Hiromi Kobayashi &Shiro KohshimaBiological Laboratory, Facultyof Bioscience andBiotechnology, Tokyo Instituteof Technology (c/o Faculty ofScience), 12-1, O-okayama2-chome, Meguro-ku, Tokyo152-8551, Japan. E-mail:hiromi@innocent.com

Received 30 October 1998Revision received29 January 2001and accepted 5 February2001

Keywords: primates, eyemorphology, sclera colour,communication, adaptation,human evolution, theory ofmind.

Unique morphology of the human eye andits adaptive meaning: comparative studieson external morphology of the primate eye

In order to clarify the morphological uniqueness of the human eyeand to obtain cues to understanding its adaptive significance, wecompared the external morphology of the primate eye by measuringnearly half of all extant primate species. The results clearly showedexceptional features of the human eye: (1) the exposed white sclera isvoid of any pigmentation, (2) humans possess the largest ratio ofexposed sclera in the eye outline, and (3) the eye outline is extraor-dinarily elongated in the horizontal direction. The close correlation ofthe parameters reflecting (2) and (3) with habitat type or body size ofthe species examined suggested that these two features are adapta-tions for extending the visual field by eyeball movement, especially inthe horizontal direction. Comparison of eye coloration and facialcoloration around the eye suggested that the dark coloration ofexposed sclera of nonhuman primates is an adaptation to camouflagethe gaze direction against other individuals and/or predators, and thatthe white sclera of the human eye is an adaptation to enhance the gazesignal. The uniqueness of human eye morphology among primatesillustrates the remarkable di!erence between human and otherprimates in the ability to communicate using gaze signals.

! 2001 Academic Press

Journal of Human Evolution (2001) 40, 419–435doi:10.1006/jhev.2001.0468Available online at http://www.idealibrary.com on

Introduction

Recognizing others’ gaze direction is one ofthe important cognitive bases for communi-cation in humans (Gibson & Pick, 1963;Kendon, 1967). To clarify the biologicalbasis of this ability, especially in relation tothe evolution of social intelligence,researchers have experimentally examinedthe cognitive ability to detect gaze directionof others in nonhuman primates (Gomez,1991; Itakura & Anderson, 1996; Tomaselloet al., 1998). However, little attention hasbeen given to external morphology of theeye, although this ability of humans mightbe supported by a unique morphology of

the human eye. For example, in humans, thewidely exposed white sclera (the white of theeye) surrounding the darker coloured irismakes it easy for others to discern the gazedirection and has been said to be a charac-teristic of humans not found in other pri-mate species (Morris, 1985). However, thishas not been examined in detail, partlybecause of the di"culty in measuring thesoft parts of living animals.

In this study, we measured the externaleye morphologies of nearly half of all extantprimate species with video camera andcomputer-aided image analysing techniquesto clarify the morphological uniqueness ofthe human eye and to understand adaptivemeanings of external eye morphology inprimates. The results clearly showed excep-tional features of the human eye in both

Address correspondence to: Hiromi Kobayashi,Ph.D., 6–8, Nakanoshima-cho, Fukakusa, Fushimi-ku,Kyoto-city, Kyoto, 612-0049, Japan. Tel.: +81 75 6441402; Fax: +81 75 644 1402.

0047–2484/01/050419+17$35.00/0 ! 2001 Academic Press

shape and coloration. In our precedingpaper (Kobayashi & Kohshima, 1997), webriefly reported the morphological unique-ness of the human eye and discussed itsadaptive meanings. In the present paper wefully analysed the results and examined thefollowing hypotheses on adaptive meaningsof primate eye morphology.

We measured width/height ratio of the eyeoutline (WHR) and an index of exposedsclera size in the eye outline (SSI) to analyseeye shape. These eye-shape parametersclosely correlated with habitat type or bodysize of the species examined. To explain thecorrelation, we postulated a hypothesis thatthese two features are adaptations forextending the visual field by eyeball move-ment, especially in the horizontal direction.This hypothesis was examined and sup-ported by analysing the eye movement ofvideo-recorded primates and comparing theway that gaze direction changes amongspecies with various body sizes and habitattypes.

To explain the unique coloration of thehuman eye with its exposed white sclera voidof any pigmentation, we postulated ahypothesis that only coloration of the humaneye is adapted to enhance the gaze signalwhile eye coloration of other primates isadapted to camouflage the gaze directionagainst other individuals and/or predators.This hypothesis was examined and sup-ported by analysing relationships among iriscoloration, sclera coloration and facial col-oration around the eye. Our results sug-gested that unique features of the human eyestarted to evolve as adaptations to large bodysize and terrestrial life and were completedas a device for communication using gazesignal.

