Anim Cogn (2015) 18:713–721 DOI 10.1007/s10071-015-0838-4

ORIGINAL PAPER

Giant pandas failed to show self-recognition

Xiaozan Ma • Yuan Jin • Bo Luo • Guiquan Zhang • Rongping Wei • Dingzhen Liu

Received: 16 July 2014 / Revised: 4 January 2015 / Accepted: 6 January 2015 / Published online: 22 January 2015 Ó Springer-Verlag Berlin Heidelberg 2015

Abstract Mirror self-recognition (MSR), i.e., the ability recognition in large , provide new information on to recognize oneself in a mirror, is considered a potential a solitary species, and will be useful for enclosure design index of self-recognition and the foundation of individual and captive animal management. development. A wealth of literature on MSR is available for social animals, such as , Asian Keywords Giant pandas Á Mirror self-recognition Á Age and , yet little is known about MSR in solitary difference Á Solitary animals Á Self-cognition mammalian species. We aimed to evaluate whether the can recognize itself in the mirror, and whether this capacity varies with age. Thirty-four captive giant Introduction pandas (F:M = 18:16; juveniles, sub-adults and adults) were subjected to four mirror tests: covered mirror tests, The ability to recognize oneself in a mirror, or mirror self- open mirror tests, water mark control tests, and mark tests. recognition (MSR), is an informative index of self-recog- The results showed that, though adult, sub-adult and nition and (Liu and Yang 2009; Yang 2005). The juvenile pandas exposed to spent similar amounts MSR paradigm has been applied to evaluate self-awareness of time in social mirror-directed behaviors (v2 = 0.719, in a wide range of species (Gallup 1970, 1994; Sua´rez and P = 0.698), none of them used the mirror to touch the Gallup 1981), although this paradigm has been questioned mark on their head, a self-directed behavior suggesting by scientists (Heyes 1994, 1995). Previous research has MSR. Individuals of all age groups initially displayed shown that animals with relatively large brains have attacking, threatening, foot scraping and backwards walk- complex cognitive skills (Lefebvre et al. 2004) and are ing behaviors when exposed to their self-images in the likely to show MSR. The ability to recognize oneself in a mirror. Our data indicate that, regardless of age, the giant mirror is often assessed empirically by exposing animals pandas did not recognize their self-image in the mirror, but previously marked on the head (or elsewhere, in a position instead considered the image to be a conspecific. Our they can see only in a mirror) to a mirror, and assessing results add to the available information on mirror self- their behavior. Animals that possess MSR typically and progressively display four stages of behavior when they face a mirror: (1) non-mirror behaviors; (2) mirror-directed X. Ma Á Y. Jin Á D. Liu (&) behaviors; (3) mirror-guided behaviors; and (4) mark- Key Laboratory of Biodiversity Science and Ecological directed or self-directed behaviors (e.g., touching the mark Engineering of Ministry of Education, College of Life Sciences, on their body, thus demonstrating that they can see it and Beijing Normal University, Beijing 100875, China e-mail: [email protected] realize that it is on themselves) (Allen and Schwartz 2008; Plotnik et al. 2006). Since the pioneering work of Gallup B. Luo Á G. Zhang Á R. Wei (1970), who subjected four chimpanzees (Pan troglodytes) Key Laboratory for Reproduction and Conservation Genetics of to a MSR test, research has been published on MSR in Endangered Wildlife of Sichuan Province, China Conservation and Research Center for the Giant Panda, Wolong, Sichuan, infants (Amsterdam 1972; Brooks-Gunn and Lewis China 1979; Mans et al. 1978), chimpanzees (Calhoun and 123 714 Anim Cogn (2015) 18:713–721

