Categorization in Macaca fascicularis.

Kanthi Arum Widayati

GRADUATE SCHOOL

BOGOR AGRICULTURAL UNIVERSITY

BOGOR

I hereby declare that dissertation entitled Categorization in Macaca fascicularis is original result of my own research supervised under advisory committee and has never been submitted in any form at any institution before. All information from other authors cited here are mentioned in the text and listed in the reference at the end part of the dissertation.

Bogor, February 1st_{, 2012}

Kanthi Arum Widayati

KANTHI ARUM WIDAYATI. Categorization in Macaca fascicularis. Supervised by BAMBANG SURYOBROTO, AKICHIKA MIKAMI, ACHMAD FARAJALLAH.

Categorization is an ability to group individuals or events into different classes mediated by conceptualized mental images. There are several levels of categorization and within a taxonomy the levels are nested. At the most concrete level of categorization, all or most members of the category shared common physical attributes that differ from other categories. The higher the level of category, the fewer common attributes between members of the group. In addition to humans, the ability to categorize has also been proposed in animals. One example of categorization in animal is species discrimination. Using matching-to-sample task, present experiment tested ability of the long-tailed macaque (Macaca fascicularis) in discriminating dichotomous-stimuli of different animals. The species has been shown to be able to see photos as representations of real object so I used facial photos of humans and animals for the stimuli. First, I tested their ability to classify humans and macaques into separate group. Second, I tested their ability to discriminate their conspecific from other macaques. And the last, I tested whether the subjects were able to discriminate non-human animals from humans. In all of these experiments I found that the subjects showed high performance in categorizing objects, even when I discarded details of visual informations, such as color and local shapes. The ability to identify objects with reduced representation of physical properties means the subjects were able to generalize attributes of members of the group. This would indicate that the subjects created a higher level abstraction. On the other hand, in discriminating intrageneric macaque species I found that they were able to extract uniqueness of each species. More over, I also found that the subjects were able to put photos of non-human animals that shared very few similarities in physical percepts into one group. I suggested that the subjects could create a more abstract concept based on non-percepts relations as a basis to put the objects into one category. Thus, I concluded that M. fascicularis were able to perform multiple levels of categorizations.

Key Words: Categorization, Conceptualized mental image, Species discrimination, Macaca fascicularis, Matching-to-sample task

KANTHI ARUM WIDAYATI. Kemampuan kategorisasi Macaca fascicularis. Dibimbing oleh BAMBANG SURYOBROTO, AKICHIKA MIKAMI, ACHMAD FARAJALLAH.

Kategorisasi adalah kemampuan seseorang untuk mengelompokkan individu-individu atau kejadian-kejadian ke dalam kelompok yang berbeda. Kemampuan kategorisasi didukung oleh konsep yang dibentuk di otak. Ada beberapa tingkat kategorisasi. Pada tingkat yang paling dasar, semua anggota kelompok memiliki banyak persamaan ciri fisik dibandingkan dengan kelompok lainnya. Semakin tinggi tingkat kategorisasi, persamaan ciri fisik di antara anggota kelompok semakin sedikit. Selain manusia, hewan diduga juga memiliki kemampuan kategorisasi. Salah satu contoh kategorisasi adalah diskriminasi spesies hewan yang berbeda. Penelitian ini bertujuan untuk mengetahui kemampuan diskriminasi spesies pada Macaca fascicularis. Monyet ini memiliki kemampuan untuk melihat foto sebagai representasi dari benda sebenarnya. Penelitian ini menggunakan stimulus berupa foto-foto wajah dari manusia dan hewan dengan metode matching-to-sample task. Saya melakukan tiga eksperimen utama. Pertama, saya ingin menguji apakah M. fascicularis dapat mengelompokkan manusia dan makaka ke dalam dua kelompok terpisah. Kedua, saya ingin mengetahui apakah monyet mampu membedakan antara individu-individu spesiesnya dengan individu-individu-individu-individu dari spesies lain. Terakhir, saya ingin menguji apakah monyet mampu membedakan antara manusia dan hewan lain non manusia. Hasil penelitian menunjukkan bahwa M. fascicularis mampu memisahkan dan mengelompokan objek-objek ke dalam kategori yang diujikan walaupun saya menyingkirkan informasi visual dari stimulus seperti warna dan bentuk. Kemampuan monyet untuk mengidentifikasi objek berdasarkan sedikitnya informasi fisik mengindikasikan adanya kemampuan dalam membentuk konsep yang lebih abstrak. Selain itu, pada eksperimen ke dua, monyet saya berhasil menemukan ciri-ciri unik dari masing-masing spesies makaka. Saya juga menemukan bahwa monyet berhasil mengelompokkan foto-foto hewan yang berbeda secara fisik ke dalam satu kelompok. Saya menduga bahwa monyet dapat membuat konsep yang lebih abstrak berdasarkan hubungan non-perseptual sebagai dasar untuk mengelompokkan objek ke dalam satu kategori. Saya menyimpulkan bahwa dalam mengkategori hewan, M. fascicularis menggunakan beberapa tingkatan abstraksi.

Key Words: Kategorisasi, Konsep, Diskriminasi spesies, Macaca fascicularis, Matching-to-sample task

KANTHI ARUM WIDAYATI. Categorization in Macaca fascicularis. Supervised by BAMBANG SURYOBROTO, AKICHIKA MIKAMI, ACHMAD FARAJALLAH.

Categorization is an ability to group individuals or events into different classes mediated by conceptualized mental image. There are several levels of categorization and within a taxonomy the levels are nested. At the most concrete level of categorization, all or most members of the category shared common physical attributes that differ from other category. The higher the level of category, the fewer common attributes between members of the group.

In addition to humans, the ability to categorize has also been proposed in animals. In addition to humans, the ability to categorize has also been proposed for animals. Being able to identify, visually or otherwise, a new object as a member of a category is an advantage for animals. It should help them to distinguish between food or non-food, or to discriminate between species of animals. This species discrimination is important to prevent hybridization among species.

Using matching-to-sample task, present experiments tested the ability of the long-tailed macaque (Macaca fascicularis) in discriminating dichotomous-stimuli of different animals. The species has been shown to be able to see photos as representations of real objects so we used facial photos of humans and animals for the stimuli. Using operant conditioning method, I trained monkeys to associate matching to sample stimuli against a distractor stimulus. First, I showed monkeys a sample stimulus as a reference to be matched. To ensure that the subjects paid attention to the sample stimulus, they had to touch it and for this they received reward that they find beneath the stimulus. Next, I presented a matching stimulus and a distractor stimulus side-by-side. The subjects must choose one of them. When the subjects chose the matching stimulus, they received a piece of food as a reward; the response was counted as a correct one. When they chose the distractor, they did not get any rewards. Subject's motivation for reward warranted the choice of the matching stimulus. The location of matching and distractor stimuli on the tray was arranged pseudorandomly. I blocked every 20 trials into one session and measured their correct rate. If the subject chose the matching stimuli higher than 90% in a session, I interpreted they were able to associate matching to sample stimuli. Logically, this may be inferred as the subject had developed dichotomic concepts of matching against distractor. When they showed this competence, they went to test phase. I expect that monkeys were able to transfer their concept learned in training phase into new stimuli by showing the same performance in both baseline and test trials. In the test phases, I introduced new matching and distractor stimuli and see their response into the stimuli. I did three major experiment. First, I tested their ability to classify human and macaques into separate group. Second, I tested their ability to discriminate their conspecific from other macaques. And the last, I tested whether the subjects able to discriminate non-human animals from human.

In all of these experiments I found that the subjects showed high performance in categorizing objects, even when I discarded details of visual informations, such as color and local features. The ability to identify objects with

generalize attributes of members of the group. This would indicate that the subjects have ability to create a higher level abstraction. On the other hand, in discriminating intragenic macaque species, I found that they were able to extract uniqueness of each species. More over, I also found that the subjects able to put photos of non-primates animals that shared very few similarities in physical properties into one group. Monkeys may able to create a logical concept such as A and non-A. I suggest that the subjects could create abstract concepts free from the physical properties as a basis to put objects into one category. Thus, I conclude that M. fascicularis are able to perform multiple levels of categorizations.

Key Words: Categorization, Conceptualized mental image, Species discrimination, Macaca fascicularis, Matching-to-sample task

All Rights Reserved

It is prohibited to cite all or a part of this dissertation without referring to and mentioning the source. Citation is permitted for the purposes of education, research, scientific paper, report, or critism writing only; and it does not defame the name and honor of Bogor Agricultural University.

It is prohibited to republish and reproduce all or a part of this dissertation without permission from Bogor Agricultural University.

