11
PRIMATES, 35(2): 155-165, April 1994 155 How Laboratory-raised Japanese Monkeys (Macacafuscata) Perceive Rotated Photographs of Monkeys: Evidence for an Inversion Effect in Face Perception MASAKI TOMONAGA Kyoto University ABSTRACT. Five laboratory-raised Japanese monkeys (Macacafuscata) were presented various types of photographs of Japanese and rhesus monkeys (Macaca mulatta) in upright, horizontal, and invert- ed orientations in a sensory-reinforcement experiment. The ratio of the duration of potential viewing time for the photographs which the subjects controlled to the interval between subject-controlled presentations of them (the D/I score) was used as a measure of preference for the photographs. When inverted photographs were presented, the D/I scores were lower than for upright photographs. The difference in D/I scores between photographs of the two species, which indicated discriminability between them, also diminished when the photographs were inverted. The results obtained suggest an inversion effect in face perception in macaque monkeys. Key Words: Sensory reinforcement; Species perception; Face perception; Inversion effect; Japanese monkeys. INTRODUCTION Primates are visual as well as social animals (FoBES & KING, 1982). Many studies conducted in the field and in the laboratory show that they recognize various types of social relationships (cf. CHENEY & SEYFARTH, 1990; DASSER, 1988). They also discriminate and identify individuals and species. For example, FUJITA (1990) investigated visual preference among photographs of various species of monkeys using a sensory-reinforcement proce- dure with laboratory-raised Japanese and rhesus monkey infants. In the sensory- reinforcement procedure, a photograph is presented for as long as the subject holds a lever down. If the subject holds the lever again within a specific interval since the last response, the same photograph is repeatedly presented, which enables the subject to control the frequency and duration of viewing it. FUJITA (1987, 1990) and FUJITA and MATSUZAWA (1986) calculated the ratio of the response duration (D) to the interresponse interval (I) and used this D/I score as a measure of preference for each photograph. A large value of the D/I score for a given photograph indicates that it is frequently observed for long durations, that is, it is preferred. Results from the experiments by FUJITA (1990) clearly demonstrated that both laboratory-raised Japanese and rhesus monkey infants showed higher D/I scores for photographs of rhesus monkeys than for those of Japanese monkeys. These findings were inconsistent with FUJITA'S previous experiments (1987), which revealed that field-born monkeys always preferred photographs of their own species. However, both studies showed that the monkeys displayed a difference in preference, as defined by the D/I scores, among photographs of various species. This conclusion also indicates that there is a difference in effectiveness of sensory reinforcement among photographs. Such a difference may derive from some difference in discriminative visual features of each species for the subjects.

How Laboratory-raised Japanese Monkeys (Macaca fuscata ...langint.pri.kyoto-u.ac.jp/ai/intra_data/MasakiTomonaga/...In addition to these photographs, two types of control slides were

  • Upload
    others

  • View
    1

  • Download
    0

Embed Size (px)

Citation preview

Page 1: How Laboratory-raised Japanese Monkeys (Macaca fuscata ...langint.pri.kyoto-u.ac.jp/ai/intra_data/MasakiTomonaga/...In addition to these photographs, two types of control slides were

PRIMATES, 35(2): 155-165, April 1994 155

How Laboratory-raised Japanese Monkeys (Macaca fuscata) Perceive Rotated Photographs of Monkeys: Evidence for an Inversion Effect in Face Perception

MASAKI TOMONAGA Kyoto University

ABSTRACT. Five laboratory-raised Japanese monkeys (Macacafuscata) were presented various types of photographs of Japanese and rhesus monkeys (Macaca mulatta) in upright, horizontal, and invert- ed orientations in a sensory-reinforcement experiment. The ratio of the duration of potential viewing time for the photographs which the subjects controlled to the interval between subject-controlled presentations of them (the D/I score) was used as a measure of preference for the photographs. When inverted photographs were presented, the D/I scores were lower than for upright photographs. The difference in D/I scores between photographs of the two species, which indicated discriminability between them, also diminished when the photographs were inverted. The results obtained suggest an inversion effect in face perception in macaque monkeys.

Key Words: Sensory reinforcement; Species perception; Face perception; Inversion effect; Japanese monkeys.

