Поиск по сайту




Пишите нам: info@ethology.ru

Follow etholog on Twitter

Система Orphus

Новости
Библиотека
Видео
Разное
Кросс-культурный метод
Старые форумы
Рекомендуем
Не в тему

список статей


Effects of belief in genetic relatedness on resemblance judgments by Japanese raters

Ryo Odaa, Akiko Matsumoto-Odab, Osamu Kurashimac

1. Introduction

2. Methods

2.1. Materials

2.2. Participants

2.3. Procedure of Experiment 1

2.4. Statistical analysis for Experiment 1

2.5. Procedure of Experiment 2

3. Results

3.1. Resemblance ratings

3.1.1. Experiment 1

3.1.2. Experiment 2

3.2. Reaction time

3.2.1. Experiment 1

3.2.2. Experiment 2

3.3. Signal detection analysis

4. Discussion

Acknowledgment

References

Copyright

1. Introduction

Kin selection (Hamilton, 1964) is important to behavioral evolution. For kin selection to operate, animals must discriminate between kin and nonkin (Fletcher & Michener, 1987). Family members often resemble each other because of similarities in genotype, and phenotype matching is one of several methods used to recognize kin.

Humans have a prolonged childhood, as compared with other primates, which makes male parental investment (Trivers, 1972) especially important. Female ovulation is concealed, however, thus, males are unable to assess paternity directly and might instead rely on indirect cues such as phenotype matching. Facial resemblance is an important cue, because the face is important to humans in the process of individual recognition (Bruce, 1988). A putative father may be able to assess his paternity via facial similarity (Alexander, 1974). Research into the facial resemblances between children and parents (Bredart & French, 1999, Bressan & Grassi, 2004, Christenfeld & Hill, 1995, Daly & Wilson, 1982, McLain et al., 2000, Nesse et al., 1990, Oda et al., 2002, Regalski & Gaulin, 1993) has examined this possibility. For humans, kinship terminology is another important cue in kin recognition. Languages typically have specific kinship terms that correspond to biological kin relationships (Brown, 1991). Such linguistic information about kinship also might affect judgments of resemblance.

Bressan and Dal Martello (2002) tried to remove this confounding factor by separating true genetic relatedness from a belief in genetic relatedness. They showed the participants pairs of photographs depicting Italian adults and children and asked them to assess their resemblance. Half of the photographs were actual parent and child pairs, while the other half were not genetically related. Orthogonally, each photograph was labeled to indicate whether the two individuals were genetically related: Half of the related and half of the unrelated pairs were labeled as related. Bressan and Dal Martello found that resemblance evaluation was affected both by the actual and the labeled genetic relationship. Inasmuch as humans have adapted to their environments through a variety of cultural mechanisms, we replicated the approach of Bressan and Dal Martello, but we used Japanese people as stimuli and as participants.

2. Methods

We followed the methodology of Experiments 1 and 2 in Bressan and Dal Martello (2002), except that we used computers to display stimuli and to capture reaction times. We used 7-point scales to rate similarities, due to a limitation in the experimental software, while Bressan and Dal Martello used an 11-point scale.

2.1. Materials

We took black-and-white photographs of the faces of 28 Japanese children in a kindergarten school and of their parents. The children were 4 to 5 years old (14 girls and 14 boys). The photographs were digitized and processed to control some of the features that might affect assessment. We omitted any body part below the neck, as well as ears and hair, and we shaded off the outline of each face.

We made four copies of each of the 28 children's photographs and two copies of each of the 56 parents' photographs, and the 224 resulting pictures were coupled to form a set of 112 child–adult pairs and saved as PICT files. Each pair was labeled in Japanese, indicating either that the two individuals were genetically related (“mother and daughter,” “mother and son,” “father and daughter,” and “father and son”) or that they were unrelated (“nonrelatives”). These 112 parent–child pairs were grouped to produce 28 each of four types of stimuli: related and so labeled; related and labeled as unrelated; unrelated and so labeled; and unrelated and labeled as related. Each unrelated pair was obtained by randomly pairing a child with another child's parent. Although the 112 pairs of photographs were the same for all participants, photo labels were counterbalanced; that is, two versions of the experiment were prepared so that each photo pair was paired with both related and unrelated labels. These yielded the two groups in Experiment 1.

