AGGRESSIVE BEHAVIOR Volume 31, pages 138–152 (2005)

Reconciliation Pattern After Aggression Among Japanese Preschool Children Keiko K. Fujisawa, Nobuyuki Kutsukake, and Toshikazu Hasegawa Department of Cognitive and Behavioral Science, Graduate School of Arts and Sciences, The University of Tokyo, Japan.

: : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : Using an ethological observation method (the PC-MC comparison method), this study investigated patterns of reconciliation after aggression among Japanese preschool children, focusing on factors that influenced its occurrence or acceptance, as well as on strategies and function. There were several factors among four-year-olds that facilitated reconciliatory attempts and acceptance, such as proximity after the aggression; among three-year-olds there were no such factors. Depending on the situation during and after aggression, children used ‘explicit’ (e.g., apologizing, compromising, and offering objects) and ’implicit’ (e.g., being friendly, talking without apologizing, and touching the opponent gently) reconciliatory strategies. Reconciliation occurred more often among friends than among nonfriends, particularly among four-year-olds. However, after considering the baseline affiliation level, the conciliatory tendency among non-friends was higher than that among friends in four-year-olds. Victims’ self-directed behavior (SDB) - a behavioral index of stress - was elevated following aggression, but decreased following reconciliation. This suggests that reconciliation functions to reduce the postconflict stress suffered by the targets of aggression. Aggress. Behav. 31:138–152, 2005. r 2005 WileyLiss, Inc.

: : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : Keywords: reconciliation; aggression; preschool

INTRODUCTION Aggressive behavior can be defined as behavior that causes harm to another party; thus aggression has a social character [Durkin, 1995]. Aggression may be seen as a subtype of conflict [Shantz, 1987], as not all conflict involves aggression. Among young children agonistic interaction has the potential to affect the relationship between opponents [Cords and Killen, 1998]. However, it does not always disrupt children’s social lives, but may have positive or beneficial consequences in organizing and stabilizing the early peer group [Strayer and Noel, 1986]. Grant support: JSPS Research Fellowships for Young Scientists (to KF and NK); JSPS 21st Century COE program ‘‘Center for Evolutionary Cognitive Sciences at the University of Tokyo’’. n Correspondence to: Keiko K. Fujisawa, Department of Cognitive and Behavioral Science, Graduate School of Arts and Sciences, The University of Tokyo, 3–8–1 Komaba, Meguro-ku, Tokyo 153–8902, Japan. E-mail: [email protected] Received 7 March 2003; amended version accepted 21 July 2003 Published online in Wiley InterScience (www.interscience.wiley.com). DOI: 10.1002/ab.20076

r 2005 Wiley-Liss, Inc.

Reconciliation Among Japanese Preschool Children

139

Despite decades of research on aggression in children, several questions remain unanswered. First, it is not clear how children interact with an opponent in the period shortly after aggression. Since most previous studies have focused on the immediate outcome of aggression [McGrew, 1972; Strayer and Strayer, 1976; Sackin and Thelen, 1984], such studies might have missed an interaction that occurred after a short-term separation. Indeed, recent studies have shown that interaction between children does not end immediately after aggression, and they still interact with opponents shortly after the aggression [Butovskaya and Kozintsev, 1999; Ljungberg et al., 1999; Verbeek and de Waal, 2001]. Second, the majority of previous studies were limited to clarifying the characteristics of the interactions in a situation involving aggression, since comparisons with the children’s normal behavior were lacking [McGrew, 1972; Strayer and Strayer, 1976; Sackin and Thelen, 1984]. These problems have been addressed, in part, by the ethological observation method used by de Waal and van Roosmalen [1979]. This method was later refined by de Waal and Yoshihara [1983], who studied post-conflict behavior among captive groups of chimpanzees and rhesus macaques, respectively, and systematically compared the behavior observed during the postconflict condition (post-conflict; PC) with that in the baseline condition (matched control; MC) (PC-MC comparison;1 see METHODS). Recently, the PC-MC comparison method has been applied to studies of children. This approach has shown that children affiliate with their former opponents more frequently after aggression than the baseline, and that they reconcile when separated briefly after aggression [U.S.: Verbeek and de Waal, 2001; Russia: Butovskaya and Kozintsev, 1999; Sweden: Ljungberg et al., 1999; Kalmyk: Butovskaya, 2001]. This study examined the behavior of Japanese preschool children following aggression, using the PC-MC comparison method. The aims of this study were threefold. First, the features of children’s behavior following aggression and the factors that affected the occurrence and acceptance of reconciliation were examined. Some episodes of aggression end with a clear win for one of the children (e.g., one child gets a toy by force from another child, while the victim gets nothing). Other episodes of aggression end in reconciliation (e.g., one child gets a toy by force from another child, but the aggressor returns it to the victim). Observations of preschool children have shown that conciliatory outcomes occur in one out of every three conflicts [Sackin and Thelen, 1984; Laursen and Hartup, 1989]. In those studies, sex combination between opponents [Sackin and Thelen, 1984], preconflict interaction, and postconflict proximity [Laursen and Hartup, 1989] were related to reconciliation. However, since few studies have conducted multiple variance analysis, it is still unclear which factors actually influence reconciliation when controlling for the effects of other factors [Verbeek and de Waal, 2001]. Among the factors involved in agonistic interaction, the effect of relationship quality on reconciliation was examined. Many previous studies have shown a strong association between reconciliation and friendship. For example, friends reconcile more frequently and more maturely than do non-friends [Hartup, 1992; Hartup et al., 1988; Nelson and Aboud, 1985]. By contrast, Butovskaya and Kozintsev [1999] found greater conciliatory tendencies between non-friends than among friends, when considering baseline affiliation level. Therefore, it is necessary to examine this issue by focusing on reconciliation itself, as well as baseline affiliation in the control condition. Note that ‘conflict’ in ‘post-conflict’ is used synonymously with aggression. This is confusing, because conflict must be carefully distinguished from aggression. In order to avoid confusion and to be consistent with previous primatology studies and other recent studies of children, it is stated from the outset that use of the term ‘post-conflict’ means ‘post-aggression.’

