Visual Preference by Chimpanzees (Pan Troglodytes) for Photos of Primates Measured by a Free Choice-Order Task: Implication for Inﬂuence of Social Experience

Primates (2003) 44: 157–165 DOI 10.1007/s10329-002-0022-8

ORIGINAL ARTICLE

Masayuki Tanaka Visual preference by chimpanzees (Pan troglodytes) for photos of primates measured by a free choice-order task: implication for inﬂuence of social experience

Received: 5 September 2002 / Accepted: 12 December 2002 / Published online: 22 February 2003 Japan Monkey Centre and Springer-Verlag 2003

Abstract With a free-choice task, visual preference was more complex central nervous system an animal has, the estimated in five adult chimpanzees (Pan troglodytes). more things that it learns would contribute to its life. The The subjects were presented with digitized color photo- present study investigated visual preference in five adult graphs of various species of primates on a CRT screen. chimpanzees (Pan troglodytes) that had been in captivity Their touching responses to the photographs were re- for many years and had close relationships with humans. inforced by food reward irrespective of which photo- Previous studies revealed that macaque monkeys graphs they touched. The results revealed that all showed differential preference for visual stimuli (Hum- chimpanzees touched the photographs of humans sig- phrey 1972, 1974; Humphrey and Keeble 1974). In nificantly more than any other species, or phylogenetic particular, macaque monkeys showed a preference for families of primates. This tendency was consistent across the visual stimuli of their own species (Sackett 1970; different stimulus sets. The results suggest that the Swartz and Rosenblum 1980). Fujita and Matsuzawa chimpanzees showed visual preference for the photo- (1986) developed a procedure to study the perceptual graphs of humans over those of their own species. The world of nonhuman animals through a sensory rein- results also suggest that the degree of this visual pref- forcement procedure. In their study, a female chimpan- erence was not in accordance with phylogenetic distance zee touched a button to see a variety of color slides. from the subjects’ species, chimpanzees. The preference Slides were presented as long as the subject kept for humans was stronger in the case of the colored touching the button. Repeated touch within 10 s after a photographs than in monochromatic ones. All of the five previous release produced the same slides again. The chimpanzees had been in captivity for at least 16 years. slide was changed if 10 s had passed after releasing the They were reared by humans from just after their birth, button. Analysis of response duration and response or at least from 1.5 years old. Their preference might interval revealed a clear difference between slides with have developed through social experience, especially that humans and those without humans, with the chimpanzee during infanthood. preferring to view the former. Using this procedure, Fujita and his colleagues (Fujita Keywords Chimpanzee Æ Free-choice Æ Sensory 1987, 1989, 1993a; Fujita and Watanabe 1995, Fujita et al. reinforcement Æ Social experience Æ Visual preference 1997) demonstrated that macaque species tend to show greater interest in the slides of monkeys of their own species. For example, Japanese macaques (Macaca fus- cata) preferred to observe slides of Japanese macaques Introduction over those of the other macaque species, such as rhesus macaques (M. mulatta). In contrast, rhesus macaques Do animals have an intrinsic preference for specific spe- preferred to observe slides of rhesus macaques over those cies, especially their own species? Alternatively, do they of the other macaques. Fujita pointed out that such dif- form their preference through social experience? The ferential interest might help prevent interbreeding among closely related species. Yoshikubo (1985, 1987) suggested M. Tanaka the ‘‘psychological reproductive isolation mechanism’’, Department of Brain and Behavioral Sciences, which means that macaques attain reproductive isolation Primate Research Institute, Kyoto University, from closely related species by means of actively choosing Kanrin, Inuyama, Aichi 484-8506, Japan E-mail: [email protected] their mate from their own gene pool. Tel.: +81-568-630548 Fujita (1990, 1993b) also revealed that social experi- Fax: +81-568-630085 ence in infanthood might influence an animal’s preference 158 for a particular species. He used the subjects with vari- subjects had already been trained in a matching-to- ously restricted social experience (i.e., either reared by sample task on the basis of identity, using the same humans, or with conspecific heterospecific peers). In this apparatus as in this study. study, rhesus macaques tended to prefer seeing rhesus macaques regardless of their age or social