I N T R O D U C T I O N

Primates are visual as well as social animals (FoBES & KING, 1982). Many studies conducted in the field and in the laboratory show that they recognize various types of social relationships (cf. CHENEY & SEYFARTH, 1990; DASSER, 1988). They also discriminate and identify individuals and species. For example, FUJITA (1990) investigated visual preference among photographs of various species of monkeys using a sensory-reinforcement proce- dure with laboratory-raised Japanese and rhesus monkey infants. In the sensory- reinforcement procedure, a photograph is presented for as long as the subject holds a lever down. I f the subject holds the lever again within a specific interval since the last response, the same photograph is repeatedly presented, which enables the subject to control the frequency and duration of viewing it. FUJITA (1987, 1990) and FUJITA and MATSUZAWA (1986) calculated the ratio of the response duration (D) to the interresponse interval (I) and used this D/ I score as a measure of preference for each photograph. A large value of the D / I score for a given photograph indicates that it is frequently observed for long durations, that is, it is preferred. Results from the experiments by FUJITA (1990) clearly demonstrated that both laboratory-raised Japanese and rhesus monkey infants showed higher D/ I scores for photographs of rhesus monkeys than for those of Japanese monkeys. These findings were inconsistent with FUJITA'S previous experiments (1987), which revealed that field-born monkeys always preferred photographs of their own species. However, both studies showed that the monkeys displayed a difference in preference, as defined by the D/ I scores, among photographs of various species. This conclusion also indicates that there is a difference in effectiveness of sensory reinforcement among photographs. Such a difference may derive from some difference in discriminative visual features of each species for the subjects.

Page 2: How Laboratory-raised Japanese Monkeys (Macaca fuscata ...langint.pri.kyoto-u.ac.jp/ai/intra_data/MasakiTomonaga/...In addition to these photographs, two types of control slides were

156 M. TOMONAGA

What are these features that influence the monkeys' viewing? In human infants, faces are known to have positive reinforcing properties for visual fixation (KAPLAN et al., 1992). Is that also true for monkeys? To address this question, FUJITA (1993) conducted an experi- ment in which pig-tailed monkeys were shown photographs of various species of the genus Macaca where the candidate features for species recognition (head, body, tail, and so on) had been masked by black circles. Although the subjects showed no change in preference among the species, they looked for a shorter duration and less frequently at photographs in which the head was masked compared to those without masks. These results suggested that the head (probably the face) is a significant feature for visual species recognition.

In primates, facial expressions are frequently used for visual communication. Primates have several facial expressions, although they are fewer and less variable than those of humans (CHEVALIER-SKOLNIKOFF, 1973; VAN HOOFF, 1967). Furthermore, faces may con- tain various types of information in a hierarchical way. From the face of another individual, primates may recognize him/her (MATSUZAWA, 1989, 1990), his/her sex (ITAKURA, 1992), his/her group membership (DASSER, 1988; FREDRICKSON & SACKETT, 1984; SACKETT FREDRICKSON, 1987), his/her species (FuJITA, 1987, 1990, 1993; YOSHIKUBO, 1985), and his/her affective state (HAMILTON ~; VERMEIRE, 1988; KANAZAWA, 1993: PERRETT MISTLIN, 1990; SACKETT, 1966). Some researchers of human face recognition suggest that humans may have independent modules for processing these kinds of information from human faces (BRUCE & YOUNG, 1986).

If faces are special for primates as well as humans, some of the phenomena specific to face processing by humans might also have counterparts in monkeys. For example, face recognition by humans is subject to what is called the inversion effect, in which subjects display a greater difficulty in recognizing inverted than upright faces (CAREY, 1981; GOLDSTEIN, 1965; KOHLER, 1940; SCAnNELLO & YARMEY, 1970; YIN, 1969). For example, YIN (1969) investigated the recognition of various types of visual stimuli, such as human faces, houses, and airplanes in two orientations, upright and inverted. After memorizing the list of photographs, the subjects were shown a picture and were asked whether it was in the previous list. The results indicated that recognition memory was significantly more impaired when the human faces were inverted than upright, while such a disruption did not occur for other stimuli.

In the present experiment, I tested the face inversion effect in laboratory-raised Japanese monkeys using a species perception task in which the face was a critical feature for species discrimination. If the subjects do exhibit an inversion effect, the preference score (i.e. D/I score) should be higher for upright than for inverted photographs. Furthermore, if this effect is specific to faces, the clarity of the faces in the photographs should modify the inversion effect. To address this question, the photographs employed in the present experi- ments were classified into three types on the basis of the clarity of the faces, and the data obtained were separately analyzed for each type.

METHOD

SUBJECTS

As shown in Table 1, five laboratory-born juvenile Japanese monkeys served as subjects. They had been isolated from their mothers within one week after birth and raised by human

Page 3: How Laboratory-raised Japanese Monkeys (Macaca fuscata ...langint.pri.kyoto-u.ac.jp/ai/intra_data/MasakiTomonaga/...In addition to these photographs, two types of control slides were

Perception of Rotated Faces in Monkeys

Table 1. Subjects of the present experiment and their social experience.

