麻豆传媒

WHEN I rang chimpanzee laboratories in the US, I heard excited screaming in
the background, but no chimp would take my call. I wanted to ask chimps what
they think about the world, and about each other. I had to speak to a bunch of
humans instead.

So what鈥檚 the story? It depends who you ask. A war has broken out in the
once-cosy world of primate research, and the outcome could leave the
intellectual reputation of our closest relative in tatters.

You might think that 40 years of field research on wild African chimps,
pioneered by Jane Goodall and then pursued by scores of other researchers, would
have settled the matter by now. Wild chimps not only use tools, they use each
other, deploying alliances and tactical deceptions that suggest a sophisticated,
Machiavellian social intelligence. Back in the lab, however, the cognitive
scientists are not convinced. They say that only experiments carried out under
controlled conditions can definitively reveal the workings of a chimp鈥檚 mind.
The battle is over who has devised a convincing IQ test for chimps, and fur is
flying.

Marshalling the forces of one camp is Daniel Povinelli of the University of
Louisiana at Lafayette. Once, Povinelli believed that chimps were much like
humans in their social understanding鈥攂ut not any more. He has been an arch
sceptic ever since the chimps he trained to beg for food failed to notice that
it was no use begging from someone with a bag over his head. In his new book,
Folk Physics for Apes (Oxford University Press), Povinelli concludes
that chimps are remarkably stupid about the physical world too. They certainly
made a mess of the exams that Povinelli set them.

But chimps are getting the chance of a resit. Not everyone thinks Povinelli
asked the right chimps the right questions. 鈥淚 could make him look
stupid in a novel situation,鈥� mutters one critic.

On the face of it, Povinelli鈥檚 results seem pretty damning. The chimps鈥�
failings hit home when, to put his primate research into perspective, he started
working with children. It took his laboratory chimps hundreds of trials to do
something that a human toddler could grasp immediately. Chimps are notoriously
obstreperous鈥斺�漷hey spend a lot of time not paying attention and trying to
get you mad,鈥� says Povinelli鈥攂ut could they be thick as well?

The chimps鈥� failure to understand the physical world is epitomised by the
hook-and-post test. Povinelli first trained nine-year-old chimps to use hooked
tools to reach through holes in a Plexiglas partition and retrieve a banana. The
banana was just out of reach, resting on a piece of wood or 鈥減latform鈥� that had
a vertical post sticking up at one end and a handy ring at the other. The chimps
were rewarded if they hooked the ring, and pulled the banana into reach. But
what happens if the piece of wood no longer has a ring? Will the chimps
understand that they can still retrieve the banana if they hook the post? To
make it easier for them, Povinelli offered them a choice between two platforms:
one with the hook already curled around the post, or one with the hook uselessly
resting on the platform. The chimps failed miserably. 鈥淭he chimp sees two
examples of contact, doesn鈥檛 notice the difference, and grabs either one,鈥� says
Povinelli. 鈥淏ut children understand instantly.鈥�

Results such as these suggest that chimps don鈥檛 understand 鈥渃onnection鈥� the
way we do, says Povinelli. For them, mere contact is enough鈥攖hey seem to
lack any understanding of underlying mechanical causes. Deceived by superficial
perceptions, they don鈥檛 notice the vital difference between, say, a rope tied to
a banana and a rope that is simply touching it. Chimps may get there in the end,
but only through trial and error. They have no notion of why the world works the
way it does, according to Povinelli. They never conceptualise what they observe,
he says.

So are we chumps to think that chimps are clever? 鈥淧ovinelli does show that
kids can do things chimps can鈥檛,鈥� says Robin Dunbar from Liverpool University.
鈥淏ut he鈥檚 gone on to conclude that the chimps are showing no evidence at all of
advanced cognition. He鈥檚 not got the whole story.鈥� Other primatologists point
out that negative results are notoriously difficult to interpret. 鈥淲ith the
evidence he has, I can see why he is making those conclusions,鈥� says Josep Call
of the Max Planck Institute for Evolutionary Anthropology in Leipzig, 鈥渂ut
different methods could change the picture completely.鈥� That鈥檚 a view shared by
Marc Hauser of Harvard University, whose own findings suggest that mere monkeys
are doing things that Povinelli鈥檚 chimps fail to do.

Hauser offered tamarin monkeys a choice of two blue canes with which to
retrieve a food reward from the other side of a partition. Only one of the canes
was positioned so that it could hook the food, and the monkeys sensibly chose
that one. Hauser went on to show that the tamarins were paying attention to the
relevant, functional aspect of the tool: despite their earlier experience, they
ignored blue canes wrongly positioned in favour of yellow ones that hooked the
food, and ignored ropes laid out in the shape of a cane. If monkeys can
understand what makes a good tool, surely chimps can?

