Animals in Translation

by Temple Grandin

  I believe the rats did better than the humans either because of weaker frontal lobes or because rats don’t have language or both. One thing we do know about humans is that the left brain, which is the conscious language part of the brain, always makes up a story to explain what’s going on. Normal people have an interpreter in their left brain that takes all the random, contradictory details of whatever they’re doing or remembering at the moment, and smoothes everything out into one coherent story. If there are details that don’t fit, a lot of times they get edited out or revised. Some left brain stories can be so far off from reality that they sound like confabulations.

  The interpreter probably got in the way on the lever-pressing experiment. The human subjects kept trying to come up with a story about the dots, and when they did come up with a story they stuck to it. Then the dot story kept them from realizing they should just forget about the dots and press the lever every time the screen changed.

  ANIMAL WELFARE: TAKING CARE OF ANIMALS THE WRONG WAY

  Working in animal welfare, I constantly have to reason with normal humans who are too smart for their own good.

  My most important contribution to the field has been to take the idea behind Hazard Analysis Critical Control Point analysis, or HACCP (pronounced hassip), and apply it to the field of animal welfare. The animal welfare audit I created for U.S. Department of Agriculture is a HACCP-type audit.

  My HACCP system works by analyzing the critical control points in a farm animal’s well-being. I define a critical control point as a single measurable element that covers a multitude of sins. For instance, when I’m auditing the animals on a farm, one thing I want to know is whether the animals’ legs are sound. There are a lot of things that can affect a cow’s ability to walk: bad genes, poor flooring, too much grain in the feed, foot rot, poor hoof care, and rough treatment of the animals. Some regulators will try to measure all of these things, because they think a good audit is a thorough audit.

  But that’s not my approach. I measure one thing only: how many cattle are limping? That’s all I need to know, just how many cattle are limping. That one measurement covers the multitude of sins that can cause cattle to go lame. If too many animals are limping, the farm fails the audit and that’s it. The only way the farm can pass the next audit is to fix whatever it is that’s making their animals lame. If management knows what the problem is, they can get busy fixing it. If they don’t know what the problem is, they have to hire someone who can tell them, and then fix it.

  For my animal welfare audit, I came up with five key measurements inspectors need to take to ensure animals receive humane treatment at a meatpacking plant:

  Percentage of animals stunned, or killed, correctly on the first attempt (this has to be at least 95 percent of the animals).

  Percentage of animals who remain unconscious after stunning (this must be 100 percent).

  Percentage of animals who vocalize (squeal, bellow, or moo, meaning “ouch!” or “you’re scaring me!”) during handling and stunning. Handling includes walking through the alleys and being held in the restraining device for stunning (no more than 3 cattle out of 100).

  Percentage of animals who fall down (animals are terrified of falling down, and this should be no more than 1 out of 100, which is still more than would fall down under good conditions, since animals never fall down if the floor is sound and dry).

  Electric prod usage (no more than 25 percent of the animals).

  I also have a list of five acts of abuse that are an automatic failure:

  Dragging a live animal with a chain.

  Running cattle on top of each other on purpose.

  Sticking prods and other objects into sensitive parts of animals.

  Slamming gates on animals on purpose.

  Losing control and beating an animal.

  This is all you need to know to rate animal welfare at a meatpacking plant. Just these ten details. You don’t need to know if the floor is slippery, something regulators always want to measure. For some reason whenever you start talking about auditing the plants everybody turns into an expert on flooring. I don’t need to know anything about the flooring. I just need to know if any of the cattle fell down. If cattle are falling down, there’s a problem with the floor, and the plant fails the audit. It’s that simple.

  The plants love it, because they can do it. The audit is totally based on things an auditor can directly observe that have objective outcomes. A steer either moos during handling or he does not.

  Another important feature of my audit: people can remember two sets of five items. That level of detail is what normal working memory is built to hold on to.21

  But I find that people in academia and often in government just don’t get it. Most language-based thinkers find it difficult to believe that such a simple audit really works. They’re like the people in the lever-pressing experiments; they think simple means wrong. They don’t see that each one of the five critical control points measures anywhere from three to ten others that all result in the same bad outcome for the animals.

  When highly verbal people get control of the audit process, they tend to make five critical mistakes:

  They write verbal auditing standards that are too subjective and vague, with requirements like “minimal use of electric prod” and “non-slip flooring.” Individual inspectors have to figure out for themselves what “minimal use” means. A good audit checklist has objective standards that anyone can see have or have not been met.