Thompson 1988; Lin et al. 1992) and (Pongo between panda-like eye-mask patterns that differ subtly in pygmaeus) (Lethmate and Du¨cker 1973). Although scien- shape (Dungl et al. 2008). These findings indicate that giant tists have tested for MSR in rhesus monkeys (Macaca pandas may have strong cognitive abilities. Typically, mulatta), java monkeys (M. fascicularis), stumptail maca- animals, including , which exhibit MSR, have lar- ques (M. speciosa), hamadryas baboons (Papio hama- ger relative brain sizes and more highly evolved social dryas), capuchin monkeys (Cebus paella), other lesser apes cognition abilities than those that show no MSR (Prior and pigeons (Columba liviadomestica), none of these spe- et al. 2008). Thus, the major aim of this study is to cies show sound evidence of MSR (Anderson 1984; investigate whether solitary giant pandas exhibit MSR. Anderson and Roeder 1989; de Waal et al. 2005; Gallup The development of self-recognition is correlated with and Suarez 1991; Mitchell and Anderson 1993; Shaffer and ontogeny and the development of other cognitive processes Renner 2000; Roma et al. 2007; Epstein et al. 1981; Toda (Gallup 1994). For example, human infants do not show and Watanabe 2008). The results of MSR studies in MSR until 18–24 months of age (Amsterdam 1972); in ( gorilla) are mixed (Allen and Schwartz 2008; chimpanzees, the age of onset of MSR appears to be Ledbetter and Basen 1982; Patterson and Cohn 2006). It is between 2 and 2.5 years of age (Lin et al. 1992), but either proposed that the results may be caused by cognitive bias a precipitous decline in self-recognition between 16 and due to the introduction of a new mirror image stimulus 20 years of age or a critical period effect in mid-life is (Keen et al. 2014) and the differences in enclosure/cage found in chimpanzees (Povinelli et al. 1993). Therefore, the structure, animal management and environmental enrich- second aim of this study is to explore the effect of age on ment for the subjects (Allen and Schwartz 2008); thus, the MSR in giant pandas. consistency and validity of MSR test protocols including To address these aims, we conducted a series of tests in environmental enrichment are key factors to produce reli- captive giant pandas. The results of this study may not only able and comparable results. Studies have shown that extend our understanding of MSR in animals with different bottlenose dolphins (Tursiops truncatus) (Reiss and Marino social structures, but also increase our knowledge of the 2001), Asian elephants (Elephas maximus) (Plotnik et al. cognitive ability of the giant panda. This knowledge can be 2006), and magpies (Pica pica) (Prior et al. 2008) also used to enhance the management and design of captive exhibit MSR, though some questions remain about the individuals’ enclosures and exhibits. evolution and adaptive value of MSR, and about the neu- rological processes involved (Suddendorf and Butler 2013). Materials and methods Most previous MSR studies were focused on social animals, and there is no published research on solitary Subjects and housing mammals. Solitary living animals will have less chance to encounter and interact with conspecifics in comparison Thirty-four giant pandas, ranging between 0.5 and 15 years with social living animals (Davies et al. 2012). However, of age and housed at the China Conservation and Research little is known whether solitary living animals have the Center for the Giant Panda, Bifengxia Base, Sichuan ability of MSR. Studies on MSR of solitary mammals Province, participated in the study. To study the effects of could throw light on the evolution of MSR, by showing age on MSR, subjects were divided into three groups: whether MSR is adaptive in solitary species. The giant adult, 5.5–15 years old; sub-adult, 1.5–5.5 years old; and panda (Ailuropoda melanoleuca) is an endangered solitary juvenile, 0.5–1.5 years old. Adults and sub-adults were species in China. Males and females communicate exten- individually maintained in enclosures consisting of an sively via scent and only occur together during the mating indoor bedroom (6 9 4 m) and an outdoor yard in an season (Schaller et al. 1985). They can identify each other irregular shape (about 100 9 60 m) with shrubs, herbs, through scent-markings (Liu et al. 2006; Swaisgood et al. climbing facilities and a small pond as the water source. 1999; Yuan et al. 2004; Zhang et al. 2008), vocalizations Their diet consisted of fresh bamboo, carrots, apples, and (Charlton et al. 2009) and cross-modal signaling (Xu et al. slices of savory panda bread (containing bamboo powder, 2012). In common with other bears, the giant panda has a soybean, and nutritional supplements). The juveniles were larger brain (relative to its body size) than other carnivore raised collectively in the panda baby center, which inclu- species (Gittleman 1986). It also has a larger Encephali- ded two enclosures each with an outdoor yard (30 9 20 m) zation Quotient (EQ) than the polar bear (Ursus maritimus) and an indoor cage (6 9 4 m). Test subjects were moved

(EQJ: 0.961 vs 0.510 for the panda and polar bear, for the duration of the test to one of the indoor enclosures, respectively) (Dong and Zhang 2011). Research has shown while the other juveniles remained together in the other that giant pandas cannot only discriminate the colors red enclosure. The diet of juveniles consisted exclusively of and green (Kelling et al. 2006), but also differentiate formula milk (EnfamilÒ, Mead Johnson Nutritionals, 123 Anim Cogn (2015) 18:713–721 715

Evansville, Indiana; or EsblilacÒ, PetAgÒ, Hamsphire, cage, beyond the pandas’ reach (Fig. 1); for juveniles, the Illinois, USA) and bamboo shoots. Water was available ad mirror was placed about 30 cm away from the outside of lib to all subjects. the wire mesh fence. Detailed information on the animal management is Based on a previous study of red and green color dis- provided elsewhere (Liu et al. 2006; Xu et al. 2012). All crimination in two pandas (Kelling et al. 2006), we chose subjects were healthy and unmedicated throughout the red paint to create a mark on the pandas’ bodies in the mark study. The tests had no negative effects on the pandas. Our tests (MAT). To make either a color mark or a water mark protocol was in accordance with legal requirements in the (or sham mark), the keeper usually called the subject over People’s Republic of China, conformed to Wildlife Pro- and fed it bamboo shoots, while the experimenter painted a tection law, and was approved by the Animal Welfare and round mark (5–10 cm in diameter) with a paintbrush on the Ethics Committee of Beijing Normal University and the individual’s forehead either with red Children’s Tempera China Conservation and Research Center for the Giant Paint (Guangzhou Magi-Wap Culture Article Co., Ltd) or Panda at Wolong. with water. The paint material used was odorless and nontoxic. The purpose of the mark is to attract the subject’s Previous mirror experience attention in the mirror; any part of the body not visible to the subject without the use of a mirror is therefore suitable A small pond (1.0 m in diameter) was present in the out- for marking (Gallup 1994). We chose to mark the forehead door yard of each adult and sub-adult panda enclosure. We instead of another frequently used body part, the ear cannot exclude the possibility that the pandas had seen (Calhoun and Thompson 1988; Gallup 1970, 1977), their own reflections in the water of the pond; however, to because the black ears were difficult to mark and to avoid the best of our knowledge, these pandas had not seen interfering with olfactory communication (ears are themselves in a mirror prior to the study. Thus, all subjects involved in pheromone cues in giant pandas). were suitable for the tests because they did not have any previous experience with mirrors. Experimental procedure