Method

Eye shape measurementsA total of 874 adult animals (88 species:Prosimii; 10, Ceboidea; 26, Cercopithe-

coidea; 43, Hominoidea; 9) were studied(Table 1). Facial images of 80 species wererecorded by video camera at the JapanMonkey Centre. Facial images of eightspecies (Microcebus (1), Loris tardigradus (2),Perodicticus potto (1), Tarsius (1), Saguinusimperator (1), Pithecia monachus (1), Cacajaorubicundus (1), Cercopithecus hamlyni (1))were collected from books (Itani & Uehara,1986; Yoshino, 1994). For humans, facialimages of 244 Japanese, 347 Caucasian and68 Afro-Caribbean adults were studied. 244Japanese, 280 Caucasian and 2 Afro-Caribbean images were recorded by videocamera, and 67 Caucasian and 66 Afro-Caribbean images were collected frombooks (Ohara, 1970; Gomi, 1994).

Two parameters were measured for eachspecies: the width/height ratio of the eyeoutline (WHR) and an index of exposedsclera size in the eye outline (SSI). Frontalfull-face images without obvious facialexpression of subjects were recorded byvideo camera. These images were processedand analysed on a Macintosh Quadra840AV computer using the public domainNIH Image program. For each image, (A)the distance between the corners of the eye,(B) the longest perpendicular line betweenthe upper and lower eyelid, (C) width of theexposed eyeball, and (D) diameter of the iriswere measured (Figure 1). WHR means(A)/(B) and SSI means (C)/(D). Data ofweight, crown–rump length and habitat typeof primates were collected from books (Itani& Uehara, 1986; Napier & Napier, 1985)since we could not get permission forphysical contact with primates. Walking-height and sitting-height of primates weremeasured in the Japan Monkey Centre usingmarks on the wall of the cages.

Eye-coloration measurementsColoration of the exposed sclera (includingthe conjunctiva to be precise), iris and facearound the eye was recorded for each of 91species by direct observation of living

420 #. $%&'(')#* '+, ). $%#)#*-'

animals (82 species) and of eyeball speci-mens (55 species, 124 animals) kept in theJapan Monkey Centre (Table 1). The scleracolour included in the term ‘‘Pale brown’’was a paler one than yellow ochre: 10YR6/7·5 of Munsell Colour System (seeFigure 9).

Eye coloration of 82 primate species wereclassified into 4 types (see Type 1–4, Figure11) by the di!erences of colour or contrastbetween sclera and iris/face. This classifica-tion was carried out by one person observingliving animals. To check reliability of thisclassification, another person independentlyclassified the face pictures of 76 primatespecies (see Figure 11). The results agreedin 70 species (92%). Disagreement was onlyobserved between Type 1 and Type 2 in 6species (8%).

Eyeball specimens of the Japanesemacaque (1 subject) and crab-eatingmacaque (2 subjects) were supplied from aco-operative program of the PrimateResearch Institute, Kyoto University,Inuyama, Aichi, Japan. Eyeballs were fixedwith 4% paraformaldehyde in 0·1 M phos-phate bu!er (pH 7·2) at 4"C overnight. Thetissue including the conjunctiva and corneaseparated from eyeballs was washed severaltimes in cold phosphate-bu!ered saline(PBS), dehydrated in an ethanolic seriesfinishing xylene and embedded in para"n.Serial sections with a 4 !m thickness werecut with disposable blades, floated on waterand placed on slides. These sections weredepara"nized in xylene, washed in ethanoland PBS and studied by light microscopy(see Figure 10).

Eye movement analysisTo analyse the movement of the eye and thehead when the animals change the directionof their look, eating food by hand in cagesvarious primates were video-recorded. Asfor humans, persons eating alone in ahamburger shop with a hamburger in theirhands were video-recorded.

To calculate the ratio of scanning per-formed only by eyeball movement, move-ments of the eyeball and the head scanningwere counted (total observation time:10,037 sec) for 29 individuals of 18 species(Table 1).

To calculate the ratio of horizontal scan-ning to vertical scanning frequency andtime duration of horizontal and verticalscanning were measured (total observationtime=12,579 sec) for 40 individuals of 26species (Table 1): arboreal species: Lemurcatta, Cebus apella, C. albifrons*, Pitheciapithecia*, Ateles belzebuth*, A. geo!royi, A.paniscus, Cercocebus galeritus, Cercopithecuscephus*, Colobus angolensis, Presbytis cristata,P. vetulus, P. francoisi*, Nasalis larvatus,Hylobates lar and H. pileatus; semi-arborealspecies: Macaca fuscata*, Cercocebus torqua-tus, Mandrillus sphinx*, M. leucophaeus*,Cercopithecus ascanius, Presbytis entellus andPan troglodytes; terrestrial species: Papiohamadryas, Erythrocebus patas and Homosapiens (*: duration time only) (Table 1).