KANTHI ARUM WIDAYATI

Dissertation

submitted in partial fulfillment of the requirements for a Doctoral Degree in Animal Bioscience Major of Graduate School of Bogor Agricultural University

GRADUATE SCHOOL

BOGOR AGRICULTURAL UNIVERSITY

BOGOR

1. Dr. Entang Iskandar (Researcher, Primate Research Center, Bogor Agricultural University)

2. Dr. Yamato Tsuji (Assistant Professor, Primate Research Institute, Kyoto University, Japan)

Examiners in the Open examination:

1. Prof. Ir. Wasmen Manalu, Ph.D. (Professor of Faculty of Veterinery Medicine, Bogor Agricultural University)

2. Dr. Harry Susianto (Senior Staff of Faculty of Psychology, University of Indonesia)

Title : Categorization in Macaca fascicularis Name : Kanthi Arum Widayati

Student ID : G362070051 Major : Animal Biosciences

Endorsed by,

Supervisory Committee

Dr. Bambang Suryobroto Chair

Prof. Dr. Akichika Mikami, MD Dr. Achmad Farajallah

Member Member

Acknowledged by,

Chair of Major of Animal Biosciences Dean of Graduate School,

Dr. Bambang Suryobroto Dr. Ir. Dahrul Syah, M.Sc.Agr.

The title of this dissertation is Categorization in Macaca fascicularis. The experiment took place in the Laboratory of Zoology Departement of Biology Faculty of Mathematics and Natural Sciences Bogor Agricultural University, Bogor and Primate Research Institute, Kyoto University, Japan.

I want to send my gratitude to Dr. Bambang Suryobroto, Dr. Achmad Farajallah and Prof. Dr. Akichika Mikami, MD as author’s advisors; to Mr. Adi Surahman, Mrs. Ani and Mr. Mamat for taking care of the monkeys. To all individuals with their unique personality in Zoo Corner (Dr. Tetri Widiyani, Sarah Nila, S.Si, Puji Rianti, M.Si, Islamul Hadi, M.Si, Eneng Nunuz R, S.Si, Elda Irma Kawulur, M.Si, Andi Darmawan, M.Si) for great discussion and their supports for the experiments.

I also want to send my appreciation to Dr. Yamato Tsuji from Primate Research Institute, Kyoto University, Dr. Harry Susianto from Faculty of Psychology, University of Indonesia, Dr. Entang Iskandar from Primate Research Center, Bogor Agricultural University and Prof. Wasmen Manalu, PhD from Faculty of Veterinary Medicine, Bogor Agricultural University for improving the dissertation.

I also want to send my gratitude to Dra. Taruni Sri Prawasti.M.Si, Dr. Dedy Duryadi Solihin, Tri Heru Widarto, M.Sc, Dr. Dyah Perwitasari, Dr. Rika Raffiudin, Dr. Tri Atmowidi, Msi for their support during the study.

I deeply send my gratitude to Yayasan SDM IPTEK HABIBIE CENTER for the scholarship for academic year 2008/2009 and 2009/2010. I also want to send my appreciation to program SANDWICH LIKE from Directorate General of Higher Education which gave me financial support when I work in Primate Research Institute, Kyoto University, Japan.

I also want to appreciate and gives love to Ucok, Sukhoi and Kerok for their great performances.

Bogor, February 1st ₂₀₁₂

Author was born in Bogor on 29th _{of September 1982 as the first child of}

four from the parent, Hidayat Banjaransari and Sri Wilujeng.

In 1999, author graduated from SMUN 3 Bogor and enrolled the Department of Biology, Faculty of Mathematics and Natural Science, Bogor Agricultural University and graduated in 2004. At the same year, author enrolled in the Graduate School, Study Program of Biology, Bogor Agricultural University. In 2007, author enrolled her Doctoral Course at Bogor Agricultural University, majoring in Animal Bioscience. In 2007, author officially accepted as a staff in the Department of Biology, Bogor Agricultural University.

Page

LIST OF FIGURES... xiv

APPENDIX …... xiv

INTRODUCTION …... 1

LITERATURE STUDY... 3

Categorization in Human ... 3

Categorization in Animal... 3

Operant Conditioning Behavior... 5

METHODS... 6

Subjects... 6

Apparatus... 6

Stimuli and Procedure... 7

a. Human vs Macaques... 10

b. Macaca fascicularis vs Other Macaques... 11

c. Human vs Non-Human... 12

Data Analysis... 13

RESULTS... 14

Human vs Macaques... 14

Macaca fascicularis vs Other Macaques... 18

Human vs Non-human... 20

DISCUSSION... 22

CONCLUSION... 27

Page 1 Schematic diagram of operant conditioning match-to-sample task of

the experiment in baseline (a) and test trials (b)... 8

2 Example of stimuli used for Human vs Macaques experiment... 9 3 Example of stimuli used for Human vs Non-Human experiment …... 13 4 Monkeys performances in categorizing human and monkeys in color

stimuli... 15

5 Monkeys performances in categorizing human and monkeys using black and white stimuli...

16

6 Monkeys performances in categorizing human and monkeys using resized stimuli...

16

7 Monkeys performances in categorizing human and monkeys using black and white blurred stimuli...

17

8 Monkeys performances in categorizing human and monkeys using resized, blurred, black and white stimuli...

17

9 Monkeys performances in categorizing conspecific and other monkeys using colored stimuli...

19

10 Monkeys performances in categorizing conspecific and other monkeys using black and white stimuli...

19

11 Monkeys performances in categorizing human and non-human using colored and black and white stimuli...

21

APPENDIX

Page 1 Monkeys performance in Human vs Macaques experiments... 32 2 Monkeys performance in M. fascicularis vs other Macaques

experiments... 34

3 Monkeys performances in Human vs Non-Human experiments... 35 4 Test of the similarity of the monkey performances between baseline

and test trials …... 36

INTRODUCTION

We live in a world full of ever-changing objects. It is impossible for us to memorize each and every new individuals we encountered. Our brain need to have an ability to deal with infinite informations that is continuously coming from the surrounding environment. One way to overcome the memory storage constraint is to represent factual objects as conceptualized mental images. A concept concludes a lot of informations that had been collected at the time we perceive the object and abstracts them to assist the process of grouping individuals or events into different categories (Rosch et al. 1976). These categories reduce the number of bits of information to manageable classes. There are several levels of categorization and within a taxonomy the levels are nested (Rosch et al. 1976). At the most concrete level of categorization, all or most members of the category shared common physical attributes that differ from other category. The higher the level of category, the fewer common attributes between members of the group.

In addition to humans, the ability to categorize has also been proposed in the animals. Being able to identify, visually or otherwise, a new object as a member of a category is an advantage for the animals. It should help them to distinguish between food or non-food, and to discriminate between species of animal. This species discrimination is important to prevent hybridization among species (Yoshikubo 1985, Fujita 1987, Fujita and Watanabe 1995, Fujita et al.

1997).

Present experiment aims to test the ability of species discrimination by primates. I did three major experiments. First, I tested whether the monkeys were able to categorize monkey individuals as a class against human individuals as another class. Second, I tested whether they were able to discriminate their conspecific from other macaque species. Third, I tested their ability to categorize non primate animal as non-human class against human individuals as human class. Previous studies showed that monkeys were able to see photos as representations of real objects (Kyes et al. 1982). I therefore used photos of various humans, monkeys and other animals as stimuli to infer their categorization

ability. The stimuli consisted of information on physical properties of the object. The physical properties represented in photos may be modified in many ways to alter the amount of information in the stimuli. These modified photos provide a way to deduce levels of categorization of the monkey. Some experiments found that the monkeys look to their conspecific longer than to non-conspecific so they used duration of visual fixation to indicate this discriminative capability (Demaria and Thierry 1988, Fujita 1993, Fujita and Watanabe 1995, Fujita et al. 1997). However, although counting the fixation time is easy to be described quantitatively, it is uncertain as to what the reaction time measures. For instance, the longer the time could represent two facts; the monkeys likes the picture in the stimuli, or it can be the opposite. In fact, in agonistic bouts, the amonkeys tend to look longer to their opponent (de Waal et al. 1976). Thus, the reaction time may better be interpreted as a measure for the attention of the monkeys and may not carry information about species discrimination. Moreover, it is not clear whether the monkeys are truly able to distinguish between species or their familiarity with their own species due to experience in their life history made them see conspecific longer. Instead of counting fixation time of each stimulus, present experiment test their categorization ability in discriminating dichotomous-stimuli of two different classes of animal.

LITERATURE STUDY

Categorization in Humans

Categorization is the ability to put an object to a group based on some internal representations of the group (Rosch et al. 1976). The ability to categorize objects and events and extending this behavior to new instances is fundamental to many human activities. I sort the objects and events around us into categories, while still being able to recognize some or all of the individuals. In general, categorization could be divided into two levels (Behl-Chadha 1996). If the detected physical properties of the individuals within a category are mostly similar, we call it as basic level of abstraction (Rosch et al. 1976). For example; we could put Asians, Africans and Caucasians into one group; that is a human group. If the connection between members of one category is not only based on perceptual similarity but more on relations between concepts, we call it as abstract level of categorization (Mervis and Rosch 1975). For example, we can put rice, bread and fruit into the food or edible group and put a chair, a plane and a soap into the non-food group. Although it is basically divided into two levels, the relation between each class often overlap and sometimes it is very difficult to differentiate between classes. These levels of categorization are basic to understand language, number and social relationship with other humans. This ability obviously needs memory, learning and reasoning.