experience. However, Japanese macaques with restricted experience Apparatus tended to prefer to see rhesus macaques over Japanese Each chimpanzee was trained and tested in an experi- macaques. mental booth (1.8 m in width, 1.8 m in depth, and 2.0 m The present study developed a new method for in height). A 53-cm (21-inch) CRT display (Totoku, evaluating visual preference using a touch-sensitive CV213PJ; 1280·960 pixels, 24-bit color, refresh rate monitor system. The present method was based on a 85 Hz) with a touch screen (MicroTouch) was installed form of sensory reinforcement. All that the subjects on one wall of the booth. A universal feeder (Biomedica, had to do was to touch stimuli on a touch-sensitive BUF-310) delivered small pieces of food reward (apples display. The stimuli that the subject touched were or raisins) into a food tray below the display. A Pentium moved inside a frame at the top of the display. In each II-300 computer controlled the equipment by software trial of the task, the subjects were giving the chance to written in Microsoft Visual Basic ver. 5. choose three times to estimate their order of preference. Food reward was often delivered irrespective of which stimulus the subject chose, but choices were not Stimuli always reinforced in order to extinguish superstitious behaviors (e.g., a choice on the basis of position of the Stimuli were 5.6·5.6-cm digitized color images (198·198 stimuli). pixels, 24-bit color bitmap file) made from colored pho- In the present study, I aimed to investigate species tographs. There were three stimulus sets, each of which preference in chimpanzees reared by humans, but with consisted of four categories: (1) human and great apes access to social interaction with fellow chimpanzees. (four genera: Homo, Pan, Gorilla, Pongo); (2) Haplorhine This study used three different levels of biological/ (four families: Hominidae, Pongidae, Hylobatidae, phylogenetic categories – species, family, or superfamily Cercopithecidae); and (3) primate sets (five superfamilies: – in order to investigate the relationship between pre- Hominoidea, Cercopithecoidea, Ceboidea, Lemuroidea ference and phylogenetic distance. and Lorisoidea). Each category in each set consisted of ten different exemplars. Each exemplar was used in only one of the stimulus sets. Only the Hominoidea category in set 3 Experiment 1 consisted of three subcategories (Hominidae, Pongidae, Hylobatidae) of ten different exemplars each (i.e., 30 Methods exemplars). That is, sets 1 and 2 consisted of 40 exemplars, and set 3 consisted of 60 exemplars. The species used in Subjects each set are listed in Table 2. Photographs of humans did not include Japanese people whom the subjects met every The subjects were five adult female chimpanzees, day. Instead, the photographs of humans included a wide named Ai, Mari, Pendesa, Popo, and Pan (23, 23, 23, variety in terms of race, age, and sex, since the focus of the 18, and 16 years old, at the time of testing, respec- present study was preference based not on familiarity, but tively). Table 1 summarizes the profiles of the subjects. on biological category. Ai and Mari were born in Africa and received at the laboratory at 1 year of age. Pendesa was born in captivity at the Japan Monkey Center, reared by humans Procedure and received at the institute at 2 years of age. Popo and Pan were born at the Primate Research Institute, An open rectangle was presented at the top of the display. Kyoto University and reared by humans from just after A trial began with the appearance of a warning stimulus their birth, but introduced to the chimpanzee group at (gray solid square, 4·4 cm) at a random position below 11 and 9 years of age. They had previously participated the rectangle. After the subject had touched the starting in various experiments testing cognitive abilities (e.g., stimulus, 12 images were presented in 12 cells randomly Hirata & Morimura 2000; Kawai and Matsuzawa 2000; chosen within a 3 rows by 5 columns matrix on the display. Matsuzawa 2001; Tanaka 2001; Tomonaga and Ma- The 12 images consisted of 3 exemplars from the 4 cate- tsuzawa 2002). They lived with six other chimpanzees gories of a particular stimulus set. In a trial, the subject in an outdoor compound and attached indoor resi- was requested to touch three photographs (cf. Fig. 1). dence. They were not deprived of food at anytime Each touch was followed by a chime, and randomly fol- during the present study. Care and use of the chim- lowed by food reward with a probability of 60–67% ir- panzees adhered to the Guide for the Care and Use of respective of which photographs the subject touched. Laboratory Primates of the Primate Research Institute, After a touch, the photograph selected by the subject Kyoto University (2002). Before the present study, all disappeared, and the same photograph was presented 159

