Exercises
Pepperberg's
Parrots
by
Mark Caldwell
Alex,
age 23, and Griffin, age 4, are hell on decor. Thanks
to them, the laboratory-cum-home of these impish African Grey
parrots looks as if a miniature tornado has blown through.
Strewn about are pieces of fruit from discarded snacks, mangled
toys, and a huge mound of cardboard cartons demolished with
great relish by Alex.
Animal behaviorist Irene Pepperberg, the birds'
caretaker, maintains an amused tolerance about her two charges,
partly because they have richly rewarded her serious scientific
inquiry into the intelligence of animals. For more than 20
years, her research has implicitly challenged the idea that
humans alone are capable of real thought and real communication.
In a series of experiments that withstand rigorous and even
hostile scrutiny, Alex and Griffin have shown themselves intelligent
enough to comprehend and juggle abstract images of the objects
that make up their world─skills once thought to be the exclusive
property of humans.
Under Peperberg's tutelage, the parrots have
learned to speak English so clearly they'd delight even a
fussy speech expert. But the value of this work goes far beyond
experiments in voice training. These parrots aren't just parroting─they associate specific words with specific objects, and
they have learned to identify a number of different colors,
shapes, and materials. Show Alex two triangles, one yellow
and one blue, and ask him what's the same about them. He'll
answer, "Shape." Ask him what's different, and he'll say, "Color!"
Until Pepperberg began this research in the
1970s, few scientists had studied intelligence in parrots,
and few do today. Most inquiries have instead focused on monkeys,
chimpanzees, gorillas, and dolphins, all of which are much
more difficult to raise, feed, and handle. Pepperberg especially
likes parrots because, like humans, they're smart, long-lived
(often up to 50 years), social animals that depend on communication
for survival. And, best of all, to communicate with people,
parrots don't need devices built with buttons to push, and
they don't need to be taught sign language: They can learn
to speak. Nobody knows why parrots can do this, or exactly
how they do it. But
they can clearly get their two cents in despite having a brain
the size of a walnut.
Even
the most skeptical visitor to Pepperberg's lab is sure to
be taken aback by what the birds can do. When Alex,
for example, wants to visit a favorite atrium near the lab,
he orders, "Go see tree!" And some of his communications are
fresh─new words and phrases he hasn't been taught. Once,
for example, as two of his student trainers prepared to leave
the lab at the end of the day, Alex admonished: "You be good.
See you tomorrow!"
It's not just what the parrots say that makes
them seem eerily human; it's the level of intelligence they
easily demonstrate. Consider Griffin's performance on a test
Pepperberg devised to see whether the birds could use a mirror
image of an object to manipulate it. Children don't typically
master that skill until they're 3 years old. In the experiment,
a nut is concealed underneath the lid of a box. The nut is
attached to a wire that runs up through a slit in the lid
and connects to a paper clip for the parrot to yank. The slit
branches out into three tracks, each of which ends in a hole
through which the nut can
be pulled. The trick is that two of the three slits are blocked
by obstructions that can only be seen by looking in a mirror
that reflects a backward view of what's inside the box. Most
humans who try to solve the puzzle are baffled, but Griffin,
watching intently from his perch on the lab counter, will
demand to be brought over, peer into the mirror for perhaps
half a second, triumphantly zip the nut down the right track,
jerk it up through the opening, and grab it.
Although Irene Pepperberg has always loved animals, she chose
chemistry as her field in graduate school at Harvard University.
There, in the early 1970s, she happened to see a television
documentary about how animals learn. Like a number of researchers
at the time, she became fascinated by the question of just
how smart animals are and whether they can learn to communicate
with humans. While finishing up her chemistry degree, she
began taking courses in animal behavior, psychology, and communication.
In 1977 she acquired 1-year-old Alex. He became the center
of her research in a small lab at Purdue University.
Much of the animal-intelligence research of
that era ultimately foundered before an onslaught of skeptics.
Most notable was Washoe, a chimpanzee who became world-famous
for her apparent mastery of American Sign Language. But observers
of Washoe objected that the researchers who worked with her
were giving subtle cues and generously interpreting ambiguous
gestures as signs for words. "It's so easy to overinterpret
when you have a bird that says, ‘I'll see you tomorrow,'"
says Pepperberg. "I've got enough anecdotal data like that
to fill a book. But what's really referential?" What, in other
words, constitutes proof that the bird knows what it is saying?
The
history of science is littered with animals that seemed to
be displaying extraordinary brainpower but were just responding
to unconscious prompts from owners and trainers.
To avoid such pitfalls, Pepperberg designs
experiments with the care of a hard-nosed skeptic. For example,
when an experiment demands that Alex learn a new word like "none," Pepperberg waits until the parrot's pronunciation
is so unambiguous that different observers agree 90 percent
of the time on what he's saying. That ensures that an experimenter
eager for Alex to succeed won't mishear a slurred random utterance
as a correct answer. She also strives to avoid the possibility
that a correct response is merely conditioned─automatic
behavior generated by the anticipation of a reward. So experimenters
vary their repertoire of questions to make sure that Alex
is responding to the content of the question and making intelligent
discriminations. And to avoid the possibility that the experimenter's
body language might be prompting Alex's answers, students
who test Alex are never the same ones who taught him the words
and concepts involved. As a result, Pepperberg's work has
won accolades for its persuasiveness from the likes of Oxford
University animal behaviorist Marian Stampl Dawkins, an authority
on animal consciousness and a skeptic about many studies in
the field.