  For some reason, highly verbal people have a tendency to measure inputs, such as maintenance schedules, employee training records, and equipment design problems, instead of outputs, which is how the animals are actually doing. A good animal welfare audit has to measure the animals, not the plant.

  Highly verbal people almost always want to make the audit way too complicated. A 100-item checklist doesn’t work nearly as well as a 10-item checklist, and I can prove it.

  Verbal people drift into paper audits, in which they audit a plant’s records instead of its animals. A good animal welfare audit has to audit the animals, not the paper and not the plant.

  Verbal people tend to lose sight of what’s important and end up treating small problems the same way they treat big problems.

  All five of these mistakes hurt the animals. When you make the audit process more complicated, the auditors veer off into all the fine detail that goes into making a humane slaughterhouse, which leads to wanting to micromanage the plants. Instead of looking at outcomes to the animals, they want to tell the plant how to build its floors. Then they want to send auditors out to inspect the construction to make sure the floors are right. The animal gets lost in the confusion. I don’t care about floors. I care about cows. Are they falling down? That’s all I need to know.

  The other thing that happens is that auditors lose track of what’s important. If you give an auditor a 100-item checklist, he’ll tend to treat 50 of the items as if they’re major, whereas maybe only 10 items are so critical that if the plant fails any one of those 10 it should fail the audit, period. When a plant fails 1 critical item out of 10, it’s easy to fail the whole plant. But when it fails that same item on a list of 100, it doesn’t look so bad.

  Even worse, an auditor working with a long, overly complicated checklist can miss the huge problems completely, even though they’re on the list. A friend of mine told me a horrible story about a plant where the stunning equipment wasn’t working right, and they had live animals hanging from hooks going down the slaughter line. The USDA inspector missed it. He got focused on some worker who was whacking the pigs too hard on the butt, and he wrote them up for that. Meanwhile the plant had a hideous, enormous problem of live animals on the slaughter line that ought to mean an automatic fail. The inspector didn’t see it, or maybe he did see it but it didn’t register on him.

  I think this kind of blindness must have to do with the limits on normal human perception. Somehow, when an inspector has to audit 100 different aspects
of a plant’s functioning, he stops seeing the lady in the gorilla suit. I’m not saying it’s okay to be whacking the pigs, of course. It’s not, and it should be corrected. But when the audit checklist is too long, auditors start hyper-focusing on small details and missing the great big details that matter the most.

  I’ve seen this happen many times. About a year ago I visited plants in Europe, where the plants and the inspectors were supposed to continuously monitor and improve 100 different items on a checklist. The plants were horrible.

  Sometimes the standards that verbal thinkers want to include aren’t even connected to reality. For instance, I’ve been working with KFC—Kentucky Fried Chicken—to raise standards for animal welfare in the poultry industry, and one of the standards an abstract, verbal thinker will want to put on the audit form is that the lights have to be off for at least four hours every night. Well, how am I going to get out to the farm at 3:00 A.M. to make sure the lights are off? I’m not. And I’m not going to trust the paperwork.

  What I need to audit isn’t the lights, it’s the outcome of turning off the lights to the chickens’ welfare. The lights have to be off because darkness slows down a baby chick’s growth. Today’s poultry chicken has been bred to grow so rapidly that its legs can collapse under the weight of its ballooning body. It’s awful. Darkness slows down the baby chick’s growth just enough to prevent this from happening, so getting those lights off is important, because lameness is a severe problem in chicken welfare. I’ve been to farms where half of the chickens are lame. When I audit a chicken farm, what I want to know is, can the chickens on this farm walk? If the chickens are lame, something is wrong, and the farm fails the audit.

  And I strongly object to paper audits, because anyone can change his paperwork if he wants to. However, a plant can’t falsify things I can directly observe. I don’t want to see the maintenance records on the stunner. If the stunner is well maintained, it’s going to work. That’s all I need to know. I’ve measured broken wings on chickens. I want to see the animals.

  The other dangerous thing about paper audits and 100-item checklists is that they can set you up for a situation of things slowly getting worse without anyone knowing it. When you drift away from the animals themselves and start auditing the paperwork, the bad can become normal pretty quickly.

  I want to stress this point. Maintaining animal welfare standards in a meatpacking plant is an ongoing responsibility. The whole principle of HACCP is that you have to keep measuring standards and compliance or everything goes bad on you. It’s kind of like maintaining your weight: you have to keep on top of it. Paper audits end up masking small, incremental declines in standards that result in very large drops in animal welfare.