Table 1 Subjects and their a b proﬁles. PRI Primate Research Name Age at test Birth Age to PRI Notes Institute, Kyoto University. JMC Japan Monkey Center, Ai 23 Africa (wild) 1:01 Reared by humans together Inuyama, Aichi 484–0081, with Mari and another male Japan infant Mari 23 Africa (wild) 1:06 Moved to JMCat the age of 9 and lived in the chimpanzee group of JMC. She returned to PRIat the age of 19 Pendesa 22 JMC 2:09 Reared by humans from just after her birth. She lived with Ai, Mari and the oth erinfants after moving to PRI Popo 17 PRI 0:0 Reared by humans and lived with her brother and sister (Pan) in her infanthood She joined the adult group at the age of 10 a Pan 16 PRI 0:0 Reared byhumans and lived The age when the subject was with her brother and sister tested b (Pan) in her infanthood. The age when the subject Shejoined the adult group arrived at the Primate Research at the age of 9 Institute, Kyoto University

Table 2 The species contained in the stimulus sets

Stimulus set 1: human and great apes Category Pan Homo Gorilla Pongo

Pan troglodytes Homo sapiens Gorilla gorilla Pongo pygmaeus

Stimulus set 2: Haplorhine Category Pongidae Hominidae Hylobatidae Cercopithecidae

Pan troglodytes Homo sapiens Hylobates lar Cercocebus torquatus Pan paniscus Hylobates concolor Chlorocebus aethiops Gorilla gorilla Hylobates hoolock Cercopithecus mitis Pongo pygmaeus Hylobates agilis Macaca nemestrina Hylobates syndactylus Macaca fascicularis Papio anubis Procolobus badius Colobus guereza Pygathirix nemaeus Trachypithecus obscurus

Stimulus set 3: primates Category Hominoidea Cercopithecoidea Ceboidea Lemuroidea and Lorisoidea

Pan troglodytes Lophocebus albigena Cebus apella Eulemur mangoz Pan paniscus Chlorocebus aethiops Ateles geoﬀroyi Eulemur fulvus Gorilla gorilla Cercopithecus diana Alouata fusca Lemur catta Pongo pygmaeus Cercopithecus mona Lagothrix lagotricha Hapalemur griseus Homo sapiens Macaca thibetana Cacajao calvus Mirza coqureli Hylobates lar Macaca radiata Saimiri sciureus Verecia variegata Hylobates concolor Papio hamadryas Chiropotes satanas Indri indri Hylobates hoolock Theropithecus gelada Saguinus oedipus Nycticebus coucang Hylobates agilis Pigathrix roxellana Leontopithecus rosalia Perodicticus potto Hylobates syndactylus Trachypithecus cristatus Cebuella pygmaea Galago senegalensis inside the rectangular frame at the top of the display, is, 360 choices were recorded from each subject for each starting on the left. After three choices, the three stimuli stimulus set. In a session, stimuli from only one of the that the subject had touched were presented inside the three sets were presented. The combination of exemplars rectangle for 1 s, then the display was cleared. After a 1-s in each category and the position of the stimuli were inter-trial interval, the next trial started. One session randomly changed in each trial. The same stimulus set was consisted of 10 trials (i.e., 30 choices), and the experiment not used in consecutive sessions to avoid habituation to was continued for 12 sessions for each stimulus set. That the stimuli. 160

Fig. 1 The chimpanzee, Pan, makes her second choice in stimulus set 1 (human and great apes). The ﬁrst choice (a photograph of a human) has been placed inside the rectangle on the top of the monitor

Data Analysis

In addition to summing up the number of choices in each opportunity, each choice was scored according to the order of choice in a trial. The ﬁrst choice scored 3 points, the second choice scored 2 points, and the third choice scored 1 point. That is, the stimuli that the subjects chose early on in a trial were taken to be those that the subjects preferred. The score for each category within each stimulus set was summed up separately for the ﬁve subjects.