In humans, categorizing behavior was predicted to start at infantile age. Quinn (1993, 2002) found that the 3-4 month old infants attended to natural objects as though they belong to groups of basic level. The experiments used the preferential looking procedure and measure the looking time of the pictures. The infants tended to look at the picture of a new categories for longer period of time. Their ability was regarded as in the basic-level of categorization.

Categorization in Animal

The field of research on the ability of animals to categorize objects was opened by the pioneering study of Herrnstein and Loveland (1964). They showed

that the pigeon were able to sort photographs on the basis of whether the image contained human being or not. It means the pigeon were able to make a class of human and a class of non-human, which is the definition of categorization. Since then, several studies have demonstrated categorization in other animals, most of them used primates as models (ex: Tanaka 2001, Jitsumori and Matsuzawa 1991, Santos 2000). Some of the researches aimed to know whether categorization processes in the animal is based on similar processess compare to human (ex: Farbe-Thorpe 2003, Freedman et al. 2002,2003, Hampson et al. 2004). In other case, some studies wanted to find out the levels of categorization that could be performed by the animals (ex: Tanaka 2000, Jitsumori and Matsuzawa 1991, Santos 2000, Inoue et al. 2008, Vonk & McDonald 2002; Vonk & McDonald 2004). In both cases, the studies used photos of biologically significant objects, such food vs non food, animals vs non-animals, and the last is own species vs non-own species or species discrimination (Tanaka 2001, Jitsumori and Matsuzawa 1991, Santos 2000, Inoue et al. 2008, Fujita et al 1995, 1997, Vonk & McDonald 2002; Vonk & McDonald 2004).

Species discrimination is important to prevent hybridization among species (Yoshikubo 1985, Fujita 1995, Fujita et al. 1997). Some studies were conducted to find out which part of the body are really important to identify and categorize the species. Those studies found that face provide information about species, sex, age and emotion of an individual (Pascalis and Bachevalier 1988, Tomonaga 2007). Fujita et al. (1995) demonstrated that Sulawesi macaques performed basic level of categorization in identifying their conspecific. The stimuli used in that experiment were photos of monkey presented sequentially. By counting the fixation time of each stimulus, they found that the Sulawesi macaques see their conspecific longer than non-conspecific. Thus, the photos carry specific information which could be categorized by conspecific or non-conspecific (Fujita et al. 1995, Pascalis and Bachevalier 1988, Tomonaga 2007). However, although the fixation time is easy to be described quantitatively, it is difficult to draw conclusion from it. For example, the longer reaction time could mean that monkey likes the picture in the stimuli, or it could be the opposite. In fact, in agonistic bout, the monkeys tend to

look longer to his/her opponents (de Waal et al. 1976).

Looking at previous work in primates, there are some data on categorization in great apes, M. mulatta, M. fuscata and Sulawesi macaques but not really in long-tailed macaques (M. fascicularis). A few studies showed that M. fascicularis were able to see photos as a representations of the real objects (Dasser 1987, Kyes et al. 1982). More over, some studies also found that those monkeys were able to identify individual within species (Dasser 1987), read facial expression from the drawing and could discriminate between drawings of some bodies of old world primate species (Dittrich 1994). However, there were no direct experiment to infer the level of categorization in this species.

Operant Conditioning Behavior

In this experiment, I used the operant conditioning methods which refers to process in which the frequency of occurrence of the particular type of behavior was modified by the consequences of the behavior (Reynolds 1975). In operant conditioning, the behavior (called operant) came from animal motivation to response to a given stimulus. The stimulus used in operant conditioning is discriminative. In this particular experiment, monkeys were trained to choose a specific stimulus (the plus stimulus, S+) from another alternative to get a reward. If they choose the other (the minus stimulus, S-), they would not get any rewards. The behavior to choose a specific card is an operant. The reward is the consequences of the behavior and it should increase animal motivation to choose the plus stimulus. The specific task in this operant conditioning method is matching to sample (Miller et al. 1996) with photographs of human, macaques and other animals as stimuli.

METHODS

Subjects

The subjects were two adult M. fascicularis males, named Ucok and Sukhoi and one female named Kerok. Ucok were born in Pangandaran National Park, with present age at least 19 years old. Kerok and Sukhoi were born in Department of Biology, Bogor Agricultural University with ages are 14 and 9 years old, respectively. Both male monkeys were maintained at the Department of Biology of Bogor Agricultural University. The female monkey was maintained at Primate Research Institute, Kyoto University. They were reared in individual cages and tested in the same cage. Experiments were conducted according to the Guide for the Care and Use of Laboratory Animals by the National Institute of Health, U.S.A. (1985), and the Guide for the Care and Use of Laboratory Primates by the Primate Research Institute, Kyoto University (1986, 2002). The training phases were held about two years for Ucok and one year for Sukhoi and two months for Kerok depended on the ability of learning of the monkeys. The test phase was conducted for few months for all monkeys.

Apparatus

A modified version of the Wisconsin General Test Apparatus (WGTA) was used for the presentation of stimuli to the macaques. The apparatus used consisted of a horizontal tray containing three shallow food-wells mounted on a portable shelf. A sample stimulus and a matching stimulus were placed over the food-wells so that they always associated with a food reward beneath. When setting up the food reward between trials, an opaque screen was placed between the cage and the experimenter to prevent animals from seeing the process. The eye of the experimenter was covered with sunglasses, so the monkey could not use eye direction of the experimenter as cue to find the matching stimulus. To control the animal’s motivation level, animals were deprived of food after 5 pm until the next morning before the day of a training session or a test session. Foods were provided either during or after the experiments.

Stimuli and Procedure

I assessed the ability of M. fascicularis in categorizing objects by conducting three experiments. First, Human vs Macaques experiment, aimed to test the ability of M. fascicularis in categorizing monkeys and humans into two different conceptual classes. Second, M. fascicularis vs Other Macaques experiment, to test the ability of M. fascicularis in categorizing conspecific differently from M. mulatta or M. fuscata. And the last one, Human vs Non Human experiment, to test the ability of M. fascicularis in categorizing various kinds of animal differently from human.

Fujita et al. (1995) demonstrated the ability of species discrimination of Sulawesi macaques using photos as stimuli. Other studies of categorization also used photos (Demaria and Thierry 1988, Fujita 1993, Fujita and Watanabe 1995, Fujita et al. 1997) but those former studies did not control feature of the photos so it is difficult to delineate which component of the photo provides informations to do categorization. Each species of the animal possess distinctive non-facial, physical features; for instance, posture and proportion of legs which may or may not be present at the photo frames. For that reason, I used facial photos and controlled the background color by changing it into green. Because of limitation of visual fixation time to infer species discrimination as noted in Introduction, present experiments tested their categorization ability based on dichotomic discrimination of one species against other. To do this, I employed match-to-sample task that often were used in working memory experiments (Miller et al.

1996, Rao et al. 1997) and later to study categorization experiments (Freedman et al. 2002; Hampson et al. 2004; Inoue et al. 2008; Tanaka 2001). Using operant conditioning method, at first, the monkeys were trained to respond only to particular stimulus, not based on previous experience in their life. Every trial consisted of two presentations. The first presentation shows sample stimulus, and the second matching and distractor stimuli side by side. The sample and matching stimuli always belonged to a same category that dichotomically differ to the distractor (Figure 1). I assumed that each stimulus was independent from each other so a stimulus could be defined as a matching stimulus in one trial and could

be a distractor in other trial depended to the sample stimulus.

I trained the monkeys to associate sample and matching stimuli against distractor stimulus. First, I showed the monkeys a sample stimulus as reference to be matched (Figure 1a). To ensure that the subjects paid attention to the sample stimulus, they had to touch the sample stimulus and for this they received reward that they may find beneath the stimulus. Next, I presented side-by-side matching and distractor stimuli which one had to be chosen by the subjects. When the subjects chose the matching stimulus, they received food as a reward; the response was counted as a correct one. When they chose the distractor, they did not get any rewards. Subject's motivation for reward warranted the choice of the matching stimulus. I repeated the trials consecutively with inter-trial intervals for about 30 second. The location of matching and distractor stimuli on the tray was arranged pseudorandomly. I blocked every 20 trials into one session and measured their correct rate. If the subjects chose the matching stimuli higher than 90% in a

session, it means they were able to associate matching to sample stimuli. Then I introduced a set of new photographs of the same human and the same monkey. I repeated this training until the same criterion was reached (90% correct rate in 3 continuous sessions). After the monkey cleared the criterion of third set, I introduced the photographs of a different human and a different monkey. This procedure was repeated for six set of new humans and monkeys. After the monkey cleared all of these training phases, I started the test session. At this stage, logically, this may be inferred as the subjects had developed dichotomic concepts of matching against distractor. When they showed this competence, they went to test phase (see below for details) where the matching photos were new ones. If the subjects were able to transfer the learned concept to novel stimuli, it proved that they could categorize.

a. Human vs Macaques

In this experiment, I assessed categorization ability of M. fascicularis to discriminate monkeys from humans using facial features. Both primates possess distinctive physical features, notably different shape of face and the presence of hair in the monkey's face which is absent from human. These distinctions would allow us to safely infer the discriminative ability on the different conceptual class of human and monkey. For monkey category, I used photos of both sexes of three species of macaques, those were M. fascicularis, M. mulatta and M. fuscata. For human category, I used female and male photos. Some of the female heads were covered by scarf. Nevertheless, all pictures showed the whole face.

