Animals in Translation (37 page)

The big problem is fence busting. Every year I get about twenty calls from lawyers about cattle getting loose on the highway and getting hit by a car. The drivers always want to sue the ranchers for inadequate fencing. I have to explain to the lawyers that there is no pasture fence on the market that can keep cattle inside a pasture once the cattle have learned how to get through it. Only a steel stockyard fence is strong enough to hold cattle in physically, and steel fences are too expensive to put up around grazing lands. The fences ranchers use keep cattle in only because the cattle don't realize that they have the power to break through them.

Is fence busting true cognition? Sometimes it is and sometimes it isn't. Usually cattle discover how to break through a fence by accident. Cattle will push on a fence to reach greener grass on the other side, and keep pushing until one day the fence falls over. Then they draw the appropriate conclusion: if I push on the fence, I can get out and go eat where I want. Animals also figure out, probably by accident, that if they run through an electric fence it's going to hurt for only a few seconds. We know this because pigs who have learned to go through electric fences often squeal
before
they hit the wire. They know what's coming.

Some cattle have learned to break through a fence through simple
trial and error, but others have started to build on what they learned by accident. There was one bull from the Arizona high country who was the champion fence buster. Bulls are the worst fence busters, and once a bull has learned to break through a fence it's difficult to keep him in. This particular bull was the champion; he took out fences faster than the U.S. Forest Service could build them. He knew how to knock over a high-quality four-strand barbed wire fence built to government standards. In one afternoon he walked through four brand-new fences. I saw him after he had been locked in a stall corral that was too strong for him to break out of.

All of us were amazed that the bull could tear out so many barbed wire fences without getting cut. His tan-and-white hide did not have a single scratch. This is where cognition is at work. He had figured out how to knock over a barbed wire fence without getting cut. Nobody ever saw him do it but he must have figured out that if he pushed over the posts with his head first, and then walked through, he would not get cut. He was careful.

Holstein steers are another story. With all the licking and tongue manipulating they do, Holsteins end up opening gate latches beef cattle never even try to open. I don't think they're really solving a problem, though; it's more like a happy accident. What starts out as a pure desire to lick and tongue things turns into the discovery that they can open gates. Still, once they figure it out they're experts. They can open just about anything, including every sliding bolt gate latch on the market. The only kind of latch that can keep a Holstein cow inside a pen is a chain hooked together with a dog leash snap. They love to get out, too. At one feedlot a group of Holsteins escaped their pen and walked up to the office to lick the windows and remove the paint from the manager's pickup.

A
RE
A
NIMALS
A
S
S
MART
A
S
P
EOPLE
?

I can't answer that question, and neither can anyone else. Researchers who believe we know for a fact that man is the crown of creation when it comes to IQ are off base. That's what researchers
think,
not what they know. I've come to the conclusion that although in many ways other mammals are similar to us, in other ways they may be to
tally alien. A lot of our tests and experiments with animals probably aren't telling us what we think they're telling us.

Dr. Pepperberg's breakthrough with Alex ought to make researchers think twice. It's not just that what we know keeps changing, but that the way we go about
finding out
how animals think sometimes changes, too. That's the moral of Dr. Pepperberg's story. The reason she finally succeeded where everyone else had failed was that she was the first person to consider that maybe it was the researchers' fault birds weren't learning anything, not the birds'.

All of the parrot studies up to then had used a classical
operant conditioning
format. Operant conditioning, also called
instrumental conditioning
or
stimulus-response teaching,
is when an animal learns
to do something
in order to get what he wants. A rat who's learned to press a lever to get food pellets has had operant conditioning. Using operant conditioning the experimenter would show the bird a red triangle and a blue triangle and say “touch blue,” then reward him with a piece of food whenever he happened to peck at the blue triangle by chance. If he happened to peck the red one he didn't get the food. After a while he was supposed to learn blue, because he had been rewarded for pecking the blue triangle every time he heard “touch blue.” That's classic behaviorism.

The problem was, no bird ever learned blue. They didn't learn red, either. They didn't learn anything, really. Apes weren't learning too much in those setups, either, but no one wanted to hear about it, because everyone thought it was much more scientific to do a stimulus-response experiment in the lab than to watch an animal learn things naturally in his normal habitat. When a few researchers began teaching apes in more naturalistic settings they were criticized for being unscientific and performing
uncontrolled
experiments. In science, there's nothing worse than an experiment that's uncontrolled.
¹⁰

Dr. Pepperberg decided to give up on operant conditioning and try a different branch of behaviorism called
social modeling theory.
Albert Bandura developed social modeling theory at Stanford University in the 1970s, based on how he thought real people and real animals probably learned in the real world.
¹¹
For years behaviorists had assumed that animals and people learn everything they know through either
operant
or
classical conditioning.
(
Classical condition
ing
works with innate, reflexive responses like eye blinks and salivation. Pavlov's dog learning to salivate at the sound of a tone is classical conditioning.)

But Dr. Bandura pointed out that the stimulus-response learning animals did in labs was just learning by trial and error. The animal does more of whatever behaviors he gets rewarded for doing, and less of whatever behaviors he's been punished or negatively reinforced for doing.

That sounds like a logical way to learn until you think what it would mean in the wild. In the real world, trial and error learning would get a lot of animals killed. If the only way a baby antelope could learn to run away from a lion was by finding out what happens if you
don't
run away from a lion, there wouldn't be any baby antelope left. Pretty soon there wouldn't be any lions left, either, since they wouldn't have baby antelope to eat.