157

Name Sex Birthday Age at onset of present experiment Cagemate

Janko Female June 10, 1986 5:2 M. mulatta Qta Male May 29, 1987 4:2 34. fuscata Tonko Female May 04, 1986 5:2 M. mulatta Hanako Female May 28, 1988 3:2 M. mulatta Ichitaro Male May 26, 1988 3:2 M. mulatta

caretakers. They were raised in a cage (70 x 70 x 70 cm) with one cagemate. Some monkeys were paired with a same species cagemate (QTA), and the others were raised with rhesus monkeys (Janko, Tonko, Hanako, and Ichitaro). In previous experiments (FuJITA, 1990), all subjects showed a preference for photographs of rhesus monkey individuals over those of Japanese monkeys. This result indicates that they can discriminate reliably between Japanese and rhesus monkeys from pictorial stimuli. In the present experiment, the subjects were not deprived of either food or water. The care and use of the subjects adhered to the "Guide for the Care and Use of Laboratory Primates" 0986) of the Primate Research Institute, Kyoto University.

APPARATUS

The apparatus was the same as that used by FUJITA (1990). Two identical experimental chambers (60 x 60 x 60 cm) were set up in the experimental room. The front panel (40 x 40 cm) of each chamber was made of clear Plexiglass. A slide projector (CABIN, model Family Cabin) was installed behind the screen and projected the slide photographs onto a 33 x 33 cm opaque screen placed 50 cm from the front panel. A single lever with a red lamp was installed in the lower center of the front panel. All experimental events were controlled and recorded by a MSX2 personal computer (TOSHIBA, model HX-34).

STIMULI

In the present experiment, I prepared two slide sets for preliminary training and two slide sets of Japanese and rhesus monkeys for the species-perception tasks. One of the prelimi- nary slide sets consisted of 100 photographs of animals and plants, while the other set consisted of 100 scenes from Japan. The two slide sets of monkeys were selected from the library of photographs collected for FUJITA'S (1990) experiments; the photographs had been taken in the monkey corrals at the Primate Research Institute, Kyoto University. The first set, called the UI ( u p r i g h t - inverted) set, contained 160 photographs and consisted of 40 different photographs in two orientations (upright and inverted) for each species. In addition to these photographs, two types of control slides were used; one was a white light slide, and the other was a no-light slide. Each type of slide appeared 20 times. Each slide was randomly arranged and set in two 100-slide carousels. The second set, called the UHI (up r igh t -ho r i zon ta l - i nve r t ed ) set, contained 252 photographs and consisted of 42 photographs in three different orientations (upright, horizontal, and inverted) for each species, and 24 white-light and 24 no-light control slides. The photographs used were differ- ent from those in the UI set. These slides were randomly assigned into three carousels. Figure 1 shows examples of the photographs employed in the present experiment.

Page 4: How Laboratory-raised Japanese Monkeys (Macaca fuscata ...langint.pri.kyoto-u.ac.jp/ai/intra_data/MasakiTomonaga/...In addition to these photographs, two types of control slides were

Fig.

1.

Exa

mpl

es o

f up

righ

t ph

otog

raph

s us

ed i

n th

e pr

esen

t ex

peri

men

t. U

pper

row

, pho

togr

aphs

of

rhe

sus

mon

keys

; lo

wer

row

, pho

togr

aphs

of

Japa

nese

mon

keys

. L

eft

colu

mn,

FA

CE

-CL

R s

lides

; ce

nter

col

umn,

FA

CE

-AM

B s

lides

; ri

ght

colu

mn,

FA

CE

-UN

C s

lides

.

o o z >

Page 5: How Laboratory-raised Japanese Monkeys (Macaca fuscata ...langint.pri.kyoto-u.ac.jp/ai/intra_data/MasakiTomonaga/...In addition to these photographs, two types of control slides were

Perception of Rotated Faces in Monkeys

Table 2. Number of slides of each category employed in the present experiment.

159

Species

UI set UHI set

FACE-CLR FACE-AMB FACE-UNC Total FACE-CLR FACE-AMB FACE-UNC Total

M. fuscata 15 15 10 40 23 9 10 42 M. mulatta 15 14 11 40 13 12 16 42

Total 30 29 21 80 36 21 26 84

The number of monkeys presented in each photograph varied from 1 to approximately 10. These photographs were divided into the following three categories on the basis of the clarity of the faces as judged by the experimenter before the onset of the present experi- ments (see Fig. 1): photographs in which front views of the faces may be seen clearly [FACE-CLR (clear)], those in which faces are seen but less clearly than in a front view, such as a side view [FACE-AMB (ambiguous)], and those in which faces were unclear or uniden- tifiable because the individuals in the photograph were looking down towards the ground and so on [FACE-UNC (unclear)]. The distribution of slides among these categories is summarized in Table 2.