Apparatus and materials A pilot study was conducted from February 26 to April 18, 2013. The actual experiment was conducted in three stages. The apparatus used in this study was a one-way glass Stage 1 consisted of three tests over 10 consecutive mirror (150 9 120 cm) supported 6 cm above the ground days: a 3-day covered mirror test (CMT), a 4-day open by a metal frame for adults and sub-adults, and a smaller mirror test (OMT) and a 3-day mark test (MAT). In the one-way glass mirror (59 9 44 cm) for juveniles. For CMT, subjects were exposed to the back of the mirror adults and sub-adults, the mirror was placed about 85 cm (non-reflective surface) on three consecutive days. In the away from the outside of the wire mesh fence of the indoor OMT and MAT, the front of mirror (reflective surface) was turned to face to the pandas on four and three consecutive days, respectively. The only difference between OMT and MAT for the pandas was that we put a red color mark on the subject’s forehead during the MAT. The mark test began immediately after we added the mark on the first day of MAT. Stage 1 was conducted from July 17 to September 19, 2013, and from November 28 to December 23, 2013 on 17 adults and 7 sub-adults. Stage 2 was an independent water control test, consist- ing of three tests over five consecutive days: a three-day open mirror test (OMT), a 1-day water mark test (WMT), and a 1-day mark test (MAT) on nine individuals (previ- ously used in Stage 1). This control test was conducted from November 28 to December 23, 2013. Stage 3 consisted of four tests over 11 consecutive days: a 3-day covered mirror test (CMT), a 4-day open mirror Fig. 1 A schematic diagram of an adult giant panda’s enclosure test (OMT), a 1-day water mark control test (WMT), and a including the neighboring panda’s enclosure, the locations of the 3-day mark test (MAT). Stage 3 was conducted from mirror and the digital video (DV) camera during the daily mirror self- recognition tests. The shape of the enclosure for keeping juveniles is February 26 to May 15, 2014, on 10 juvenile pandas similar to that of an adult but a smaller size (Table 1). 123 716 Anim Cogn (2015) 18:713–721

Table 1 Information on stages, tests, and animals in this study Stage Period Tests Subjects

1 July 17–September 19, 2013 CMT-OMT-MAT 16 (Adult: Sub-adult = 11:5) November 28–December 23, 2013 8 (Adult: Sub-adult = 6:2) 2 November 28-December 23, 2013 OMT-WMT-MAT 9 (Previously used in July–September of 2013, Adult: Sub-adult = 7:2) 3 February 26–May 15, 2014 CMT-OMT-WMT-MAT 10 juveniles

Within each test, the procedure was as follows: The Table 2 Definitions and categorization of coded behaviors mirror was placed in position each day at 07:30 h. In Behavior Definition Stages 1 and 2, each daily trial spanned 15 min, yielding 150 min of observation time for each individual in Stage 1 Non-mirror behavior and 75 min in Stage 2 (water mark control test) (each Body-directed Behaviors shown to meet physiological individual was tested only once in each stage). For tests of requirements without looking at the mirror juveniles in Stage 3, each daily trial spanned 10 min, Mirror-directed behavior yielding 110 min of observation time in total for each Viewing The eyes target the mirror for two or more seconds individual. After the daily 15 or 10 min trial had ended for Investigating Attempting to touch the mirror the animal, the mirror was moved to an indoor area sepa- Aggressive Attacking: Rushing toward the front of mirror from behavior within the indoor cage rating two neighboring cages that was not accessible to the Threatening: Barking and howling pandas, to keep it out of the panda’s sight (Fig. 1). We ran MSR tests on 2–3 animals each day with an Backwards walking: Retreating five or more steps interval of at least 15 min between each animal’s test trial Foot scraping: Scraping the substrate with a backward motion of the hind paws during Stages 1 and 3. We ran tests on 4–5 animals daily Playing Attempting to interact with the mirror in an during Stage 2. amicable manner