140

Fujisawa et al.

Second, reconciliation strategies were investigated. As most previous studies lacked a comparison with control behavior, it is not clear whether the affiliative behaviors involved in reconciliation differ from affiliation in a control situation. Moreover, it has been noted that some reconciliation strategies are difficult to recognize [Ross and Conant, 1992]. In such cases, the concept of implicit reconciliation may offer a way of identifying such strategies, although it has yet not been examined sufficiently. We predicted that children would reconcile using both the form of affiliation that characteristically follows aggression and the form of affiliation that also occurs in the control condition. Third, the functional aspects of reconciliation were investigated. The ‘uncertainty reduction’ hypothesis [Aureli and van Schaik, 1991], in particular, was tested, using a noninvasive index of stress. This hypothesis predicts that a reconciliatory interaction reduces the anxiety or stress that the opponents suffer following aggression. This hypothesis has been well confirmed in non-human primates [Aureli and van Schaik, 1991; Castles and Whiten, 1998; Das et al., 1998; Kutsukake and Castles, 2001]. It was found in these studies that the frequency of self-directed behavior (SDB), which is used as an index of stress or anxiety [Maestripieri et al., 1992], occurs more frequently following aggression, and that its occurrence decreases after reconciliation. This hypothesis was investigated in human children by Ljungberg et al. [1999]. They examined the rate of displacement activities (e.g., jumping up and down, screaming, running around) of children involved in aggression, as an index of stress. They found that the rate decreased when reconciliation was accepted, suggesting that reconciliatory behaviors reduced children’s stress after aggression. However, there is a possibility that the displacement activities examined in this study may themselves have constituted aggressive behavior. Moreover, the observations in their study were made in a small laboratory room; this might have affected the results, in that this unnatural situation may have caused the displacement behavior in the children. The present study attempted to test this hypothesis by measuring the rate of SDB during free play, for the first time. Human SDB, such as touching one’s own hand or face, is known as a reliable ethological index of increased emotional and physiological arousal [Troisi et al., 2000]. For example, patients with more pronounced alexithymic features show more frequent SDB during interviews [Troisi et al., 2000]; college students perform SDB when solving difficult problems more frequently than when solving easy ones [Hrubes and Feldman, 2001]; subjects who have difficulty identifying and describing their feelings show frequent SDB, indicating tension and anxiety [Troisi et al., 1996]. Thus SDB was expected to be useful for investigating changes in the stress level of children following aggression.

METHOD Participants The study was conducted in two classrooms at a public preschool in Tokyo, Japan. The class of three-year-olds consisted of nine boys and six girls, and the class of four-year-olds consisted of 12 boys and 10 girls. One three-year-old girl was not included in the study because she had an autistic disorder that might have affected the results. All of the children were Japanese. The mean ages of the participants at the start of the study were 49 months for the class of three-year-olds (range: 44 to 54 months) and 61 months for the class of four-yearolds (range: 56 to 66 months). Observation was conducted during the latter half of the school

Reconciliation Among Japanese Preschool Children

141

year in Japan, so some members of the three-year-old class reached four years of age and some members of the four-year-old class reached five years of age. However, children’s age refers to the class unit in the preschool. That is, the former was regarded as the three-yearolds and the latter was regarded as the four-year-olds. Observation Methods Observations were conducted during free-play time between 9:00 a.m. and 12:00 noon, in a classroom or playground, between October 2001 and March 2002. In order to minimize the disturbance effect of multiple observers, the first author (KF), who is a native Japanese and had established a good rapport with the children and teachers before starting this study, made all the observations. Following de Waal and Yoshihara [1983], in post-conflict observations (PC), the observer scanned for aggression and carried out a 10–min focal observation of the child who had been the target of aggression, immediately after the aggression ended. Aggression was regarded as taking place when either of the following two types of behavior occurred: physical threats or deliberate pushing/pulling, hitting, kicking, biting [Strayer and Strayer, 1976]; or verbal threats, insults, or derogatory comments [Verbeek and de Waal, 2001]. These behavioral categories were chosen for the present study because they show very clearly whether aggression occurred verbally (e.g., while looking at a child’s picture, one child said to the other, ‘‘Your picture is ugly!’’) or physically (e.g., one child threw a toy at another child); however, some forms of aggression were difficult to distinguish. In an aggressive interaction, the following items were recorded: (1) aggression type (physical or verbal), the names of the initiator and victim, and the timing of its occurrence; (2) the distance between opponents following aggression (together: within 1 m, or separate: more than 1 m apart); (3) the presence or absence of intervention by a teacher; (4) the presence or absence of intervention by bystander(s); and (5) location (classroom or playground). PC was compared with the 10–min baseline observation (matched control: MC) of the same child. MCs were collected only when the following conditions were met: (1) MC observations were begun within 15 min (either side) of the start time of the PC, e.g., if a PC observation was conducted at 10:00 a.m., then the 10–min MC observation was made between 9:45 a.m. and 10:15 a.m., on another day; (2) the focal child had not been involved in any aggression for at least 10 min before a scheduled MC; (3) the former opponent was from 1 to 10 m away from the focal child, to exclude the possibility that former opponents interacted in the MC period simply because they were close to each other, as compared to the PC, or the possibility that they did not interact in the MC because of the distance between them; (4) the locations (classroom or playground) were the same. If it was not possible to conduct a MC observation that met these requirements within five days of the PC, the PC was discarded. In PCs and MCs, these items were recorded: (1) renewed aggression between the opponents, its direction, and the timing of its occurrence; (2) any affiliative attempt between former opponents, the timing of its occurrence, and the initiator (the aggressor or victim); (3) whether the affiliative attempt was accepted (yes or no); (4) the distance between opponents (together: within 1 m, or separate; more than 1 m apart). Eight exclusive affiliation types were recorded: apology (in PC), object offer, symbolic offer (e.g., giving a turn to the opponent), compromise, invitation to play, non-aggressive close interaction and talking that did not