PROCEDURE

In the present experiment, the sensory-reinforcement procedure was employed (FuJITA

MATSUZAWA, 1986; FUJITA, 1987, 1990, 1993). Figure 2 gives a schematic diagram of typical trials. When the subject held down the lever in the presence of a red light on the lever, the slide was projected onto the screen during the subject's holding of the lever, with the red light off. Since the presentation of the slide was contingent upon the subject's responses, it was procedurally considered to be a sensory reinforcer. The reinforcing strength of the slides can be identified by measuring the duration of lever holding (D) and the inter- response interval (I: the time from the release of the lever to the next press). The slide presentation was terminated when the subject released the lever or when lever holding bad lasted l0 sec. I f the subject held the lever again within 10 sec after the previous release of the lever, the same slide was again presented. I f the subject did not press the lever for l0 sec, the next slide was set up for presentation. The time from onset of lever holding through lever releasing to the next onset of holding was defined as a response cycle. The

SLIDE SLIDE SLIDE CHANGE CHANGE CHANGE

; 1 n

:,,...R! ..~I .... ~3 .... i J .~!-- ..i ,R.!-I-~R~I ..... R4 .......

: i i ; D I I1 ~ I2 1)3 I3 : D I I1 1 D1 * D 2 1 D 3 ~ D 4 I4

D2 I 1 II I2 I3

R: RESPONSE CYCLE D: RESPONSE DURATION I: INTERRESPONSE INTERVAL

Fig. 2. Schematic diagram of the sensory-reinforcement procedure.

Page 6: How Laboratory-raised Japanese Monkeys (Macaca fuscata ...langint.pri.kyoto-u.ac.jp/ai/intra_data/MasakiTomonaga/...In addition to these photographs, two types of control slides were

160 M. TOMONAGA

period in which the same slide was presented once or more than once in succession was defined as a trial.

Testing with each slide set was continued until each slide had been repeated five times. A daily session continued for approximately six hours. Each subject was initially given the UI set, and then the UHI set. Before starting the experimental sessions with each set, the subjects underwent a single training session with the preliminary slide sets.

D A T A ANALYSIS

As a measure of the reinforcing strength of each slide, D/I scores were calculated (FuJITA, 1990). The mean duration of lever holding (D) and the mean interresponse inter- val (I) were calculated for each slide across five repeated trials. The D/I score for each slide was obtained on the basis of these values. An increase in the D/ I score indicates that the subject held the lever longer in the presence of the slide. From this, it was concluded that the slide has a greater reinforcing strength, and that the subject "prefers" that slide. Since there were individual differences in the means and standard deviations of the distributions of responses used to calculate the D/ I score, the data for each subject were normalized into Z-scores in which the mean was 50 and the standard deviation was 10. In the present experi- ment, these normalized D/I scores were employed for analysis.

RESULTS

Figure 3 summarizes the results for the UI set. Each point represents the mean normal- ized D/ I score for each category of slide across the subjects. Mean normalized D/I scores for control slides are also shown in the left part of the figure. Data from each subject are listed in Table 3.

As shown in Figure 3 and Table 3, all subjects including QTA who had a different social experience before the onset of the present experiment (see Table 1), held the lever longer

u.l r r o o O9

121 a H/ N

<

r r O z

65

60

55

5O

45

(~. 0 M. mulatta �9 �9 M. fuscata

' ' u ' ' u ' o ' ' 40 N W P INV P INV P INV

CONTROL FACE-CLR FACE-AMB FACE-UNC

ORIENTATION OF PHOTOGRAPH Fig. 3. Mean normalized D/I scores for each type of photographs in the UI set. The leftmost data set shows the results for the control slides (N: no light; W: white light); the second set, the category FACE-CLR; the third set, FACE-AMB; and the rightmost set, FACE-UNC. Open circles indicate the mean normalized D/I scores of the slides with rhesus monkeys; filled circles indicate those with Japanese monkeys. The vertical bar for each point indicates the standard deviation across subjects.

Page 7: How Laboratory-raised Japanese Monkeys (Macaca fuscata ...langint.pri.kyoto-u.ac.jp/ai/intra_data/MasakiTomonaga/...In addition to these photographs, two types of control slides were

Perception of Rotated Faces in Monkeys

Table 3. Normalized D/I scores for each subject (UI set).