2.2. Participants

Sixty participants (30 men, 30 women), ranging in age from 18 to 25 years old (median age: 20), were recruited on the campuses of two universities for Experiment 1. We recruited a new sample of 60 participants (30 men, 30 women), ranging in age from 18 to 26 years old (median age: 20), on the same campuses for Experiment 2. All participants were native Japanese speakers.

2.3. Procedure of Experiment 1

The participants were divided into two groups (15 men, 15 women); one group was allocated to Experimental Group 1 and the other to Experimental Group 2. Each block of photographs was presented on an 18-in. liquid crystal monitor (EIZO FlexScan L685EX, resolution: 1280×960), and the presentation was software controlled (Cedrus SuperLab Ver. 1.4). The label, the adult's photograph, and the child's photograph were arranged lengthwise, in that order. The order of presentation of photograph pairs was randomized for each rater.

The participants were asked to read aloud the label that accompanied each pair of photographs and then to judge the degree of facial similarity between the child and the adult. Ratings were made on a scale of one to seven. Once rating for a pair was indicated by a keyboard input, the next pair of photographs was presented. There was no time limit, and participants were instructed to complete the task at their own pace.

2.4. Statistical analysis for Experiment 1

Because a preliminary analysis of variance (ANOVA) showed no significant effect of group (F<1), the data from the two groups were pooled. We evaluated the effects on resemblance ratings of the rater's gender (between-subjects variable) and relatedness between adult and child, as well as label, gender of depicted adult, and gender of depicted child (all within-subjects variables). We calculated the mean estimated resemblance for each of 16 combinations of within-subjects variables, and a five-way ANOVA was performed.

2.5. Procedure of Experiment 2

The procedure was the same as in Experiment 1, except that there were no labels on the stimulus pairs. Participants were asked to rate the perceived resemblance of the pair members on a scale of one to seven.

3. Results

3.1. Resemblance ratings

3.1.1. Experiment 1

The F ratios and p values of the main effects and interactions are in Table 1. There was a significant effect on resemblance ratings, of both genetic relationship and the belief that there was a genetic relationship, but no significant interaction (Fig. 1). Children were judged to look more similar to their parents (M=3.81) than to unrelated adults (M=3.39). Children were also judged to look more similar to adults when they were labeled as related (M=3.92) than when they were labeled as unrelated (M=3.28). The effect size was larger for label (d=0.74) than for genetic relatedness (d=0.47).

Table 1.

F ratios and p values of main effects and interactions of the five-way ANOVA

Factor Experiment 1 Experiment 2
Fa p Fa p
Main effects Gender of rater (GR) 2.56 .1151 0.89 .3491
Relatedness (RE) 160.29 <.0001 128.42 <.0001
Label (LA) 61.05 <.0001
Gender of adult (GA) 8.39 .0053 8.82 .0043
Gender of child (GC) 13.31 .0006 0.879 .3523
Interactions GR×RE 4.54 .0373 2.76 .1022
GR×GA 0.37 .5468 0.04 .8506
GR×GC 2.73 .1037 0.03 .8628
RE×GA 0.34 .5645 0.94 .3352
RE×GC 59.66 <.0001 34.01 <.0001
GA×GC 20.47 <.0001 32.29 <.0001
GR×LA 0.08 .7747
RE×LA 0.30 .5860
LA×GA 1.24 .2700
LA×GC 1.13 .2926
GR×RE×GA 0.13 .7252 0.29 .5949
GR×RE×GC 0.37 .5445 0.46 .5009
GR×GA×GC 0.11 .7364 0.12 .7320
RE×GA×GC 26.84 <.0001 7.79 .0071
GR×LA×GA 0.01 .9189
GR×LA×GC 1.10 .2988
RE×LA×GA 1.31 .2566
RE×LA×GC 0.31 .5826
GR×RE×LA 0.32 .5753
LA×GA×GC 1.88 .1751
GR×RE×GC×GA 0.02 .8846 0.41 .5240
GR×RE×LA×GA 1.02 .3171
GR×RE×LA×GC 1.07 .3055
GR×GA×GC×LA 0.26 .6125
LA×GA×GC×RE 0.53 .4712
LA×GA×GC×RE×GR 0.22 .6442
a