Rather than draw a big line between the cognitive capabilities of humans and
non-humans, Dunbar is inclined to a more gradualist view. 鈥淭he real difference
may be the size of the computer you can bring to bear on social and
environmental problem-solving,鈥� he says. 鈥淭he bigger your mental computer, the
more able you are to link events in causal chains or to mentally rehearse a
solution, and chimps are significantly better endowed than monkeys are. Let
chimps look briefly at a puzzle box one day, and they solve the puzzle more
quickly the next day. They鈥檙e mulling it over鈥攕omething鈥檚 been going on in
the interim.鈥� Dunbar, like most people who work with apes, suspects that chimps
understand the physical world a lot better than Povinelli鈥檚 findings might
suggest.

Illogical like us

So what about their social intelligence? Povinelli鈥檚 chimps don鈥檛 even seem
to understand basic concepts such as the importance of eye contact. They
continued to beg for food from researchers who had their backs turned, were
blindfolded, had bags over their heads, or whose eyes were closed. He concludes
that chimps can have little understanding of another鈥檚 point of view if they
behave like this. But we humans also do illogical things, such as gesticulating
for emphasis when speaking on the phone and shouting at our computers. Might
chimps be far more human-like than Povinelli鈥檚 experiments suggest? Most
researchers agree with Call when he says the answer is 鈥渟till up in the air鈥�.
Even Andrew Whiten of St Andrews University, a pioneer of the idea of
Machiavellian intelligence
(麻豆传媒, 14 February 1998, p 22),
admits that 鈥渢here鈥檚 all still to play for鈥� when it comes to establishing
chimps鈥� true social understanding.

Povinelli may conclude that chimps are as thick as two short planks, but
researchers like Brian Hare from Harvard University reckon we have simply failed
to give them the best opportunity to demonstrate their abilities. 鈥淚t is easy to
demonstrate that an animal is stupid,鈥� he says. 鈥淭he trick is to find situations
in which they can demonstrate what they can do, not what they can鈥檛.鈥� According
to Hare, the secret is to remember just one thing about chimps: 鈥淭hey are
super-super-super competitive, especially when it comes to monopolising food.鈥�
So chances are that the chimpanzee鈥檚 most sophisticated cognitive abilities will
have evolved to outcompete other chimps.

Working with Mike Tomasello and Call at the Max Planck Institute, Hare has
devised a testing situation that makes instant sense to the chimps. The team pit
two chimps, one subordinate and one dominant, against each other in a contest to
win a morsel of food. In a row of three cages, the competitors are in the end
cages. Food is placed in the middle cage. If the subordinate can get to it
first, he can keep it鈥攊t鈥檚 a law of primate society.

One ingenious set of experiments, published earlier this year by the trio,
and Bryan Agnetta of Emory University in Atlanta, Georgia, suggests that chimps
are indeed capable of understanding a competitor鈥檚 point of view (Animal
Behaviour, vol 59, p 771). Two bananas were placed in the middle cage, but
a barrier hid one from the dominant chimp. The subordinate could see both. What
did he do? He preferentially targeted the fruit hidden from the dominant even
though both bananas looked the same from where he was standing. 鈥淭he experiment
demonstrates that chimps can take the visual perspective of another, and are
aware of what another can and can鈥檛 see,鈥� says Hare.

In experiments soon to be published, the team used two barriers and one piece
of food. Both competitors could see the food being placed behind one barrier,
then the door to the dominant鈥檚 cage was closed, and only the subordinate saw
the food being moved behind the other barrier. In the control situation,
however, both animals saw the switch. By and large, the subordinate only ran in
and got the food if the dominant animal had not seen the switch take place. And
in a third experiment, the researchers sometimes substituted the dominant
individual with another dominant chimp after food had been placed in the middle
cage. When the doors opened again the subordinate could see a new competitor who
knew nothing about the food鈥檚 whereabouts. What would the subordinate do? This
time, he made a dash for the food.

鈥淲e want to be very cautious,鈥� says Call, 鈥渂ut you could say that this is
evidence of understanding 鈥榢nowing鈥� in others鈥攊t is the first hint that
the chimp is doing something more than just 鈥榮eeing鈥�.鈥� Hare, on the other hand,
points to these findings as evidence of that holy grail of animal cognition
research: the attribution of 鈥渇alse belief鈥� to another individual. The
subordinate who dashes into the middle cage on seeing a 鈥渘aive鈥� dominant, or
after a food swap that only he has seen, could plausibly be said to think that
the dominant has a 鈥渇alse belief鈥� about the food鈥檚 whereabouts. The attribution
of false belief to another鈥斺�滻 believe that you think something鈥檚 the
case鈥濃�攈as become a widely accepted benchmark of a 鈥渢heory of mind鈥�. Most
children acquire this ability, also known as second-order intentionality, at
about four or five (see 鈥淣atural born mind-readers鈥�).

Given these findings, what should we make of Povinelli鈥檚 results with the
begging chimps? 鈥淗is experiments are not necessarily in total contradiction to
ours,鈥� says Call. 鈥淗e looked at whether chimps understood that eyes are what
allow us to see. That鈥檚 interesting, but perhaps comparable to asking someone in
the street how our visual system works. An incorrect answer doesn鈥檛 necessarily
mean that someone doesn鈥檛 know about the experience of seeing.鈥� Hare goes
further: he suspects that were Povinelli to make the begging experiments
competitive, the chimps would get it right.

