before or after; proximity before but not after; and no proximity before but proximity after. Table 3 shows that the most frequent of these was proximity both before and after a bout (38%).
HOLE
124
Table 2 Overall Sex Differences in How Play Ended (21-60 days) Category of bout ending
MI-MR
Sexes of interactants FI-FR MI-FR
I moves away
R moves away I is distracted
R is distracted I sniffs ground R sniffs ground
I self-grooms R self-grooms
I sniffs R R sniffs I sexual behaviour other
FI-MR 24.4% (40) 17.1% (28) 12.8% (21) 3. I% (5) 20.1% (33) 10.4% (17) 1.8% (3) 4.3% (7) I .8%
(3) I .2% (2) 0.6% (1) 2.4% (4)
* (Percent of the total number of play bouts involving sexes of interactants concerned, combining data from all age periods. “M” = male, “F” = female, other codes as in Table 1).
Age Changes in Proximity of Players Table 3 shows the data on the proximity ofplayers before and after play, broken down by age but combining the data for both sexes of interactant. Examination of Table 3 suggests that in the first three age periods, the overall pattern described in the previous section was shown, with approximately 60% of play bouts occurring between pups who were already close together. In the final age period, however, there appears to be a reversal of this pattern with 60% of bouts occurring between animals who were not close to each other. (A more detailed analysis for age-related sex differences revealed that this reversal in the final age period was shown by all four permutations of interactant, but most markedly so in bouts initiated by females with males, with 72% of bouts [I3 out of 181 occurring between players who were not already close to each other). The frequencies with which players remained close together after play had ended were very consistent across age periods; in all cases, approximately 60% of play bouts were followed by the participants remaining in close proximity.
Overall Sex Differences in Proximity of Players Table 4 shows the relative frequencies of the proximity measures described above, analyzed according to the participants’ sexes but summing the data over
PROXIMITY MEASURES OF SOCIAL PLAY IN THE RAT
125
Table 3 Overall Age Changes in Frequency of Proximity between Players Before and After Play (Males and Females Combined). Proximity, before and after play
age period 41-50
2 1-30
31-40
P before, P after
36.0% (54)
42.6% (75)
NP before, NP after
22.7% (34)
25.0% (44)
P before, NP after
21.3% (32)
13.6% (24)
NP before, P after
20.0% (30)
18.8%
(33)
17.4% (47)
26.9% (28)
100.0% (150)
100.0% ( 176)
100.0% (270)
100.0% (104)
100.0% (700)
Total P before
57.3% (86)
56.3% (99)
40.4% (42)
56.1% (393)
Total NP before
42.7% (64)
43.7% (77)
59.6% (62)
43.9% (307)
Total P after
56.0% (84)
57.4% (101)
58.6% (61)
Total NP after
44.0% (66)
42.6% (75)
35.4% (43)
Total
21.1% (57)
5 1-60
21-60
32.7% (34)
~~~
* (Percent of the total number of bouts observed in each age period. “P” “NP”
=
= close proximity;
no proximity)
all age periods. In play bouts involving female recipients (i.e., FI/FR and MI/FR play), the interactants were as likely to be apart from each other before or after play as they were to be together. In contrast, when play involved a male recipient, the players were more likely to be in close proximity before and after play than they were to be apart (60%). In other words, male recipients were more likely to be involved in play with an animal near them, and were more likely to remain close to that animal once the play had ended. In all cases, the most commonly observed permutation of proximity scores was proximity both before and after play. However, in play involving a female recipient, the “no proximity before or after” category occurred almost equally as frequently.
Who Approached Whom Before Play Commenced Overall Frequency of Categories of Approach Behaviour Figure 1 gives the overall frequencies with which each of the possible categories of approach behaviour was observed to occur, combining the data for both
126
HOLE
Table 4 Overall Sex Differences in Frequency of Proximity between Players Before and After Play (21-60 Days). MI-MR
P before, P after
41.9% (75) 21.2% (38) 17.9% (32) 19.0% (34) 100.0% ( 179)
31.1% (37) 16.0% (19) 19.3% (23) 100.0% (1 19)
MI-MR
FI-FR
NP before, NP after P before, NP after NP before, P after Total:
Total P before Total NP before Total P after Total NP after
59.8% (107) 40.2% (72) 60.9% (109) 39.1% (70)
FI-FR
Sex of Players MI-FR
Proximity:
FI-MR
33.6% (40)
MI-FR
49.6% (59) 50.4% (60) 52.9% (63) 47.1% (56)
FI-MR 59.8% (98) 40.2% (66) 64.6% (106) 35.4% (58)
* (Percent of the total number of bouts performed by each permutation of players’ sexes)
sexes and all age periods. Nearly half (46%) of all bouts were preceded by I approaching R. The next most common category, accounting for 30% of all bouts, was for both pups to be close to each other already. 18% of bouts occurred after R approached 1. It was relatively uncommon for both partners to approach each other, only 5% of all bouts resulting from this type of encounter. Age-Differences in the Frequency of Categories of Approach Behaviour
Figure 1 shows the relative frequencies of the different categories of approach behaviour in each of the four age periods, combining the data for both sexes. There is some indication of an age-related trend, in that with increasing age, play was more frequently preceded by the initiator approaching the recipient. There was a corresponding decrease with age in the frequency with which play occurred between two animals already in close proximity.