Results

Although the chimpanzees randomly received food reward irrespective of which photographs they touched, all the chimpanzees showed a major difference in number of Fig. 2 Mean of choices for each choice opportunity (1, 2, 3) for the choices in terms of stimulus categories. Figure 2 shows respective categories. Cerco. on the x-axis indicates Cercopitheci- dae/Cercopithecoidea category. Asterisks indicate significant mean frequency of choices for each category within the differences between the scores of two categories: *** P<0.001, respective stimulus sets. In each set, the difference in the ** P<0.01, * P<0.05 number of choices was larger in the first choices than in the following choices. In stimulus set 1, the photographs of Homo were cho- In stimulus set 2, photographs of Hominidae were sen more often than photographs from other categories. A chosen much more often than those of the other cate- Category (4) · Choice (3) ANOVA revealed that the main gories. A Category (4) · Choice (3) ANOVA revealed effect of Category was significant [F(3,48)=19.9, P<0.001] that the main effect of Category was significant and that the interaction between Category and Choice [F(3,48)=24.9, P<0.001] and that the interaction was was also significant [F(6,48)=6.19, P<0.001]. Tukey’s significant [F(6,48)=5.92, P<0.001]. Tukey’s HSD test HSD test revealed that the average number of choices of revealed that averaged number of choices of Hominidae Homo (38.3) was significantly higher than those of the (43.7) was significantly higher than those of the other other categories (Pan 32.3, P<0.05; Gorilla 22.4, categories (Pongidae: 25.0, P<0.001; Hylobatidae: 22.0, P<0.001; Pongo 26.9, P<0.001), and that the average P<0.001; Cercopithecidae: 29.2, P<>0.01). number of choices of Pan was significantly more than that In stimulus set 3, photographs of Hominoidea and of Gorilla (P<0.001). Cercopithecoidea were chosen relatively more often than 161 those of the other two categories. A Category (4) · cates the score of the category from which the chance Choice (3) ANOVA revealed that the main effect of level score was subtracted. The chance level score was Category was significant [F(3,48)=4.88, P<0.01] and calculated as follows: If the subject touched the stimuli that the interaction was significant [F(6,48)=2.44, randomly, each category should have been chosen P<0.04]. Tukey’s HSD test revealed that the average 30 times as the first choice, the second choice, and the number of choices of Hominoidea (32.1) and Cercopith- third choice over the 12 sessions. Consequently, the ecoidea (32.3) were significantly higher than that of the chance level score was 180 (i.e., 30·3+30·2+30·1). In prosimian (Lemuroidea and Lorisoidea: 27.0, P<0.02). set 1, every subject except Mari chose the photographs of Homo (i.e., human) much more often than those in Analyses based on differentiated scoring the other categories. Every subject chose photographs of Figure 3 shows the score for each category within the gorillas far below chance level. A one-way ANOVA of respective stimulus sets for each subject. Each bar indi- stimulus category was conducted and revealed that the main effect of stimulus category was significant [F(3,16)=12.7, P<0.001]. Tukey’s HSD test revealed that the score of Homo (mean 247.6) was significantly higher than that of Pan (mean: 187.0, P<0.05), Gorilla (mean: 123.8, P<0.001), and Pongo (mean: 161.6, P<0.01). Tukey’s test also revealed that the score of Pan was significantly higher than that of Gorilla (P<0.04). In set 2, every subject chose the photographs of Ho- minidae (i.e., human) much more often than those of the other categories. A one-way ANOVA of stimulus category was conducted and revealed that the main effect of stimulus category was significant [F(3,16)=16.9, P<0.0001]. Tukey’s HSD test revealed that the score of Hominidae (mean 286.0) was significantly higher than that of Pongidae (mean: 146.6, P<0.001), Hylobatidae (mean: 120.8, P<0.001), and Cercopithecidae (mean: 166.6, P<0.01). There was no difference among the scores of the other three categories. In set 3, the difference among the categories was the smallest among the three sets. Two subjects, Pendesa and Popo showed a preference for Hominoidea, but two other subjects, Ai and Pan, preferred Cercopithecoidea over Hominoidea. The last subject, Mari, showed a preference for both Hominoidea and Cercopithecoidea. A one-way ANOVA revealed that the main effect of stimulus category was significant [F(3,16)=7.23, P<0.01]. Tukey’s HSD test revealed that the scores of Hominoidea (mean: 198.8) and Cercopithecoidea (mean: 198.2) were significantly higher than that of prosimian (Lemuroidea and Lorisoidea, mean: 155.2, P<0.01). The scores of Homi- noidea and Cercopithecoidea were nearly significantly higher than Ceboidea (mean: 167.8, P<0.08).