An animal鈥檚 personal history can also profoundly shape its cognitive
performance, according to Sally Boysen of Ohio State University. Some of the
chimps in Povinelli鈥檚 experiments came from commercial chimp labs and were
raised in fairly austere conditions, she says. 鈥淐himps that have grown up in
close proximity to people鈥攃himps that know how toilets flush and have
trusting relationships with human caregivers鈥攁re very different beings,鈥�
Boysen asserts. 鈥淭here is an exponential change in their capacity to process
information and take on problem-solving.鈥�

Being a pal

Boysen has 11 such chimps at the OSU Chimpanzee Center in Columbus. She
describes an experiment that indicates they are able to see a situation from
another鈥檚 perspective. The subjects were three pairs of chimps sharing a strong
social bond, including Kermit and Darrell, 22-year-old males who have always
been together and are good friends. The experiment entailed transferring the
chimp pairs to a new cage, where sometimes danger lurked in the person of Boysen
armed with a tranquillising dart gun. If only one friend, Darrell for example,
was aware of the threat, all hell broke lose. His reaction鈥攈air erect,
fear grimace and strangled screams鈥攃aused Kermit to stop dead in his
tracks. But if both were aware of the threat, neither bothered to put on such a
display.

鈥淚 don鈥檛 think you can explain Darrell鈥檚 behaviour without some inferential
understanding along the lines of 鈥業 know but he doesn鈥檛鈥�,鈥� says Boysen. But she
argues that the animals鈥� deep friendship makes all the difference. 鈥淚f you
repeated the experiment with other chimps, I suspect you could get a graded
response depending on the depth of the social relationships between the pairs.鈥�
In the wild, friendships and alliances are extremely important to these animals,
so it makes sense to suppose that in such situations chimps might be on their
mettle. Great ape brains have evolved to cope with the demands of complex,
shifting social groups, says Dunbar.

So what should we make of the fact that chimps seem to be 鈥渁 strange mixture
of intelligence and stupidity鈥�, as Whiten puts it? 鈥淚鈥檝e seen chimps be
incredibly stupid about physical objects,鈥� says Hare. 鈥淭hey might bring you a
ball, a foot in diameter, and try to stick it through a hole an inch square.鈥�
But, he adds, asking why they have such limitations is the wrong question. 鈥淚t
would be better if humans had wings and we could fly to work, but we
don鈥檛鈥攚e just need to keep up with the Joneses. The cognitive abilities
[chimps] have do what they need to do.鈥�

鈥淲e鈥檝e barely scratched the surface in understanding this species that has
shared our pathway for so long,鈥� says Boysen. Povinelli predicts that chimps
鈥渁re going to turn out to be extraordinarily different from us鈥�. Perhaps no one
will disagree when he adds: 鈥淐himpanzees are going to be chimpanzees.鈥�

Without language, chimps have little chance of developing an understanding of
how another individual thinks, suspects Celia Heyes, a comparative psychologist
at University College London. She suggests that we learn our theory of
mind鈥攁ll those beliefs and desires we spend so much time talking
about鈥攊n the same way that we learn language.

It could be rather like our theory of gravitation, she argues. Five hundred
years ago the theory didn鈥檛 exist鈥攊t is a recent product of our culture.
Perhaps the theory of mind itself was devised thousands of years ago, as early
humans tried out different ways of predicting what people will do. 鈥淧erhaps,
when trying to predict what people do, they hit on this idea: maybe there are
ghostly inner states and forces鈥攖houghts, beliefs and desires鈥攖hat
drive behaviour.鈥�

Evidence in support of Heyes鈥檚 delightfully contentious idea comes from
Australian research on deaf children born to hearing parents. Because the
parents themselves are not adept at sign language, they can鈥檛 use it to
communicate ideas about unobservables鈥攅verything from heaven, ghosts and
fairies to mental states. The researchers found that deaf children brought up in
this situation were slower to develop a theory of mind.

Chimps may understand about false beliefs in others, but that doesn鈥檛 put
them in the same league as humans. Most five-year-olds are capable of this
鈥渟econd-order intentionality鈥�, and by the age of 13 or 14 we reach the normal
adult upper limit of fourth-order intentionality鈥斺�滻 believe that you think
that I want you to understand that something鈥檚 the case.鈥�

Robin Dunbar and his colleagues at the University of Liverpool know this
because they have been asking people to listen to 200-word stories incorporating
tangled layers of attribution. Up to fourth-order, error rates on recall are
low鈥攁bout 10 per cent. But error rates rocket to more than 60 per cent at
fifth-order. Most of us can鈥檛 hold all that social reflexivity in our minds at
once. It鈥檚 not a memory problem, says Dunbar, because we have no trouble with
equally complex causal chains of physical events. The difficulty seems to lie in
processing information about mental states.