Behavioral observations We recorded the durations the pandas engaged in all Immediately after the mirror had been placed in position, a mirror-directed behaviors: Viewing (the eyes targeting the subject was released into its indoor cage (where it could see mirror for two or more seconds), investigating (reaching the mirror) and its behaviors were observed by the first toward the mirror with the paws), playing (attempting to author and recorded by a digital video camera (Sony DCR- interact with the mirror image in an amicable manner), and SR68E, Sony, Shanghai, China) throughout the trial. When aggressive behavior (including attacking–rushing toward each test trial video recording was played back with the front of the mirror from within the indoor cage; Windows Media Player, data were recorded in Microsoft threatening–barking and howling; backwards walking– Office Excel (version 2010) by one observer (the first retreating five or more steps; and foot scraping–scraping author). Information about the date and animal identity was the substrate with a backward motion of the hind paws), removed from the video recordings by the second author during the 10 or 15 min trial. before they were played back, so that analysis could be Based on the pilot study results, we first assessed the conducted blind. Behavioral data were collected from the intra-observer reliability of behavioral data recording for videos by focal sampling and continuous recording meth- four major behaviors (viewing, investigating, aggressive ods (Martin and Bateson 1986). behavior, and playing). To do this, an observer (blind to the The behaviors recorded in this study, and their defini- experimental design and procedure but familiar with giant tions are listed in Table 2. For subsequent analysis, all panda behavior) repeatedly viewed recordings of 79 test behaviors were grouped into two categories: non-mirror trials (20 % of the total, 395 trials) carried out on all 34 behaviors, including eating, resting, sniffing, scent-mark- different subjects, at intervals of 24 h. The 79 trials were ing, grooming, and vocalizing; and mirror-directed randomly chosen from the 395 trials and included CMT, behaviors, including viewing, investigating, attacking, OMT, and MAT trials. The intra-observer reliability was threatening, backwards walking, foot scraping (Allen and calculated as the Kappa coefficient, k, in which the Schwartz 2008) and playing. Attacking, threatening, observed agreement between two observations was cor- backwards walking, and foot scraping were further grouped rected for agreement by chance alone (Martin and Bateson into aggressive behavior (Table 2). 1986). The Kappa coefficients were 0.81, 0.73, 1.0, and

123 Anim Cogn (2015) 18:713–721 717

7.5 0.89 for playing, viewing, investigating, and aggressive Adult behavior, respectively. We concluded that all behaviors Sub-adult could be recorded in a repeatable way by one observer. To Juvenile do this, the first author repeatedly viewed all the video 5.0 recordings and classified the pandas’ behavior for analysis. We first calculated the duration (percentage of time) of each behavior for each individual trial and then grouped 2.5 viewing, investigating, playing, and aggressive behavior into the category mirror-directed behavior. The percentage Percentage of time spent of time spent in mirror-directed behavior and the percent- in mirror-directed behaviors 0.0 age of time spent in each independent behavioral category CMT OMT MAT were subsequently analyzed. Based on the results of the pilot study, we expected pandas to show significant daily Fig. 2 No age difference was found in percentage of time adult (n = 17), sub-adult (n = 7) and juvenile (n = 10) pandas engaged in variation in behaviors during the CMT, OMT, and MAT, mirror-directed behaviors during the covered mirror tests (CMT), so we recorded and compared their behavioral responses to open mirror tests (OMT) and mark tests (MAT) in Stage 1 and 3 the mirror by day, instead of pooling and averaging the data over the different days. (OMT) was distributed as follows: viewing (9.4 %), play- Data analysis ing (0.1 %), investigating (\0.1 %), and aggressive behavior (0.5 %). Data were not normally distributed, so we used the Krus- kal–Wallis test followed by post hoc Mann–Whitney Does age affect pandas’ behavioral responses U tests to compare the behavioral responses of pandas in to mirrors? the three age groups to the covered mirror (CMT), mirror image (OMT), and marked-mirror image (MAT), respec- No age-related difference was found in mirror-directed tively. As no significant differences among the three age behavior in response to the covered mirrors or the open groups were found (see below), we pooled the data from mirrors throughout the three stages, including in the CMT, the three age groups together. We then employed the OMT, and MAT (Fig. 2). Adult giant pandas (n = 17) Friedman test to compare the pandas’ daily behavioral spent more time engaged in mirror-directed behavior than responses during CMT, OMT, and MAT, followed by post did either sub-adults (n = 7) or juveniles (n = 10) during hoc of Wilcoxon signed-rank tests. OMT and MAT, and juveniles spent more time in mirror- For the analysis of water mark control tests (WMT) and directed behavior during CMT, but these small differences mark tests (MAT), we compared only the percentage of were statistically similar during CMT (v2 = 0.222, time spent in mirror-directed behaviors. For consistency, P = 0.895), OMT (v2 = 0.719, P = 0.698), and MAT we submitted the behavioral data from the nine adults and (v2 = 0.187, P = 0.911; Fig. 2). sub-adults in Stage 2 and the 10 juveniles in Stage 3 to a Wilcoxon signed-rank test to compare the percentages of Can pandas recognize themselves in the mirror? time spent in mirror-directed behavior on the first and the third day of OMT, on the third day of OMT and WMT, and All 34 subjects in three age groups could not recognize on the WMT and the first day of MAT, respectively (Heth themselves in the mirror, but rather regarded their mirror et al. 1999). images as conspecifics. During the CMT, the percentage of All statistical analyses were conducted using SPSS 17.0 time subjects engaged in mirror-directed behavior, for Windows (SPSS Inc., Chicago, USA). The data are including viewing, investigating, aggressive behavior, and shown as the mean ± SEM. Alpha was set at 0.05, and all playing decreased gradually over time (Figs. 3, 4A–D); tests were two-tailed. they spent on average only 7.6 s viewing on the third day (Z =-4.871, P = 0.000), which meant that the subjects became habituated to the covered mirror (Z =-1.44, Results P = 0.15) (Fig. 4A). Presentation of the open mirror led to a threefold increase in mirror-directed behavior over that Throughout the entire study, subjects of all three age observed during the CMT (Fig. 3): The largest differences groups spent more than 80 % of their time engaged in non- were observed in viewing, aggressive behavior, and mirror behaviors. The percentage of the total time spent in investigating, yet no significant difference was found in mirror-directed behaviors during the mirror exposure viewing behavior from the third day of OMT (Fig. 4A) or 123 718 Anim Cogn (2015) 18:713–721 in viewing behavior between the first day of the MAT and P = 0.073; Fig. 4A). Aggressive behavior was predomi- the last day of OMT (Z =-1.805, P = 0.071; Fig. 4A). nately seen in the OMT (Fig. 4C) and playing was seen in There was no significant difference in viewing behavior in both CMT and OMT (Fig. 4D). all the daily trials during the MAT (Z =-1.793, Do pandas respond to the red color mark?