142

Fujisawa et al.

include an apology, touching the opponent’s body in a friendly manner, and joining in the opponent’s play [Sackin and Thelen, 1984; Ljungberg et al., 1999]. Acceptance of the affiliative attempt was identified as the receiver nodding, smiling, or saying ‘yes’ [Ljungberg, et al., 1999]. Bouts of self-directed behavior (SDB) in both PCs and MCs were counted as an index of stress. SDB was defined as touching the own face, head, or clothes, fingering their hands or feet, and thumb sucking [Waxer, 1977; Fairbanks et al., 1982). Since the above behaviors are continuous, a behavior was scored as distinct when the interval between two behaviors exceeded 5 sec. Reliabilities After collecting the data referred to in this study, observation of the participants continued in order to investigate longitudinal changes in reconciliation patterns. In the subsequent study, video data was used to assess the reliability of coding. Although the data analyzed were not the data that were the basis of the present study, it is highly likely that the reliability calculated for the data in the subsequent study would also apply to the present study, as the observations were a continuation of the present study and were conducted by the same observer, under the same conditions, using the same observation procedure (PC-MC comparison method). Two coders, who were unaware of the conditions relating to any specific encounter, and the first author, coded the video script using the same coding as that used in the present study. Cohen’s Kappa value for each category was as follows: location, 1.0; aggression type, .80; distance following aggression, .77; apology, .77; object offer, .88; symbolic offer, .75; compromise, .78; invitation to play, .81; non-aggressive close interaction and talking, .84; touching the opponent’s body, .64; joining in the opponent’s play, .78; teacher’s intervention, .60; bystander’s intervention, .80; SDB, .90. These Kappa values suggest good to excellent inter-rater reliability for each rating. Definition of Friendship Friendship was assessed using two methods. One constituted an ethological approach and the other was by teachers’ nomination. The instantaneous scan sampling method [Martin and Bateson, 1985], was used to record, for every child, the partners within 1 m, every 15 min during the observation period (threeyear-olds: n = 178, four-year-olds: n = 191). Note that this scan sampling was done independently of PC and MC recording. Then, proximity score was calculated for every possible combination in each age group as follows: (the proximity score of a pair) = (the number of times that the pair was observed in proximity) / (the total data for a child). Since observations were undertaken in both a classroom and a playground, it is possible that the higher density of children in the classroom biased the results. However, the Wilcoxon signed ranks test did not show significant difference between the proximity scores for the classroom and playground (n = 36, z =
.408, p = .683, ns). This suggests that the location did not affect the results. Therefore, data were pooled in this analysis. Pairs with proximity scores exceeding the mean of the proximity score in each age group, plus one standard deviation, were regarded as friends. The mean (7SD) was 0.063 (70.046) for three-year-olds and 0.067 (70.11) for four-year-olds. As a result, seven of 78 potential pairs of three -year-olds and 17 of 231 potential pairs of four-year-olds were labeled as ‘friends’. Among the three-year-olds, four pairs were boy-boy pairs and three pairs were

Reconciliation Among Japanese Preschool Children

143

girl-girl pairs. Among the four-year-olds, nine pairs were boy-boy pairs, seven pairs were girlgirl pairs, and one pair was a boy-girl pair. In the teachers’ friendship nomination approach, the teachers in charge of each age group were asked to list each child’s friends. The consistency of the result of scan sampling and the teachers’ list by investigating the correlation between the matrixes were compared. Proximity scores were correlated with the teacher’s notion of friendships (Mantel test, 2,000 permutations; three-year-olds: Z = 11, po.0001, four-year-olds: Z = 22, po.0001), suggesting that this evaluation of friendship was reliable. Analysis Occurrence of reconciliation after aggression. Two commonly used methods of analysis were used to investigate the occurrence of reconciliation. The first method used was the PC-MC comparison [de Waal and Yoshihara, 1983]. In this method, a PC-MC pair forms one data point. A PC-MC pair is labeled ‘attractive’ if the former opponents affiliate only in PC, or earlier in PC than in MC. Similarly, a PC-MC pair is labeled ‘dispersed’ if the former opponents affiliate only in MC, or earlier in MC than in PC. A PC-MC pair is labeled ‘neutral’ if the opponents do not affiliate in either PC or MC, or if affiliation occurs at the same time in both PC and MC. The proportion of ‘attractive PCMC pairs’ with the proportion of ‘dispersed PC-MC pairs’ was compared, using the revised measure of conciliatory tendency of Veenema et al. [1994] and the ‘relative reconciliation measure’ of Butovskaya and Kozintsev [1999]. The former index is calculated as follows: the number of dispersed pairs (d) is subtracted from the number of attractive pairs (a) and then divided by the total number of conflicts involving each individual. The latter is calculated as follows: (a
d)/(a þ d). The relative reconciliation measure has an advantage in that it is independent of the baseline level of affiliation. This is because the relative reconciliation measure excludes the number of neutral pairs from the calculation of conciliatory tendency, while the absolute conciliatory tendency takes them into account. Therefore, if the baseline affiliation level is low in a group, the group will necessarily show low reconciliation measures and cannot be used in comparisons of groups with different baseline affiliation levels [see Butovskaya, 2001]. Second, the ‘time-rule’ method [Aureli et al., 1989] was used. For each PC and MC, the minute during which the first affiliation between the former opponents occurred was determined. Then, the frequency distribution of the first affiliative contact in PCs with that in MCs, was compared using the Kolmogorov-Smirnov test. To examine the generality of the results at the individual level, this part of the analysis was based on the PC-MC pairs involving children who had at least three qualifying PC-MC pairs (n = 22, mean: 6.8 PC-MC pairs per child, range: 3 to 11 PC-MC pairs). Factors influencing the occurrence and acceptance of reconciliatory attempts. To investigate the factors influencing the occurrence and acceptance of reconciliatory attempts, the association between each variable (occurrence and acceptance) and seven variables: aggression type (physical or verbal), distance between opponents following aggression (together or separate), sex combination (same or opposite sex), relationship (friend or nonfriend), teacher intervention (present or absent), bystander intervention (present or absent), and location (classroom or playground) was examined. Moreover, using logistic regression analysis, the extent to which each of the factors contributed to the occurrence and acceptance of reconciliatory attempts when other factors were adjusted was determined. This part of the

