Human and chimpanzee face processing in chimpanzees

Julie Martin-Malivel1 and Kazunori Okada2

1Yerkes National Primate Research Center, Emory University2Computer Science Department, San Francisco State University

Predisposition for processing conspecific faces?

• Monkeys reared in isolation: preference for pictures of conspecific infants; fear reactions to threat pictures (Sackett, 1966)

• Sackett (1970): monkeys have an innate predisposition to respond to conspecifics, suggesting the possible existence of a species-specific recognition system

• Preference for pictures of their own species (Fujita, 1987), even when reared without experience with conspecifics (Fujita, 1990, 1993). E.g., rhesus monkeys reared with Japanese monkeys still showed a preference for pictures of their genetic species, rhesus monkeys (Fujita, 1990)

Species-specific effect

• Laboratory monkeys daily exposed to human faces discriminated between pictures of monkey faces but not between pictures of human faces (Pascalis & Bachevalier, 1998)

• Use internal features to discriminate conspecifics but not to discriminate humans (monkeys: Martin-Malivel et al., 2001; sheep: Peirce, Leigh, & Kendrick, 2000; Peirce et al., 2001).

• Right hemisphere advantage for conspecifics but for human faces (Peirce et al., 2000, 2001)

Subjects

• Four chimpanzees from the Language Research Center (GSU): Lana, Sherman, Panzee, Mercury

– Exposure to chimpanzees: • Three conspecifics in their daily lives• About 12 total in their whole lives

– Exposure to humans: • High (caretakers, researchers, vets, visitors, etc.) • Free visible faces (no masks and shields)• Direct interactions (extensive training for

‘language’/lexigrams and multiple cognitive tasks)

Do they recognize one species better than the other?

• Baseline trials: scores with human pictures versus scores with chimpanzee pictures

• Categorical Perception: morphs between humans; morphs between chimpanzees

Delayed Matching-To-SampleStimuli

• Eight unknown chimpanzees; Eight unknown humans

Delayed Matching-To-SampleProcedure

Results – Baseline trialsHuman versus chimpanzee pictures

Scores

0102030405060708090

100

LRC Yerkes

perc

enta

ge c

orre

ct (

%)

p=.013* p=.266

500

1000

1500

2000

2500

3000

LRC Yerkes

chimp

human

center factor: p < .001*** -------------

p=.48

p=.86

Are the human pictures more different than the chimpanzee pictures?

Computational simulation(based on artificial face recognition system)

0.850.8250.800.775

Similarity Coefficients from Computational Simulation

100

95

90

85

80

75

70

Sco

res

LR

C S

ub

ject

s

humanchimphumanchimp

species

R Sq Linear = 0.528

R Sq Linear = 0.337

Correlation between Computational Simulation and

chimpanzees’ scoresHuman pictures

Chimpanzee pictures

LRC

r(21) = -.627, , p = .001**

• The LRC chimpanzees are better at recognizing human than chimpanzee pictorial faces

• Computational simulation is consistent with subjects’ performance

• Do subjects perceive faces in a categorical way?

• Emergence categorical perception effect for both humans and chimpanzees?

Theoretical boundary

100ch1/0ch2 80ch1/20ch2 60ch1/40ch2 40ch1/60ch2 20ch1/80ch2 0ch1/100ch2

|_____P1____| |_____P2____| |_____P3____| |_____P4_____| |_____P5_____|

Ch1 Ch2

Hu1 Hu2

Categorical perception

Cross-categories Within-categoryWithin-category

12 chimpanzee morphs

12 human morphs

Results – Categorical Perception

0

20

40

60

80

chimp humanLRC

perc

enta

ge c

orre

ct (

%)

within-catego

cross-catego

p=.027*p=.698

Conclusions

• Chimpanzees with higher exposure to human faces than conspecific faces can recognize human faces more efficiently

• Categorical perception emerged only with species for which they have a high expertise

Merci!

