When does play panting occur during social play in wild chimpanzees

ORIGINAL ARTICLE

Takahisa Matsusaka

When does play panting occur during social playin wild chimpanzees?

Received: 18 September 2002 / Accepted: 6 April 2004 / Published online: 7 July 2004� Japan Monkey Centre and Springer-Verlag 2004

Abstract To clarify the social functions of play panting inchimpanzees, I investigated when they emitted playpanting in social play and how the interactions were af-fected by the occurrence of play panting. The subjectswere the M-group chimpanzees living in Mahale,Tanzania. The following observations were made: (1)chimpanzees emitted play panting when they were tickledor chased but rarely did so when they tickled or chasedothers. Chimpanzee play panting does not have thefunction of a play signal communicating that these‘‘aggressive’’ actions are performed not as aggression butas play. (2) Chimpanzees emitted play panting more oftenwhen they received ‘‘aggressive’’ actions that supposedlyelicited higher arousal. (3) A chimpanzee tended to con-tinue to perform ‘‘aggressive’’ actions when the targetemitted play panting. Play panting activates the interac-tion of social play by encouraging the performer to con-tinue tickling or chasing. These results can besummarized as showing that chimpanzee play pantingserves as positive feedback to the play partner forcontinuing somewhat fragile interactions, which maycontain the risk of excessive arousal and the risk of con-fusing ‘‘defensive’’ actions by the target of the ‘‘aggres-sive’’ actions with real efforts to escape the situation.

Keywords Chimpanzees Æ Social play Æ Play panting ÆEvolution of laughter Æ Play signals

Introduction

It has traditionally been believed that humans are theonly creatures who can laugh. However, our closest

relatives—chimpanzees, gorillas, and orangutans—alsoutter laugh-like sounds when they are tickled, as Darwin(1872) first reported. Such panting or grunting soundsemitted by playing chimpanzees are sometimes called‘‘laughter’’ or ‘‘laughing’’ by some researchers, but I usethe more neutral term ‘‘play panting’’ in this article,following Nishida et al. (1999). Similar sounds are alsoheard during the social play of bonobos (de Waal 1988)and some other primate species such as Barbary maca-ques (Kipper and Todt 2002). Van Hooff (1972)hypothesized that play panting and play face in primateswere the phylogenetic origins of human laughter. Hecompared the structure and context of facial displays ofhumans and other primates and suggested that humansmiling was homologous to ‘‘silent bared-teeth display’’(i.e., grin face), which was widely seen when a primatewas threatened or aggressed against, and that humanlaughter was homologous to ‘‘relaxed open-mouth dis-play’’ (i.e., play face, often accompanied by play pantingin some species), which occurred in primate social play.Some studies have followed van Hooff’s hypothesis andhave investigated situational variations and phylogeneticdistribution of these displays in primates (e.g., Pre-uschoft 1992; Preuschoft and van Hooff 1997). How-ever, these previous studies from evolutionaryperspectives mainly investigated the facial expressions ofgrin face and play face. Perhaps the low amplitudes ofthe vocalizations made it difficult to conduct detailedanalyses on play panting in most of the primate species(Kipper and Todt 2002). So far, very few studies havefocused on play panting in primates.

Some previous works have described the play pantingof chimpanzees, including studies of the social play ofchimpanzees in the wild (Hayaki 1985; Plooij 1979) andin captivity (Loizos 1969; Mendoza-Granados andSommer 1995; Markus and Croft 1995) and studies ofthe vocal expressions of wild chimpanzees (Van Lawick-Goodall 1968; Marler 1969; Marler and Tenaza 1977).These studies commonly reported that chimpanzees of-ten emitted play panting while they played with others,especially when they wrestled vigorously or when they

T. MatsusakaLaboratory of Human Evolution Studies,Graduate School of Science, Kyoto University,Kitashirakawa-Oiwake-cho, Sakyo-ku,Kyoto 606-8502, JapanE-mail: [email protected].: +81-75-7534085Fax: +81-75-7534115

Primates (2004) 45:221–229DOI 10.1007/s10329-004-0090-z

were being tickled. Provine (1996) compared the soundstructures of human laughter and chimpanzee playpanting. Although they had significant similarity in theirrhythmic structures, he emphasized two differences.First, chimpanzee play panting lacked the clear har-monic structure observed in human laughter. Second,while the discrete notes of human laughter (‘‘ha’’) wereproduced by chopping a single exhalation, the notes ofchimpanzee play panting were produced during eachexhalation and inhalation (Provine 1996). However,Bachorowski et al. (2001) claimed that previous workson human laughter examined only typical voicedlaughter, and they discussed two other kinds of unvoicedlaughter, a grunt-like (pant-like) laugh and a snort-likelaugh. The difference between human laughter andchimpanzee play panting that Provine (1996) empha-sized seems less valid when we think of these types ofunvoiced human laughter, although Bachorowski et al.(2001) did not compare the sound structure of thesevocalizations. One study investigated cardiac andbehavioral reactions of infant chimpanzees to played-back conspecific play panting and revealed that itevoked notable cardiac acceleration and ‘‘threat-like’’staccato vocalization (Berntson et al. 1989). However,neither study focused on the roles of play panting insocial play interactions, nor have its social functionsbeen elucidated.