Sex Differences in Approach Behaviouv No sex differences were found in approach behaviour; the pattern described above was shown by both sexes at all ages.
PROXIMITY MEASURES OF SOCIAL PLAY IN THE RAT TYPE OF APPROACH BEHAVIOUR
I to R R to I mutual
( A ) 21-30 DAYS:
6.P ..
ltoR Rtol
( 8 ) 31-40 DAYS
(C) 41-50 DAYS:
b
77 28
8 64
-
mutual B.P. ItoR 1
(9) 51-60 DAYS: r!%aIB.P.
I to R
21-60 DAYS:
,",ttal B.P.
(O'JEFtALL)
127
~
17
~~~
62 28
e
18
308
F 1
123
36 204
0 I0 20 30 40 50 60 K TOTAL NO.BOUTS PER AGE-PERIOD
Fig. 1 . Relative frequencies of different types of approach behaviour (data for males and females combined).
Location of Play Within Environment Proximity of Players to Cover During Play Figure 2 shows the overall frequency with which play was observed to occur in each of the three proximity zones used, combining the data from a11 ages and 80
w
TOTAL NO.BOUTS
7
495
60 I I I
40
119
20
OL c10 cm
10-20 crn
,20 cm
PROXIMITY TO CAGE WALL Fig. 2. Relative frequency with which play occurred in proximity to the cage wall.
128
HOLE Z TOTAL NOBOUTS 20
16
10
5
0
FL
FM
FR
ML
MIDDLE
MR
HVITER
NM
FOOD
LOCATION OF PLAY-BOUTS Fig. 3.
Relative frequency of play in different locations within the cage
both sexes. These observed data are shown compared to the frequencies which would be expected if social play was located at random within the environment. In calculating the expected frequencies, account had to be taken of the different areas encompassed by each of the zones: 64% of the cage floor area was contained within the inner zone (i.e., over 20 cm from the wall), 17% by the middle zone, and 19% by the strip nearest the walls, so one might expect the animals to be observed mainly within the central area. In contrast, animals exhibited a very clear preference for playing within 10 cm of the nearest wall. These data underestimate the extent of the preference for cover during play, since many of the bouts scored under “no proximity” were located close to large bundles of nest material or within deep depressions excavated by the animals (not necessarily those that were playing) in the inner regions of the environment. The number of bouts occurring away from any cover was very small, suggesting that rats strongly prefer not to play in exposed areas-that is, that during play, they maintain the marked thigmotaxis that they display during more “serious” activities. No changes in this pattern were found in relation to age or sex of players. Clearly, increased body size of the subjects (so that, relative to the animals, the environment becomes smaller) and increased familiarity with the environment with increasing age did nothing to change the animals’ preference for cover during play.
Locution of Play Bouts in Environment Figure 3 shows the observed frequency of play in each region of the cage. Data were collected only for 95 bouts from L1 (between days 41 and 60), and 92 bouts from L2 (between days 21-30 and 41-60), so a full age and sex analysis was not possible. Fig. 3 suggests that although play was not distributed randomly around the perimeter of the cage, neither was it especially associated with areas of the environment which contained resources such as food and water; play appears to have occurred where the animals were most active.