144

Fujisawa et al.

analysis used all the PC data. In order to examine the generality of the results at the individual level, the Wilcoxon signed ranks test on the PCs from children who had more than three PCs that involved the factor under consideration were used. Strategies for reconciliation; explicit vs. implicit. The proportions of eight types of affiliative behavior between PCs and MCs were compared to examine the behavioral specificity of affiliation in PCs. If an affiliative behavior occurs more often in PCs than in MCs, it can be regarded as a specific behavior in PCs. Otherwise, the behavior can be regarded as occurring in both PCs and MCs. Therefore, the former were defined as ‘explicit’ reconciliatory affiliation and the latter as ‘implicit’ reconciliatory affiliation. In addition, the association of affiliation type (explicit or implicit) and six variables: initiator (aggressor or target), distance, sex combination, relation, bystander intervention, and teacher intervention. This part of the analysis used all the PC data was examined. In order to examine the generality of the results at the individual level, the Wilcoxon signed ranks test was conducted on the PCs involving children who had more than three PCs that involved the factor under consideration. Post-conflict self-directed behavior. The methods of previous studies on nonhuman primates [Aureli and van Schaik, 1991; Castles and Whiten, 1998; Kutsukake and Castles, 2001] were followed to analyze self-directed behavior (SDB). First, the PC data were excluded in which a focal child experienced reconciliation or further aggression, because such events after aggression might influence the rate of SDB. Then, the temporal distribution of SDB was calculated, during every minute, in the PCs. Next, the mean rate of SDB (and 95% confidence interval) over the 10–min MC period was calculated. This constituted the baseline of SDB for each individual, because the frequency of SDB during the MC was not expected to show any temporal distribution. The mean rate of SDB without reconciliation, or further aggression, was compared with the one before reconciliation in reconciled PC, to assess the effects of reconciliation on the SDB rate. Next, the mean SDB rate before reconciliation in reconciled PC was compared with the one after reconciliation in reconciled PC. Then, the mean rate of SDB after reconciliation in reconciled PC was compared with the one during MC. RESULTS Sixty-five pairs were observed for three-year-olds (mean: 5.9 PC-MC pairs per child, range: 1 to 9 PC-MC pairs, same sex: 51 PC-MC pairs, other sex: 14 pairs) and 92 PC-MC pairs for four-year-olds (mean: 5.4 PC-MC pairs per child, range: 1 to 11 PC-MC pairs, same sex: 80 PC-MC pairs, other sex: 12 pairs). Eighty-seven cases constituted physical aggression (pushing/pulling: 46.0%, hitting: 43.7%, kicking: 5.7%, others: 4.6%) and 70 cases constituted verbal aggression (verbal threats: 48.6%, insults: 22.9%, derogatory comments: 21.4%, others: 7.1%). Occurrence of reconciliation after aggression. The overall conciliatory tendencies are shown in Table I. No age differences were found (Mann-Whitney U-test; nthree-year-olds = 8, nfour-yearolds = 14, V: U = 51.5, p = .971, ns, R: U = 44.5, p = .578, ns). According to these measures, the proportion of attractive pairs was significantly higher than that of dispersed pairs (60.1 vs. 20.8%, Wilcoxon signed ranks test; n = 22, z =
3.726, po.0001). Therefore, it was revealed that Japanese preschool children who were targets of aggression affiliated with former opponents significantly more in PCs than in MCs. Moreover, the aggressors initiated

Reconciliation Among Japanese Preschool Children

145

TABLE I. Conciliatory Tendencies overall

3-year-olds 4-year-olds

friends

non-friends

V

R

V

R

V

R

38.3 35.2

43.4 45.7

63.6 15.9

63.6 20

32.7 53.2

38.1 71.4

Note. V: conciliatory tendency based on Veenema, et al. [1994]. R: relative conciliatory tendency based on Butovskaya and Kozintsev [1999].

affiliation more often than the victims (75.4 vs. 24.6%, Wilcoxon signed ranks test; n = 22, z =
3.385, po.001). The greatest difference in the cumulative observations of reconciliation between PC and MC was found after 5 min in the classroom, while it was found after 4 min in the playground. However, the timing of the first affiliations did not differ significantly between PC and MC, in both classroom and playground (Kolmogorov-Smirnov test; classroom: D = 0.175, p 4.1, ns, playground: D = 0.202, po.10, ns). It was still possible that the increased affiliative interaction between former opponents after aggression could simply have been due to a general indiscriminate increase in affiliation in PCs to all or many members of a group, by the focal children. This possibility was eliminated using the following analysis. The focal children affiliated with significantly fewer partners in the PCs than in MCs (PCs: 0.29 vs. MCs: 0.41 affiliative partners/min, Wilcoxon signed ranks test; n = 22, z =
2.896, p = .004), while the proportion in which the affiliative partner was the former opponent was greater in the PCs than in the MCs (33.5 vs. 13.2%, Wilcoxon signed ranks test; n = 22, z =
4.010, po.0001). These results show that former opponents interacted affiliatively earlier in PCs than in MCs, and the affiliation in PCs was selective. Factors influencing the occurrence and acceptance of reconciliatory attempts. In both age groups, reconciliatory attempts occurred more often in PCs when the opponents were in close proximity after aggression than in PCs when they were separate after aggression (together vs. separate – three-year-olds: 23/26 cases vs. 28/39 cases, w2(1) = 3.842, po.05, four-year-olds: 46/50 cases vs. 26/42 cases, w2(1) = 12.152, po.001). This was confirmed at the individual level (92.0 vs. 66.7%; Wilcoxon signed ranks test; n = 13, z =
2.051, p = .040). Moreover, both three- and four-year-olds initiated reconciliatory affiliation more often between friends than between non-friends (friends vs. non-friends – three-year-olds: 13/ 13 cases vs. 38/52 cases, w2(1) = 4.461, p = .05, four-year-olds: 38/43 cases vs. 34/49 cases, w2(1) = 4.852, po.05). In four-year-olds, reconciliatory attempts occurred more often (1) when a bystander(s) intervened than when one did not (presence: 24/26 cases vs. absence: 48/66 cases, w2(1) = 4.203, po.05); and (2) when a teacher did not intervene than when a teacher did (presence: 18/31 cases vs. absence: 54/61 cases, w2(1) = 11.209, po.001). Three-year-olds initiated reconciliation more often when the opponent was the same sex than the other sex (same sex: 44/52 cases vs. other sex: 7/13 cases, w2(1) = 5.826, po.05). Other factors were not associated with the occurrence of reconciliatory attempts in either age group. Four-year-olds accepted reconciliatory attempts more frequently (1) when the opponents were together than when they were separate after aggression (together: 37/50 cases vs. separate: 16/42 cases, w2(1) = 12.05, po.001); (2) from friends than from non-friends