Human laughter is generally regarded as a socialsignal that plays important roles in social interactionssuch as conversation and social play (Weisfeld 1993;Provine 1996). If we accept van Hooff’s notion thatchimpanzee play panting is homologous to humanlaughter, the common functions of the two vocalizationscan be thought of as the original features of humanlaughter, which now occurs in various social contexts.Investigating social functions of chimpanzee playpanting might help us to understand the origin andevolution of our own laughter.

The purpose of this study is to clarify the socialfunctions of chimpanzee play panting during social playinteractions. Toward this aim, I examined the followinghypotheses. These were formulated to explain the func-tion of vocalizations or specific actions that occur duringsocial play in primates or mammals in general, and thefunction of human laughter and chimpanzee play pant-ing in particular.

Non-aggression hypothesis

Observers of social play in mammals have pointed outthat many behavioral patterns performed in play (i.e.,chasing or biting) are also performed in aggression orpredation (e.g., Bekoff and Allen 1998). These actionsmay lead to misinterpretation by their targets: the tar-gets may confuse the actions for aggression. Therefore, itis suggested that specific play signals have evolved tocommunicate that the ‘‘aggressive’’ behavior is per-formed as play and not as aggression (Bekoff and Allen

1998). Previous works have revealed that these playsignals occur especially at the initiation of social play(e.g., Bekoff 1974) and at the ‘‘aggressive’’ actions dur-ing playing (Bekoff 1995). This explanation has beenproposed for both chimpanzee play panting (Loizos1969; van Hooff 1972) and human laughter (BlurtonJones 1967; Humphreys and Smith 1984).

From this hypothesis, we can derive two specificpredictions. First, chimpanzees are expected to emit playpanting slightly before the initiation of social play.Second, the player who directs ‘‘aggressive’’ actiontoward the other is expected to emit play panting.

Play activation hypothesis

Studies of primate play suggest that play vocalizationmay function to prolong the play interaction (squirrelmonkey: Biben and Symmes 1986; cotton-top tamarin:Goedeking and Immelmann 1986). A more specificfunction of human laughter has been proposed, wherelaughter encourages the partner to reproduce tickling orother kinds of arousal stimulus during play and activatesthe interaction (Rothbart 1973; Harris 1999). This wasalso hypothesized for the function of chimpanzee playpanting in tickling interaction (Harris 1999) and can beconsidered applicable for play panting during social playinteraction in general.

In her review of laughter in young children, Rothbart(1973) stated that child laughter occurred when thearousal states of the children were heightened by stim-ulation from the caretakers. At the same time, shepointed out that the same stimulation could sometimesmake children cry rather than laugh; that is, stimulationin play interaction contains risky or thrilling propertiesthat can collapse the interaction if it exceeds the child’stolerable limits of arousal. [These points were also dis-cussed by Plooij (1979) concerning chimpanzee playpanting in mother–infant play.] Rothbart proposed thatchild laughter would function to encourage a caretakerto reproduce stimulation by providing feedback to thecaretaker that the given stimulation was within thechild’s tolerable limits of arousal and was pleasurablefor the child.

When we examine this hypothesis for the function ofplay panting in chimpanzees, the following predictionsneed to be considered.

1. Relationship between received stimulation and playpanting. This hypothesis predicts that chimpanzeesemit play panting more often when they receivestimulation eliciting greater arousal, that is, when thesituation contains a greater risk of excessive arousaland needs positive feedback for continuing interac-tion. I examined this in terms of types of ‘‘aggressive’’actions and the age class of the performer. It is sup-posed that chimpanzees should have higher arousalwhen they play with adolescents or mature individ-uals than with infants or juveniles. It is also supposed

juve es, dol ado esce ts, at atu e d v dua s

222

that chimpanzees should have higher arousal duringrisky actions such as mouthing (a synonym of playbiting). Additional analysis was conducted on therelationship between the occurrence of screaming andthe conditions mentioned above to examine thevalidity of these suppositions.

2. Effect of play panting on succession of ‘‘aggressive’’actions. This hypothesis also predicts that a chimpan-zee continues to perform ‘‘aggressive’’ actions moreoften when the target individual emits play panting.

Methods

Subjects

I conducted this study in the Mahale Mountains Na-tional Park, Tanzania. The subjects were 51–54 chim-panzees of the M group living in the Park. Three infantswere born during the study period. In Mahale, a long-term study on chimpanzees has continued since 1965(Nishida 1990), and all individuals are identified andwell habituated.