Sometimes, human error offers the birds an
opportunity to show how smart they are. Roughly 5 percent
of the time, for example, student questioners slip up and
scold Alex with a "No!" when he has in fact given the correct
answer. When
this occurs, Alex tends to stick to his guns and repeat the
correct answer. Eventually the examiner comes to
her senses, and Alex gets the reward he deserves.
Pepperberg is careful to point out that such
experiments have stimulated Alex and Griffin to master intellectual
tasks both far different from─and possibly harder than─those they might achieve in the wild. And that is perhaps
the most promising aspect
of her research in the long term. "I don't want to say they're
learning English," she says, "because we can't get inside
an animal's brain, and we can't know if its experience of
English is exactly the same as ours. But they have learned
a communication system that's completely alien to them." She
argues that this achievement alone is significant─and not
just because it challenges dogma about animal intelligence.
The way she has trained parrots to master skills outside their
natural repertoire─which she calls "exceptional learning"─also serves as an interesting model for human instruction.
Her training technique is interactive, using
a model/rival mode. And it is simple: Both animals and people
learn more readily when they can observe and even compete
with others learning at the same time. The best results come
if, rather than being harangued by a teacher, the pupil watches
a tutor coaching someone else. The pupil learns by noting
that the teacher rewards the student for correct answers and
gives rebukes for wrong ones. The technique, pioneered by
Dietmar Todt at the University of Freiburg, Germany, in the
1970s, worked quickly when Pepperberg tried it with Alex.
He caught on rapidly as she taught the undergraduate who served
as a model. Alex began competing with the student for Pepperberg's
attention and approval. The model/rival strategy worked so
well for Alex that it has been successfully adapted for helping
developmentally disabled children learn.
By the late 1980s, Alex
had learned the names of more than 50 different objects,
five shapes, and seven colors. He's also learned what "same"
and "different" mean—a
step so crucial in human intellectual development that it
has become a staple of Sesame Street shtick. Show
Alex two objects of different shape, color, and material,
and then ask him, "What's same?" He'll answer, "None!" Thus
he appears to understand not only similarities and differences
but also an even more abstract concept—absence. Alex has
become so skilled that, in one of Pepperberg's projects, he
has been promoted to the position of tutor. "We want to see
if Alex can work as a tutor for Griffin," Pepperberg explains.
"Partly we got Griffin to replicate our work with Alex, to
prove he wasn't some kid of avian Einstein."
To demonstrate, Justin, one of Pepperberg's
undergraduate assistants, instructs Griffin and Alex to hop
up on the work perch. As in many a human classroom, things
get off to a balky start. Griffin seems cranky, insistently
croaking, "Wanna go back." He has spent most of this morning
idly shoving a metal spoon back and forth across the countertop
and wants to return to the lab counter to play. Justin displays
a ring and asks: "What toy, Griffin?" Griffin answers "Wood!"
and Alex unhelpfully pipes up with "Rock! Rock! Rock!" A couple
more trials follow, in which Griffin correctly identifies
cork, wood, and wool (which he pronounces with a sirenlike
flourish :"Wooh-ull!" ). Alex sits taciturnly by. As the experimental
session progresses, both Alex and Griffin seem to slide into
the spirit of things, falling almost naturally into the model/rival
mode. Justin displays a key and asks, "What toy?" Griffin
doesn't answer at first, but when Alex barks, "Key!" Griffin
follows suit. When Griffin doesn't respond to a toy truck,
Alex chimes in with a quite fed-up-sounding "Truck! Truck!"
Pepperberg's instructional techniques have been so successful
that they've now attracted the interest of researchers in
artificial intelligence. She recently worked with computer
scientists at the Massachusetts Institute of Technology to
help develop software that can learn or that can be used to
help others learn. "It's a lot simpler to model a parrot's
learning than a person's, " Pepperberg says. But she believes
that computer software could be written to respond to─maybe
even "learn" from─the interplay of correct answers and mistakes
that makes up the model/rival system.
Whether its promise is for people or computers,
Pepperberg's research is forcing psychologists to rethink
the boundaries between animals and humans. The thought processes
of Alex and Griffin may approximate our own a lot or a little,
says Pepperberg, but it's getting well-nigh impossible to
deny their intelligence.
A recent incident caught on videotape during
a model/rival session illustrates the point: Griffin, while
trying to say "paper," splutters "ay-uhr." Alex, seemingly
pushed to the limits of his patience, peremptorily orders
Griffin to "Talk clearly!"
(1 952 words)
(From Discover, January 2000 )
Text
|