PROXIMITY MEASURES OF SOCIAL PLAY IN THE RAT
129
Discussion Relationship of Play t o Aggression Various data reported here suggest that social play in this species occurs in a very different context to that of either juvenile or adult aggression, and that it has different immediate consequences, despite the superficially similar appearance of play to aggressive behaviours. The same conclusion has been drawn on the grounds of the morphology of the behaviour patterns in play and aggression (Hole & Einon, 1984), and on the basis of a consideration of the differing body sites of attack and defence during play and aggression (Pellis & Pellis, 1987, 1988a,b; review in Pellis, 1988). In the present study, the most direct evidence of a lack of association between play and agonistic activities comes from the data on how play bouts ended. Playful wrestling always ended in an apparently amicable fashion, and was never seen to escalate into “serious” fighting. Although most play bouts were terminated by physical separation of the interactants, this was not by the animal underneath fleeing; in at least half of the observed bouts, play was ended by the animal on top simply moving away from his partner. There were cases in which the recipient of the play ran away (often into the nest, and into the pile of resting pups, which was apparently an effective way of avoiding a pursuer), but more often than not play was simply “switched off ’: it became apparent that one or both of the pups had simply “lost interest” in the encounter, becoming unreceptive to attempts by the partner to continue the playful interaction. Moving away-in many cases amounting to little more than one of the players turning his head away from the partner-effectively acted as a “play marker” (Chalmers, 1980) that the interaction was finished. Other forms of bout ending were for one of the players to transfer his attentions from playing to sniffing the surrounding environment or his playmate; and for one of the players to be distracted by another animal. Again, both of these were “amicable” forms of bout ending. The data on the proximity of the pups before and after play, and who approached whom before play occurred, also imply that play is essentially nonagonistic, whatever its superficial appearance. Most bouts involved the animals being in close proximity before and after. Just over half of all bouts involved the interactants who were near to each other before play began; and nearly 60% of bouts were followed by the players remaining in close proximity to each other. It appears that play had little effect on the proximity of pups either before or after it; it certainly did not lead to immediate separation of the interactants in the same way as would adult fighting. Similarly, if play were linked to defence of resources (such as the food hopper, water spouts or incipient nesting sites) one might expect it to have occurred more often in those sites than in other parts of the environment, and to be initiated by the animal who was already there rather than by the “intruder” (thus representing incipient defence of resources). However, this was found not to be the case. Although certain areas of the environment appeared to be preferred for play, these were not necessarily areas containing resources; and the single largest category of approach behaviour was for the initiator to approach the recipient, rather than for the initiator to lay in wait for an “intruder.” This is different from what happens in the wrestling interactions of hamsters
130
HOLE
and gerbils; in these species, juvenile wrestling occurs in particular locations (at the nest, food hopper, or water bottle) and is almost always initiated by the animal who is approached (Einon, unpublished). Hole and Einon (1984) considered that although the wrestling interactions of hamsters, gerbils and other rodents superficially resembled those of Raftus species, the formers’ behaviour should not be regarded as being “play” of the same kind as occurs in the latter. Note that this is not to say that other rodents’ wrestling interactions are not some form of “play;” merely that their “play” seems to differ in certain respects from the behaviour that has been described as “play” in rats. Pellis (1988) has shown that the juvenile interactions of Golden hamsters differ in their sites of attack from adult fighting, so that “play” and “aggression” can be clearly differentiated in this species, as it can in the rat. Pellis points out that rat and hamster play differ in the sites chosen for attack; the present data suggest that they may also differ in the context within which they take place. The present data confirm Hole and Einon’s (1984) suggestion that one objective way to substantiate the intuitively-based differentiation between these two types of behaviour may be in terms of where the bouts take place. The overall conclusion from the analyses of how bouts end, where they occur, and how the interactants come together, is that play in this species does not appear to involve practice of aggression-related behaviours, either in terms of practice of the context of adult aggression (for example, “pretend” defense of resources) or in terms of the use of covert damaging tactics against the play partner.
Duration of Play Bouts Hole (1988) found that play bout durations in the rat were randomly determined; the data presented here suggest that this apparent randomness may arise because play bouts cease for a variety of reasons. Some bouts are ended by factors intrinsic to the animals themselves while others are terminated for reasons which are extrinsic to the animals and largely attributable to chance factors (e.g., distraction by other animals). Fagen (1976, 1981) has suggested that the duration of play bouts may be dictated by the function which the play serves-for example, play may have useful exercise and physical training functions, and the length of play bouts may therefore correspond to the optimum length of time for which exercise should be performed. The data reported here, however, suggest that even if there were any patterns in bout duration which resulted directly from play serving some particular function (e.g., exercise), these patterns would in practice be very difficult to detect; bouts whose terminations are dictated by functional considerations may be associated with those terminated for reasons unrelated to any features of the play itself. In short, there are a whole variety of reasons for play to end, many of them probably having little to do with the function of the play itself. In all cases, initiators were responsible for ending more bouts than recipients (60%). The simplest explanation for this is that, since most play bouts involve little reciprocity (Hole, 1988), the animal that initiates the interaction usually remains on top for the duration of the encounter; it is easier for him to terminate the play bout because the animal underneath would have to extricate himself from the pinning before ending the bout.