146

Fujisawa et al.

(friends: 34/44 cases vs. non-friends: 19/48 cases, w2(1) = 13.353, po.001); and (3) when bystander(s) intervention occurred than when it did not (presence: 21/26 cases vs. absence: 32/66 cases, w2(1) = 7.961, po.01). No variables were associated with the acceptance of reconciliatory attempts in three-year-olds. By logistic regression analysis, it was found that, while no variables affected the occurrence and acceptance of reconciliatory attempts in three-year-olds, close distance after aggression (together) influenced the occurrence and acceptance of reconciliatory attempts in four-yearolds. Furthermore, the absence of teacher intervention affected the occurrence of reconciliatory attempts, and friendship affected the acceptance of reconciliatory attempts in four-year-olds (see Tables II and III). Strategies for Reconciliation; Explicit vs. Implicit As shown in Table IV, there were marked differences in the first affiliative behavior in PCs and MCs. Apology, object offer, symbolic offer, compromise, and invitation to play occurred relatively frequently in PCs, as compared to MCs. These behaviors were defined as explicit reconciliatory affiliation. Joining the opponent’s play, and touching the opponent, did not differ in frequency between PCs and MCs. Non-aggressive close proximity, and talking without apology, occurred more often in MCs than in PCs. These behaviors were defined as implicit reconciliatory affiliation. These results suggest that children not only reconcile using the manner of affiliation that characteristically follows agonistic interactions, but also using the same manner of affiliation as in control conditions.

TABLE II. Logistic Regression Analyses: Factors Influencing on the Occurrence of Reconciliatory Attempts 3-year-olds

4-year-olds

Factor

df

B

SE

Wald

p

B

SE

Wald

p

sex combination relation distance teacher’s intervention bystanders’s intervention aggression type location

1 1 1 1 1 1 1

1.56 7.88 1.11
0.85 8.1 0.18
0.56

0.87 41.34 0.94 0.8 43.5 0.73 0.76

3.24 1.04 1.39 1.13 0.03 0.06 0.54

ns. ns. ns. ns. ns. ns. ns.

1.37 0.83 1.85 0.61 0.91
0.07 0.12

0.91 0.66 0.75 0.61 0.91 0.64 0.64

2.27 1.6 6.05 4.24 0.21 0.01 0.04

ns. ns. 0.01 0.04 ns. ns. ns.

TABLE III. Logistic Regression Analyses: Factors Predicting the Acceptance of Reconciliatory Attempts 3-year-olds

4-year-olds

Factor

df

B

SE

Wald

p

B

SE

Wald

p

sex combination relation distance teacher’s intervention bystanders’s intervention aggression type location

1 1 1 1 1 1 1

1.16 0.29 0.93
0.62
0.16 0.49
0.33

0.77 0.77 0.64 0.59 0.79 0.58 0.59

2.27 0.14 2.12 1.09 0.04 0.72 0.31

ns. ns. ns. ns. ns. ns. ns.

1.07 1.58 1.5
0.02 1.01 0.58
0.27

0.81 0.54 0.58 0.57 0.66 0.54 0.54

1.75 8.68 6.85 0.008 2.37 1.16 0.25

ns. 0.003 0.008 ns. ns. ns. ns.

Reconciliation Among Japanese Preschool Children

147

TABLE IV. The Proportion of Various Types of Affiliative Behavior in PCs and in MCs Behavior

PC(%; N = 94)

MC(%; N = 54)

x2

p

36.2 4.3 12.8 8.5 11.7 6.4 10.6 9.6

0 0 7.4 0 3.7 14.8 13 61.1

38.12 4.052 4.215 8.209 6.5 0.299 0.56 15.832

o0.001 0.044 0.04 0.004 0.011 ns. ns. o0.001n

apology object offer symbolic offer compromise invitation to play join the opponent’s play touch the opponent’s body nonaggrresive close and talk n