All these videotape analyses were carriedout by one person. To check reliability of theanalyses, a second person scored every0·5 sec randomly sampled 20% of the video-tapes independently. Agreement betweenthe persons was 82% on average. TheCohen’s kappa (Bakeman & Gobbman,1997) was 0·63 on average.

Results and discussion

Unique shape of the human eyeFigure 2(a) shows that human eyes have thelargest exposed sclera area and showextraordinary horizontal elongation of theeye outline among primates. SSI increasedin the following order: Prosimii (prim-itive type) <Ceboidea<Cercopithecoidea<Hominoidea. WHR increased in the follow-ing order: Prosimii (primitive type)<Ceboidea<Cercopithecoidea and Homi-noidea [the di!erence between phylogenicgroups was significant with analysis of

421.+*/.0 -%12#%3%4( %5 6#0 #.-'+ 0(0

Table 1A

Eye shape[Figure 2(a)]

Eye colour(Figure 9) Eye movement

Living animals(Figure 11)

Eyeballspecimens (Figure 5)

(Figure 7—time)

(Figure 7—frequency)

ProsimiiMicrocebus 1 3Lemur catta 3 3 1 1 1L. macaco 2 2 2L. fulvus 2 2Varecia variegata 3 6 1Loris tardigradus 2 5Perodicticus potto 1 2Galago senegalensis 3 2 1Otolemur crassicaudatus 1 2 2Tarsius 1 2

CeboideaCallimico goeldii 1 2 1Callithrix jacchus 3 3 2C. argentata 2 2C. humeralifer 2 2G. geo!royi 3 3 2C. penicillata 2 1Cebuella pygmaea 3 4 6Saguinus midas 1 2 2S. weddelli 1 1S. imperator 1 1 1S. labiatus 1 1 3S. mystax 2 2 2S. oedipus 1 4 3Cebus capucinus 3 3 3C. albifrons 1 4 1 2C. apella 2 3 1 1 1C. nigrivittatus 1Aotus trivirgatus 3 4 2Callicebus moloch 2 5 2Saimiri sciureus 2 10 2Pithecia pithecia 2 2 1 1P. monachus 1Cacajao rubicundus 1Alouatta caraya 1 1 6Ateles paniscus 1 2 1 1 1A. belzebuth 2 4 1A. geo!royi 3 5 2 2 2 2Lagothrix lagothricha 2 4

CercopithecoideaMacaca sylvanus 3 3M. silenus 2 3 1M. maurus 1 1M. nemestrina 3 3 2M. nigra 1 1M. fascicularis 3 3 5M. fuscata 6 7 1 1M. fuscata yakui 3 3 2M. mulatta 4 18 17

422 #. $%&'(')#* '+, ). $%#)#*-'

Table 1B

Eye shape[Figure 2(a)]

Eye colour(Figure 9) Eye movement

Living animals(Figure 11)

Eyeballspecimens (Figure 5)

(Figure 7—time)

(Figure 7—frequency)

M. cyclopis 5 10M. sinica 3 4 1M. assamensis 1 1 1M. radiata 2 9 1M. thibetana 3 4M. arctoides 1 2Cercocebus torquatus 2 2 2 2 2C. torquatus lunulatus 2 3C. atys 1 1 1C. galeritus agilis 1 1 1 1 1 1C. galeritus chrysogaster 2 3C. albigena 1Papio papio 3 3 2P. anubis 12 50 2P. cynocephalus 1 3P. hamadryas 3 9 1 1 1 1Mandrillus sphinx 3 4 1M. leucophaeus 2 4 1 2Theropithecus gelada 1 1 1Cercopithecus hamlyni 1 1C. neglectus 3 3C. mitis 2 5C. ascanius schmidti 2 2C. cephus 2 2 2C. ascanius 2 3 2 2 2C. mona 2 2 1C. petaurista 1 1 1C. aethiops 2 5Miopithecus talapoin 2 2Allenopithecus nigroviridis 1 3Erythrocebus patas 4 4 2 2 2Colobus angolensis 5 6 1 1 1C. guereza 2 2Presbytis cristata 2 8 7 4 4 4P. francoisi 2 4 1 1P. pileata 1 1 1P. vetulus 2 2 1 1 1P. entellus 4 7 2 1 1 1P. obscura 2Nasalis larvatus 2 2 2 2 2

HominoideaHylobates lar 2 2 1 1 1H. agilis 1 1 2H. pileatus 2 2 1 1 1H. syndactylus 2 2H. klossii 1Pongo pygmaeus 2 3 1Pan troglodytes 9 13 2 3 3 3P. paniscus 1 1Gorilla gorilla 4 6Homo sapiens 659 247 2 2 2