Discussion

The results suggest that the chimpanzees had differential visual preference for specific categories in each stimulus set. In stimulus set 1, the subjects showed clear differential preference among the phylogenetically closely related four species. The results suggest that four out of Fig. 3 Standardized score of each category in the three stimulus five chimpanzee subjects tended to choose the photo- sets. Each bar indicates the score after subtraction of the chance graphs of humans and to avoid choosing those of go- level score (180) from the raw score. Cerco. on the x-axis indicates Cercopithecidae/Cercopithecoidea category. Asterisks indicate rillas. Interestingly, the subjects showed clear differential significant differences between the scores of two categories: visual preference between the photographs of chimpan- *** P<0.001, ** P<0.01, * P<0.05, + P<0.08 zees and gorillas although photographs of the two 162 closely related species looked similar to human observers Stimuli (e.g., black fur and skin). The results in set 2 clearly suggest that all subjects The general format of the stimuli was the same as in tended to choose the photographs of humans (Homini- experiment 1. The stimuli were a subset of those used as dae) more than the other three phylogenetical families. stimulus set 3 (Primates) in experiment 1. Ten images The difference in numbers of choices and scores was were used from Pongidae, Hominidae, Hylobatidae, and wider between Hominidae and its closely related families Cercopithecidae categories to create a new set: (1) (Pongidae and Hylobatidae) than between Hominidae Haplorhine – each image was processed to monochrome and Cercopithecidae. The results suggest that differential format to make another stimulus set; (2) Haplorhine visual preference may be more prominent in phyloge- (black and white); finally, the background of each image cetically more closely related families. from stimulus set 1 was erased to create the third stim- In set 3, there was no difference in the scores of ulus set; (3) Haplorhine (without background) – that is, Hominoidea and Cercopithecoidea. This may be due to each image from set 3 appeared against a white back- the difference among the three families of the Hominoi- ground, while the subject individuals remained colored. dea category. That is, although the subjects tended to choose the photographs of Hominidae, they avoided Procedure and data analysis Pongidae and Hylobatidae, much as they did in set 2. However, the subjects tended to choose the photographs The procedure was the same as in experiment 1, except of Hominoidea and Cercopithecoidea more than those for the stimuli. Data analysis was also the same as in of prosimians (i.e., Lemuroidea and Lorisoidea). The experiment 1. Each choice was counted and scored. The results suggest that the chimpanzees might not have a number of choices and the scores were summed up for preference for phylogenetically distant species as they do each category separately. for species more closely related. The chimpanzee subjects showed a consistent tendency to choose the photographs of humans in three Results different sets. This may have been due to previous long- term social experience with humans as suggested by In each condition, the order of the number of choices Fujita (1990, 1993b). However, it is possible that the showed the same pattern as in experiment 1. Table 3 photographs used in the experiment were biased to in- shows the mean number of choices in the first, second crease preference for humans. For example, almost all and third opportunity in the respective stimulus sets. photographs of nonhuman primates had a green back- Especially in stimulus set 1, the photographs of Ho- ground (i.e., tree leaves or plants), but the photographs minidae were chosen much more often than those of the of humans had various types of background and were other categories. A Category (4) · Choice (3) ANOVA thus more colorful. Although this explanation is unlikely revealed that the main effect of Category was significant because many of the monkey and prosimian photo- [F(3,48)=26.4, P<0.001] and that the interaction was graphs also appeared colorful, another experiment was also significant [F(6,48)=3.04, P<0.02]. Tukey’s HSD conducted to examine this possibility. test revealed that averaged number of choices of Ho- minidae (51.8) was significantly higher than those of the other categories (Pongidae: 23.5, P<0.001; Hylobatidae: Experiment 2 16.7, P<0.001; Cercopithecidae: 28.0, P<0.01). Although color information was deleted in stimulus Experiment 2 aimed to make clear the factors contrib- set 2, the photographs of Hominidae were still chosen uting to the preference for the photographs of humans more often than those of the other categories. A Cate- obtained in experiment 1. In experiment 2, stimulus set 3 gory (4) · Choice (3) ANOVA revealed a main effect of used in experiment 1 was used again for the other cat- Category was significant [F(3,48)=9.48, P<0.001] but egory level (i.e., Haplorhine) in order to examine that the interaction was not significant [F(6,48)=0.731, whether the same results as in experiment 1 would be n.s.]. Tukey’s HSD test revealed that the average num- obtained with the different set, within which the subjects ber of choices of Hominidae (39.9) was significantly showed clear preference to a specific category. In addi- higher than those of the other categories (Pongidae: 26.1, tion, in experiment 2, processed stimuli (i.e., black-and- P<0.001; Hylobatidae: 23.9, P<0.001; Cercopithecidae: white, without background) were used to examine the 30.1, P<0.03). effects of perceptual features. Similarly, although the background had been deleted in the photographs of stimulus set 3, the photographs of Hominidae were again chosen much more often than the Methods other categories in stimulus set 3. A Category (4) · Subjects and apparatus Choice (3) ANOVA revealed that the main effect of Category was significant [F(3,48)=17.3, P<0.001] but The subjects and the apparatus were the same as in that the interaction was not significant [F(6,48)=1.58, experiment 1. n.s.]. Tukey’s HSD test revealed that the average num- 163