PROXIMITY MEASURES OF SOCIAL PLAY 1N THE RAT
131
Age and Sex Differences in Play As in previous analyses of the features of play behaviour of these animals (Hole, 1988; Hole, submitted), no major age- or sex-related differences were found. It appears that any ontogenetic changes in play in this particular species are confined largely to the morphology, sequencing, and topography of play (Poole & Fish, 1975, 1976; Meaney & Stewart, 1981; Pellis & Pellis, 1987). However, two age-related changes in the measures described here are worthy of comment despite not being amenable to statistical analysis. With increasing age, there was a steady rise in the frequency with which play ended by the initiator sniffing the recipient (although the frequency of this behaviour compared to other types of bout ending was always small). This probably reflects the maturation of sexual behaviour in this species from 40-50 days of age onwards (Meaney & Stewart, 1981) as does the fact that sexual activities were seen to end play during the 41-50 day age period. Sexual maturation results in a period of a few days when male rats are highly active sexually but also very indiscriminate, mounting males and unreceptive females as well as receptive females (Hole, personal observations). The fact that this flurry of sexual activity is explicitly reflected in the endings of only 2% of play bouts is evidence of the strong dissociation between play and any “serious” social activities, due to play occurring only in a “relaxed field” (Bally, 1945). Finally, between 51-60 days, the pattern of approach behaviour appears to have changed; initiators of bouts were more likely to approach recipients rather than playing with an animal who was already in close proximity. The explanation for this may be that by this age, rats are rapidly-becoming less playful, so that a player has to range more widely around his environment in order to find a partner receptive to play. (Note that, in contrast, there were no apparent ontogenetic changes in the frequency with which players remained close to each other after play).
Survival Costs of Play Young rats were found to show a very marked avoidance of playing in the center of the cage; that is, play occurs close to physical cover and away from exposed areas. Players are also very sensitive to external disturbance and to the introduction of novel objects into their environment; either of these factors will suppress play, at least temporarily (Hole, unpublished observations). If the same features are true of wild animals, then these behaviours presumably minimize the costs of playing by reducing players’ exposure to predators and ensuring that play only occurs in safe conditions. Claims that play has important benefits to young animals have often relied heavily on the assumption that play must have significant advantages in order to compensate for the high survival costs of playing-energy expenditure, risk of injury, and risk of predation due to increased conspicuity (Fagen, 1981). However, this argument has not been supported by detailed examinations of the actual costs of playing as opposed to the potential costs. First, the energetic costs of playing have been shown to be overestimated (Martin, 1984; Miiller-Schwarze, Stagge, & Muller-Schwarze, 1982). Second, various anecdotal and experimental observa-
132
HOLE
tions suggest that the risk of injury to players has been overestimated. For example, rhesus macaques play only in the lower branches of trees, thus minimizing injury if they fall (Symons, 1978);and in a study using separated groups of adults and 1yr-old juvenile monkeys caged outdoors, adult females were found to become more vigilant during periods of spontaneous play among juveniles (Biben, Symmes, & Bernhards, 1989). The present data on play suggest that the supposed risk to players of predation might also have been overstated. A further implication of these results is that it may well be misguided to look at the structure of play as though it were solely dictated by the function or functions that play is considered to fulfil. Much of play might consist of “secondary” adaptations designed to minimize the costs of performing the behaviour, rather than representing “primary” adaptations towards the “main” function of the play itself (whatever that might be).
Conclusion The analyses presented suggest two main conclusions. First, the data on how play bouts end and on the spatial relationships between players lend quantitative support to the intuitive impression that play in this species is essentially an amicable activity, or at least one which is unrelated in any important way to the adult aggressive behaviours which it superficially resembles. They also provide a preliminary demonstration of a technique that has the potential to provide an objective means of differentiating social play from other similar but nonplayful contact behaviours. Second, the data demonstrating that play occurs close to cover suggest that, instead of emphasizing the potential costs of play, one should instead look at ways in which these costs are minimized in the actual performance of the behaviour concerned. The distractibility of playing animals, the diurnal patterning of play bouts and their location within the environment may all be at least partially interpretable as adaptations designed to reduce the costs of playing. This has a number of implications: Many of the observed features of play (including the morphology of its constituent behaviour patterns) may reflect the operation of a number of factors. It may be too simplistic to regard the apperance of play as solely being the reflection of adaptations designed to benefit whatever function it is that play provides. Another consequence, in conjunction with other evidence which suggests that the costs of play have been overemphasized, is that the major argument for supposing that play has important functions (i.e., that play must have benefits to compensate for its high costs) is empirically unsubstantiated. There is no reason why one should not continue to take the stance that play does have functions; but it must be appreciated that this is a working hypothesis for which there is no objective evidence at present. It is, of course, very unwise to extrapolate from captive to wild animals, even of the same species, and anything other than highly tentative considerations of the survival costs of play must await studies performed in more natural habitats. However, these data suggest that systematic and quantitative investigation of aspects of play other than the behaviour patterns concerned may be useful in throwing some new light on this enigmatic behaviour.