For training phase, sample and matching stimuli were the same and monkey's stimuli were always M. fascicularis (see Figure 2b and 2d). For practical reason, I used 90% correct rate as threshold to infer subject's competence in discriminating human from monkey categories. Furthermore, to make sure they have achieved this ability, they have to show it in three successive sessions.

To test if a subject could apply the learned concept of human and monkeys to new individuals, I changed the pictures of matching stimulus with different individuals that belonged to the same category as the sample stimuli (eg. Figure 2c , 2e, 2g); for example, if the sample stimulus is monkey b then the matching stimulus is monkey d. The subject should compare the novel matching stimulus to the available distractor. If the subject associate the sample and the different-picture of matching stimuli, I may infer that they put those two different-pictures into one class that dichotomically differs from the class of distractor. This would evidenced they transfer the human and monkey concepts to novel stimuli. This would prove their categorization ability. For practical reason, the test sessions consisted of baseline and test trials in certain proportion. In the baseline trials, the stimuli is the same as in training trials so this would provide a reference to compare the test trials.

The stimulus photos give subjects various informations about color and shape of each species. Thus, I expected that subjects used informations from detailed physical properties of sample and matching stimuli to perform basic level

of categorization. To test whether the subjects has levels of abstraction, I extended the experiment to see whether the subjects would still have the ability to identify objects if the informations on the physical properties of the stimuli were reduced. For this reason, I deviced extended experiments that stripped off certain visual information from the stimuli. First, I took color off the photos and presented it in black and white. In the next step, I tried to test whether size of the stimuli could affect the performance of the subjects in categorizing humans differently from monkeys. I thought that altering the size of the photos would also perspectively change the shape of the figures. I reduced the size of the stimuli to one quarter of the original and showed it to monkeys. In this case, I tested them using pairs of photos of colored human and M. fascicularis. I continued the test by giving the subjects the blurred, black and white, original size photos of human and macaques. These photo manipulations aimed to test whether the subjects could categorize human differently from monkeys though the stimuli lack informations about color and shape. In my last attempt to ascertain whether the subjects could categorize human differently from macaques, I reduced the size of the black and white, blurred photos of human and monkeys to one quarter of the original size. The complete and modified physical properties of the photo stimulate the varied amount of information perceived on the objects. Thus, by comparing subject's responds to the stimuli, I may interpret their level of categorization. In total, I used 82 photos as the stimuli for the test.

b. Macaca fascicularis vs Other Macaques

In this experiment, I assessed categorization ability of M. fascicularis to discriminate their own species from other species of macaques. In this experiment, I used several facial photos of M. fascicularis, M. mulatta and M. fuscata. I did this in two conditions, that is in color and black and white.

After the first experiment of Human v Macaques, I assured that monkeys have mastered to do maching-to-sample task. In this intra-generic experiment I used two paradigms, that is, M. fascicularis vs M. mulatta and M. fascicularis vs M. fuscata. Since those macaques shared almost similar facial feature, I wondered

whether subjects were able to find the difference between each species and use it to group them into separate class. In test phase, I changed the pictures of matching stimulus with different individuals that belonged to the same species. Same as the first experiment, the test session consisted of certain proportion of baseline and test trials. In baseline trials of both paradigms, the photos of matching stimuli were the same to sample stimuli. In total, we used 40 photos as the stimuli for the test.

c. Human vs Non-Human

In this experiment, I assessed categorization ability of M. fascicularis to discriminate various kinds of animals from human using facial features. In this experiment, I used several photos of human (see the first experiment) and non-primate animals, such as mammals, reptiles and amphibians (Figure 3). I wondered whether subjects were able to ignore the physical difference of various animals and creating a new class, that is non-human group. As in the second experiment, I did this in color and black and white. In test trials, I changed the pictures of matching stimulus with different individuals that belonged to the same category as the sample stimuli; for example, if the sample stimulus was monkey 2b then the matching stimulus was animal 3a. The subjects should compare the novel matching stimulus to the available distractor. If the subjects could associate the sample and the different-picture of matching stimuli, I may infer that they put those two pictures into one class that dichotomically differs from the class of distractor. This would mean they are able to categorize non-human animals differently from human. The baseline stimuli were the same stimuli used in Human vs Macaques experiment (see Figure 2). In total, I used 68 photos as the stimuli for the test.

Data Analysis

In this experiment, monkeys have to respond to particular stimuli. I defined the response of monkeys as correct when they choose matching stimulus, and false when they choose distractor stimulus. This binomial responses was dependent on stimuli presented in baseline and test trials; I therefore treat baseline and test trials as independent variables and the effect of individuals in the analysis as random effect. Thus, I used Generalized Linear Mixed Model (GLMM, Venables and Ripley 2002) using R software version 2.10.1 (R Development Core Team 2010) to analyze the discrete, binomial data.

In test phase, I also conducted several baseline trials using stimuli that used in training phase and test trials using new stimuli. I expect that monkeys were able to transfer their concept learned in training phase into new stimuli by showing the same performance in both baseline and test trials. For that reason, using GlmmPQL, I compare their results in both trials.

RESULTS

Human vs Macaques

As mentioned before, this experiment aimed to test the ability of monkeys in categorizing monkeys and humans into two different conceptual classes. The two classes were presented dichotomously as matching and distractor stimuli. In baseline trials, the photos of matching stimuli were the same to sample stimuli. As expected, in this trials all subjects could associate matching to sample stimuli. They showed high performance of choosing the correct stimuli at the proportion around 90% of the trials. Because the matching stimuli were always of a different category to the distractor stimuli, it might mean they developed different and mutually exclusive concepts of human and monkeys. These results provided a reference to test whether subjects could transfer the newly developed mental concepts to identify new individuals. I did this by changing the matching stimulus with pictures of different humans and different kinds of monkeys. In this test, I found that all subjects associate sample stimuli to new individuals of matching stimuli (Figure 4, Appendix 1, Appendix 4). This same result to baseline (glmmPQL, p=0.15; Venables and Ripley 2002) would indicate that the subjects were able to correctly identify new individuals as members of its own class or of human class. This would mean that monkeys categorized humans differently from monkeys.

Since the stimuli used in the above experiment contained information on physical properties (that is, shape and color) of the objects, I suggested that the subjects used those properties to create a concept (see Discussion for detailed). I was wondering whether the subjects would still have the ability to identify objects if the informations of the physical properties were reduced. I expected that the subjects will be able to gain informations from the stimuli with a reduced physical properties and combine them with previous concepts they learned before and this would lead them to create higher level of abstraction. For this reason, I deviced three experiments that stripped off certain visual information from the stimuli.

First, I took color off from the photos and presented it in black and white. I found that the subject's performance in both baseline and test phases were the same (glmmPQL, p=0.19) (Figure 5, Appendix 1). This result indicates that even without color, the subjects categorized humans differently from monkeys.

In the next step, I tested whether size of the stimuli could affect the performance of the subjects in categorizing humans differently from monkeys. I thought that altering the size of the photos would also perspectively change the shape of the figures. I reduced the size of the stimuli to one quarter of the original and showed it to monkeys. In this case, I tested them using pairs of photos of colored human and M. fascicularis. I found that the subject's performance in both baseline and test phase were the same (glmmPQL, p=0.85)(Figure 6, Appendix 1, Appendix 4). It means the subjects did not affected by the reduction in size of the stimuli.

I continued the test by giving the subjects the blurred and black and white original size photos of human and macaques. These photo manipulations tested whether the subjects could categorize human differently from monkeys though the stimuli lack informations about color and shape. I found that the subjects were able to categorize human separately from monkeys. Their performance in both baseline and test phases were the same (glmmPQL, p=0.53) (Figure 7, Appendix 1, Appendix 4).

In my last attempt to find out whether the subjects could categorize human differently from macaques, I reduced the size of the black and white, blurred photos of human and monkeys to one quarter of the original size. I found that even in a condition lacking important physical informations, such as color, shape and size, their performance in both baseline and test phases were the same (glmmPQL, p=0.57) (Figure 8, Appendix 1, Appendix 4). It means that the subjects could categorize human differently from monkeys. Overall, by seeing the high performance of the subjects when tested by the manipulated stimuli, I concluded that they were able to develop higher level of abstraction based on available concepts they have learned before.