Data collection and analysis

The study period was from October 2000 to February2001. I conducted observation by following a focalchimpanzee all day or as long as possible. During theobservation, I recorded ‘‘social play’’ interactions (de-fined below) that occurred around the focal individualwith a video camera (SONY: DCR-PC7). Some of theseinteractions did not include the focal individual. I madefocal observation since it enabled me to collect muchdata from the start of the play interactions. Most of thevideo data were shot about 5 m from chimpanzees.Some of the video data that did not record vocalizationsvery well were excluded from analysis. The total dura-tion of the analyzed social play was 638.1 min, and theseobservations included 8 mature males and 15 maturefemales, 5 adolescent males and 2 adolescent females, 2juvenile males and 3 juvenile females, and 2 infant malesand 9 infant females. For age classes, I defined ‘‘infants’’as individuals 0–3 years old, ‘‘juveniles’’ as those 4–7 years old, ‘‘adolescents’’ as those 8–13 years old, and‘‘mature individuals’’ as those over 14 years old. Al-though most of the previous studies defined ‘‘infants’’ asindividuals 0–4 years old (e.g., Hiraiwa-Hasegawa et al.1984; Nishida et al. 1990), two 4-year-old individualswere classified as ‘‘juveniles’’ because they usuallymoved independently and did not ride on their mothers’backs.

Terms are defined as follows:

Play panting Panting or grunting sounds emitted byplaying chimpanzees. Play panting has some variancefrom small breathy panting to large throaty grunt-ing, but I did not differentiate these variations in thisstudy.

Social play ‘‘Social play’’ was defined as interactions ofwrestling (including tickling) and chasing [includinground-chasing in Hayaki (1985)] (Nishida et al. 1999). Iexcluded from analysis ‘‘play mothering’’ that was notaccompanied by tickling. I also excluded ‘‘gregariouslocomotor play,’’ in which two or more infants gatheredin a tree crown and leaped between branches or swungon a branch repeatedly, since this did not involve chas-ing or wrestling and could be regarded as simultaneousinstances of solitary play.

Play bout I analyzed only social play between two indi-viduals. I defined a play bout as beginning when bothplayers start to engage in the interaction of social playand ending when it stops for more than 10 s (in otherwords, I included brief breaks of 10 s or less in a playbout). If one of the players was swapped for anotherindividual, a different play bout was counted. Social playbetween two individuals sometimes shifts to that betweenthree or more individuals and vice versa. Even in suchcases, I analyzed dyadic play bouts, which start when thenumber of players becomes two and end when a thirdparty joins the interaction. Unsuccessful invitations tosocial play, which were called ‘‘para-play bouts’’ inHayaki (1985), are not included in the definition.

‘‘Aggressive’’ action During social play interaction, wecan often observe that some kinds of actions that alsooccur in aggression or predation are performed by one ofthe players toward the other. I call the former individual‘‘performer’’ and the latter ‘‘target.’’ The six identifiedtypes of such ‘‘aggressive’’ actions are listed below.Behavioral patterns are termed after Nishida et al.(1999). I recorded each occurrence of these actions andrecorded whether play panting or screaming occurredand which player (i.e., performer or target) emitted themin each action. (It was sometimes difficult to discern theemitter of play panting only from the video-recordedvocalization, especially when the players belonged to thesame age class. Therefore, I recorded the name of theemitter by dictation during video shooting.)

Mouthing The performer puts its open mouthagainst the target’s body and nibbles.This is thought to be a kind of ticklingamong chimpanzees. An action beginswhen the performer puts its mouth onthe target and ends when it takes itsmouth away or changes the body part itmouths. The performer sometimes bitesthe neck or arm of the target and pulls. Iincluded these instances in this category.

Tickle The performer presses the fingers of oneor both hands on the target’s body. Thisis often accompanied by a succession ofalternating finger movements of ‘‘flexingand stretching.’’ An action begins whenthe performer puts its fingers on thetarget and ends when it takes the fingersaway or changes the body part it tickles.

223

Chase The performer runs after the target orfollows the target at walking speed,during which the performer oftenreaches for and tries to catch the targetby the leg. Even in round chasing (i.e.,chasing around a tree trunk or shrubs),one of the players (i.e., the performer)often approaches the other (i.e., thetarget) from behind and tries to catchthe other. An action ends when theperformer catches the target.

Grab The performer catches the arm or leg ofthe target, who tries to move away. Thisaction often occurs during chasing.

Slap/stamp The performer hits the target with thepalm of hand or sole. If the interval ofsuccessive hitting actions is less than 1 s,I counted those actions as one action.

Drag The performer grasps the arm or leg ofthe target and pulls its body along theground.

In addition to these one-way ‘‘aggressive’’ actions, othertypes of interactions also occurred during play bouts:mutual ‘‘aggressive’’ interactions such as struggling,mutual tickling or round chasing, ‘‘non-aggressive’’interactions such as gentle finger wrestling, and briefbreaks. However, in examining the hypotheses, I re-stricted the analyses to the one-way ‘‘aggressive’’ actionslisted above, where I could clearly identify which indi-vidual was the performer and which was the target.