423.+*/.0 -%12#%3%4( %5 6#0 #.-'+ 0(0

variance (ANOVA) (SSI: F (3,90)=14·68,P<0·01, WHR: F (3,90)=32·77, P<0·01).In multiple comparison (LSD), the di!er-ence between phylogenic groups, exceptWHR between Cercopithecoidea andHominoidea, was significant (SSI: MSe=0·02, P<0·01, WHR: MSe=0·04, P<0·01)].Even in Hominoidea species, SSI and WHRof human were exceptionally high [Figure2(b): the di!erence between phylogenicgroups was significant with ANOVA (SSI: F(4,675)=79·82, P<0·01, WHR: F (4,676)=33·92, P<0·01). In multiple comparison(LSD), the SSI di!erence between speciesof Hominoidea, except between orang-utanand gorilla and between orang-utan andchimpanzee, was significant (MSe=0·02,P<0·01). The WHR di!erence betweenhuman and others was significant (MSe=0·19, P<0·01). We measured both sexes ofthree human races: Mongoloid, Caucasianand Afro-Caribbean, and sexual and racialdi!erence were slight in these parametersrelative to interspecies di!erence.

It is possible that the di!erence in theseeye shape parameters reflect some di!erencein visual function and/or adaptation tosome environmental or physiological factorsuch as the habitat and body size of thespecies. Thus, we analysed relationshipsbetween the eye shape parameters and thesefactors.

Relationship between eye shape and visualfunctionPrimates are mammals with well developedvisual function. Many primate species have(1) cone cells for colour vision, (2) the fovea(a dense concentration of cones in the retinafocused on the centre of gaze) for a highresolution visual image, (3) forward-facingeyes for wide stereoscopic vision by botheyes, and (4) a well developed postorbitalplate behind the eyes. It is possible that thedi!erence in eye shape parameters amongphylogenic groups has some relationshipwith these anatomical structures for welldeveloped visual functions. However, sincemost primate groups except the prosimianshave all these anatomical structures, thedi!erence in eye shape parameters amongphylogenic groups cannot be explained bythe di!erence in these visual functions. Forexample, some nocturnal prosimian speciesare the only primates which lack both conecell and a fovea in their eyes (Wolin &Massopust, 1970; Alfieri et al., 1976;Webb & Kass, 1976; Debruyn et al., 1980;Castenholtz, 1984). Figure 3 shows the re-lationship of eye shape parameters withorbital axis angle measured from thecranium (Shigehara, 1996) which reflectsthe ability of stereoscopic vision. In thisfigure Prosimians with high values of orbitalaxis angle showed low values of SSI andWHR. However, among simians with moreforward-facing eyes and developed post-orbital closure, SSI and WHR spread overvarious values and no significant correlationis observed. These facts suggest that thevariation in the eye shape parametersof each phylogenic group does not reflectevolutionary trends in these visual functions.

Relationship between body size and SSISSI correlates well with various body sizeparameters (weight: r=0·59, P<0·001,crown–rump length: r=0·59, P<0·001,sitting height: r=0·65, P<0·001, walkingheight: r=0·72, P<0·001). The best

Figure 1. (A): the distinace between the corners of theeye, (B): the longest perpendicular line to (A) betweenthe upper eyelid and lower eyelid, (C): width ofthe exposed eyeball, and (D): diameter of the iriswere measured. WHR means (A)/(B) and SSI means(C)/(D).

424 #. $%&'(')#* '+, ). $%#)#*-'

Figure 2. Variation of WHR (width/height ratio of the eye outline) and SSI (index of exposed sclera sizein the eye outline) among the phylogenic groups of primates (a), and in Hominoidea (b).

425.+*/.0 -%12#%3%4( %5 6#0 #.-'+ 0(0

correlation was observed with walkingheight (Figure 4). This means that specieswith larger body size have a larger exposedscleral area.

A larger SSI means a smaller iris relativeto the eye outline and probably a greaterability for visual field extension by eyeballmovement; in eyes with a large SSI the smalliris has a wider space to move within theopen eye outline. In mammalian animals,only primates have central foveae necessaryfor fine vision. Therefore, eyeball and/orhead movement are important, especially for

primates, to adjust the images to the centralfovea.

If we suppose that a larger exposed sclerais an adaptation for extending the visual fieldby eyeball movement, the correlationbetween SSI and body size can be explainedby the theory of scaling. This is because, asbody size becomes larger, visual field exten-sion by eyeball movement becomes moree!ective than that by head or body move-ment. This is so because as body heightbecomes greater, the weight of the head andbody increases proportionally to the cube of

Figure 3. Orbital axis angle and WHR/SSI. Orbital axis angle is the angle formed when right and leftorbital axes meet. Orbital axis is the line between the lowest point of the optic canal and the centre of theorbital width (Shigehara, 1996). Circles, prosimians; squares, simians.