Macaca fascicularis vs other macaques

This experiment aimed to test the ability of monkeys in categorizing M. fascicularis differently from other macaque species. In this experiment I used two paradigms, that is, M. fascicularis vs M.mulatta and M. fascicularis vs M. fuscata. In baseline trials of both paradigms, the photos of matching stimuli were the same to sample stimuli. Thus, all subjects could associate matching to sample stimuli at around 90% of trials. Similar with human vs macaques experiment, the matching stimuli were always of a different category to the distractor stimuli, so this result might mean they developed different and mutually exclusive concepts between

M. fascicularis and the other macaques; that is M. mulatta and M. fuscata. This result provided a reference to test whether the subjects could use the developed concepts of M. mulatta to identify new individuals of the species; the same also true for M. fuscata. I did this by changing the matching stimulus with pictures of

different individual of monkeys. I found that all subjects associate sample stimulus to new individuals of matching stimuli (Figure 9, Appendix 2, Appendix 4). This same result to baseline (glmmPQL, p=1) indicate that the subjects were able to correctly identify new individuals as members of their respective class. This would mean that the subjects categorized M. fascicularis differently from the two other macaque species.

As in the first experiment, I ascertained this categorization ability by taking color off the photos used in the stimuli and presented it in black and white color. I found that all subjects associate sample stimulus to black and white version of the matching stimuli (Figure 10, Appendix 2, Appendix 4). This same result to baseline (glmmPQL, p=1) would indicate that even without color, the subjects could also categorize M. fascicularis differently from other macaques.

Human vs Non-Human

This experiment aimed to test ability of monkeys in categorizing human differently from non-primate animals. In this experiment, I used several photos of human and non-primate animals, such as mammals, birds and reptiles and amphibians. I tested the subjects using both color and black and white photos. For baseline trials, I used same photos of human and M. fascicularis that used in baseline trials of Human vs Macaques experiment. I found that the subjects did high performance in test trials, similar to that in baseline trials (glmmPQL, p=0.65 for colored stimuli and p=0.70 for black and white stimuli) (Figure 11, Appendix 3, Appendix 4). It means that they were able to maintain concept of human and developed new concept about non-human.

DISCUSSION

Fujita and his coworkers (1993, 1995, 1997) found that monkeys see their conspecific longer compared to other species, and this were thought to lead to species discrimination. His works using various kinds of pictures of macaques give us an insight into how monkey learn to categorize via species discrimination. However, the methods that he used could be doubted; for example, there were several interpretation for how monkeys see conspecific in longer duration. First, the longer reaction time might mean that monkey's attention varied in response to different pictures regardless of their species membership. Second, they were mostly wild born so they have already familiar with their own species since opportunities to learn facial properties of members of its own species exist in the life history of the monkeys. Thus, monkeys may prefer to look longer at a picture that is similar to their group mate and, by doing so, exclude pictures of another species. Although it can be considered as species discrimination, it's still not clear if monkeys can discriminate various species excluding their own. Third, since most experiments (Demaria and Thierry 1988, Fujita 1993, Fujita and Watanabe 1995, Fujita et al. 1997) did not control features, such as backgrounds, of the stimulus, it is hard to conclude which part of the stimulus attracts the monkeys attention.

The method of matching to sample task I used might be more reliable than counting perceptual duration for several reasons. The task of associating matching and sample solves the first and second problems of Fujita (see experiments on M. fascicularis vs non M. fascicularis below). More over, by controlling the background of stimuli, and the use of a specific body part (that is, face), I tried to reduce information noise which may distract my deduction. Thus, since my stimuli represent real objects (Dasser 1987, Kyes et al. 1982), this experiment showed that monkey could discriminate species. Therefore, my method should make a strong conclusion about the ability of categorization in M. fascicularis.

The first experiment to show the ability of categorization in M. fascicularis was the discrimination of human vs monkey. In the training phase, I

introduced pictures of human and monkey to the subjects. There were consistent similarities and differences of physical properties between human and monkey pictures. It is not unreasonable to conclude that the subjects used these similarities and differences to create a conceptualized mental image of human that differs to that of monkey. There are some physical properties that provide information to create concepts. First is shape. The global and local shapes of human's and monkey's faces are very different. While human faces are oval, monkey's faces are rounded with hair on it. Another possibility is that the subjects detected the presence of the eye. This mechanism, called “eye direction detector”, is important to understand facial emotion (Farroni et al. 2002, Baron-Cohen et al. 1999). Given the eyes, its angles with nose and lips of both species are different. The second is color. In this experiment, I used colored photographs. As seen in stimulus photos (Figure 2), compared to uniformly presented background color, the global color of the monkeys was different to that of human. Indeed, this had been shown by Santos (2001) that colors are used as information to categorize object. Those informations may help the subjects to recognize and discriminate between categorically human and monkey.

The resulting concepts of human and monkey learned from training phase were used by M. fascicularis individuals as basis to categorize pictures in the test phase. In every trial of test phase I changed the baseline pictures with new ones. This would prove the ability of the subjects to transfer its concepts to respond to novel stimuli. For the monkey pictures, I used three different species of macaques. I found that proceeding from baseline to test trials performance of the subjects remained the same. These indicate that the subjects categorized monkey as monkey and human as human. It is interesting to note that although two of the subjects (Ucok and Sukhoi) had never saw other species (M. fuscata and M. mulatta) for their entire life, they categorize photos of those species as monkey instead of human which in captivity they see everyday. It might be concluded that they had concepts of human and monkey based on different color and shape of the stimuli. I may interpret this fact as the monkeys recognized natural discontinuity between the two class; Rosch et al. 1976 called this as basic level of abstraction.

As this kind of categorization ability was based on perceptual similarity of physical attributes of the photos discussed above I may also describe this as the concrete level of abstraction (Vonk and MacDonald 2002).

In nature, sometimes there are conditions (such as at long distance or less light conditions) where monkeys could not see detailed information of object to do categorization. Thus, they have to be able to categorize objects based on broad, generalized concept that is developed on concept with complete and detailed physical percepts. According to Rosch et al. (1976), the world is structured because real-world properties do not occur independently of each other. Because of this correlation, few properties of an object would suffice to predict the whole properties. After seeing the results of the above test (which were positive), I wondered whether the subjects would still have the ability to identify objects if the informations on the physical properties of the stimuli were reduced. This would simulate natural conditions and further proof the ability to transfer the learned concept in my subjects. I then manipulated the stimuli by taking color off the matching and distractor stimuli. The result showed that the ability of all of the subjects to categorize between human and monkey was not hampered. They could successfully transferred their mental image to the black and white photos. This fact showed that the subjects did not depend on color and use other informations to discriminate the stimuli. Some studies showed that monkey may use global shape (Dittrich 1994) and local features (Demaria and Thierry 1988) of object as informations in categorizing object. Jitsumori and Matsuzawa (1991) showed that

M. mulatta and M. cyclopsis wereable to classify human shilouettes as human and assume that monkeys used global shape to recognize human. Rosch et al. (1976) explained that levels of abstraction are nested. At the basic level, objects within category shared most similarity in concrete properties. When it goes to the more abstract levels, it share less properties among each other. Thus, we may follow that the use of global shape is one way monkey used to generalize the physical properties of an object. To test whether the subjects might be able to generalize their physical percepts, I blurred or resized the stimuli to reduce and/or to change the local shape informations carried by the stimuli. Again, in all of these extended

tests, the M. fascicularis subjects showed high performance of discrimination. This ability to classify blurred or resized photos of human as human and blurred or resized photos of monkey as monkey demonstrated that the subjects were looking for features of blurred or resized matching stimulus that have common properties with those of sample stimulus. It seemed that the subjects subtracted the local shape differences to get a generalized concept of object. Compared to basic level of abstraction proofed before, this generalized level of abstraction showed that the subjects performs at least two levels of categorization by learning from their experience in training and test phases.

For second experiment, I tested the ability of M. fascicularis in discriminating M. fuscata or M. mulatta against their conspecific. Those macaques belong to fascicularis group (Fooden 1969) so they shared many similar physical properties of face, such as the existence of facial hair and other local shapes. Thus, the monkeys have to extract the differences from total facial properties to get the uniqueness of each species. Result of the experiment showed that monkeys could discriminate their conspecific from other species. They may used color as information, since the hair color of M. fascicularis in the stimuli were lighter than that of M. mulatta and darker from M. fuscata. More over, the facial color of M. fascicularis is relatively brownish while M. mulatta paler and M. fuscata reddish. In Human vs Macaques experiments, when color information was discarded, the monkeys used (whether global or local) shape informations to categorize the two different classes. Thus, I took the color off the stimuli and found that the subjects again showed high performance of discriminations. This adds to my conclusion that the subjects were able to detect the differences in physical properties of each species and use it as basis in discrimination. This experiment showed the categorization ability the other way around from the first experiment; that is, the transfer of concept from general to specific.