Table 3 The mean number of choices in the ﬁrst, second, and Categories of stimulus third opportunity Stimulus set 1 Pongidae Hominidae Hylobatidae Cercopithecidae

First choice 22.0 64.2 11.4 22.4 Second choice 24.0 52.4 18.4 25.2 Third choice 24.4 38.8 20.4 36.4

Stimulus set 2 Pongidae Hominidae Hylobatidae Cercopithecidae

First choice 23.6 42.0 24.2 30.2 Second choice 27.8 42.6 22.8 26.8 Third choice 26.8 35.0 24.8 33.4

Stimulus set 3 Pongidae Hominidae Hylobatidae Cercopithecidae

First choice 24.8 58.2 12.4 24.6 Second choice 27.0 46.6 17.4 29.0 Third choice 28.8 38.2 20.2 32.8

ber of choices of Hominidae (47.7) was significantly higher than those of Pongidae (mean: 153.2, P<0.05) and higher than those of the other categories (Pongidae: 26.9, Hylobatidae (mean: 143.0, P<0.03), but not significantly P<0.001; Hylobatidae: 16.7, P<0.001; Cercopithecidae: different from that of Cercopithecidae (mean: 177.6). 30.1, P<0.001). In set 3, the results were very similar to those in set 1. That is, four of the five subjects chose the photographs of Hominidae more often than those of the other cate- Analyses based on differentiated scoring gories. A one-way ANOVA of stimulus category was conducted and revealed that the main effect of stimulus The score of Hominidae was the highest in most of the category was significant [F(3,16)=8.09, P<0.01]. Tukey’s stimulus sets and most of the subjects. Figure 4 shows HSD test revealed that the score of Hominidae (mean: the score for each subject. Each bar indicates the score 306.0) was significantly higher than those of Pongidae of the category from which the chance level score was (mean: 157.2, P<0.05), Hylobatidae (mean: 92.2, subtracted. Chance level was calculated in the same way P<0.01), and Cercopithecidae (mean: 164.6, P<0.05). as in experiment 1. In set 1, the score of the Hominidae There was no difference among the scores of the other category (i.e., human) was much higher than those of the three categories. other categories for every subject. A one-way ANOVA of stimulus category was conducted and revealed that the main effect of stimulus category was significant Discussion [F(3,16)=8.09, P<0.01]. Tukey’s HSD test revealed that the score of Hominidae (mean: 336.2) was significantly The results in experiment 2 revealed that the visual higher than those of Pongidae (mean: 138.4, P<0.01), preference for the photographs of humans shown in Hylobatidae (mean: 91.4, P<0.001), and Cercopitheci- experiment 1 was not limited for the particular stimulus dae (mean: 154.0, P<0.01). There was no difference sets, but was consistent across the stimulus sets. The among the scores of the other three categories. The results for stimulus set 1 were the almost same as those results revealed that the subjects tended to choose in stimulus set 2 in experiment 1 although the stimulus photographs of humans (i.e., Hominidae). sets were different. The results revealed that the subjects In the case of stimulus set 2, only one subject, Pan, had a consistent tendency to choose the photographs of showed a very clear tendency to choose photographs of humans (i.e., Hominidae) more often than those of the Hominidae. Four of the five subjects did not show a clear other categories. Concerning the order of number of tendency to choose photographs from a specific category. choices, Hylobatidae were behind Cercopithecidae, which However, three chimpanzees, Mari, Pendesa, and are phylogenetically more distant from the subjects’ Popo, chose photographs in the Hominidae category more species (Pan troglodytes). This pattern also seemed often than those in the other categories. A one-way consistent across the stimulus sets. ANOVA of stimulus category was conducted and The results for stimulus set 2 suggest that color in- revealed that the main effect of stimulus category was formation may have influenced the preference for pho- significant [F(3,16)=4.20, P<0.05]. Tukey’s HSD test tographs of humans. However, the results of set 2 revealed that the score of Hominidae (mean: 246.2) was showed a similar tendency as that seen in set 1. Even 164

1. Every subject tended to choose photographs of humans more often than any other category of primates. The preference was consistent across different stimulus sets varying in color and in background. 2. The degree of visual preference was not in accordance with phylogenetic distance from chimpanzees. 3. The subjects’ preference for the photographs of humans was reduced in the case of monochromatic photographs in comparison with colored ones.