426 #. $%&'(')#* '+, ). $%#)#*-'

body height. In contrast, the force requiredfor movement increases only with the squareof body height as it depends on the size ofthe muscle cross-section. Moreover, sincethe relative growth of the eyes to body heightis smaller than that of head and body size,comparative eyeball size becomes smaller inlarger animals (Schultz, 1940). Thus, tosave energy when changing the direction ofgaze, a large-sized species would move theeyeball more often than a small-sizedspecies, and have a larger exposed scleralarea. Besides, in small species with com-paratively large eyeballs in a small skull,space for muscles moving the eyeball may beseriously limited.

To examine this hypothesis, we video-recorded various primates (18 species, 29individuals) eating food by hand in cages,

and counted the movements of head andeyeball when they changed the direction inwhich they were looking. The resultsshowed that the proportion of scanning per-formed only by eyeball movement was cor-related with SSI (Figure 5, r=0·73,P<0·001). It was exceptionally high inhumans (61#28% of horizontal scan, n=5)compared with other primates (4·3–24·4%,mean=10·6%). The highest proportionin nonhuman primates was observed inchimpanzees (20–35%, n=3), the largestnonhuman species examined. These resultsagreed with our hypothesis.

Relationship between habitat type and WHRThe mean value of WHR is greatest interrestrial species, moderate in semi-arboreal species and lowest in arboreal

Figure 4. SSI and walking height.

427.+*/.0 -%12#%3%4( %5 6#0 #.-'+ 0(0

species [Figure 6: Di!erence among habitattypes was significant with ANOVA and withmultiple comparison (ANOVA: F (2,127)=24·63, P<0·01, LSD: MSe=0·052,P<0·01)]. This result suggests that a hori-zontally elongated eye outline is adaptive toterrestrial life in some way. We speculatedthat horizontal elongation of the eye outline

is adaptive in extending the visual field hori-zontally by eyeball movement, and terres-trial life needs more horizontal scanningthan vertical scanning.

To examine this hypothesis, we observedvarious primates eating food by hand incages and measured the time and frequencyof horizontal scanning and vertical scanning.

Figure 5. SSI and eye movement. EyeMove/TotalMove=frequency of gaze direction change only by eyemovement/frequency of all gaze direction change by head movement or/and eye movement.

Figure 6. WHR and habitat type.

428 #. $%&'(')#* '+, ). $%#)#*-'

The result shows that the ratio of horizontalscanning to vertical scanning is higher interrestrial species than in arboreal species[Figure 7(a): the di!erence among habitattypes was significant with ANOVA (time: F(2,23)=14·5, P<0·01, frequency: F (2,15)=4·53, P<0·05). The di!erence betweenterrestrial species and arboreal ones wassignificant in multiple comparison (LSD)(time: MSe=20·76, P<0·01, frequency:MSe=20·05, P<0·01)]. These ratios werealso correlated with WHR [Figure 7(b),time: r=0·74, P<0·001, frequency: r=0·88,P<0·001]. The results support our hypoth-esis. These investigations suggest that theshape of the eye outline and relative size ofthe exposed scleral area are the result of an

adaptation for visual field extension byeyeball movement.

Unique coloration of the human eye

The colour of exposed sclera. The followingfour colorations of exposed sclera wereobserved (Figures 8 and 9); (a) in almost allnonhuman primates (85 species out of 92species, or 92%) the exposed part of thesclera is uniformly brown or dark brown, (b)Macaca sylvanus and M. nemestrina with apale brown body colour had sclera colouredpale brown, (c) Saguinus midas, S. labiatus,Callithrix argentata and Callimico goeldii hadbrown sclera with a white part in corner ofthe eye, (d) humans were the only primates

Figure 7. (a) Habitat type and the ratio of horizontal scanning to vertical scanning. (b) Ratio of horizontalscanning to vertical scanning and WHR.

429.+*/.0 -%12#%3%4( %5 6#0 #.-'+ 0(0

having white sclera without any pigmenta-tion. Microscopic analysis of the eyeballsection specimens from the Japanesemacaque and crab-eating macaque revealed

Figure 8. Three types of sclera coloration in nonhumanprimates.

Figure 9. Variation of scleral colour. The shaded portion of the nonhuman primate eyeballs shows thegeneral area where colour was noted. The solid line surrounding the cornea represents the eye outline.

Figure 10. Eyeball section of crab-eating macaque.