Comparison with another chimpanzee population

• Four chimpanzees from Yerkes (Katrina, Scott, Lamar, Jarred)

– Exposure to chimpanzees: Yerkes population is large, but limited visual contact. Only direct interaction with 1 conspecific since many years (pair-housed )

– Exposure to humans: wear PPE

Results - Baseline trialsHuman versus chimpanzee pictures

Reaction times

500

1000

1500

2000

2500

3000

LRC Yerkes

resp

onse

tim

es (

ms)

chimp

human

center factor: p < .001*** -------------

p=.48

p=.86

Results - Baseline trialsHuman versus chimpanzee pictures

Scores

0102030405060708090

100

LRC Yerkes

perc

enta

ge c

orre

ct (

%)

p=.013* p=.266

0.850.8250.800.775

Similarity Coefficients from Computational Simulation

100

95

90

85

80

75

70

Sco

res

LR

C S

ub

ject

s

humanchimphumanchimp

species

R Sq Linear = 0.528

R Sq Linear = 0.337

Correlation Computational Simulation with

chimpanzees’ scores

Human pictures

Chimpanzee pictures

LRC

r(21) = -.627, , p = .001**

0.850.8250.800.775

Similarity Coefficients from Computational Simulation

100

98

96

94

92

90

88

86

84

82

80

Sco

res

Yer

kes

Su

bje

cts

humanchimphumanchimp

species

R Sq Linear = 0.615

R Sq Linear = 0.713

Yerkes

r(21) = -.817, p < .001***

Results – Categorical Perception

0

20

40

60

80

chimp humanLRC

perc

enta

ge c

orre

ct (

%)

within-catego

cross-catego

p=.027*p=.698

0

20

40

60

80

chimp humanYerkes

p=.11p=.61

Conclusions

• Chimpanzees with higher exposure to human faces than conspecific faces can recognize human faces more efficiently (LRC)

• Yerkes recognize both human and chimpanzee faces (consistent with Parr et al., 1998), but maybe not as fine tuned as LRC chimpanzees (no CP effect)

Predisposition for processing conspecific faces?

• Rhesus monkeys reared in isolation showed a preference for pictures of conspecific infants in comparison with other pictures, and showed fear reactions to threat pictures (Sackett, 1966).

• Sackett (1970) has argued that monkeys have an innate predisposition to respond to the individual features of conspecifics, suggesting the possible existence of a species-specific recognition system.

• Monkeys prefer to look at pictures of their own species than pictures of other species (Fujita, 1987), even in subjects reared without experience with individuals of their own species (Fujita, 1990, 1993). E.g., rhesus monkeys reared with Japanese monkeys still showed a preference for pictures of their genetic species, rhesus monkeys (Fujita, 1990).

Other-species effect

• Laboratory monkeys daily exposed to human faces presented novelty preference for monkey faces but not for human faces (Pascalis and Bachevalier, 1998)

• In some studies, monkeys (Martin-Malivel et al., 2001) and sheep (Peirce et al., 2001) did not use internal features to discriminate pictures of human faces.

• In contrast, sheep used internal features to discriminate sheep faces (Peirce, Leigh, & Kendrick, 2000). A right hemisphere advantage was found in sheep with sheep faces but not with human faces (Peirce et al., 2000, 2001),

Categorical Perception

Theoretical boundary

100ch1/0ch2 80ch1/20ch2 60ch1/40ch2 40ch1/60ch2 20ch1/80ch2 0ch1/100ch2

|_____P1____| |_____P2____| |_____P3____| |_____P4_____| |_____P5_____|

Ch1 Ch2

Hu1 Hu2

100hu1/0hu2 80hu1/20hu2 60hu1/40hu2 40hu1/60hu2 20hu1/80hu2 0hu1/100hu2

|_____P1____| |_____P2____| |_____P3_____| |_____P4_____| |_____P5_____|

Categorical perception

Cross-categories Within-categoryWithin-category

12 chimpanzee morphs

12 human morphs