The results thus did not show that the chimpanzees had a visual preference for the photographs of their own species. This is in contrast with the findings of Fujita and his colleagues (Fujita 1987, 1990; Fujita and Watanabe 1995, Fujita et al. 1997). It is likely that the results in the present study were partially owing to the subjects’ history. Fujita (1990, 1993b) suggested that social experience in infanthood might influence preference. In his studies, preference for conspecifics disappeared in Japanese macaques that were raised with other species. This means that the preference for one’s own species may be acquired through social experience in infanthood in Japanese macaques. Every subject in the present study had been raised from infancy by human caretakers for many years and had taken part in various behavioral and cognitive studies. That is, apart from other members of their captive chimpanzee community, humans were the species most familiar to the subjects. Such social experience may have promoted preference for the photographs of humans. At present, there are three chimpanzee infants at the Primate Research Institute, all of which are being raised by their mothers. These infants will be tested for visual preference in the future, in order to clarify the influence of social experience in infanthood. The present study revealed that there was a difference within the Pongidae category, to which the subjects be- Fig. 4 Standardized score of each category in three stimulus sets. longed. In particular, the chimpanzees chose the pho- Each bar indicates the score after subtraction of the chance level tographs of chimpanzees significantly more often than score (180) from the raw score. Cerco. on the x-axis indicates those of gorillas, their closest phylogenetical relatives. Cercopithecidae/Cercopithecoidea category. Asterisks indicate sig- The score of orangutans was intermediate between nificant differences between the scores of two categories: *** P<0.001, ** P<0.01, * P<0.05 chimpanzees and gorillas. Thus, the subjects’ preference did not correspond to phylogenetic classification on the basis of morphological similarity to humans (i.e. tax- with the background deleted, the subjects’ preference for onomy). In other words, their preference was not gen- the photographs of humans was not reduced at all. The eralized as a function of morphological similarity. results suggest that the color information that had an However, the subjects tended to choose the photographs effect on this preference was contained not in the of Hominoidea and Cercopithecoidea more often than background but in the humans themselves. those of Ceboidea, or prosimians in experiment 1. The results suggest that visual preference in the chimpanzees might be contrastive among phylogenetically close spe- General discussion cies, but indifferent among phylogenetically distant species. Tomonaga et al. (1993) reported that a chim- With a free-choice task, five adult chimpanzees were panzee was able to name the conspecifics more accu- examined for visual preference for various levels of rately than humans. Ueno (1994) revealed that tufted categories within the order primates. The present study capuchins were able to discriminate odors of various revealed the following three points. species of New World monkeys more accurately than 165 those of Old World monkeys. These results suggest that Fujita K (1993a). Role of some physical characteristics in species primates may have a mechanism for emphasizing dif- recognition by pigtail monkeys. Primates 34:133–140 Fujita K (1993b) Development of visual preference for closely references between phylogenetically closer species. lated species by infant and juvenile macaques with restricted The results in the present study suggest that color social experience. Primates 34:141–150 information contained in the humans themselves may Fujita K, Matsuzawa T (1986) A new procedure to study the have an effect on preference by the chimpanzees. The perceptual world of animals with sensory reinforcement: recognition of humans by a chimpanzee. Primates 27:283–291 photographs of humans included a wide variety in terms Fujita K, Watanabe K (1995) Visual preference for closely related of race, age, and sex, since the focus of the present study species by Sulawesi macaques. Am J Primatol 37:253–261 was preference based not on familiarity, but on biolog- Fujita K, Watanabe K, Widarto TH, Suryobroto B (1997) Dis- ical category. Therefore, the ‘‘human’’ category had a crimination of macaques: the case of Sulawesi species. Primates much