:PCoMC

Three-year-olds used explicit reconciliatory attempts more often when aggressors initiated the reconciliation than when victims did (aggressor: 27/36 cases vs. victim: 4/15 cases, w2(1) = 9.223, po.05). Moreover, they affiliated more explicitly when together than when they were separate after aggression (together: 22/23 cases vs. separate: 9/28 cases, w2(1) = 21.367, po.001). The generality of these results was confirmed at the individual level (Wilcoxon signed ranks test; initiator: 75 vs. 35.4%, n = 8, z =
2.046, p = .041, distance: 96.4 vs. 36.3%, n = 6, z =
2.333, p = .020). Other factors were not associated with affiliation type in three-year-olds. Four-year-olds did not show these differences, although they used different types of affiliative behavior with friends and non-friends. Four-year-olds affiliated implicitly when the opponent was a friend more often than they did when the opponent was a non-friend (friend: 24/39 cases vs. non-friends: 11/33 cases, w2(1) = 6.817, po.01). Other factors were not associated with affiliation type in four-year-olds. Friendship and Reconciliation As shown in Table I, the conciliatory tendencies between friends based on Veenema et al. [1994] and Butovskaya and Kozintsev [1999] did not significantly differ from those between non-friends for three-year-olds (Mann-Whitney U-test; V: U = 10, p = .30, ns, R: U = 11.5, p = .441, ns). By contrast, in four-year-olds these conciliatory tendencies were significantly more frequent between non-friends than between friends (Mann-Whitney U-test; V: U = 42.5, p = .045, R: U = 33.5, p = .043). Post-Conflict Self-Directed Behavior As shown in Figure 1, following aggression, the focal children’s rates of SDB increased above the control level in PC with no reconciliation. The generality of the phenomenon was confirmed, as individuals’ mean rate of SDB in PC exceeded that in MC (1.08 vs. 0.18 bouts/ min, Wilcoxon signed ranks test; n = 14, z =
3.637, po.0001), suggesting that the victims suffered relatively high levels of stress following aggression. As shown in Figure 2, the mean rate of SDB in PCs with no reconciliation or further aggression did not differ from that before reconciliation in reconciled PCs (1.08 vs. 1.51 bouts/min; Wilcoxon signed ranks test: n = 14, z =
1.538, p = .123, ns). The SDB rate after reconciliation was lower than that before reconciliation (0.29 vs. 1.51 bouts/min, Wilcoxon signed ranks test; n = 14, z =
3.296, p = .001). Moreover, the mean rate of SDB after reconciliation did not differ from that in MC (0.29 vs. 0.18 bouts/min, Wilcoxon signed ranks test; n = 14, z =
1.412, p = .158, ns). Therefore, these results suggest that reconciliation

148

Fujisawa et al.

Fig. 1. The frequency of SDB rate in the first 10 minutes of PCs and MCs without reconciliation or further aggression.

Fig. 2. The frequency of SDB rate without/before/after reconciliation, and in MCs.

decreased the SDB rate of the victim. This indicates that reconciliation functions to reduce the victim’s stress.

DISCUSSION Reconciliation Pattern Among Japanese Children The PC-MC comparison method revealed that Japanese children affiliated with their former opponents more frequently, and selectively, following aggression than they did in the control condition. This reconciliation pattern was confirmed for children who were separated

Reconciliation Among Japanese Preschool Children

149

briefly after aggression. These results are in good agreement with recent studies that have used the PC-MC comparison method [Butovskaya, 2001; Butovskaya and Kozintsev, 1999; Ljungberg et al., 1999; Verbeek and de Waal, 2001]. They are also consistent with the reconciliation pattern among nonhuman primates [Aureli and de Waal, 2000; Aureli et al. 2002]. Earlier studies focused on the immediate outcome [see Verbeek et al., 2000], and the separation of combatants was not viewed as a cooling-off period before better future interactions, but rather as a primitive ‘‘out of sight, out of mind’’, here-and-now solution [Shantz, 1987]. However, our study and those recent studies support that children still interact after aggression and they do reconcile, a result that earlier studies, which examined only the immediate termination of aggression, failed to show. The conciliatory tendency of Japanese children was about 35%. Differences in the conciliatory tendency have been found in previous studies using the PC-MC comparison method [40, 47, 60, and 70% in children from Russia, the U.S., Sweden, and Kalmyk, respectively, in Butovskaya et al., 2000, p. 247, Table 12.1]. Butovskaya [2001] suggested that cross-cultural differences are present in children’s reconciliation frequency and practices. Combining these results, the conciliatory tendency was lowest in Japanese children. Due to the difference in age range (this sample was of three- and four-year-olds, while other studies examined three- to seven-year-olds), we cannot conclude whether the low conciliatory tendency of Japanese children was due to their younger age or the influence of Japanese culture. With respect to factors influencing the occurrence and acceptance of reconciliation, reconciliatory attempts occurred when opponents were together after aggression and when they were friends in both age groups. For three-year-olds, sex combination was related to the occurrence of reconciliation, but no factor significantly influenced the occurrence and acceptance of reconciliation after adjusting for other factors. For four-year-olds, a close distance after aggression was associated with the acceptance of reconciliation. The effect of distance on the occurrence of reconciliatory attempts, and their acceptance, was confirmed after adjusting for other factors. The importance of proximity after aggression has been found in previous studies [Laursen and Hartup, 1989; Verbeek and de Waal, 2001]. This suggests that proximity after aggression is still important for reconciling and restarting interactions, although opponents can also reconcile after they have been separated briefly. It is worth noting that teacher intervention did not have a positive effect on the occurrence of reconciliatory attempts. There is little research on the effects of adult intervention in peer aggression, but a few studies have found similar results [Killen and Sueyoshi, 1995; Killen and Turiel, 1991; Verbeek and de Waal, 2001]. The reason for this may be beyond the scope of this study; however, perhaps teachers intervene in inconsistent and ineffective ways, because they do not know the cause of the aggression in most cases. This may also be associated with the fact that teachers in Japanese preschools encourage children to work out conflicts on their own [Killen and Sueyoshi, 1995]. The comparison of the frequency of the occurrence of various affiliative behaviors after aggression with the control condition showed that some affiliative attempts were characteristic of the period following aggression, and others occurred in both the period following aggression and the control condition. We found that both clearly helped children to reconcile. Some previous studies concluded that children often end their conflicts or aggressive interaction without active resolution [Eisenberg and Garvey, 1981; Killen and Turiel, 1991; Walton and Sedlak, 1982]. In light of these results, however, it is probable that these previous studies overlooked implicit reconciliatory attempts. In addition, it is likely that reconciliatory behaviors, explicit or implicit, are always necessary for reconciliation.