161

FACE-CLR FACE-AMB FACE-UNC

Subject Control slides Species Up lnv Up Inv Up Inv

Janko N 37.32 Mf 50.21 53.93 48.97 50.04 48.28 53.69 W 40.97 Mm 61.79 59.73 53.44 53.81 50.05 52.51

Qta N 39.24 Mf 52.48 53.34 52.06 46.04 51.70 46.45 W 47.26 Mm 62.67 54.02 57.24 46.92 52.79 48.94

Tonko N 40.27 Mf 48.52 48.42 46.54 46.17 49.32 47.93 W 42.30 Mm 57.97 55.85 57.62 53.96 59.09 55.28

Hanako N 43.13 Mf 52.21 46.12 51.17 44.30 54.76 46.59 W 42.22 Mm 60.79 51.53 61.87 50.75 52.56 50.46

Ichitaro N 46.03 Mf 48.96 53.00 47.61 46.11 49.20 50.68 W 46.49 Mm 64.85 54.65 49.88 49.35 53.42 48.07

N: No-light slides; W: white-light slides.

for the slides with rhesus monkeys than for those with Japanese monkeys. Furthermore, in categories FACE-CLR and FACE-AMB, upright slides were preferred to inverted slides. Two-way (Species x Orientation of the photograph) randomized-block analyses of variance (ANOVAs) with subjects as repeated factors were conducted separately for the data from each slide category (FACE-CLR, FACE-AMB, and FACE-UNC). With FACE-CLR slides, both main effects were significant [Species, F(1,12)=47.59, p<.01; Orientation, F(1,12)=7.21, p<.05] , and the interaction was also significant [F(1,12)=9.75, p<.01]. Tukey's HSD test revealed significant differences between upright Japanese and rhesus monkeys, and between upright and inverted rhesus monkeys (p's <.05). With FACE-AMB slides, both main effects were significant [Species, F(1,12)=15.19, p<.01; Orientation, F(1,12) = 7.39, p < .05], but their interaction was not significant IF(l,12)= 0.65]. With FACE- UNC slides, no significant main effects or interaction were observed [F's < 3].

Figure 4 summarizes the results for the UHI set. Table 4 lists each subject's data. The subjects showed almost the same tendency as for the UI set. Separate ANOVAs, for FACE- CLR slides, indicated that both main effects were significant [Species, F(1,20)=7.08, p< .05 ; Orientation, F(2,20)=5.48, p<.05] , but their interaction was not [F(2,20)=2.48, p=.109]. Tukey's HSD test revealed significant differences between upright Japanese and rhesus monkeys and between upright and inverted rhesus monkeys (p's < .05). For FACE-

55

S -'3 50

rr 45 O Z

40 L I

NW

0 M. mulatta (~ �9 M. fuscata

UIp I I i I I I HOR INV UP HOR INV UP HOR INV

C O N T R O L FACE-CLR FACE-AMB FACE-UNC

ORIENTATION OF PHOTOGRAPH Fig. 4. Mean normalized D/I scores for each type of photographs in the UHI set.

Page 8: How Laboratory-raised Japanese Monkeys (Macaca fuscata ...langint.pri.kyoto-u.ac.jp/ai/intra_data/MasakiTomonaga/...In addition to these photographs, two types of control slides were

162

Table 4. Normalized D/I scores for each subject (UHI set).

M. TOMONAGA

FACE-CLR FACE-AMB FACE-UNC Subject Control slides Species Up Hor Inv Up Hor Inv Up Hor Inv Janko N 36.67 Mf 48.81 52.19 51.76 47.72 51.84 48.63 47.10 50.76 47.48

W 41.08 Mm 62.32 59.43 53.28 52.63 55.31 51.93 52.68 56.50 48.59 Qta N 41.96 Mf 55.53 52.65 48 .09 56.41 48.25 48 .24 52.85 48.33 43.56

W 45.10 Mm 62.81 53.68 49.57 56.52 52.26 46.61 54 .06 47.39 49.79 Tonko N 37.90 Mf 52.81 50.55 54.91 46.95 50.30 47.94 48.54 46.58 50.98

W 41.16 Mm 58.36 54.29 51.12 57.29 54.67 49.66 56 .07 55.44 52.46 Hanako N 47.38 Mf 53.09 53.94 51.44 55.46 50.85 45 .19 52.85 48.36 48.32

W 43.79 Mm 57.45 49.85 46.66 53.28 52.60 50.14 50.26 46.47 51.95 Ichitaro N 47.33 Mf 52.46 50.66 47.21 51.50 45.77 48.35 47.23 47.00 47.13

W 45.07 Mm 56.86 50.01 58.64 53.20 50.99 54.99 49.24 52.83 51.75

A M B slides, bo th ma in effects were s ignif icant [Species, F(1,20)= 9.08, p <.01; Or ien ta t ion , F(2,20) = 4.44, p < .05], but their in terac t ion was not [F(2,20)= 0.06]. For F A C E - U N C slides, only the ma in effect o f Species was s ignif icant IF( I ,20)= 9.00, p < . 0 1 ] .