df=1,58.


View full-size image.

Fig. 1. Mean estimated resemblance as a function of genetic relatedness (child/parent vs. child/nonparent) and belief in relatedness (“related” label vs. “unrelated” label) in Experiment 1. Thick lines show the mean estimated resemblance in Experiment 2 (when there was not information about relatedness). Bars indicate the standard error of the mean.


There were also significant effects of both adult and child gender. Children were judged to look more similar to adult females (M=3.67) than to adult males (M=3.53). Boys were judged to more closely resemble adults (M=3.67) as compared with girls (M=3.54). The interaction of those two factors was also significant. There was also a significant interaction between the gender of the child and relatedness, and there was a significant three-way interaction (Table 1). As Fig. 2 indicates, when comparing children with their parents, boys were judged to more closely resemble their fathers than were girls, while there were no differences in the other combination.


View full-size image.

Fig. 2. Mean estimate resemblance as a combination of genetic relatedness (child/parent vs. child/nonparent), gender of adult, and gender of child in Experiment 1. Bars indicate the standard error of the mean.


Finally, our data show no significant effect of rater's gender and no significant interactions with rater's gender (Table 1).

3.1.2. Experiment 2

As in Experiment 1, children were judged to more closely resemble their parents (M=4.09) than unrelated adults (M=3.67) and were judged to be more similar to adult females (M=3.94) than to adult males (M=3.82). We also found significant interactions between the gender of adults and gender of children and between relatedness and the gender of the children. Moreover, interactions among those three variables were also significant (Table 1). Whether or not stimulus pairs were labeled, boys were judged to more closely resemble their fathers than were the girls (Fig. 3).


View full-size image.

Fig. 3. Mean estimate resemblance as a combination of genetic relatedness (child/parent vs. child/nonparent), gender of adult, and gender of child in Experiment 2. Bars indicate the standard error of the mean.


3.2. Reaction time

3.2.1. Experiment 1

The mean reaction time was 7.6 s (range: 2.2–40.3 s, S.D.=4.4). Using the same analysis strategy on reaction times as on resemblance ratings revealed that only the main effect of label was significant [F(1,58)=20.7, p<.0001]. Reaction times for the pairs labeled as parent and child (M=8.2 s) were longer than those for the pairs labeled as nonrelatives (M=7.1 s). We considered that the time to read the labels might influence assessment speed; however, each of the five kinds of labels revealing relationships consisted of six Japanese moras (“mother and daughter” is “ha-ha-to-mu-su-me,” “mother and son” is “ha-ha-to-mu-su-ko,” “father and daughter” is “chi-chi-to-mu-su-me,” “father and son” is “chi-chi-to-mu-su-ko,” and “nonrelatives” is “ta-ni-n-do-u-shi), and pronunciation time should be almost the same.

There was no significant interaction between label and genetic relatedness.

3.2.2. Experiment 2

The mean reaction time was 6.8 s (range: 1.9–21.8 s, S.D.=2.9). There were no significant effects on the reaction time.

3.3. Signal detection analysis

Following Bressan and Dal Martello (2002), we performed separate signal detection analyses (Gescheider, 1997) for each experiment. We dichotomized the resemblance ratings by considering ratings from one to three as “low” and ratings from four to seven as “high.” A hit rate (how often each rater estimated “high” when the pairs were genetically related) and a false alarm rate (how often each rater estimated “high” when the pairs were unrelated) were calculated.