150

Fujisawa et al.

The reconciliatory strategies differed among three- and four-year-olds. Three-year-olds made explicit reconciliatory attempts when the aggressors initiated them or when the opponents were nearby, after aggression, while four-year-olds showed no difference with respect to these points. It might be easier for younger children to understand the intent of explicit reconciliatory attempts, as compared to implicit ones, because such attempts are very clear and characteristic in the situation following aggression. To reconcile, it is necessary to understand and appreciate the other’s needs and goals [Dunn and Slomkowski, 1992]. The differences in the reconciliatory strategies of three- and four-year-olds suggest that younger children are halfway towards developing these skills. Through reconciliation, children develop awareness of others’ needs, desires, and expectations of social interaction [Cords and Killen, 1998]. Such knowledge contributes to the development of children’s social competence [Shantz and Hartup, 1992; Cords and Killen, 1998]. Reconciliation and Friendship With respect to friendship and reconciliation in developmental terms, little research has examined how the interaction of age and friendship in younger children determines reconciliation strategies and outcomes [Hartup, 1992]. When we considered only the period following aggression, we found that reconciliatory attempts occurred more frequently among friends than among non-friends, for both three- and four-year-olds. In addition, attempts succeeded more often among friends than among non-friends in four-year-olds. The effect of relationship on the acceptance of reconciliation was confirmed after adjusting for other factors in four-year-olds. This is consistent with previous studies that showed that friends mutually disengaged from their disagreements more readily than did non-friends, and that equal or partially equal outcomes were more common between friends than between nonfriends [Nelson and Aboud, 1985; Hartup et al., 1988; Dunn and Slomkowski, 1992; Hartup, 1992; Vespo and Calpan, 1993]. In four-year-old children, however, the PC-MC comparison method revealed that conciliatory tendencies [V: Veenema et al., 1994; R: Butovskaya and Kozintsev, 1999], in which the baseline affiliation level was concerned, were more prevalent among non-friends than among friends. Butovskaya and Kozintsev [1999] studied conflicts in Russian primary school children (six- and seven-year-olds) and found the same tendency. They also argued that affiliative activities after aggression between non-friends might be used to restore a relationship damaged by the aggression, while those between friends did not carry as much meaning as those among non-friends, because the bonds between friends are strong and, if damaged, their restoration requires relatively little effort. This explanation may be applicable to younger children (four-year-olds) in Japan. The difference in the meaning of reconciliatory affiliation between friends and non-friends may be explained by our finding that friends use implicit reconciliatory strategies rather than explicit ones, while the reverse was true between non-friends in the case of four-year-olds. Since implicit reconciliatory attempts do not differ from affiliation in the normal condition, reconciliatory affiliation between friends might have essentially the same meaning as in the normal condition. Implicit reconciliatory attempts do not clearly convey the intent of reconciliation, as compared with explicit reconciliatory attempts, so the receiver needs to be more tolerant of the other’s perspective. Friends have a greater investment in one another than do non-friends, and this investment leads them to minimize the ensuing damage when there is a risk of aggression with continued interaction [Hartup, 1992]. Therefore, friends can reconcile tolerantly.

Reconciliation Among Japanese Preschool Children

151

Stress and the Functional Aspect of Reconciliation Although some forms of aggression are physically and psychologically stressful for the individuals so engaged [Cords and Killen, 1998], few studies have observed stress in children involved in aggression. This study is the first to examine the rate of self-directed behavior (SDB) in children who were the target of aggression. We found that the rate of SDB during the period following aggression when there was no reconciliation was much higher than in the normal condition. Moreover, the SDB rate after reconciliation was lower than that before reconciliation, and the rate decreased to almost the same level as that during the MC. These results indicate that reconciliatory interaction reduces the anxiety or stress that the opponents suffer following aggression. Thus, the ‘uncertainty reduction’ hypothesis [Aureli and van Schaik, 1991] was confirmed in preschool children. It means that reconciliation after aggression functions to reduce the victim’s stress or anxiety. Although the sample size in the present study was not large, we would like to emphasize the utility of the ethological observation method for understanding the sociality of children in many respects. For example, the stress levels of children in natural settings have proven very difficult to estimate without the benefit of the behavioral index of stress (e.g., SDB). In addition, ethological methods of observation made it possible to examine not only the immediate outcome of aggression but also interactions after aggression, such as reconciliatory strategies, by comparing the post-aggression and control situation, or the functional aspects of reconciliation, in naturalistic settings. Future studies based on larger sample sizes using the ethological observation method (PC-MC comparison method) will help in understanding reconciliatory behavior among young children.

ACKNOWLEDGEMENT We would like to thank Ryoko Kohata, Akemi Suzuki, Kazuko Kagawa, Hiroyo Kasuda, and Hisae Kanai for permitting this study and their great cooperation at the preschool, as well as all the children and their parents. C. Hemelrijk kindly provided the program of Matrixtester, for which we are grateful.

REFERENCES Aureli F, Cords M, van Schaik CP 2002. Conflict resolution following aggression in gregarious animals: a predictive framework. Anim Behav 64:325–343. Aureli F, de Waal FBM. 2000. Natural conflict resolution. London, UK: University of California Press. Aureli F, van Schiak CP. 1991. Post-conflict behaviour in long-tailed macaques (Macaca fascicularis): II. coping with the uncertainty. Ethology 89:101–114. Aureli F, van Schiak CP, van Hooff JARAM. 1989. Functional aspects of reconciliation among captive long-tailed macaques (Macaca fascicularis). Am J Primatol 19:39–51. Butovskaya ML. 2001. Reconciliation after conflicts: ethological analysis of post-conflict interactions in Kalmyk children. In: Remirez JM, Richardson DS, editors. Cross-cultural approaches to aggression and

reconciliation. Huntington, NY: Nova Science Publishers. p 167–190. Butovskaya ML, Kozintsev AG. 1999. Aggression, Friendship, and reconciliation in Russian primary schoolchildren. Aggress Behav 25:125–139. Butovskaya ML, Verbeek P, Ljungberg T, Lunardini A. 2000. A multicultural view of peacemaking among young children. In: Aureli F, de Waal FBM, editors. Natural conflict resolution. London, UK: University of California Press. p 243–258. Castles DL, Whiten A. 1998. Post-conflict behaviour of wild olive baboons. II. stress and self-directed behaviour. Ethology 104:148–160. Cords M, Killen M. 1998. Conflict resolution in human and nonhuman primates. In: Langer J, Killen M, editors. Piaget, evolution, and development. New Jersey: Lawrence Erlbaum Associates Publishes. p 193–218.