D I S C U S S I O N

The aim of the present exper iment was to examine whether or not l abora to ry- ra i sed Japanese monkeys showed a preference between pho tog raphs o f Japanese and rhesus monkeys when they were presented in upr ight versus inverted or ienta t ions . The first m a j o r f inding was tha t the subjects ma in t a ined the presenta t ions o f upr igh t pho tog raphs for longer dura t ions and more f requent ly than those o f inverted ones, especial ly in the case o f rhesus monkey slides and F A C E - C L R and F A C E - A M B slides. This suggests tha t the visual features which de te rmine the reinforcing s t rength of a s t imulus may be or ien ta t ion- specific. JITSUMORI and MATSUZAWA (1991) found that it was easy for monkeys to establish the concepts of " u p r i g h t n e s s " and " inve r t ednes s " of na tura l i s t ic st imuli , implying tha t the monkeys may see someth ing meaningfu l in upr igh t versus inverted pho tographs . In the present exper iment , the effects o f the o r i en ta t ion var iable d i sappea red with FACE-UNC slides, suggest ing tha t faces have an or ien ta t ion-spec i f ic effectiveness o f sensory rein- forcement for monkeys, and tha t faces are more meaningfu l t han the other features o f the p h o t o g r a p h s such as their colors, b a c k g r o u n d scenes, and o ther body par ts shown.

A second i m p o r t a n t f inding was that the d i sc r imina t ion between species was d is rupted when the slides were presented ups ide down. As noted earlier, d i f ferent ia l effects of sensory re inforcement a m o n g the different types o f pho tog raphs imply tha t these pho tog raphs are visual ly d i sc r iminab le in monkeys. As FUJITA (1993) revealed, monkeys uti l ize the faces as a dis t inct feature for species d iscr iminat ion . The second f inding f rom the present experi- men t suggests tha t the features for species d i sc r imina t ion con ta ined in the faces are dependen t on their o r ien ta t ions or tha t d i sc r iminabi l i ty o f faces depends on their or ienta- t ion. I f the faces are presented with an a b n o r m a l or ien ta t ion , such as inversion, monkeys canno t d i scr imina te them.

In the present exper iment , the inversion effects were s t ronger when the faces were c lear ly ident i f ied than when they were less clearly ident i f ied as descr ibed above. This could imp ly tha t these effects are de te rmined by the faces in the p h o t o g r a p h s to some degree. The present results can thus be considered as evidence for an inversion effect o f face percep-

Page 9: How Laboratory-raised Japanese Monkeys (Macaca fuscata ...langint.pri.kyoto-u.ac.jp/ai/intra_data/MasakiTomonaga/...In addition to these photographs, two types of control slides were

Perception of Rotated Faces in Monkeys 163

tion in nonhuman primates. Some studies on discrimination learning with faces as stimuli in primates have failed to show an inversion effect (BRUCE, 1982; DITTRICH, 1990; ROSENFELD t~r VAN HOESEN, 1979; TOMONAGA et al., in press), but in other recent studies an inversion effect was observed (KEATING & KEATING, 1993; PERRETT et al., 1988). One reason for the discrepancy with the data reported here may be related to the difference in stimulus functions between the previous (discriminative) and present experiments (reinforc- ing). A second reason is that the monkeys might utilize a different strategy for face process- ing, i.e. piecemeal or configural processing, depending on different task requirements. For example, KEATING and KEATING (1993) trained four monkeys to discriminate between stan- dard human faces and faces made up of each part (forehead, eye, eyebrow, nose, lips, and chin) selected randomly from elements stored in an apparatus for creating montages of the face (Identi-Kit). After learning this discrimination, monkeys were shown inverted faces and required to judge whether they were standard or not. Three of the four monkeys revealed a decrease in accuracy when inverted faces were presented. Furthermore, KEATING and KEATING (1993) investigated the fixation pattern of the monkeys and found that one monkey who displayed a strong inversion effect (007o correct for inverted faces) solved this task in a configural way, employing the principal: if the right eye is not in the specificposi- tion (upper left), reject it as standard. In contrast, the fixation patterns of another monkey who showed no inversion effect suggested that this monkey utilized a piecemeal strategy: if the familiar left eye appears in any position, accept it as standard. These results suggested that an inversion effect was observed when the subjects used configural or holistic cues for the discrimination of faces. This hypothesis can explain the inconsistency in findings from previous experiments. For example, BRUCE (1982) employed a set of single photo- graphs for each face A versus face B discrimination problem and failed to obtain an inversion effect, whereas OVERMAN and DOTY 0982) employed "tr ial-unique" stimuli for face versus no-face discrimination and obtained an inversion effect. It is clear that the use of a number of stimuli can affect the inversion effect. In the present experiment, each monkey was shown about 80 different photographs. This procedure not only encouraged categorization of the photographs on the basis of species, but resulted in the emergence of an inversion effect.