430 #. $%&'(')#* '+, ). $%#)#*-'

that the brown coloration of the exposedsclera was due to pigmentation in the epi-thelium cornea, conjunctiva and sclera(Figure 10). External observations of othernonhuman primate eyes suggest that theirdark coloration is also due to similar pig-mentation. Humans have transparentconjunctiva and white sclera without pig-mentation. The inner part of the sclera innonhuman primates was also white like thatof humans.

Adaptive meaning of dark-coloured sclera.Nonhuman primates have sclera colouredbrown. As pigmentation costs some energy,the dark coloration of the exposed scleraprobably has some adaptive function. Browncoloration of the exposed sclera wasobserved in many other mammal speciestoo, and the following two hypotheses havebeen proposed on the adaptive meanings ofsclera colour. (1) Anti-glare theory: it waspointed out that the pigmentation may be ananti-glare device because it seemed to beabsent in nocturnal or crepuscular species(Duke-Elder, 1985). However, our resultson primate species were contrary to thisexpectation: nocturnal species (Galago sen-egalensis, Tarsius syrichta, Perodicticus potto,Nycticebus coucang and Aotus trivirgatus) alsohad coloured sclera and diurnal humans hadno pigmentation. Therefore, our resultscannot be explained by the ‘‘anti-glaretheory’’. (2) Gaze camouflage theory: inmany nonhuman primates, gaze direction isimportant in intraspecific communication(Chance, 1962; van Hoo!, 1962; Andrew,1964). For example, direct eye contact isassociated with gestures predominantlyshowing a tendency to attack in manymonkeys. In macaques, it is reportedthat sclera pigmentation obscures gazedirection and may be adaptive for escapingthe attacks of other individuals (Perrett &Mistlin, 1990). Coloured sclera obscuringgaze direction may serve to deceivenatural predators too, by making it di"cult

for predators to know if the prey hasthem in their gaze. If prey animals canmake it known to the predator thatthey already know of its presence, theirchances of survival may increase (Sherman,1977).

To examine the gaze camouflage theory,we analysed the relationship between scleracolour, iris colour and face colour aroundthe eye. If the dark coloration of exposedsclera is adaptation for gaze camouflage, thecolour of exposed sclera should be similar tothe colour of iris and/or face around theeyes, to make it di"cult to detect the pos-ition of iris in the eye outline and/or the eyeoutline in the face.

Relationship between sclera colour, iris colourand face colourFigure 11 shows the relationship betweensclera colour, iris colour and face colouraround the eye. 82 primate species observedwere classified into the following four typesby the di!erence of colour or contrastbetween sclera and iris/face.

Type (1) face!sclera!iris (43 species):darkness of sclera colour is similar to thatof face and iris; both eye outline in the faceand iris position in the eye outline areunclear.Type (2) face<sclera!iris (37 species):sclera colour is darker than face colour butsclera colour is similar to iris colour; eyeoutline in the face is clear but iris position inthe eye outline is unclear.Type (3) face!sclera>iris (1 species,ru!ed lemur): sclera colour is darker thaniris colour but sclera colour is similarto face colour; eye outline in the face isunclear but iris position in the eye outline isclear.Type (4) face>sclera<iris (1 species, Homosapiens): sclera colour is paler than facecolour and iris colour; both eye outline inthe face and iris position in the eye outlineare clear.

431.+*/.0 -%12#%3%4( %5 6#0 #.-'+ 0(0

Figure 11. Colour patterns of sclera, iris and face skin around the eye.

Figure 12. Di!erence of gaze stimulus between human and orang-utan.

Almost all nonhuman primate speciesobserved (80 out of 81 species) belonged toType 1 or Type 2 coloration. In thesecoloration types, the position of the irisin the eye outline was unclear because ofsimilarity between sclera colour andiris colour (‘‘Gaze camouflage type’’). InType 1 coloration (43 species), the positionof the eye outline in the face was alsounclear because of similarity betweensclera colour and face colour around theeyes. In addition, the ru!ed lemur (Vareciavariegata), the only species that hadType 3 coloration, has a very small exposedsclera area (SSI=1·08) and almost all thearea of its eye outline is occupied bythe iris. Thus, the ru!ed lemur’s eye alsocan be seen as a ‘‘gaze camouflagetype’’. The results thus support the ‘‘gazecamouflage theory’’.