We found no difference in pandas’ behavior in response to 20 e CMT their mirror image and in the control tests: During all the OMT WMT in Stages 2 and 3, the pandas were never seen 16 MAT touching the water mark on their forehead. The subjects

12 gradually became habituated to the mirror image by the third day of OMT (Z =-3.349, P = 0.001, n = 19; 8 Fig. 5). There was no difference in mirror-directed a behavior between the third day of OMT and the WMT ab 4 (Z =-0.059, P = 0.953, n = 19), or between the third c bc cd cd c d cd day of OMT and the first day of MAT (Z =-1.516, Percentage of time spent in mirror-directed behaviors 0 P = 0.129, n = 19; Fig. 5). The subjects did display sig- 1 2 3 4 5 6 7 8 9 10 nificantly more mirror-directed behavior in MAT than in Days (9) (10)(11) WMT (Z =-2.567, P = 0.010, n = 19; Fig. 5). Fig. 3 Percentage of time pandas (n = 34) engaged in mirror- directed behaviors during the covered mirror tests (CMT), open mirror tests (OMT) and mark tests (MAT) in Stages 1 and 3. Data are Discussion mean ? SEM. Bars sharing the same letters are not significantly different from one another. The numbers under x-axis show the testing day numbers in Stage 1 (upper) and Stage 3 (lower inside the In general, self-recognition ability is restricted to animals brackets), respectively. Same in the following figures with relatively large brains and highly evolved social

A 17.0 d C CMT 2.5 OMT a CMT 13.6 MAT OMT 2.0 MAT 10.2 1.5

6.8 1.0 a ab Percentage of time spent in viewing 3.4 ab b 0.5 bc bc b b c

Percentage of time spent in aggressive behavior a a 0.0 0.0 1 2 3 4 5 6 7 8 9 10 1 2 3 4 5 6 7 8 9 10 (9) (10)(11) Days (9) (10) (11) Days

B 0.27 D 0.20 b CMT a CMT a OMT OMT MAT a MAT 0.15 0.18 ab 0.10

0.09 a ab ab ab Percentage of time spent in playing 0.05 Percentage of time spent in investigating ac ac 0.00 0.00 1 2 3 4 5 6 7 8 9 10 1 2 3 4 5 6 7 8 9 10 (9) (10) (11) (9) Days Days (10)(11)

Fig. 4 Percentages of time pandas engaged in viewing the mirror (A), investigating the mirror (B), aggressive behavior toward the mirror image (C), and playing with the mirror image (D) in Stages 1 and 3. Abbreviations are as in Figs. 2 and 3

123 Anim Cogn (2015) 18:713–721 719

modifying behaviors. Therefore, our data suggest that 0.095 ** OMT WMT pandas recognize their mirror images as conspecifics, MAT 0.076 rather than as illusory images or as themselves. These results are consistent with previous studies in monkeys 0.057 (Gallup 1977; Gallup and Suarez 1991). Our water mark NS control test results provide further evidence that pandas 0.038 ** recognize their images only as conspecifics, since they NS showed increased mirror-directed behavior to the red- 0.019 marked image than to the water-marked image in the