In humans, face recognition and the inversion effect are known to be functionally later- alized to the right hemisphere (LEEHEY et aI., 1978; MILNER, 1968; YIN, 1970). Although some studies with primates have also demonstrated such a hemispheric asymmetry in face processing (HAMILTON 8r VERMEIRE, 1988), many studies reported a lack of such asym- metry (e.g. HAMILTON 8Z VERMEIRE, 1983; OVERMAN r162 DOTY, 1982; PERRETT et al., 1988). In humans, the right hemisphere is specialized for spatial discrimination (TAYLOR & WARRINGTON, 1973). This implies that a configural or holistic strategy of face processing is closely related to the right hemispheric advantage of spatial information. Recently, primatologists and neuropsychologists have shown much interest in primate laterality or hemispheric specialization (e.g. WARD 8,:: HOPKINS, 1993). The hemispheric specialization of face processing observed in nonhuman primates should be further examined when discussing the evolution of the specificity of face processing in humans.

Acknowledgments. I thank Dr. K. FUJITA for his helpful advice concerning this study and kind permission to utilize his photograph libraries. Thanks are also due to SUMIHARU NAGUMO for his technical support and to IVER IVERSEN for his critical reading of an earlier version of the English manuscript.

Page 10: How Laboratory-raised Japanese Monkeys (Macaca fuscata ...langint.pri.kyoto-u.ac.jp/ai/intra_data/MasakiTomonaga/...In addition to these photographs, two types of control slides were

164 M. TOMONAGA

R E F E R E N C E S

BRUCE, C., 1982. Face recognition by monkeys: absence of inversion effect. NeuropsychoL, 20: 515-521.

BRUCE, V. & A. YOUNG, 1986. Understanding face recognition. Brit. J. Psyehol., 77: 305- 327. CAREY, S., 1981. The development of face perception. In: Perceiving and Remembering Faces,

G. DAVIES, H. ELLIS, & J. SHEPHERS (eds.), Academic Press, New York, pp. 9 - 3 8 . CHENEV, D. L. & R. M. SEYFARTH, 1990. How Monkeys See the WorM. Univ. of Chicago Press,

Chicago, Illinois. CHEVALmR-SKOLNmOFG S., 1973. Facial expression of emotion in nonhuman primates. In: Darwin

and Facial Expression, P. EKMAN (ed.), Academic Press, New York, pp. 11-89. DASSER, V., 1988. A social concept in Java monkeys. Anim. Behav., 36: 225-230. DITTRICH, W., 1990. Representation of faces in longtailed macaques (Macaca fascicularis). Ethol.,

85:265 - 278. FOBES, J. L. & J. E. KING, 1982. Vision: the dominant primate modality. In: Primate Behavior,

J. L. FOBES & J. E. K~NG (eds.), Academic Press, New York, pp. 245-270. FREDRICKSON, W. T. & G. P. SACKETT, 1984. Kin preference in primates (Macaca nemestrina): related-

ness or familiarity? J. Comp. PsychoL, 98: 29-34 . FUJ1TA, K., 1987. Species recognition by five macaque monkeys. Primates, 28: 353-366.

, 1990. Species preference by infant macaques with controlled social experience. Int. J. Primatol., 11: 553-573.

, 1993. Role of some physical characteristics in species recognition by pigtailed macaques. Primates, 34:133 - 140.

- - & T. MATSUZAWA, 1986. A new procedure to study the perceptual world of animals with sensory reinforcement: recognition of humans by a chimpanzee. Primates, 27: 283-291.

GOLDSTEIN, A. G., 1965. Learning of inverted and normally oriented faces in children and adults. Psychonom. Sci., 3: 447-448.

HAMILTON, C. R. & B. A. VERMEIRE, 1983. Discrimination of monkey faces by split-brain monkeys. Behav. Brain Res., 9: 263-275.

& - - , 1988. Complementary hemispheric specialization in monkeys. Science, 242: 1691 - 1694.

ITAKURA, S., 1992. Sex discrimination of photographs of humans by a chimpanzee. Percept. Motor Skills, 74: 475- 478.

JITSUMORI, M. & T. MATSUZAWA, 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.