Succession of ‘‘aggressive’’ action

‘‘Aggressive’’ actions often occurred in series. The fre-quency distribution of intervals of ‘‘aggressive’’ actionswithin each play bout is shown in Fig. 1. It was highlyprobable that an ‘‘aggressive’’ action occurred within10 s after termination of the previous action. Therefore,I operationally defined that a succession of ‘‘aggressive’’action occurs when an ‘‘aggressive’’ action is followed bythe next action at an interval of 10 s or less. Successionof ‘‘aggressive’’ actions included two situations. One was‘‘continuation,’’ where the performer continued to per-form an ‘‘aggressive’’ action (either the same as or dif-ferent from the previous action), and the other was ‘‘rolechange,’’ where the former target directed an ‘‘aggres-sive’’ action toward the former performer. I excluded an‘‘aggressive’’ action from this analysis if I lost goodobservation within 10 s after termination of the action. Ialso excluded an ‘‘aggressive’’ action if the participantsof social play changed within 10 s after the termination,because the new player may have a disturbing effect onthe dyadic play interaction.

Statistics

In the following analyses, I calculated the percentages of‘‘aggressive’’ actions in which the targets emitted play

panting and the percentages of ‘‘aggressive’’ actions afterwhich the performers continued to perform an ‘‘aggres-sive’’ action. In such calculations, the reliability of avalue would decrease if the denominator were too small.Therefore, for statistical comparison, I used only the dataof individuals whose number of ‘‘aggressive’’ actions wasfive or more. A two-tailed Wilcoxon’s signed-rank pairedtest was employed to test the difference.

Results

Did play panting occur before the initiationof play bouts?

The total number of play bouts observed was 793. Age-class combinations of play bouts are shown in Table 1.Some of the play bouts were incomplete: they hadalready begun when I started to record and I could notobserve from 10 s before the initiation (112 cases), or Ihad lost good observation and I could not observe up to10 s after the termination (90 cases). I included theseincomplete play bouts in the analyses, unless otherwisenoted.

I examined whether play panting occurred during the10 s before the initiation of play bouts. I excluded 112play bouts from this analysis because I could not recordthem from 10 s before the initiation. In addition, I ex-cluded 116 play bouts because they shifted from socialplay of a different combination or from that betweenthree or more individuals, with or without brief breaksof 10 s or less. Among the remaining 565 play bouts, inonly 3 cases was the initiation preceded by the occur-rence of play panting. In two of these cases, the indi-viduals who emitted play panting were making solicitingbehaviors in ‘‘defensive’’ postures at the same time,including ‘‘stand with the hip to the partner, look back

Fig. 1 Frequency distribution of intervals between ‘‘aggressive’’actions

224

and swing the hind leg in front of him’’ and ‘‘touch andrun away.’’ These individuals were chased or mouthedafter a few seconds.

Who emitted play panting during ‘‘aggressive’’ actions?

The total number of ‘‘aggressive’’ actions observed was2,290 (Table 1), among which 1,307 actions wereaccompanied by play panting. Total duration of‘‘aggressive’’ actions was 195.6 min (30.7% of totalduration of play bouts). In 16 ‘‘aggressive’’ actions, itwas impossible to discern which individual emitted playpanting, so these were excluded from analysis.

Figure 2 compares the rates of occurrence of playpanting between when chimpanzees were performersand when they were targets of ‘‘aggressive’’ actions.Chimpanzees emitted play panting more often whenthey were targets of ‘‘aggressive’’ actions than when theywere performers. This tendency was significant in infantsand mature individuals (P<0.01 and P<0.05, respec-tively) and approached significance in juveniles andadolescents (0.05<P<0.10). This clearly was thecase regardless of the kind of ‘‘aggressive’’ action(percentages of actions accompanied by play panting bythe performer and by the target, respectively, were asfollows: 6.1% by the performer and 63.1% by the targetduring 1,225 cases of ‘‘mouthing’’; 3.0 and 38.2% during267 cases of ‘‘tickle’’; 5.0 and 52.3% during 384 cases of‘‘chase’’; 6.0 and 58.2% during 261 cases of ‘‘grab’’; 3.3and 25.9% during 116 cases of ‘‘stamp/slap’’; and 0 and42.9% during 21 cases of ‘‘drag’’).

Relationship between received stimulationand screaming

During ‘‘aggressive’’ actions, 13 incidents of screamingby the targets were observed. Among these 13 incidents,11 incidents occurred when the performer of the actionwas an adolescent or mature individual (1,309 of 2,290‘‘aggressive’’ actions were performed by adolescent ormature individuals). In addition, 10 incidents occurred

when the kind of ‘‘aggressive’’ action was ‘‘mouthing’’(1,225 of 2,290 ‘‘aggressive’’ actions were ‘‘mouthing’’).

Relationship between received stimulationand play panting

I examined the effects of the performer’s age class andthe type of ‘‘aggressive’’ action on play pantingoccurrence. Figure 3 shows the relationship betweenthe performer’s age class and the occurrence of playpanting by the target. Target chimpanzees tended toemit play panting more often when the performer ofthe ‘‘aggressive’’ action was an adolescent or matureindividual than when the performer was an infantor juvenile. This tendency was statistically significant ininfant targets (P<0.01). However, there were notenough samples for statistical analysis in age classesother than infants.