The Human vs Macaques and M. fascicularis vs Other Macaques experiments showed that M. fascicularis were able to develop concept using informations from similarity and difference of physical properties to put objects into one particular class. However, in some conditions, it takes more than physical

similarity to put objects into one group. One object may relate to others based on relations that could not be detected easily by sensory properties. Thus, in this case, the subjects need the ability to associate different conceptualized mental images to discriminate stimuli lacking in physical clues. The class that is created does not rely on perceivable features and may reflect a more conceptual understanding of category belongingness. This kind of concept assisted higher level of categorization. To test whether M. fascicularis were able to create such a more abstract concept, I did Human vs Non-Human experiment. In two previous experiments, there are physical similarities between all sample and matching stimuli. In Human vs Non-Human experiment, I dismissed physical similarities within one category; that is, Non-Human class consisted of photos of animals such as mammal, bird, reptile and amphibian which were different in many ways. At first I tested them with colored stimuli. Our subjects showed high performance in discriminating between human and non-human classes. Since there are no similarity within non-human category, I may suspect that the subjects use color information to do categorization. Still, if I took color off the stimuli, the subjects discriminated non-human from human. I assumed that the monkeys created a new concept, that is concept of non-human that differ from concept of human. Thus, the result clearly showed that the subjects could perform higher level of abstraction.

CONCLUSION

These studies showed that M. fascicularis were able to discriminate several species of animals. They did it by creating concepts that were based on common features of the stimuli. Thus, I concluded that M. fascicularis were able to perform multiple levels of categorizations.

REFERENCES

Baron-Cohen S, Ring HA, Wheelwright S, Bullmore ET, Brammer MJ, Simmons A, and Williams SCR. 1999. Social intelligence in the normal and autistic brain: an fMRI study. European J. of Neurosci. 11: 1891-1898.

Behl-Chadha. 1996. Basic level and superordinate-like categorical representations in early infancy. Cognition 60 :105-141.

Dasser V. 1987. Slides of Group Members as Representations of the Real Animals (Macaca fascicularis). Ethology. 76: 65–73.

Demaria C, Thierry B. 1988. Responses to Animal Stimulus Photographs in Stumptailed Macaques (Macaca arctoides). Primates 29: 237-244.

de Waal FBM, van Hooff, Netto W. 1976. An Ethological Analysis of Types of Agonistic Interaction in a Captive Group of Java-monkeys (Macaca fascicularis). Primates 17: 257-290.

Dittrich W. 1994. How monkeys see others : Discrimination and recognition of monkeys' shape. Behav. Processes 33: 139-154.

Eimas PE and Quinn PC. 1994. Studies on the formation of perceptually based basic-level categories in young infants. Child Development65:903–917. Fabre-Thorpe M. 2010. Visual categorization: accessing abstraction in

non−human primates. Phil. Trans. R. Soc. Lond. B 358: 1215-1223.

Farroni T, Csibra G, Simion F, Johnson MH. 2002. Eye contact detection in humans from birth. Proc. Natl. Acad. Sci. 99: 9602–9605.

Freedman DJ, Riesenhuber M, Poggio T, Miller EK. 2002. Visual categorization and the primate prefrontal cortex: neurophysiology and behavior. J Neurophysiol 88:929-941.

Freedman DJ, Riesenhuber M, Poggio T, Miller EK. 2003. A comparison of primate prefrontal and inferior temporal cortices during visual categorization. The Journal of Neuroscience 23: 5235-5246.

Fujita K. 1987. Species recognition by five macaque monkeys. Primates 28: 353-366.

Pigtail Monkeys. Primates 32: 133-140.

Fujita K and Watanabe K. 1995. Visual Preference for Closely Related Species by Sulawesi Macaques. Am. J. Primatol. 37: 253-261.

Fujita K, Watanabe K, Widarti TH, Suryobroto B. 1997. Discrimination of macaques by macaques: the case of Sulawesi species. Volume 38: 233-245. Hampson RE, Pons TP, Stanford TR, Deadwyler SA. 2004. Categorization in the

monkey hippocampus: A possible mechanism for enconding information into memory. Proc. Natl. Acad. Sci 101: 3184-3189.

Herrnstein RJ and Loveland DH. 1964. Complex visual concept in the pigeon. Science 146 : 549-551.

Inoue T, Hasegaway T, Takara S, Lukatz B, Mizuno M, Aou S. 2008. Categorization of biologicallsignificant objects, food and gender, in rhesus monkeys I.. Behavioral study. Neuroscience Research 61: 70-78.

Jitsumori M and Matsuzawa T. 1991. Picture perception in monkeys and pigeons: transfer of rightside-up versus upside-down discrimination of photographic objects across conceptual categories. Primates 32: 473-482.

Kyes RC, Mayer E, Bunnel BN. 1992. Perception of stimuli presented as photographic slides in cynomolgus macaques (Macaca fascicularis). Primates 33 :407-412.

Mervis CB and Rosch E. 1981. Categorization of Natural Objects. Ann. Rev. Psychol 32:89-115.

Miller EK, Erickson CA, Desimone R. 1996. Neural mechanisms of visual working memory in prefrontal cortex of the macaque. The Journal of Neuroscience 16 :5154-5167.

Pascalis O and Bachevalier J.1998. Face recognition in primates : a cross-species study. Behav. Processes 43: 87-96.

Quinn P. 2002. Category representation in young infants. Current Directions of Psychol.l Sci. 11:66-70.

Quinn PC, Eimas PD, Rosenkrantz SL. 1993. Evidence for representations of perceptually similar natural categories by 3-month-old and 4-month-old infans. Perception 22:463-475.

R Development Core Team. 2010. R: A language and environment for statistical computing. R Foundation for Statistical Computing, Vienna, ISBN 3-900051-07-0. http://www.R-project.org/. [August 10, 2010].

Rao SC, Rainer G, Miller EK. 1997. Integration of What and Where in the Primate Prefrontal Cortex. Science 276 :821-824.

Reynold GS. 1975. A Primer of Operant Conditioning. Illinois: Scott, Foresman and Company.

Rosch E, Mervis CB, Gray WD, Johnson DM, Boyes-Braem P. 1976. Basic object in natural categories. Cogn Psychol 8: 382-439.

Santos LR, Hauser MD, Spelke ES. 2001.Recognition and categorization of biologically significant objects by rhesus monkeys (Macaca mulatta): the domain of food. Cognition 82 :127–155.

Tanaka M. 2001. Discrimination and categorization of photographs of natural objects by chimpanzees (Pan troglodytes). Animal Cognition 4:201-211. Tomonaga M. 2007. Visual search for orientation of faces by a chimpanzee (Pan

troglodytes): face-specific upright superiority and the role of facial configural properties. Primates 48:1–12.

Venables WN and Ripley, B. D. 2002. Modern applied statistics with S. Fourth edition. USA: Springer.

Vonk J and MacDonald. 2002. Natural concept formation in a juvenile gorilla (Gorilla gorilla gorilla) at three levels of abstraction. J. Exp. Analysis of Behav. 78:315-332.

Vonk J and MacDonald. 2004. Levels of abstraction in Orang Utan (Pongo abelii)

Categorization. J. Comparative Psychol. 118:3-13.

Yoshikubo S. 1985. Species discrimination and concept formation by rhesus monkeys (Macaca mulatta). Primates 26: 285-299.

Appendix 1 Monkeys performance in Human vs Macaques experiments

Tabel 1 Monkeys performance in categorizing human and monkeys in Human vs Monkey experiments. In every trials, correct response was defined as monkey chose matching stimulus in association with sample stimulus. Responses in test and baseline trials of all sessions were pooled to get the performance rates. Based on result of the training phase, we set 90% rate as threshold to define that the subject could discriminate human and monkey as mutually exclusive concepts. Performances of the subject were assessed using binomial test. Confidence Levels in the table indicate 95% confident intervals.

Monkeys Experiment Trials Correct (%) Confidence Levels Colored Stimuli

Sukhoi Baseline 97.04 94.25 - 98.71

Test 94.44 91.00 - 96.86

Baseline 91.67 85.21 - 95.93

Test 90.83 84.19 - 95.33

Baseline 95.00 89.43 - 98.14

Test 95.00 89.43 - 98.14

Ucok Baseline 95.39 93.91 - 96.59

Test 94.12 90.15 - 96.83

Baseline 87.50 80.22 - 92.83

Test 89.17 82.19 - 94.10

Baseline 87.78 79.18 - 93.74

Test 90.00 81.86 - 95.32

Kerok Baseline 95.36 92.19 - 97.51

Test 96.43 93.53 - 98.27

Baseline 98.33 94.11 - 100.0

Test 99.17 95.44 - 99.98

Baseline 95.00 89.44 - 98.14

Test 92.50 86.24 - 96.51

Black and White Stimuli

Sukhoi Baseline 96.43 91.86 - 98.83

Test 96.43 91.86 - 98.83

Baseline 98.75 93.23 - 99.97

Test 93.75 86.01 - 97.94

Baseline 95.00 87.69 - 98.62

Test 96.25 89.43 - 99.22

Ucok Baseline 95.97 94.10 - 97.37

Test 98.57 94.93 - 99.83

Baseline 96.25 89.43 - 99.22

Test 97.50 91.26 - 99.70

Baseline 97.50 91.26 - 99.70

Test 98.75 93.23 - 99.97

Kerok Baseline 98.57 95.88 - 99.70

Test 98.57 95.88 - 99.70

Baseline 99.17 95.44 - 99.98

Test 100.0 96.97 - 100.0

Baseline 97.50 92.87 - 99.48

Test 99.17 95.44 - 99.98

Human vs M. fascicularis Human vs M. mullatta Human vs M. fuscata Human vs M. fascicularis Human vs M. mullatta Human vs M. fuscata Human vs M. fascicularis Human vs M. mullatta Human vs M. fuscata

Human vs M. fascicularis Human vs M. mullatta Human vs M. fuscata Human vs M. fascicularis Human vs M. mullatta Human vs M. fuscata Human vs M. fascicularis Human vs M. mullatta Human vs M. fuscata

Table 1 continued

Monkeys Experiment Trials Correct (%) Confidence Levels Resized Stimuli

Sukhoi Baseline 93.33 83.80 - 98.15

Test 96.67 88.47 - 99.59

Ucok Baseline 97.78 96.15 - 98.84

Test 96.67 88.47 - 99.59

Kerok Baseline 97.78 92.20 - 99.73

Test 100.0 95.98 - 100.0

Black and white, blurred stimuli.