Percentage of time spent control tests (Fig. 5), though they did not show any mark- in mirror-directed behaviors 0.000 1 2 3 4 5 touching behavior. This suggests that they recognize the (5) (6) (7) (8) (9) difference in appearance between a red color-marked Days image and a water-marked individual image yet cannot Fig. 5 Percentages of time pandas (n = 19) engaged in mirror- identify the red color-marked individual as themselves on directed behaviors during the open mirror tests (OMT), water mark the third day of OMT (Fig. 5). A previous study shows that control tests (WMT) and mark tests (MAT) in Stages 2 (n = 9) and 3 giant pandas can discriminate between panda-like eye- (n = 10). Data are shown as mean ? SEM. **P \ 0.01, NS no mask patterns that vary only slightly in shape (Dungl et al. significant difference 2008). However, giant pandas appear to be incapable of learning to use mirrors to integrate features of their own cognition (Prior et al. 2008). As a species of the Ursidae, reflection and respond appropriately. This result may be the giant panda has a large relative brain size and EQ, and a related to or determined by their living style in the field: highly developed and evolved brain (Dong and Zhang The giant panda is a solitary species in the wild. Both 2011; Xie et al. 1984). Yet we found no evidence that the adults and sub-adults only occur together with other pandas giant panda could recognize its self-image in the mirror. during the mating season. For individual recognition and Moreover, there were no age-related differences in self- discrimination, including between mother and offspring image recognition capability, so we can conclude that the during the lactation period, they use more olfactory and lack of MSR is not related to developmental stage or auditory cues than visual cues (Xu et al. 2012; Yuan et al. maturity. 2004; Zhang et al. 2008). However, the visual ability of When presented with a mirror, animals typically captive giant pandas is estimated to equal to humans’ respond in one of three ways: (1) they behave as though the 0.3–0.4 standard vision, i.e., they are nearsighted (Lin et al. mirror image is a conspecific or another animal and show unpublished data). From an evolutionary perspective, MSR aggressive behavior toward it; (2) they behave as though may not be a requisite adaptation in this species: Animals the mirror image is illusory and ignore it, showing interest lacking the capacity for MSR are not hindered by reflective only in the smooth surface; (3) they recognize themselves surfaces, such as standing pools of water. Giant pandas do in the mirror, then decorate or groom themselves in front of not appear to need the capacity for MSR to survive. the mirror (Wang 2010) or show self-directed behavior by The present study indicates that giant pandas did not using the mirror to respond to themselves (Gallup 1994). If recognize themselves in the mirror; instead, they perceived subjects touch a mark on their forehead more than under the mirror image as an unfamiliar conspecific, and thus baseline conditions (without a mirror), or show decorating displayed aggressive behavior toward the mirror image or grooming behaviors when they look at the mirror, or use during the OMT (Fig. 4C). However, they rapidly realized the mirror to investigate parts of their bodies that they had that the conspecific posed no threat to them and reduced not seen before, MSR ability is inferred. Otherwise, they their vigilance due to the fact that they were habituated to show no MSR ability (Allen and Schwartz 2008; Gallup the presence of a neighboring individual in next door in 1970, 1994; Lethmate and Du¨cker 1973). captivity (Shang 2005; Ji et al. 1996). This result suggests During the first day of the OMT, the giant pandas that when keeping giant pandas in captivity, frequently directed strong aggressive behaviors toward the mirror switching them between enclosures may promote com- image, including attacking, threatening, backwards walk- munication and familiarity between conspecifics. This may ing, and foot scraping. Subsequently, they habituated to the be useful for both captive breeding and the reintroduction mirror image. However, they were never observed dis- of captive giant pandas into the wild. playing mark-touching, decorating or self-directed We found no significant differences in the percentage of response behaviors. In addition, during a prolonged test time spent in mirror-directed behavior by pandas of the (14 days in total) of three subjects exposed to a mirror in three age groups (Fig. 2), but decreasing trends in mirror- the pilot study, observations revealed no increase in self- directed behavior were found from adult to sub-adult to 123 720 Anim Cogn (2015) 18:713–721 juveniles. This may be related to the decreasing need, with No. 2012BAC01B06), and the International Collaborative Project on decreasing age, for pandas to defend a territory (Schaller the Conservation for the Giant Panda to D Liu (Grant No. Beijing- 2010-02) and R Wei (Grant No. SD0630). et al. 1985), since the pandas recognized their mirror image as a conspecific. With respect to the behavioral changes, Conflict of interest The authors declare that they have no conflict adults and sub-adults tended to show more investigative of interest. and aggressive behaviors, and juveniles were more focused Ethical standards The experiments