The values of d′ (the sensitivity parameter) varied little across conditions: 0.59 when the pairs were labeled as related, 0.60 when they were labeled as unrelated, and 0.62 when there was no label. However, the beta (bias parameter) was different among the three trials: 1.01 when the photos were labeled as related, 1.52 when the photos were labeled as unrelated, and 1.05 when there was no label. The increasing pattern was the same as in Bressan and Dal Martello (2002).

4. Discussion

Most of our results replicated those of Bressan and Dal Martello (2002), although the children that we used in our photographs were roughly half as old (Table 2). In particular, we replicated all the main effects found by Bressan and Dal Martello, and some of the interactions. Our results indicated that both genetic and labeled relatedness independently affected resemblance ratings. Children were judged to be more similar to their labeled parents than to labeled nonrelatives, regardless of the actual genetic relationship. A larger effect size for label than for relatedness suggests that assertions about a relationship can affect adult–child resemblance judgments.

Table 2.

Comparison of the methods and results between two studies

Bressan and Dal Martello (2002) This study
Methods
Materials 10 Italian families 28 Japanese families
Child age: 8 years old Child age: 4 to 6 years old
Color, Not modified B/W, Modified
Participants 60 Italians (30 men, 30 women) 60 Japanese (30 men, 30 women)
19–65 years old 18–26 years old
Procedure Pictures shown in albums Pictures shown on computer display
Rating scale: 0–10 Rating scale: 1–7
Results
Effects of genetic relatedness and labels Both genetic relatedness and labels significantly affect the ratings. Both genetic relatedness and labels significantly affect the ratings.
The rating of the pairs without labels fell exactly between the mean values of those with “related” and “unrelated” labels. The rating of the pairs without labels was closer to that of the pairs with “related” labels.
Effects of gender Children were judged to look more similar to the adult female. Children were judged to look more similar to the adult female.
Girls were more similar to their mother. Boys were more similar to their father.
There was a marginally significant effect on gender of rater. There was no significant effect on gender of rater.

As presented in Fig. 1, photo pairs that were unlabeled were judged almost as similar as pairs labeled as related, whereas Bressan and Dal Martello (2002) found that unlabeled pairs were judged as intermediate in resemblance between labeled relative and labeled nonrelatives. In other words, we found that belief in nonrelatedness decreased the estimated resemblance among pair members more than a belief in their relatedness increased it.

It is surprising that genetically related but allegedly unrelated pairs were judged less similar to each other than genetically unrelated pairs (Fig. 1). The possible underlying process is suggested by reaction time and signal detection analyses. Reaction time was shorter when the pairs were labeled unrelated than when the individuals were allegedly related. While the sensitivity parameters were almost equivalent for photos labeled as related, unrelated, and unlabeled, the bias parameter when photos were labeled unrelated was the largest among the three conditions and greater than 1.0. This implies a large degree of conservative bias when assessing the resemblance between two allegedly unrelated individuals. Participants might have employed a rule of thumb that enabled them to judge more efficiently, such that if two individuals were labeled as unrelated, they were most likely to be dissimilar.

Children were judged to more closely resemble women than men. This replicates Bressan and Dal Martello (2002), but we also found that boys were judged to be more similar to their fathers than were girls. This occurred in assessments both with and without labels. One explanation is that we might have sampled father–son pairs who resemble each other more than general ones. Our stimulus photographs were of children in kindergarten, taken on a parents' visiting day. We informed parents of the purposes of our study and asked their permission to take photographs of both them and their children. For this reason, cases of doubted paternity might be underrepresented in our sample. Another possible reason for this effect is that there was some kind of inherited feature common among fathers and their sons (e.g., a Y-chromosome trait) that participants used in their judgments. According to Bressan and Dal Martello, Italians hold a popular belief that girls take after fathers and boys after their mothers. Interestingly, Japanese people also have the same belief (Oda et al., 2002), but both our results and that of Bressan and Dal Martello inconsistent with this cultural belief.