In contrast, humans were the only speciesthat had Type 4 coloration, in which botheye outline in the face and iris position inthe eye outline were very clear because thecolour of the exposed sclera is paler thanthat of the facial skin and iris (‘‘gazesignalling type’’). The human was theonly species with sclera much paler thanthe facial skin. Because of this coloration,it is very easy to discern the gaze directionin human, in contrast to the gaze-camouflaging eyes of the other primates.Figure 12 shows the contrast between scleracolour and face/iris colour of human andorang-utan (Pongo pygmaeus). In this figure,darkness of colours was shown by 256 stepsgrey scale number (white=0, black=255,Figure 12). Even great apes that have SSIand WHR near that of humans [Figure2(b)] had ‘‘gaze camouflage eye’’ type withbrown sclera, brown facial skin and browniris, in which the eye position and iris pos-ition is unclear. In contrast to the gaze-camouflaging eyes of orang-utan, the humansclera is remarkably paler than the facial skinand the iris, and it is very easy to discern thegaze direction.

Evolution of unique morphology of the humaneyeOur results suggest that the unique shape ofhuman eye (largest SSI and WHR in pri-mates) is a result of adaptations to extendthe visual field by eyeball movement,especially in the horizontal direction. Inhuman evolution, the ratio of exposed sclerain the eye outline might increase because thevisual field extension by eyeball movementbecame more e!ective as body sizeincreased. And the eye outline might behorizontally elongated because the terrestriallife of humans needed more horizontalscanning than vertical scanning by eyeballmovement.

Why have only humans discarded sclerapigmentation? It may be because the neces-sity for gaze camouflage decreased and thatof gaze-signal enhancement increased inhuman evolution. The predation risk mightdecrease because of the enlarged body sizeand the use of tools and fire. Gaze-signalenhancement might aid the conspecificcommunication required for increasedco-operative and mutualistic behaviours toallow group hunting and scavenging.Co-operative and mutualistic behavioursmight need refined communication systems,such as language, to inform one’s intentionto other members of the group. The humaneye, with a large scleral area surrounding theiris and a great ability of eyeball movement,would have provided a chance for a drasticchange from ‘‘the gaze-camouflaging eyes’’into ‘‘the gaze-signalling eyes’’ through asmall change in scleral coloration. The SSIand WHR of human eyes are even greaterthan those of gorillas, the largest primate,which suggests adaptation for gaze-signalenhancement.

Eye morphology and gaze-signalcommunicationBaron-Cohen (1995) postulated a neuralmechanism (eye direction detector) inhuman brains specialized to detect others’

433.+*/.0 -%12#%3%4( %5 6#0 #.-'+ 0(0

eye direction, and discussed the possibilitythat such a mechanism might be related withthe evolution of a ‘‘theory of mind’’. Inrecent years, many studies have been carriedout to examine the cognitive ability to detectothers’ gaze direction in various nonhumanprimates. The studies have suggested thelimited ability to detect others’ gaze direc-tion in nonhuman primates (Itakura &Anderson, 1996; Tomasello et al., 1998).However, there seems to be some confusionin defining ‘‘gaze direction’’ in these studies.Since gaze direction can be changed byeyeball movement, by head movement andby body movement, it should be defined byconsidering all those factors: eyeball direc-tion, head direction and body direction.Most of these studies, however, defined‘‘gaze direction’’ mainly by eyeball direction.Our results (Figure 5) suggest that the con-tribution of eyeball movement to the changein gaze direction is extremely high inhumans compared with other primatespecies. The gaze signalling eye coloration ofhumans also suggests that the contributionof eyeball direction to the gaze signal isexceptionally high in humans. Therefore, fornonhuman primates, head direction andbody direction might be more important todetect others’ gaze direction than eyeballdirection. We should pay more attention tohead direction and body direction in futureanalyses of gaze-signals in nonhumanprimates.

Summary

Comparative analysis of the external mor-phology of the primate eye revealed excep-tional features of the human eye: (1) theexposed white sclera is void of any pigmen-tation, (2) humans possess the largest ratioof exposed sclera in the eye outline, and (3)the eye outline is extraordinarily elongatedin the horizontal direction. The close corre-lation of the parameters reflecting (2) and(3) with habitat type or body size of the

species examined suggested that these twofeatures are adaptations for extendingthe visual field by eyeball movement,especially in the horizontal direction.Comparison of eye coloration and facialcoloration around the eye suggested thatthe dark coloration of exposed sclera ofnonhuman primates is an adaptation tocamouflage the gaze direction against otherindividuals and/or predators, and that thewhite sclera of humans is an adaptation toenhance the gaze signal.

Acknowledgements

We would like to thank Shigetaka Koderafor his interest in our study and forallowing us to observe the animals of theJapan Monkey Centre. It is a pleasureto acknowledge the hospitality and encour-agement of the members of JMC. Wewish to express our gratitude to TetsuroMatsuzawa, Takashi Kageyama andNobuo Shigehara for stimulating discussionunder the co-operative research programmein Primate Research Institute, KyotoUniversity, and to Manabu Ogiso,Nobuyuki Saitoh and Teruhiko Hamanakafor providing eyeball samples. We areindebted to a number of our colleagues atthe Tokyo Institute of Technology andthe Primate Research Institute, KyotoUniversity, especially to Mitsue Nomura,Michael A. Hu!man, Sou Kanazawa,Masami Yamaguchi and KazuhideHashiya, for their constructive criticism onthis paper.