PROXIMITY MEASURES OF SOCIAL PLAY IN THE RAT
133
Notes I would like to thank Dr. H. C. Plotkin and my anonymous reviewers for their helpful comments on a previous version of this manuscript.
References Aldis, 0. (1975). Play $fighting. New York: Academic Press. Bally, G. (1945). Vom ursprung and von den grenzen der freiheit, eine deutung des spieles bei tier und mensch. Basel: Schwabe. Bekoff, M., & Byers, J. A. (1981). A critical reanalysis of the ontogeny and phylogeny of mammalian social and locomotor play: An ethological hornet’s nest. In K. Immelmann, G. W. Barlow, L. Petrinovich, & M. Main (Eds.), Behuuiorul development: The Bielefeld interdisciplinary project Cambridge: Cambridge University Press, pp. 296-337. Biben, M., Symmes, D., & Bernhards, D. (1989). Vigilance during play in squirrel monkeys. American Journal of Primatology, 17, 41-49. Byers, J. A. (1977). Terrain preferences in the play behavior of Siberian ibex kids (Capra ibex sibirica). Zeitschrift fur Tierpsychologie, 45, 199-209. Chalmers, N. R. (1980). The ontogeny of play in feral olive baboons (Papio anubis). AnimalBehaviour, 28, 570-585. Fagen, R. (1976). Exercise, play and physical training in animals. In P. P. G. Bateson & P. H. Klopfer (Eds.), Perspectives in ethology, Vol. 2 New York: Plenum Press, pp. 189-219. Fagen, R. (1981). Animal play behavior. London: Oxford University Press. Hole, G. (1988). Temporal features of social play in the laboratory rat. Ethology, 78, 1-20. Hole, G. J., & Einon, D. F. (1984). Play in rodents, In P. K. Smith (Ed.), Play in animals and humans (pp. 95-1 17). Oxford: Basil Blackwell. Martin, P. (1984). The (four) whys and wherefores of play in cats: A review of functional, evolutionary, developmental and causal issues. In P. K. Smith (Ed.), Play in animals and humans (pp. 71-94). Oxford: Basil Blackwell. Martin, P., & Caro, T. M. (1985). On the functions of play and its role in behavioral development. Advances in the Study of Behavior, 15, 59-103. Meaney, M. J., & Stewart, J. (1981). A descriptive study of social behaviour in the rat Rattus norvegicus. Animal Behaviour 29, 34-45. Miiller-Schwarze, D., Stagge, B., & Miiller-Schwarze. C. (1982). Play behavior: Persistence, decrease and energetic compensations during food shortage in deer fawns. Science, 215, 85-87. Pellis, S . M. (1988). Agonistic versus amicable targets of attack and defense: Consequences for the origin, function, and descriptive classification of play-fighting. Aggressive Behavior 14, 85-104. Pellis, S. M., & Pellis, V. C. (1987). Play-fighting differs from serious fighting in both target of attack and tactics of fighting in the laboratory rat Ratfus norvegicus. Aggressive Behavior, 13, 227-242. Pellis, S . M., & Pellis, V. C. (1988a). Identification of the possible origin of the body target that differentiates play fighting from serious fighting in Syrian golden hamsters (Mesocricetus auratus). Aggressive Behavior, 14, 437-449. Pellis, S . M., & Pellis, V. C. (1988b). Play-fighting in the Syrian golden hamster Mesocricetus auratus waterhouse, and its relationship to serious fighting during post-weaning development. Deuelopmental Psychobiology, 21, 323-337. Poole, T. B., & Fish, J. (1975). An investigation of playful behaviour in Rattus norvegicus and Mus musculus (Mammalia). Journal of the Zoological Society of London, 175, 61-71. Poole, T. B., & Fish, J. (1976). An investigation of individual, age and sexual differences in the play of Rattus norvegicus (Mammalia: Rodentia). Journal of the Zoological Society of London 179, 249-260. Smith, P. K. (1982). Does play matter? Functional and evolutionary aspects of animal and human play. Behavioral & Brain Sciences, 5 , 139-184. Smith, P. K. (1984). Play in animals and humans. Oxford: Basil Blackwell. Symons, D. (1978). Play and aggression: A study ofrhesus monkeys. New York: Columbia University Press.