comply with the current laws on playing, such as rolling, when they encountered the of the People’s Republic of China. mirror during OMT and MAT (0.0006 ± 0.00059, 0 ± 0 and 0.004 ± 0.004, for adults, sub-adults and juveniles, respectively). The observed age difference in playing may References be caused by the captive management practices used for those animals. Adults and sub-adults were kept individu- Allen M, Schwartz BL (2008) Mirror self-recognition in a Gorilla ally in cages neighboring those of another heterosex indi- (Gorilla gorilla gorilla). J Integr Biosci 5:19–24 Amsterdam B (1972) Mirror self-image reactions before age two. Dev vidual, while juveniles were kept together in the baby Psychobiol 5:297–305. doi:10.1002/dev.420050403 center. The presence of a cage-mate may have not only Anderson JR (1984) Monkeys with mirrors: some questions for impaired the subjects’ interest in the mirrors, but may also primate psychology. Int J Primatol 5:81–98. doi:10.1007/ have led to increased social behavior, including playing, in BF02735149 Anderson JR, Roeder J-J (1989) Responses of capuchin monkeys juveniles (Gallup 1994). This result in juvenile may pro- (Cebus apella) to different conditions of mirror-image stimula- vide further evidence that giant pandas do not have MSR tion. Primates 30:581–587. doi:10.1007/BF02380884 ability. Brooks-Gunn J, Lewis M (1979) ‘‘Why mama and papa?’’ The We are aware of some limitations of our study. We did development of social labels. Child Dev 50:1203–1206. doi:10. 2307/1129349 not control for luminosity in the panda cages. Wild pandas Calhoun S, Thompson RL (1988) Long-term retention of self- typically occur under the dense forest canopy, but we recognition by chimpanzees. Am J Primatol 15:361–365. doi:10. turned on the lights in the indoor cages if we felt it was too 1002/ajp.1350150409 dark to conduct the test. We did not randomize the order in Charlton BD, Zhang Z, Snyder RJ (2009) Vocal cues to identity and relatedness in giant pandas (Ailuropoda melanoleuca). J Acoust which the tests were conducted, and Stage 2 took place Soc Am 126:2721–2732. doi:10.1121/1.3224720 3 months after Stage 1. The water mark control tests had to Davies NB, Krebs JR, West SA (2012) An introduction to behavioural take place before the mark tests, because it was difficult to ecology (4th edn). Wiley, Oxford, pp 147–178 remove the red paint from the pandas’ fur. Fortunately, the de Waal FB, Dindo M, Freeman CA, Hall MJ (2005) The in the mirror: hardly a stranger. Proc Natl Acad Sci USA stages were conducted independently and the data were 102:11140–11147. doi:10.1073/pnas.0503935102 analyzed separately, which might reduce the effects of the Dong W, Zhang J (2011) Evolution of cranial cavities in giant pandas lack of randomization. (Ailuropoda melanoleuca, Carnivora, Mammalia). Vertebrata An extended test in a naturalistic open area instead of in Palasiatica 49:241–252 Dungl E, Schratter D, Huber L (2008) Discrimination of face-like front of a wire fence may improve the ecological validity patterns in the giant panda (Ailuropoda melanoleuca). J Comp of the test results. Further, additional considerations of self- Psychol 122:335–343. doi:10.1037/0735-7036.122.4.335 recognition, including the development of additional test Epstein R, Lanza RP, Skinner BF (1981) Self-awareness in the measures, may provide a more extensive empirical pigeon. Science 212:695–696. doi:10.1126/science.212.4495. 695 framework for the development of theories of self-recog- Gallup GG (1970) Chimpanzees: self-recognition. Science 167: nition (Swartz and Evans 1991). 86–87. doi:10.1126/science.167.3914.86 In summary, our study indicates that giant pandas are Gallup GG (1977) Absence of self-recognition in a monkey (Macaca not able to recognize their own images in a mirror. Instead, fascicularis) following prolonged exposure to a mirror. Dev Psychobiol 10:281–284. doi:10.1002/dev.420100312 pandas of all ages perceive a mirror image as an unfamiliar Gallup GG (1994) Self-recognition: research strategies and experi- conspecific. This study provides valuable information mental design. In: Parker ST, Mitchell RW, Boccia ML (eds) regarding recognition and cognition in large mammals, Self-awareness in animals and humans: developmental perspec- particularly in solitary species. In addition, the results may tives. Cambridge University Press, Cambridge, pp 35–50 Gallup GG, Suarez SD (1991) Social responding to mirrors in rhesus be useful in the management and environmental enrich- monkeys (Macaca mulatta): effects of temporary mirror ment of captive giant pandas. removal. J Comp Psychol 105:376–379. doi:10.1037/0735- 7036.105.4.376 Acknowledgments We thank staff at the China Conservation and Gittleman JL (1986) Carnivore brain size, behavioral ecology, and Research Center for the Giant Panda for their assistance in data phylogeny. J 67:23–36. doi:10.2307/1380998 collection and animal management. This work was supported by Heth G, Todrank J, Johnston RE (1999) Similarity in the qualities of grants from the Natural Science Foundation of China NSFC (Grant individual odors among kin and species in Turkish (Mesocricetus No. 31172097, 31472009), the National Key Technology R & D brandti) and golden (Mesocricetus auratus) hamsters. J Comp Program of the Ministry of Science and Technology of China (Grant Psychol 113:321–326. doi:10.1037/0735-7036.113.3.321