In conclusion, despite some methodological differences, our study largely replicates the results of Bressan and Dal Martello (2002) in a very different cultural group.

Acknowledgment

We would like to thank Teranishi Kindergarten School and the participant families for their help, and two anonymous reviewers and S. Gaulin for their helpful comments.

References

Alexander, 1974 1.Alexander RD. The evolution of social behavior. Annual Review of Ecology and Systematics. 1994;5:325–383.

Bredart & French, 1999 2.Bredart S, French RM. Do babies resemble their fathers more than their mothers? A failure to replicate Christenfeld and Hill (1995). Evolution and Human Behavior. 1999;20:129–135.

Bressan & Dal Martello, 2002 3.Bressan P, Dal Martello MF. Talis pater, talis filius: Perceived resemblance and the belief in genetic relatedness. Psychological Science. 2002;13:213–218. MEDLINE

Bressan & Grassi, 2004 4.Bressan P, Grassi M. Parental resemblance in 1-year-olds and the Gaussian curve. Evolution and Human Behavior. 2004;25:133–141.

Brown, 1991 5.Brown D. Human universals. New York: McGraw-Hills; 1991;.

Bruce, 1988 6.Bruce V. Recognizing faces. Mahwah, NJ: Lawrence Erlbaum; 1988;.

Christenfeld & Hill, 1995 7.Christenfeld N, Hill E. Whose baby are you?. Nature. 1995;378:669. MEDLINE | CrossRef

Daly & Wilson, 1982 8.Daly M, Wilson M. Whom are newborn babies said to resemble?. Ethology and Sociobiology. 1982;3:69–78.

Fletcher & Michener, 1987 9.In:  Fletcher DJC,  Michener CD editor. Kin recognition in animals. Chichester: Wiley; 1987;.

Gescheider, 1997 10.Gescheider GA. Psychophysics: The fundamentals. 3rd ed. Mahwah, NJ: Laurence Erlbaum; 1997;.

Hamilton, 1964 11.Hamilton WD. The genetical evolution of social behaviour I, II. Journal of Theoretical Biology. 1964;7:1–52. MEDLINE | CrossRef

McLain et al., 2000 12.McLain DK, Setters D, Moulton MP, Pratt AE. Ascription of resemblance of newborns by parents and nonrelatives. Evolution and Human Behavior. 2000;21:11–23.

Nesse et al., 1990 13.Nesse RM, Silverman A, Bortz A. Sex differences in ability to recognize family resemblance. Ethology and Sociobilogy. 1990;11:11–21.

Oda et al., 2002 14.Oda R, Matsumoto-Oda A, Kurashima O. Facial resemblance of Japanese children to their parents. Journal of Ethology. 2002;20:81–85.

Regalski & Gaulin, 1993 15.Regalski JM, Gaulin SJ. Whom are Mexican infants said to resemble? Monitoring and fostering paternal confidence in the Yucatan. Ethology and Sociobiology. 1993;14:97–113.

Trivers, 1972 16.Trivers R. Parental investment and sexual selection. In:  Campbell B editors. Sexual selection and the descent of man. Chicago: Aldine de Gruyter; 1972;p. 136–179.

a Graduate School of Engineering, Nagoya Institute of Technology, Gokiso-cho, Showa-ku, Nagoya 466-8555, Japan

b Department of Welfare and Culture, Okinawa University, 555 Kokuba, Naha 902-8521, Japan

c Graduate School of Agricultural and Life Sciences, University of Tokyo, 1-1-1 Yayoi, Bunkyo-ku, Tokyo 113-8657, Japan

Corresponding author. Tel/fax: +81 52 735 5112.

PII: S1090-5138(05)00022-X

doi:10.1016/j.evolhumbehav.2005.04.002



2007:11:26