Table 1 Age-classcombinations of play bouts and‘‘aggressive’’ actions. Infinfants, Juv juveniles, Adoladolescents, Mat matureindividuals. Younger fi older‘‘aggressive’’ actions performedby younger individuals towardolder ones, same age class‘‘aggressive’’ actions betweenindividuals of same age class,older fi younger ‘‘aggressive’’actions performed by olderindividuals toward youngerones

No. ofplay bouts

Play boutduration (min)

No. of ‘‘aggressive’’ actions

Younger fi older Same ageclass

Older fi younger Total

Inf–Inf 174 98.1 244 244Inf–Juv 112 88.7 78 206 284Inf–Adol 63 69.8 67 201 268Inf–Mat 132 113.7 64 347 411Juv–Juv 72 45.6 201 201Juv–Adol 80 75.5 97 224 321Juv–Mat 42 57.4 24 174 198Adol–Adol 81 48.8 222 222Adol–Mat 25 21.7 35 55 90Mat–Mat 12 18.8 51 51Total 793 638.1 365 718 1,207 2,290

play panting

Fig. 2 Rates of occurrence of play panting when chimpanzees wereperformers and when they were targets of ‘‘aggressive’’ actions.Only the data of individuals whose numbers of ‘‘aggressive’’actions in both of the two conditions were five or more wereanalyzed. The numbers of analyzed individuals are in parentheses.**P<0.01; *P<0.05; star 0.05<P<0.1. Error bars show the SDs.Inf infants, Juv juveniles, Adol adolescents, Mat mature individuals

225

Next, I examined the relationship between type of‘‘aggressive’’ action and occurrence of play panting bythe target. Infant targets emitted play panting moreoften when they were mouthed than when subjected toother kinds of ‘‘aggressive’’ actions (Table 2). Thistendency was especially strong among infants 0–1 yearold, who hardly ever emitted play panting in situationsother than being mouthed. However, this tendency wasnot clearly seen among targets of other age classes; thatis, these targets often emitted play panting even whenthey were chased or grabbed.

These results seem to indicate that infants emitted playpanting more often when they received stimulation elic-iting higher arousal. However, other explanations are still

possible. For example, adolescent andmature individualstended tomouth longer, and this could have increased theopportunities of play panting. In fact, longer mouthingactions were performed by adolescent and mature indi-viduals more often than by infants and juveniles(Table 3). However, regardless of duration of mouthingactions, infants emitted play panting more often whenthey were mouthed by adolescents or mature individuals(Table 3). Another possible explanation is that infant andjuvenile performers are relatively unskillful and tend tomouth less ticklish body parts of infants. In fact, therewere some differences among age classes of the performerin the distribution of body parts where they oftenmouthed (Table 4). Infant and juvenile performersmouthed the arms and legs more often and the bellyand underarms less often than older individuals. Thearms and legs seemed to be less ticklish than other bodyparts. However, regardless of the mouthed body part,infants emitted play panting more often when the per-former was an adolescent or mature individual (Table 4).

Effect of play panting on successionof ‘‘aggressive’’ actions

I then examined whether play panting of the targetpromotes the succession of ‘‘aggressive’’ actions. Fromthe total of 2,274 ‘‘aggressive’’ actions in which I could

Fig. 3 Effect of performer’s age class (infant and juvenile vsadolescent and mature) on play panting by the targets. Thenumbers of analyzed individuals (targets) are in parentheses. Onlythe data of targets whose numbers of ‘‘aggressive’’ actions in bothof the two conditions were five or more were used. **P<0.01, N.S.not significant. Error bars show the SDs. Inf infants, Juv juveniles,Adol adolescents, Mat mature individuals

Table 2 Effects of type of ‘‘aggressive’’ action on play panting bythe targets. Percentages given are those of ‘‘aggressive’’ actionsaccompanied by play panting by the targets. Total number of ac-tions in each condition (100%) are in parentheses. Data are pooled

because there were not enough samples for statistical analysis.‘‘Drag’’ and ‘‘slap/stamp’’ were omitted because they occurred lessoften than the actions listed here. Inf infants, Juv juveniles, Adoladolescents, Mat mature individuals

Target Performer Mouthed Tickled Chased Grabbed

Early infa Adol–mat 56.3% (32) 4.5% (22) – (0) 0% (5)Inf–juv 12.1% (33) 0% (8) 9.1% (11) 0% (14)

Late infb Adol–mat 72.3% (282) 31.2% (109) 46.2% (13) 25.0% (48)Inf–juv 34.1% (167) 26.7% (30) 24.1% (79) 44.4% (63)

Juv Adol–mat 81.0% (210) 57.4% (54) 72.5% (69) 94.2% (52)Inf–juv 63.9% (158) 66.7% (9) 59.0% (61) 75.8% (33)

Adol Adol–mat 80.5% (118) 60.0% (15) 57.4% (108) 83.3% (24)Inf–juv 61.8% (110) 66.7% (9) 40.9% (22) 70.0% (10)

Mat Adol–mat 79.5% (44) 60.0% (10) 88.2% (17) 100.0% (8)Inf–juv 29.6% (71) 100.0% (1) 75.0% (4) 75.0% (4)

a Infants 0–1 year old (four individuals)b Infants 2–3 years old (seven individuals)