References

Alfieri, R., Pariente, G. & Sole, P. (1976). Dynamicelectroretinography in monochromatic lights andfluorescence electroretinography in lemurs. Doc.Ophthal. Proc. Ser. 10, 169–178.

Andrew, R. J. (1964). The displays in the Primates. In(J. Buiettner-Janusch, Ed.) Evolutionary and GeneticBiology of the Primates, vol. 2. New York: AcademicPress.

434 #. $%&'(')#* '+, ). $%#)#*-'

Bakeman, R. & Gottman, M. (1997). Observing inter-action: an introduction to sequential analysis (2nd ed.).Cambridge: Cambridge University Press.

Baron-Cohen, S. (1995). Mindblindness: An essay onautism and theory of mind. Cambridge: MIT Press.

Castenholtz, A. (1984). The eye of Tarsius. In(C. Niemitz, Ed.) Biology of Tarsiers, pp. 303–318.Stuttgart: Gustav Fischer.

Chance, M. R. A. (1962). An interpretation of someagonistic postures: the role of ‘‘cut-o!’’ acts andpostures. Symp. Zool. Soc. Lond. 8, 81–89.

Debruyn, E. J., Wise, V. L. & Casagrande, V. A.(1980). The size and topographic arrangement ofretinal ganglion cells in the Galago. Vision Res. 20,315–327.

Duke-Elder, S. S. (1985). The eye in evolution. In(S. S. Duke-Elder, Ed.) System of Ophthalmology, p.453. London: Henry Kimpton.

Gibson, J. J. & Pick, A. D. (1963). Perception ofanother person’s looking behaviour. Am. J. Psychol.76, 386–394.

Gomez, J. C. (1991). Visual behaviour as a window forreading the mind of others in primates. In (A.Whiten, Ed.) Natural Theories of Mind, pp. 195–207.Oxford: Blackwell.

Gomi, A. (1994). Americans 1.0 1994 Los Angeles.Tokyo: Fuga Shobo.

Itakura, S. & Anderson, L. R. (1996). Learning to useexperimenter-given cues during an object-choice taskby a capuchin monkey. Cahiers De PsychologieCognitive-Current Psychology of Cognition 15, 103–112.

Itani, J. & Uehara, S. (1986). Primates. In (D. W.Macdonald, Ed.) The Encyclopaedia of Animals, vol.3. Tokyo: Heibonsha Limited.

Kobayashi, H. & Kohshima, S. (1997). Unique mor-phology of the human eye. Nature 387, 767–768.

Morris, D. (1985). Body Watching. Oxford: EquinoxLtd.

Napier, J. R. & Napier, P. H. (1985). The NaturalHistory of the Primates. Cambridge, MA: MIT Press.

Ohara, K. (1970). One. Tokyo: Tsukui Shokan.Perrett, D. I. & Mistlin, A. J. (1990). Perception of

facial characteristics by monkeys. In (W. C. Stebbins& M. A. Berkley, Eds) Comparative Perception–complex signals, pp. 187–215. New York: John Wiley& Sons, Inc.

Schultz, A. H. (1940). The size of the orbit and of theeye in primates. AJPA Old Series 26, 389–408.

Sherman, P. W. (1977). Nepotism and the evolution ofalarm call. Science 197, 1246–1253.

Shigehara, N. (1996). Metrical study of the direction ofthe orbits in primates. Primates Res. 12, 165–178.

Tomasello, M., Call, J. & Hare, B. (1998). Five pri-mate species follow the visual gaze of conspecifics.Animal Behaviour 55, 1063–1069.

Van Hoo!, J. A. R. A. M. (1962). Facial expressions inthe higher primates. Symposia of the Zool. Soc. Lond.8, 97–125.

Webb, S. V. & Kaas, J. H. (1976). The size anddistribution of ganglion cells in the retina of the owlmonkey, Aotus trivirgatus. Vision Res. 16, 1247–1254.

Wolin, L. R. & Massopust, L. C. Jr (1970). Mor-phology of the primate retina. In (C. R. Noback & W.Montagna, Eds) The Primate Brain, pp. 1–27. NewYork: Appleton Century Crofts.

Yoshino, S. (1994). Animal Face. Tokyo: NikkeiSaiensu.

435.+*/.0 -%12#%3%4( %5 6#0 #.-'+ 0(0