Sukhoi Baseline 100.0 83.15 - 100.0

Test 95.00 75.12 - 99.87

Baseline 95.00 75.12 - 99.87

Test 95.00 75.12 - 99.87

Baseline 100.0 83.15 - 100.0

Test 100.0 83.15 - 100.0

Ucok Baseline 100.0 83.15 - 100.0

Test 100.0 83.15 - 100.0

Baseline 90.00 68.30 - 98.76

Test 95.00 75.12 - 99.87

Baseline 100.0 83.15 - 100.0

Test 100.0 83.15 - 100.0

Kerok Baseline 93.33 77.93 - 99.18

Test 96.67 82.78 - 99.91

Baseline 100.0 88.43 - 100.0

Test 100.0 88.43 - 100.0

Baseline 96.67 82.78 - 99.91

Test 100.0 88.43 - 100.0

Resized, black and white blurred stimuli

Sukhoi Baseline 100.0 83.15 - 100.0

Test 100.0 83.15 - 100.0

Baseline 100.0 83.15 - 100.0

Test 100.0 83.15 - 100.0

Baseline 100.0 83.15 - 100.0

Test 100.0 83.15 - 100.0

Ucok Baseline 95.00 75.12 - 99.87

Test 95.00 75.12 - 99.87

Baseline 100.0 83.15 - 100.0

Test 100.0 83.15 - 100.0

Baseline 100.0 83.15 - 100.0

Test 100.0 83.15 - 100.0

Kerok Baseline 100.0 88.43 - 100.0

Test 100.0 88.43 - 100.0

Baseline 100.0 88.43 - 100.0

Test 100.0 88.43 - 100.0

Baseline 100.0 88.43 - 100.0

Test 100.0 88.43 - 100.0

Human vs M. fascicularis Human vs M. fascicularis Human vs M. fascicularis

Human vs M. fascicularis Human vs M. mullatta Human vs M. fuscata Human vs M. fascicularis Human vs M. mullatta Human vs M. fuscata Human vs M. fascicularis Human vs M. mullatta Human vs M. fuscata

Human vs M. fascicularis Human vs M. mullatta Human vs M. fuscata Human vs M. fascicularis Human vs M. mullatta Human vs M. fuscata Human vs M. fascicularis Human vs M. mullatta Human vs M. fuscata

Appendix 2 Monkeys performance in M. fascicularis vs other Macaques experiments

Tabel 2 Monkeys performances in categorizing conspecifics and other monkeys in M. fascicularis vs other macaques experiments. In every trials, correct response was defined as monkey chose matching stimulus in association with sample stimulus. Responses in test and baseline trials of all sessions were pooled to get the performance rates. Based on result of the training phase, we set 90% rate as threshold to define that the subject could discriminate conspecifics and other macaques as mutually exclusive concepts. Performances of the subject were assessed using binomial test. Lower and Upper in the table indicate 95% confident intervals.

Monkeys Experiment Trials Correct (%) Confidence Levels Colored Stimuli

Sukhoi Baseline 100.0 91.19 - 100.0

Test 100.0 91.19 - 100.0

Baseline 100.0 91.19 - 100.0

Test 95.00 83.08 - 99.39

Ucok Baseline 87.50 73.20 - 95.81

Test 92.50 79.61 - 98.43

Baseline 100.0 91.19 - 100.0

Test 100.0 91.19 - 100.0

Black and White Stimuli

Sukhoi Baseline 97.50 86.84 - 99.00

Test 100.0 91.19 - 100.0

Baseline 100.0 91.19 - 100.0

Test 100.0 91.19 - 100.0

Ucok Baseline 100.0 91.19 - 100.0

Test 100.0 91.19 - 100.0

Baseline 100.0 91.19 - 100.0

Test 100.0 91.19 - 100.0

M. fascicularis vs M. mullatta M. fascicularis vs M. fuscata M. fascicularis vs M. mullatta M. fascicularis vs M. fuscata

M. fascicularis vs M. mullatta M. fascicularis vs M. fuscata M. fascicularis vs M. mullatta M. fascicularis vs M. fuscata

Appendix 3 Monkeys performances in Human vs Non-Human experiments

Tabel 3 Monkeys performances in categorizing human and non-human animal in Human vs Non-Human experiments. In every trials, correct response was defined as monkey chose matching stimulus in association with sample stimulus. Responses in test and baseline trials of all sessions were pooled to get the performance rates. Based on result of the training phase, we set 90% rate as threshold to define that the subject could discriminate human and non-human animal as mutually exclusive concepts. Performances of the subject were assessed using binomial test. Lower and Upper in the table indicate 95% confident intervals.

Monkeys Experiment Trials Correct (%) Confidence Levels Colored Stimuli

Sukhoi Baseline 99.33 96.34 - 99.98

Test 98.67 95.26 - 99.84

Ucok Baseline 99.33 96.34 - 99.98

Test 99.33 96.34 - 99.98

Black and White Stimuli

Sukhoi Baseline 99.33 96.34 - 99.98

Test 98.67 95.26 - 99.84

Ucok Baseline 98.00 94.26 - 99.58

Test 98.67 95.26 - 99.84

Human vs Non-Human Human vs Non-Human

Human vs Non-Human Human vs Non-Human

Appendix 4 Test of the similarity of the monkey performances between baseline and test trials

Tabel 4 Test of the similarity of the monkey performances (shown in Appendix 1 to Appendix 3) between baseline and test trials (GLMM, Venables and Ripley 2002). Values indicate statistical probability that performances in test trials were the same as in baseline trials. All values are higher than 0.05, it means the monkeys performances in both baseline and test trials are the same.

.

Experiment Values

1. Human vs Macaques

a. Colored stimuli 0.15 0.52 0.76 0.88

b. Black and White 0.19 0.17 0.53 0.29

c. Resized stimuli 0.86

d. Blured Black and White stimuli 0.53 1.00 0.59 1.00

e. Blurred, Small, Black and White stimuli 0.57 1.00 1.00 1.00

a. Colored stimuli 1.00 0.31 1.00

b. Black and White stimuli 1.00 1.00 1.00

3. Human vs Non Human

a. Colored stimuli 0.66

b. Black and White stimuli 0.71 1. Human vs M. fascicularis

2. Human vs M. mulatta 3. Human vs M. fuscata

1. Human vs M. fascicularis 2. Human vs M. mulatta 3. Human vs M. fuscata

1. Human vs M. fascicularis 2. Human vs M. mulatta 3. Human vs M. fuscata

1. Human vs M. fascicularis 2. Human vs M. mulatta 3. Human vs M. fuscata

2. Macaca fascicularis vs other macaques

1. M. fascicularis vs M. mulatta 2. M. fascicularis vs M. mulatta

1. M. fascicularis vs M. mulatta 2. M. fascicularis vs M. mulatta

Appendix 1 Monkeys performance in Human vs Macaques experiments

Tabel 1

Monkeys performance in categorizing human and monkeys in Human vs

Monkey experiments. In every trials, correct response was defined as

monkey chose matching stimulus in association with sample stimulus.

Responses in test and baseline trials of all sessions were pooled to get

the performance rates. Based on result of the training phase, we set 90%

rate as threshold to define that the subject could discriminate human and

monkey as mutually exclusive concepts. Performances of the subject

were assessed using binomial test. Confidence Levels in the table

indicate 95% confident intervals.