wider variation in color of hair, skin, or clothes 38:233–245 Hirata S, Matsuzawa T (2001) Tactics to obtain a hidden food item than the great apes. This variation might have promoted in chimpanzee pairs (Pan troglodytes). Anim Cogn 4:285–295 visual preference in the subjects. Although the Ceroco- Humphrey NK (1972) ‘‘Interest’’ and ‘‘pleasure’’: Two determi- pithecidae/Cercropithecoidea, Ceboidea, and prosimian nants of a monkey visual preference. Perception 1:395–416 categories also contained a wide variety in terms of the Humphrey NK (1974) Species and individuals in the perceptual world of monkeys. Perception 3:105–114 color of fur, the scores of these categories were much Humphrey NK, Keeble GR (1974) The reaction of monkeys to lower than that of humans. Therefore, wide variation in ‘‘fearsome’’ pictures. Nature 251:500–502 color within a category is not sufficient to elicit visual Itakura S, Tanaka M (1998) Use of experimenter-given cues during preference. object-choice tasks by chimpanzees (Pan troglodytes), an orangutan (Pongo pygmaeus) and human infants (Homo sapi- There may have been another factor promoting ens). J Comp Psychol 112:119–126 preference for the photographs of humans. In almost all Kawai N, Matsuzawa T (2000) Numerical memory span in a the photographs, humans appeared to gaze at the sub- chimpanzee. Nature 403:39–40 jects. Kobayashi and Koshima (1997, 2001) found that Kobayashi H, Koshima S (1997) Unique morphology of the human human eyes were exceptionally conspicuous in both eye. Nature 387:767–768 Kobayashi H, Koshima S (2001) Evolution of the human eye as a shape and coloration. They suggested that human eyes device for communication. In Matsuzawa T (ed) Primate ori- had evolved to enable humans to use gaze signals in gins of human cognition and behavior. Springer, Tokyo Berlin communicating with conspecifics. Previous studies have Heidelberg, pp 383–401 shown that chimpanzees are also able to use human gaze Matsuzawa T (ed) (2001) Primate origins of human cognition and behavior. Springer, Tokyo Berlin Heidelberg signals—this has been demonstrated not only in adult Okamoto S, Tomonaga M, Ishii K, Kawai N, Tanaka M, Ma- great apes (Itakura and Tanaka 1998) but also in a tsuzawa T (2002) An infant chimpanzee (Pan troglodytes) fol- 1-year-old infant (Okamoto et al. 2002). These previous lows human gaze. Anim Cogn 5:107–114 findings lead to the hypothesis that human eye gaze Sackett GP (1970) Unlearned responses, differential rearing expe- riences, and the development of social attachments by rhesus might be ‘‘supernatural’’ and conspicuous for great apes, monkeys. In: Rosenblum LA (ed) Primate behavior: Develop- especially human-reared enculturated individuals. ments in fields and laboratory research, vol. 1. Academic Press, New York, pp 111–140 Acknowledgements The present study was financially supported by Swartz KB & Rosenblum LA (1980) Operant responding by bonnet Grants-in-Aid for Scientific Research, Ministry of Education, macaques for color videotaped recordings of social stimuli. Science, and Culture 09207105, 12002009, 12710037, and Anim Learn Behav 8:311–321 10CE2005. I wish to thank Drs. T. Matsuzawa, K. Fujita, and M. Tanaka M (2001) Discrimination and categorization of photo- Tomonaga for their valuable comments, Dr. D. Biro for her graphs of natural objects by chimpanzees (Pan troglodytes). reading of and comments on earlier drafts, and Mr. S. Nagumo for Anim Cogn 4:201–211 his technical help. I also thank the staff members of the Center for Tomonaga M, Itakura S, Matsuzawa T (1993) Superiority of Human Evolution Modeling Research, Primate Research Institute, conspecific faces and reduced inversion effect in face recognition Kyoto University for caring the chimpanzees. by a chimpanzee. Folia Primatol 61:110–114 Tomonaga M, Matsuzawa T (2002) Enumeration of briefly presented items by the chimpanzee (Pan troglodytes) and humans (Homo sapiens). Anim Learn Behav 30:143–157 References Ueno Y (1994) Olfactory discrimination of urine odors from five species by tufted capuchin (Cebus apella). Primates 35:311–323 Yoshikubo S (1985) Species discrimination and concept formation Fujita K (1987) Species recognition by five macaque monkeys. by the rhesus monkeys (Macaca mulatta). Primates 26:285–299 Primates 28:353–366 Yoshikubo S (1987) A possible reproductive isolation through a Fujita K (1990) Species preference by infant macaques with con- species discrimination learning in genus Macaca. Primate Res trolled social experience. Int J Primatol 11:553–573 3:43–47 (in Japanese with English summary)