Table 3 Duration of mouthing actions and play panting by infants.Percentages given are those of mouthing actions accompanied byplay panting by infant targets. Number of actions are in paren-theses. Data are pooled because there were not enough samples forstatistical analysis. Inf infants, Juv juveniles, Adol adolescents, Matmature individuals

Performer 1–5 s 6–10 s 11+ s

Adol–mat 64.5% (203) 84.8% (66) 88.9% (45)Inf–juv 29.2% (161) 25.0% (28) 63.6% (11)

226

discern the emitters of play panting, 105 ‘‘aggressive’’actions had to be excluded from this analysis because Ilost good observation or the participants of social playchanged within 10 s after termination of social play.Among the remaining 2,169 ‘‘aggressive’’ actions, suc-cession of ‘‘aggressive’’ actions occurred in 1,263 cases.‘‘Continuations’’ occurred after 1,067 actions, and ‘‘rolechanges’’ occurred after 196 actions.

Figure 4 shows the rates of continuation of ‘‘aggres-sive’’ actions by performers. Most of the lines rise on theright side, which indicates that most of the performers‘‘continued’’ to perform an ‘‘aggressive’’ action moreoften when the target emitted play panting during theprevious action. This tendency was statistically signifi-cant in the case of mature performers (P<0.05) andapproached significance in infant performers(0.05<P<0.10). As for ‘‘role changes,’’ no significanteffect of targets’ play panting was found (Fig. 5,P>0.05).

Discussion

Social functions of play panting

Non-aggression hypothesis

Before the initiation of social play, chimpanzees rarelyemitted play panting. During ‘‘aggressive’’ actions,chimpanzees often emitted play panting when they werethe targets but rarely did so when they performed‘‘aggressive’’ actions. Therefore, we can conclude thatchimpanzee play panting does not have the function ofcommunicating that aggression-like behaviors are per-formed as play and not as aggression. This kind ofmessage may be transmitted by other kinds of behaviors,such as exaggerated approaching [e.g., ‘‘play walk’’ or‘‘gamboling-type run’’ described in Van Lawick-Goodall(1968)] or ‘‘play face.’’

Play activation hypothesis

Infant chimpanzees emitted play panting more oftenwhen they received ‘‘aggressive’’ actions from adolescentor mature individuals than from other infants or juve-niles. Furthermore, infants emitted play panting moreoften when they were ‘‘mouthed’’ than when subjectedto other kinds of ‘‘aggressive’’ actions. This occurredmore often when adolescent or mature individuals

mouthed them, regardless of duration of actions ormouthed body parts. The additional analysis ofscreaming during ‘‘aggressive’’ actions showed that mostscreaming occurred when the performer was an adoles-cent or mature individual and when the kind of actionwas mouthing. These results suggest that infants emitplay panting more often when they receive stimulationeliciting higher arousal, which could be considered morerisky or thrilling.

This tendency was especially striking among infants0–1 year old. However, this was not so clear in indi-viduals older than juveniles; they often emitted playpanting even when they were chased or grabbed. Thisresult possibly indicates that ‘‘anticipatory’’ arousaloccurs in these older individuals who know generalbehavioral sequences in which the performer chases andcatches the target and then mouths the target. However,concerning the effect of age classes of the performers of‘‘aggressive’’ actions, even the targets older than juve-niles showed a tendency to emit play panting when

Table 4 Mouthed body parts and infant play panting. Percentagesgiven are of mouthing actions accompanied by play panting byinfant targets. Number of actions are in parentheses. Excluded arethose mouthing actions in which the mouthed body part was

unclear. Data are pooled because there were not enough samplesfor statistical analysis. Inf infants, Juv juveniles, Adol adolescents,Mat mature individuals

Performer Belly/underarm Back Head/neck Arm/leg

Adol–mat 87.5% (96) 68.9% (61) 76.0% (100) 34.8% (46)Inf–juv 32.4% (37) 35.3% (51) 34.5% (55) 18.9% (53)

Fig. 4 Rate of ‘‘continuation’’ of an ‘‘aggressive’’ action inresponse to play panting by the targets. The y-axis shows thepercentage of ‘‘aggressive’’ actions after which the performer‘‘continued’’ to perform an ‘‘aggressive’’ action within 10 s.Straight lines represent values of each individual performer, andbars represent mean values. Only the data of performers whosenumbers of ‘‘aggressive’’ actions with and without the target’s playpanting were each five or more were analyzed. The numbers ofanalyzed individuals are in parentheses. *P<0.05, star0.05<P<0.1. Inf infants, Juv juveniles, Adol adolescents, Matmature individuals

227

the performers were adolescent or mature (Table 2),although I could not conduct statistical analysis.

The performers ‘‘continued’’ to perform an ‘‘aggres-sive’’ action more often when the target emitted playpanting. This tendency seemed to be the case regardlessof age class of the performers. Since the play panting ofthe target had no effect on the occurrence of ‘‘rolechange,’’ it can be concluded that play panting in thetarget chimpanzee functions to encourage the performerto chase or tickle more.