Monkeys Experiment Trials Correct (%) Confidence Levels Colored Stimuli

Sukhoi Baseline 97.04 94.25 - 98.71

Test 94.44 91.00 - 96.86

Baseline 91.67 85.21 - 95.93

Test 90.83 84.19 - 95.33

Baseline 95.00 89.43 - 98.14

Test 95.00 89.43 - 98.14

Ucok Baseline 95.39 93.91 - 96.59

Test 94.12 90.15 - 96.83

Baseline 87.50 80.22 - 92.83

Test 89.17 82.19 - 94.10

Baseline 87.78 79.18 - 93.74

Test 90.00 81.86 - 95.32

Kerok Baseline 95.36 92.19 - 97.51

Test 96.43 93.53 - 98.27

Baseline 98.33 94.11 - 100.0

Test 99.17 95.44 - 99.98

Baseline 95.00 89.44 - 98.14

Test 92.50 86.24 - 96.51

Black and White Stimuli

Sukhoi Baseline 96.43 91.86 - 98.83

Test 96.43 91.86 - 98.83

Baseline 98.75 93.23 - 99.97

Test 93.75 86.01 - 97.94

Baseline 95.00 87.69 - 98.62

Test 96.25 89.43 - 99.22

Ucok Baseline 95.97 94.10 - 97.37

Test 98.57 94.93 - 99.83

Baseline 96.25 89.43 - 99.22

Test 97.50 91.26 - 99.70

Baseline 97.50 91.26 - 99.70

Test 98.75 93.23 - 99.97

Kerok Baseline 98.57 95.88 - 99.70

Test 98.57 95.88 - 99.70

Baseline 99.17 95.44 - 99.98

Test 100.0 96.97 - 100.0

Baseline 97.50 92.87 - 99.48

Test 99.17 95.44 - 99.98

Human vs M. fascicularis

Human vs M. mullatta

Human vs M. fuscata

Human vs M. fascicularis

Human vs M. mullatta

Human vs M. fuscata

Human vs M. fascicularis

Human vs M. mullatta

Human vs M. fuscata

Human vs M. fascicularis

Human vs M. mullatta

Human vs M. fuscata

Human vs M. fascicularis

Human vs M. mullatta

Human vs M. fuscata

Human vs M. fascicularis

Human vs M. mullatta

Table 1 continued

Monkeys Experiment Trials Correct (%) Confidence Levels Resized Stimuli

Sukhoi Baseline 93.33 83.80 - 98.15

Test 96.67 88.47 - 99.59

Ucok Baseline 97.78 96.15 - 98.84

Test 96.67 88.47 - 99.59

Kerok Baseline 97.78 92.20 - 99.73

Test 100.0 95.98 - 100.0

Black and white, blurred stimuli.

Sukhoi Baseline 100.0 83.15 - 100.0

Test 95.00 75.12 - 99.87

Baseline 95.00 75.12 - 99.87

Test 95.00 75.12 - 99.87

Baseline 100.0 83.15 - 100.0

Test 100.0 83.15 - 100.0

Ucok Baseline 100.0 83.15 - 100.0

Test 100.0 83.15 - 100.0

Baseline 90.00 68.30 - 98.76

Test 95.00 75.12 - 99.87

Baseline 100.0 83.15 - 100.0

Test 100.0 83.15 - 100.0

Kerok Baseline 93.33 77.93 - 99.18

Test 96.67 82.78 - 99.91

Baseline 100.0 88.43 - 100.0

Test 100.0 88.43 - 100.0

Baseline 96.67 82.78 - 99.91

Test 100.0 88.43 - 100.0

Resized, black and white blurred stimuli

Sukhoi Baseline 100.0 83.15 - 100.0

Test 100.0 83.15 - 100.0

Baseline 100.0 83.15 - 100.0

Test 100.0 83.15 - 100.0

Baseline 100.0 83.15 - 100.0

Test 100.0 83.15 - 100.0

Ucok Baseline 95.00 75.12 - 99.87

Test 95.00 75.12 - 99.87

Baseline 100.0 83.15 - 100.0

Test 100.0 83.15 - 100.0

Baseline 100.0 83.15 - 100.0

Test 100.0 83.15 - 100.0

Kerok Baseline 100.0 88.43 - 100.0

Test 100.0 88.43 - 100.0

Baseline 100.0 88.43 - 100.0

Test 100.0 88.43 - 100.0

Baseline 100.0 88.43 - 100.0

Test 100.0 88.43 - 100.0

Human vs M. fascicularis

Human vs M. fascicularis

Human vs M. fascicularis

Human vs M. fascicularis

Human vs M. mullatta

Human vs M. fuscata

Human vs M. fascicularis

Human vs M. mullatta

Human vs M. fuscata

Human vs M. fascicularis

Human vs M. mullatta

Human vs M. fuscata

Human vs M. fascicularis

Human vs M. mullatta

Human vs M. fuscata

Human vs M. fascicularis

Human vs M. mullatta

Human vs M. fuscata

Human vs M. fascicularis

Human vs M. mullatta

Appendix 2 Monkeys performance in

M. fascicularis

vs other Macaques

experiments

Tabel 2 Monkeys performances in categorizing conspecifics and other monkeys in

M. fascicularis vs other macaques experiments. In every trials, correct

response was defined as monkey chose matching stimulus in association

with sample stimulus. Responses in test and baseline trials of all

sessions were pooled to get the performance rates. Based on result of the

training phase, we set 90% rate as threshold to define that the subject

could discriminate conspecifics and other macaques as mutually

exclusive concepts. Performances of the subject were assessed using

binomial test. Lower and Upper in the table indicate 95% confident

intervals.

Monkeys Experiment Trials Correct (%) Confidence Levels Colored Stimuli

Sukhoi Baseline 100.0 91.19 - 100.0

Test 100.0 91.19 - 100.0

Baseline 100.0 91.19 - 100.0

Test 95.00 83.08 - 99.39

Ucok Baseline 87.50 73.20 - 95.81

Test 92.50 79.61 - 98.43

Baseline 100.0 91.19 - 100.0

Test 100.0 91.19 - 100.0

Black and White Stimuli

Sukhoi Baseline 97.50 86.84 - 99.00

Test 100.0 91.19 - 100.0

Baseline 100.0 91.19 - 100.0

Test 100.0 91.19 - 100.0

Ucok Baseline 100.0 91.19 - 100.0

Test 100.0 91.19 - 100.0

Baseline 100.0 91.19 - 100.0

Test 100.0 91.19 - 100.0

M. fascicularis vs M. mullatta M. fascicularis vs M. fuscata M. fascicularis vs M. mullatta M. fascicularis vs M. fuscata

M. fascicularis vs M. mullatta M. fascicularis vs M. fuscata M. fascicularis vs M. mullatta M. fascicularis vs M. fuscata

Appendix 3 Monkeys performances in Human vs Non-Human experiments

Tabel 3 Monkeys performances in categorizing human and non-human animal in

Human vs Non-Human experiments. In every trials, correct response

was defined as monkey chose matching stimulus in association with

sample stimulus. Responses in test and baseline trials of all sessions

were pooled to get the performance rates. Based on result of the training

phase, we set 90% rate as threshold to define that the subject could

discriminate human and non-human animal as mutually exclusive

concepts. Performances of the subject were assessed using binomial test.

Lower and Upper in the table indicate 95% confident intervals.

Monkeys Experiment Trials Correct (%) Confidence Levels Colored Stimuli

Sukhoi Baseline 99.33 96.34 - 99.98

Test 98.67 95.26 - 99.84

Ucok Baseline 99.33 96.34 - 99.98

Test 99.33 96.34 - 99.98

Black and White Stimuli

Sukhoi Baseline 99.33 96.34 - 99.98

Test 98.67 95.26 - 99.84

Ucok Baseline 98.00 94.26 - 99.58

Test 98.67 95.26 - 99.84

Human vs Non-Human Human vs Non-Human

Human vs Non-Human Human vs Non-Human

Appendix 4 Test of the similarity of the monkey performances between

baseline and test trials

Tabel 4 Test of the similarity of the monkey performances (shown in Appendix 1

to Appendix 3) between baseline and test trials

(GLMM, Venables and

Ripley 2002). Values indicate statistical probability that performances in

test trials were the same as in baseline trials. All values are higher than

0.05, it means the monkeys performances in both baseline and test trials

are the same.

.

Experiment Values

1. Human vs Macaques

a. Colored stimuli 0.15

0.52 0.76 0.88

b. Black and White 0.19

0.17 0.53 0.29

c. Resized stimuli 0.86

d. Blured Black and White stimuli 0.53 1.00 0.59 1.00 e. Blurred, Small, Black and White stimuli 0.57 1.00 1.00 1.00

a. Colored stimuli 1.00

0.31 1.00

b. Black and White stimuli 1.00

1.00 1.00 3. Human vs Non Human

a. Colored stimuli 0.66

b. Black and White stimuli 0.71 1. Human vs M.fascicularis

2. Human vs M. mulatta

3. Human vs M. fuscata

1. Human vs M.fascicularis

2. Human vs M. mulatta

3. Human vs M. fuscata

1. Human vs M.fascicularis

2. Human vs M. mulatta

3. Human vs M. fuscata

1. Human vs M.fascicularis

2. Human vs M. mulatta

3. Human vs M. fuscata

2. Macaca fascicularis vs other macaques

1. M. fascicularis vs M. mulatta

2. M. fascicularis vs M. mulatta

1. M. fascicularis vs M. mulatta