152

Fujisawa et al.

Das M, Penke Z, van Hooff JARAM. 1998. Postconflict affiliation and stress-related behavior of long-tailed macaque aggressors. Int J Primatol 19:53–71. de Waal FBM, van Roosmalen. 1979. Reconciliation and consolation among chimpanzees. Behav Ecol Sociobiol 5:55–66. de Waal FBM, Yoshihara D. 1983. Reconciliation and redirected affection in rhesus monkeys. Behaviour 85:224–241. Dunn J, Slomkowski C. 1992. Conflict and the development of social understanding. In: Shantz CU, Hartup WW, editors. Conflict in child and adolescent development. Cambridge, UK: Cambridge University Press. p 70–92. Durkin K. 1995. Developmental social psychology. Cambridge, Mass.: Blackwell Publishing. Eisenberg AR, Garvey C. 1981. Children’s use of verbal strategies in resolving conflicts. Discourse Process 4: 149–170. Fairbanks LA, McGuire MT, Harris CJ. 1982. Nonverbal interaction of patients and therapists during psychiatric interviews. J Abnorm Psychol 91:109– 119. Hartup WW. 1992. Conflict and friendship relations. In: Shantz CU, Hartup WW, editors. Conflict in child and adolescent development. Cambridge, UK: Cambridge University Press. p 186–215. Hartup WW, Laursen B, Stewart MI, Eastenson A. 1988. Conflict and the friendship relations of young children. Child Dev 59:1590–1600. Hrubes D, Feldman RS. 2001. Nonverbal displays as indicants of task difficulty. Contemp Educ Psychol 26:267–276. Killen M, Sueyoshi L. 1995. Conflict resolution in Japanese social interactions. Early Educ Dev 6:317– 334. Killen M, Turiel E. 1991. Conflict resolution in preschool social interaction. Early Educ Dev 2:240–255. Kutsukake N, Castles DL. 2001. Reconciliation and variation in post-conflict stress in Japanese macaques (Macaca fuscata fuscata): testing the integrated hypothesis. Anim Cogn 4:259–268. Laursen B, Hartup WW. 1989. The dynamics of preschool children’s conflicts. Merrill Palmer Q 35:281–297. Ljungberg T, Westlund K, Foresberg AJL. 1999. Conflict resolution in 5–year-old boys: does postconflict affiliative behaviour have a reconciliatory role? Anim Behav 58:1007–1025. McGrew WC. 1972. An ethological study of children’s behaviour. New York: Academic Press. Martin P, Bateson P. 1985. Measuring behavior. Cambridge, UK: Cambridge University Press. Maestripieri D, Schino G, Aureli F, Troisi A. 1992. A modest proposal: displacement activities as an

indicator of emotions in primates. Anim Behav 44:967–979. Nelson J, Aboud FE. 1985. The resolution of social conflict between friends. Child Dev 56:1009–1017. Ross HS, Conant CL. 1992. The social structure of early conflict, interaction, relationships and alliances. In: Shantz CU, Hartup WW, editors. Conflict in child and adolescent development. Cambridge, UK: Cambridge University Press. p 153–185. Sackin S, Thelen E. 1984. An ethological study of peaceful associative outcomes to conflict in preschool children. Child Dev 55:1098–1102. Shantz CU. 1987. Conflicts between children. Child Dev 58:283–305. Shantz CU, Hartup WW. 1992. Conflict in child and adolescent development. Cambridge, UK: Cambridge University Press. Strayer FF, Strayer J. 1976. An ethological analysis of social agonism and dominance relations among preschool children. Child Dev 47:980–989. Strayer FF, Noel JM. 1986. The prosocial and antisocial functions of preschool aggression: an etholocial study of triadic conflict among young children. In: Zahn-Waxler C, Cummings EM, Iannotti R, editors. Altruism and aggression. Cambridge, UK: Cambridge University Press. p 107–131. Troisi A, Belsanti S, Bucci AR, Mosco C, Sinti F, Verucci M. 2000. Affect regulation in alexithymiaAn ethological study of displacement behavior during psychiatric interviews. J Nerv Ment Dis 188:12–18. Troisi A, DelleChiaie R, Russo F, Russo MA, Mosco C, Pasini A. 1996. Nonverbal behavior and alexithymic traits in normal subjects- individual differences in encoding emotions. J Nerv Ment Dis 184:561–566. Veenema HC, Das M, Aureli F. 1994. Methodological improvements for the study of reconciliation. Behav Processes 31:29–38. Vespo JE, Calpan M. 1993. Preschooler’s differential conflict behavior with friends and acquaintances. Early Educ Dev 4:45–53. Verbeek P, de Waal FBM. 2001. Peacemaking among preschool children. Journal of peace psychology 7:5– 28. Verbeek P, Hartup WW, Collins WA. 2000. Conflict management in children and adolescents. In: Aureli F, de Waal FBM, editors. Natural conflict resolution. London, UK: University of California Press. p 34–53. Walton MD, Sedlak AJ. 1982. Making amends: a grammar-based analysis of children’s social interaction. Merrill Palmer Q 28:389–412. Waxer PH. 1977. Nonverbal cue for anxiety: an examination of emotional leakage. J Abnorm Psychol 86:306–314.