KANAZAWA, S,, 1993. Experimental Research on the Recognition of Facial Expressions in Man and Japanese Monkey. Unpublished Master thesis, Kyoto University.

KAPLAN, P. S., K. B. FOX, & E. R. HUCKEBY, 1992. Faces as reinforcers: effects of pairing condition and facial expression. Devel. PsychobioL, 25: 299-312.

KEAT1NG, C. F. & E. G. KEATING, 1993. Monkeys and mug shots: cues used by rhesus monkeys (Macaca mulatta) to recognize a human face. J. Comp. Psychol., 107:131-139.

KOHLER, W., 1940. Dynamics in Psychology. Liveright, New York. LEEHEY, S. C., S. CAREY, R. DIAMOND, & A. CAHN, 1978. Upright and inverted faces: the right

hemisphere knows the difference. Cortex, 14:411-419. MATSUZAWA, T., 1989. Spontaneous pattern construction in a chimpanzee. In: Understanding

Chimpanzees, P. G. HELTNE & L. A. MARQUARDT (eds.), Harvard Univ. Press, Cambridge, Massachusetts, pp. 252-265.

, 1990. Form perception and visual acuity in a chimpanzee. Folia PrimatoL, 55: 24-32 . MILNER, B., 1968. Visual recognition and recall after right temporal lobe excision in man. Neuropsy-

ehol., 6:191-209. OVERMAN, W. H. JR. & R. W. DOTY, 1982. Hemispheric specialization displayed by man but not

macaques for analysis of faces. Neuropsychol., 20: 619-635. PERRETT, D. I. & A. J. MISTLIN, 1990. Perception of facial characteristics by monkeys. In: Compara-

tive Perception, Vol. 2, W. C. STEBmNS & M. A. BERKLEV (eds.), Wiley, New York, pp. 187 -215.

Page 11: How Laboratory-raised Japanese Monkeys (Macaca fuscata ...langint.pri.kyoto-u.ac.jp/ai/intra_data/MasakiTomonaga/...In addition to these photographs, two types of control slides were

Perception of Rotated Faces in Monkeys 165

- - , A. J. CH1TTV, E A. J. SMITH, D. D. POTTER, R. BROENNIMANN, t~ M. HARRIES, 1988. Specialized face processing and hemispheric asymmetry in man and monkey: evidence from single unit and reaction time studies. Behav. Brain Res., 29: 245- 258.

PRIMATE RESEARCH INSTITUTE, 1986. Guide for the Care and Use o f Laboratory Primates. Primate Research Institute, Kyoto Univ., Aichi, Japan.

ROSENFELD, S. A. & G. W. VAN HOESEN, 1979. Face recognition in the rhesus monkey. Neuropsychol., 17:503 - 509.

SACKETT, G. E, 1966. Monkeys reared in visual isolation with pictures as visual input: evidence for innate releasing mechanism. Science, 154: 1468-1472.

- - & W. T. FREDRICKSON, 1987. Social preferences by pigtailed macaques: familiarity versus degree and type of kinship. Anim. Behav., 35: 603-606.

SCAPINELLO, K. E & A. D. YARMEV, 1970. The role of familiarity and orientation in immediate and delayed recognition of pictorial stimuli. Psychonom. Sci., 21: 329-331.

TAYLOR, A. M. & E. K. WARRINCTON, 1973. Visual discrimination in patients with localized cerebral lesions. Cortex, 9: 82-93.

TOMONA~A, M., S. ITA~URA, & T. MATSUZAWA, in press. Superiority of conspecific faces and reduced inversion effect in face perception by a chimpanzee. Folia Primatol.

VAN HOOFF, J. A. R. A. M., 1967. The facial displays of catarrhine monkeys and apes. In: Primate Ethology, D. MORRIS (ed.), Weidenfield & Nicolson, London, pp. 7 - 68 .

WARt), J. P. & W. D. HOPK1NS (eds.), 1993. Primate Laterality: Current Behavioral Evidence o f Primate Asymmetries. Springer, Berlin.

YIN, R. K., 1969. Looking at upside-down faces. J. Exp. Psychol., 81:141-145. - - , 1970. Face recognition by brain-injured patients: a dissociable ability? Neuropsychol., 8:

395 - 402. YOSHIKUBO, S., 1985. Species discrimination and concept formation by rhesus monkeys (Macaca

mulatta). Primates, 26: 285 - 299.

- - Received: May 8, 1993; Accepted: November 7, 1993

Author's Name and Address: MASAr~I TOMONA~A, Department of Behavioral and Brain Sciences, Primate Research Institute, Kyoto University, lnuyama, Aichi 484, Japan.