123 Anim Cogn (2015) 18:713–721 721

Heyes CM (1994) Reflections on self-recognition in primates. Anim Povinelli DJ, Rulf AB, Landau KR, Bierschwale DT (1993) Self- Behav 47:909–919. doi:10.1006/anbe.1994.1123 recognition in chimpanzees (Pan troglodytes): distribution, Heyes CM (1995) Self-recognition in primates: further reflections ontogeny, and patterns of emergence. J Comp Psychol create a hall of mirrors. Anim Behav 50:1533–1542. doi:10. 107:347–372. doi:10.1037/0735-7036.107.4.347 1016/0003-3472(95)80009-3 Prior H, Schwarz A, Gu¨ntu¨rku¨n O (2008) Mirror-induced behavior in Ji H, Guo D, Tao S (1996) Habituation and recovery of visual the magpie (Pica pica): evidence of self-recognition. PLoS Biol attention in normal infants. Chin Behav Med Sci 5:5–6 (In 6:1642–1650. doi:10.1371/journal.pbio.0060202 Chinese with English abstract) Reiss D, Marino L (2001) Mirror self-recognition in the bottlenose Keen HA, Nelson OL, Robbins CT, Evans M, Shepherdson DJ, : a case of cognitive convergence. Proc Natl Acad Sci Newberry RC (2014) Validation of a novel cognitive bias task USA 98:5937–5942. doi:10.1073/pnas.101086398 based on difference in quantity of reinforcement for assessing Roma PG, Silberberg A, Huntsberry ME, Christensen CJ, Ruggiero environmental enrichment. Anim Cogn 17:529–541. doi:10. AM, Suomi SJ (2007) Mark tests for mirror self-recognition in 1007/s10071-013-0684-1 capuchin monkeys (Cebus apella) trained to touch marks. Am J Kelling AS, Snyder RJ, Marr MJ, Bloomsmith MA, Gardner W, Primatol 69:989–1000. doi:10.1002/ajp.20404 Maple TL (2006) Color vision in the giant panda (Ailuropoda Schaller GB, Hu J, Pan W, Zhu J (1985) The giant pandas of wolong. melanoleuca). Learn Behav 34:154–161. doi:10.3758/BF031 University of Chicago press, Chicago 93191 Shaffer VA, Renner MJ (2000) Black-and-white colobus monkeys Ledbetter DH, Basen JA (1982) Failure to demonstrate self- (Colobus guereza) do not show mirror self-recognition. Int J recognition in gorillas. Am J Primatol 2:307–310. doi:10.1002/ Comp Psychol 13:154–160 ajp.1350020309 Shang Y (2005) Animal habituation behavior. Bull Biol 40:9–11. Lefebvre L, Reader SM, Sol D (2004) Brains, innovations and doi:10.3969/j.issn.0006-3193.2005.10.004 evolution in birds and primates. Brain Behav Evol 63:233–246. Sua´rez SD, Gallup GG (1981) Self-recognition in chimpanzees and doi:10.1159/000076784 orangutans, but not gorillas. J Hum Evol 10:175–188 Lethmate J, Du¨cker G (1973) Untersuchungen zum selbsterkennen im Suddendorf T, Butler DL (2013) The nature of visual self-recognition. spiegel bei orangutans und einigen anderen affenarten. Z Tier- Trends Cognit Sci 17:121–127. doi:10.1016/j.tics.2013.01.004 psychol 33:248–269 Swaisgood RR, Lindburg DG, Zhou XP (1999) Giant pandas Lin AC, Bard KA, Anderson JR (1992) Development of self- discriminate individual differences in conspecific scent. Anim recognition in chimpanzees (Pan troglodytes). J Comp Psychol Behav 57:1045–1053. doi:10.1006/anbe.1998.1070 106:120–127. doi:10.1037//0735-7036.106.2.120 Swartz KB, Evans S (1991) Not all chimpanzees (Pan troglodytes) Liu L, Yang LZ (2009) Research progress in self-face recognition. show self-recognition. Primates 32:483–496. doi:10.1007/BF023 J Liaoning Normal Univ (Social Science Edition) 32:52–55. 81939 doi:10.3969/j.issn.1000-1751.2009.03.014 (In Chinese with Toda K, Watanabe S (2008) Discrimination of moving video images English abstract) of self by pigeons (Columba livia). Anim Cogn 11:699–705. Liu D, Yuan H, Tian H, Wei RP, Zhang GQ, Sun LX, Wang LW, Sun doi:10.1007/s10071-008-0161-4 RY (2006) Do anogenital gland secretions of giant panda code Wang J (2010) What are animal friends thinking. Hubei Children for their sexual ability? Chin Sci Bull 51:1986–1995. doi:10. Press, Hubei (In Chinese) 1007/s11434-006-2088-y Xie J, Pirlot P, Jiao S, Wang Y, Zhang G, Zhang S (1984) Quantitative Mans L, Cicchetti D, Sroufe LA (1978) Mirror reactions of down’s morphology of the giant panda brain in comparison with those of syndrome infants and : cognitive underpinnings of self- the American bear and the Raccoon. J Beijing Second Med recognition. Child Dev 49:1247–1250. doi:10.1111/j.1467-8624. College 2:97–101 (In Chinese with English abstract) 1978.tb04100.x Xu M, Wang ZP, Liu DZ, Wei RP, Zhang GQ, Zhang HM, Zhou XP, Martin P, Bateson P (1986) Measuring behaviour. Cambridge Li DS (2012) Cross-modal signaling in giant pandas. Chin Sci University Press, Cambridge Bull 56:1–5. doi:10.1007/s11434-011-4843-y Mitchell RW, Anderson JR (1993) Discrimination learning of Yang H (2005) On the uniqueness of self-face recognition. Psychol scratching, but failure to obtain imitation and self-recognition Sci 28:1517–1519. doi:10.3969/j.issn.1671-6981.2007.05.026 in a long-tailed macaque. Primates 34:301–309. doi:10.1007/ Yuan H, Liu DZ, Sun LX, Wei RP, Zhang GQ, Sun RY (2004) BF02382625 Anogenital gland secretions code for sex and age in the giant Patterson FG, Cohn RH (2006) Self-recognition and self-awareness in panda, Ailuropoda melanoleuca. Can J Zool 82:1596–1604. lowland gorillas. In: Parker ST, Mitchell RW, Boccia ML (eds) doi:10.1139/z04-144 Self-awareness in animals and humans: Developmental perspec- Zhang J-X, Liu DZ, Sun LX, Wei RP, Zhang GQ, Wu HL, Zhang tives. Cambridge University Press, Cambridge, pp 273–290 HM, Zhao C (2008) Potential chemosignals in the anogenital Plotnik JM, De Waal FB, Reiss D (2006) Self-recognition in an Asian gland secretion of giant pandas, Ailuropoda melanoleuca, . Proc Natl Acad Sci USA 103:17053–17057. doi:10. associated with sex and individual identity. J Chem Ecol 1073/pnas.0608062103 34:398–407. doi:10.1007/s10886-008-9441-3

123