In summary, these results show that chimpanzee playpanting functions to activate the interaction of socialplay, especially when the interaction contains misleadingproperties concerning risk of excessive arousal and,therefore, when it needs positive feedback for continu-ation. Furthermore, another misleading property can befound in individuals emitting play panting. In ticklinginteraction, the targets emitted play panting and at thesame time they usually ‘‘tried to escape’’ from the ticklestimulation, as if they did not want it; they squirmed andwarded off the stimulation or defended the tickled partwith the hands or legs, as Plooij (1979) described. Sim-ilarly in chasing, the targets emitted play panting whilethey ‘‘tried to escape’’ from the performer. The samesituation is also described in human tickling interaction(McGrew 1972; Harris 1999). It is assumed that theplayful interaction of tickling or chasing would neveroccur if one did not react or try to escape. In otherwords, ‘‘trying to escape’’ may be one of the main fac-tors to ‘‘form’’ social play, which is characterized by the

interaction between ‘‘trying to escape’’ and ‘‘trying tocatch and tickle.’’ However, this situation can be con-fusing for the performer, who may confuse whether thetarget wants to quit playing and really go away or wantsto keep on being tickled or chased. Play panting by thetarget chimpanzees may also serve as positive feedbackin this situation, indicating that they are trying to escapeonly as play and do not really want to quit playing. Inthis sense, play panting can be called a play signal.

This function of prompting the partner to keep onproviding playful stimulation was also suggested for thefunction of human laughter in tickling (Harris 1999), inarousal games between infants and caretakers (Rothbart1973), and even laughter at humor (Weisfeld 1993). Thiskind of activation of affinitive interaction may be thecommon and fundamental function of human laughterand chimpanzee play panting.

Relationship between play panting and play face

Although van Hooff (1972) thought that play pantingoften ‘‘accompanies’’ play face, chimpanzees sometimesemitted play panting with the mouth only slightlyopened, which is apparently different from the typicalplay face that occurs in a playful encounter (Matsusaka,personal observations). There seemed to be a differencein the context of occurrence between play panting andplay face. In social play interactions, mainly the targetsof ‘‘aggressive’’ actions emitted play panting, and playpanting rarely occurred before the initiation of socialplay. However, it has been reported that chimpanzeesoften present play face before they mouth the targets(Plooij 1979) and that play face sometimes occurs beforethe initiation of social play (Hayaki 1985). Van Hooff(1972) stated that this could be regarded as a ritualizedintention movement of mouthing, and similarly Plooij(1979) called it ‘‘ready-to-bite face.’’ The difference incontext of occurrence between play face and play pant-ing was also suggested for Barbary macaques (Skirl et al.1996; Kipper and Todt 2002). Therefore, it is plausibleto assume that play panting and play face contain dif-ferent motivational states and may have different roles insocial play interaction. To discuss the phylogeny andevolution of laughter and smiling, further investigationsare needed to clarify the relationship between vocaliza-tions and facial expressions during play in humans, apes,and other primates.

Acknowledgements I would like to thank the Tanzanian Commis-sion for Science and Technology, Serengeti Wildlife ResearchInstitute, Mahale Mountains Wildlife Research Center, and Tan-zania National Parks for permission to do this research and fortheir support while I was in Tanzania; T. Nishida for his guidanceand encouragement throughout this study; S. Preuschoft forreading the manuscript and giving me many important suggestions;three anonymous referees for useful comments that improved themanuscript; S. Uehara, K. Hosaka, M. Nakamura, S. Fujita,J. Wakibara, N. Kutsukake, and Watongwe research assistants fortheir cooperation in the field; J. Yamagiwa, S. Suzuki, and othermembers of the Laboratory of Human Evolution Studies, Kyoto

Fig. 5 Rate of occurrence of ‘‘role change’’ in response to playpanting in the targets. The y-axis shows the percentage of‘‘aggressive’’ actions after which the former target in turnperformed an ‘‘aggressive’’ action within 10 s. Straight linesrepresent values of each individual performer and bars representmean values. Only the data of performers whose numbers of‘‘aggressive’’ actions with and without the target’s play pantingwere each five or more were analyzed. The numbers of analyzedindividuals are in parentheses. Inf infants, Juv juveniles, Adoladolescents, Mat mature individuals

228

University for their critical comments and useful suggestions. Thisstudy was financially supported by a Grant-in-Aid for ScientificResearch (A) of the MEXT (No. 12375003 to T. Nishida), a Grant-in-Aid for the COE Research (No. 10CE2005 to O. Takenaka) anda Grant-in-Aid for 21st Century COE Research Kyoto University(No. A14).