In Dr. Bandura's view, animals and people had to do a huge amount of observational learning. He thought that a baby antelope would learn to run away from lions by watching other antelope run away from lions and doing the same thing. Today we know Dr. Bandura was right, partly thanks to Susan Mineka's research on monkeys and snakes.

Dr. Bandura had obviously hit on something with social modeling theory, but it didn't occur to anyone to try using it in their research on animal learning. That was Dr. Pepperberg's innovation. She set up a social modeling situation for Alex. Instead of teaching Alex one-on-one she taught him two-on-one, two people to one bird. And instead of teaching Alex directly, she
taught the other person,
while Alex sat on his perch and watched. No one had ever done that before.

She also used items a parrot really, really wants, like a nice, crunchy piece of bark, for her learning materials. Animals and people both pay more attention to things that are important to them, like food, and you have to pay attention to learn. A parrot in the wild doesn't care about blue triangles, so why should he care about blue triangles in the lab? He doesn't.

So if Dr. Pepperberg wanted Alex to learn the color blue, she took a nice, crunchy piece of bark and painted it blue. Then she'd sit down with Alex and her research assistant and ask the assistant, “What color?”

If the assistant got the answer right, he got to play with the bark.
If the assistant got the answer wrong, he didn't get to play with the bark. All Alex got to do was watch. Dr. Pepperberg called her technique
model/rival,
because the assistant was a
model
for Alex to copy and also a
rival
for whatever item Dr. Pepperberg was using in her lesson. She set up a competition for scarce resources between Alex and the assistant.

Using modeling theory was the breakthrough. Alex learned so much that he started asking questions on his own! One day he looked at his reflection in the mirror and asked Dr. Pepperberg, “What color?”

After he'd asked about his own color six different times, and heard answers like “That's gray; you're a gray parrot” six different times, he knew gray as a category. From then on he could tell his trainer whether or not any object she showed him was gray.

This is nothing short of miraculous as far as I'm concerned. Alex was never taught to ask questions; he just did so on his own, spontaneously. That's incredible, because question asking seems to be a separate skill from making statements, judging by the language of autistic children. Autistic children who can talk rarely ask questions; some of them never do. I know a mom whose sixteen-year-old has been talking since the age of two, and she says to this day she can count on one hand the number of questions he has asked.

Question asking is so important that Bob and Lynn Koegel, of the Autism Research and Training Center at the University of California, Santa Barbara, made major breakthroughs in their autism clinic when they started teaching autistic children to ask questions.
¹²
I wonder whether we would have major breakthroughs in language comprehension with apes and dolphins if we taught them to ask questions, instead of just having them answer questions all the time.

L
EARNING
T
HAT'S
E
ASY FOR
P
EOPLE
, H
ARD FOR
A
NIMALS

Most birds and animals are almost certainly smarter than we know, but that doesn't mean they don't have some limitations that humans don't. (Humans have limitations animals don't, too. I'll get to that in the next chapter.)

I've said several times now that one of the major differences between people and our fellow mammals is that we have larger, better-developed frontal lobes. One of the benefits of having bigger frontal lobes is that we have more working memory. Since working memory is an important factor in general intelligence, if animals have less working memory overall, that's going to make a difference in their general cognitive abilities.

The question is, what differences are you going to see in a person or animal with lots of working memory versus a person or animal with a lot less working memory? I think my own brain is a good place to start, since I have terrible working memory. If I were a computer I would have a huge hard drive memory and a very small microprocessor. As a result, I have a hard time doing things that involve multitasking, like trying to make change and talk at the same time. Another problem area for me: mental arithmetic. I can't hold one number in memory while I manipulate another. For me to try to add up two two-digit numbers inside my head would be a stretch, and I couldn't even begin to add two three-digit numbers together without writing them down where I can see them.

Since we never ask animals to multitask or add numbers in their heads, one of the main places you can see this difference is in situations that require an animal to be good at
sequencing.
(I'm talking about primates and domestic animals, not birds and sea mammals like dolphins. Birds and dolphins have different brain structures from ours, and I don't know enough about their sequencing abilities to comment.) Animals are not good at sequencing. A good example is dogs getting tangled up in leashes or tie-outs. Owners are always amazed at how helpless a dog is once he's gotten his tie-out wrapped around a tree.

A big part of the problem is that he can't remember the sequence of events that got him to where he is, so he can't retrace his steps. He has the same problem if he just tries to start fresh and figure it out. If one move doesn't work he has to be able to hold that failure in mind while testing other moves. A dog probably doesn't have enough working memory to do that. He's like a person who gets mixed up driving unfamiliar streets after dark. A normal person with an excellent working memory can end up going around in circles in that situation, because
he's hit the limits of his working memory. He can't hold all of the different routes he's tried in working memory while he tries new ones, so he keeps going over the same route all over again without realizing it until he ends up back where he started.

Dogs can learn sequences, like the ones working dogs perform at show, with a lot of direct training. However, I think it's probably as hard for a dog to learn show sequences as it was for me to learn the sequence of events that take place in a large meatpacking plant. When I first went into a big plant the place looked so complicated I was amazed the managers were able to keep track of all the complex procedures. I didn't know how anyone could understand and remember anything so intricate.

In the early 1970s I visited a big meatpacking plant every Tuesday afternoon for three years. I used to stand for hours on a catwalk overlooking the floor where the carcasses were processed and dressed by about a hundred employees altogether. The place was a mass of visual details, and every Tuesday afternoon I downloaded more details into my brain.