References

Bachorowski J-A, Smoski MJ, Owren MJ (2001) The acousticfeatures of human laughter. J Acoust Soc Am 110:1581–1597

Bekoff M (1974) Social play and play-soliciting by infant canids.Am Zool 14:323–340

Bekoff M (1995) Play signals as punctuation: the structure of socialplay in canids. Behaviour 132:419–429

Bekoff M, Allen C (1998) Intentional communication and socialplay. In: Bekoff M, Byers JA (eds) Animal play: evolutionary,comparative, and ecological. Perspectives, Cambridge Univer-sity Press, Cambridge, pp 97–114

Berntson GG, Boysen ST, Bauer HR, Torello MS (1989) Con-specific screams and laughter: cardiac and behavioral reactionsof infant chimpanzees. Dev Psychobiol 22:771–787

Biben M, Symmes D (1986) Play vocalizations of squirrel monkeys(saimiri sciureus). Folia Primatol 46:173–182

Blurton Jones NG (1967) An ethological study of some aspects ofsocial behaviour of children in nursery school. In: Morris D (ed)Primate ethology. Aldine, Chicago, pp 347–368

Darwin C (1872/1921) The expression of emotions in man andanimals. In: Darwin F (ed) J. Murray, London

Goedeking P, Immelmann K (1986) Vocal cues in cotton-toptamarin play vocalizations. Ethology 73:219–224

Harris C (1999) The mystery of ticklish laughter. Am Scientist87:344–351

Hayaki H (1985) Social play of juvenile and adolescent chimpan-zees in the Mahale Mountains National Park, Tanzania. Pri-mates 26:343–360

Hiraiwa-Hasegawa M, Hasegawa T, Nishida T (1984) Demo-graphic study of a large-sized unit-group of chimpanzees in theMahale Mountains, Tanzania: a preliminary report. Primates25:401–413

Hooff JARAM van (1972) A comparative approach to the phy-logeny of laughter and smiling. In: Hinde RA (ed) Non-verbalcommunication. Cambridge University Press, Cambridge, pp209–223

Humphreys AP, Smith PK (1984) Rough-and-tumble in preschooland playground. In: Smith PK (ed) Play: in animals and hu-mans. Blackwell, Oxford, pp 241–266

Kipper S, Todt D (2002) The use of vocal signals in the social playof Barbary macaques. Primates 43:3–17

Lawick-Goodall J van (1968) A preliminary report on expres-sive movements and communication in the Gombe Stream

chimpanzees. In: Jay PC (ed) Primates: studies in adaptationand variability. Holt, Rinehart and Winston, New York, pp313–374

Loizos C (1969) An ethological study of chimpanzee play. In:Proceedings of the 2nd international congress on primatology,vol 1. Karger, Basel, pp 87–93

Markus N, Croft DB (1995) Play behaviour and its effects on socialdevelopment of common chimpanzees (Pan troglodytes).Primates 36:213–225

Marler P (1969) Vocalizations of wild chimpanzees: an introduc-tion. In: Proceedings of the second International Congress onPrimatology, vol 1. Karger, Basel, pp 94–100

Marler P, Tenaza R (1977) Signaling behavior of apes with specialreference to vocalization. In: Sebeok TA (ed) How animalscommunicate. Indiana University Press, Bloomington, pp 965–1033

McGrew WC (1972) An ethological study of children’s behavior.Academic, New York

Mendoza-Granados D, Sommer V (1995) Play in chimpanzees ofthe Arnhem zoo: self-serving compromises. Primates 36:57–68

Nishida T (1990) A quarter century of research in the MahaleMountains: an overview. In: Nishida T (ed) The chimpanzees ofthe Mahale Mountains. University of Tokyo Press, Tokyo, pp3–35

Nishida T, Takasaki H, Takahata Y (1990) Demography andreproductive profiles. In: Nishida T (ed) The chimpanzees of theMahale Mountains. University of Tokyo Press, Tokyo, pp 63–97

Nishida T, Kano T, Goodall J, McGrew WC, Nakamura M (1999)Ethogram and ethnography of Mahale chimpanzees. AnthropolSci 107:141–188

Plooij F (1979) How wild chimpanzee babies trigger the onset ofmother–infant play: and what the mother makes of it. In:Bullowa M (ed) Before speech: the beginning of interpersonalcommunication. Cambridge University Press, Cambridge, pp223–243

Preuschoft S (1992) ‘‘Laughter’’ and ‘‘smiling’’ in barbary maca-ques (Macaca sylvanus). Ethology 91:220–236

Preuschoft S, van Hooff JARAM (1997) The social function of‘‘smile’’ and ‘‘laughter’’: variations across primate species andsocieties. In: Segerstrale U, Molnar P (eds) Nonverbal com-munication: where nature meets culture. Erlbaum, Mahwah,N.J., pp 171–189

Provine RR (1996) Laughter. Am Scientist 84:38–45Rothbart MK (1973) Laughter in young children. Psychol Bull

80:247–256Skirl J, Todt D, Kipper S, Kießelbach C (1996) Laughter in re-

sponse to tickling. IPS/ASP Congress Abstract No. 496Waal F de (1988) The communicative repertoire of captive bono-

bos (Pan paniscus), compared to that of chimpanzees. Behav-iour 106:183–251

Weisfeld GE (1993) The adaptive value of humor and laughter.Ethol Sociobiol 14:141–169

229