  Unfortunately, to an abstract verbal thinker, a list with 100 different animal welfare items sounds more caring than a list with only 5. But I can prove beyond question that animals in plants undergoing 10-question audits are handled much more humanely than animals in plants undergoing 100-question audits. And it’s not just that plants using my checklist do well on the big details. They also do better on the smaller details, because the smaller details are part of the big ones.

  Even though my list contains only five critical control points, it is so strict that most plants thought they wouldn’t be able to pass. But then McDonald’s started auditing the plants. In 1999 they threw a major plant off the approved supplier list for flunking the audit, and they suspended some other plants. After that the industry got religion, and boy has the cattle handling changed. Let me tell you, you go out there now and they’re handling the cattle nice. All of the plants being audited using my list treat their animals better than plants using 100-item checklists.

  Most large plants are now audited by restaurant chains like McDonald’s, Burger King, and Wendy’s International. Just four years after McDonald’s began requiring its suppliers to audit their plants according to my standards, almost every plant is passing easily. Now when you go into a plant it’s like a magical change. I think of all the years up to 1999 as the pre-McDonald’s era and the years since then as the post-McDonald’s era. Up until 1999 the plants might buy the best equipment, but they didn’t manage it. They’d let stuff break, and they didn’t spend enough time and money training and supervising their staff or firing people who needed firing. Then as soon as McDonald’s started auditing, they were hitting up my Web page to learn the stuff they had to do. There were light-years of change.

  For the twenty-five years up to 1999 I’d been putting equipment into plants. Some of the plants used it right, but others just tore it up and ruined it. Now my equipment is perfectly maintained and there’s nothing broken on it anywhere. For the first twenty-five years of my career I was a hardware engineer; now finally I’m installing the management software. Training those auditors: that’s the software installation for the hardware I put into half the plants in North America.

  My simple five-point checklist works beautifully. But even though it works, and even though I can show that animals being audited by 100-point checklists are being handled poorly, I have to fight constantly to keep it in place.

  DO ANIMALS TALK TO EACH OTHER THE WAY PEOPLE DO?

  Those are fighting words in the fields of animal and linguistic research. A lot of people are emotionally invested in the idea that language is the one thing that makes human beings unique. Language is sacrosanct. It’s the last boundary standing between man and beast.

  Now even this final boundary is being challenged. Con Slobodchikoff at Northern Arizona University has done some of the most amazing studies in animal communication and cognition.22 Using sonograms to analyze the distress calls of Gunnison’s prairie dog, one of five species of prairie dogs found in the U.S. and Mexico, he has found that prairie dog colonies have a communication system that includes nouns, verbs, and adjectives. They can tell one another what kind of predator is approaching—man, hawk, coyote, dog (noun)—and they can tell each other how fast it’s moving (verb). They can say whether a human is carrying a gun or not.

  They can also identify individual coyotes and tell one another which one is coming. They can tell the other prairie dogs that the approaching coyote is the one who likes to walk straight through the colony and then suddenly lunge at a prairie dog who’s gotten too far away from the entrance to his burrow, or the one who likes to lie patiently by the side of a hole for an hour and wait for his dinner to appear. If the prairie dogs are signaling the approach of a person, they can tell one another something about what color clothing the person is wearing, as well as something about his size and shape (adjectives). They also have a lot of other calls that have not been deciphered.

  Dr. Slobodchikoff was able to interpret the calls by videotaping everything, analyzing the sound spectrum, and then watching the video to see what the prairie dog making a distress call was reacting to when he made it. He also watched to see how the other prairie dogs responded. That was an important clue, because he found that the prairie dogs reacted differently to different warnings. If the warning was about a hawk making a dive, all the prairie dogs raced to their burrows and vanished down into holes. But if the hawk was circling overhead, the prairie dogs stopped foraging, stood up in an alert posture, and waited to see what happened next. If the call warned about a human, the prairie dogs all ran for their burrows no matter how fast the human was coming.

  Dr. Slobodchikoff also found evidence that prairie dogs aren’t born knowing the calls, the way a baby is born knowing how to cry. They have to learn them. He bases this on the fact that the different prairie dog colonies around Flagstaff all have different dialects. Since genetically these animals are almost identical, Dr. Slobodchikoff argues that genetic differences can’t explain the differences in the calls. That means the calls have been created by the individual colonies and passed on from one generation to the next.

  Is this “real” language? A philosopher of language might say no, but the case against animal language is getting weaker. Different linguists have so
mewhat different definitions of language, but everyone agrees that language has to have meaning, productivity (you can use the same words to make an infinite number of new communications), and displacement (you can use language to talk about things that aren’t present).

  Prairie dogs use their language to refer to real dangers in the real world, so it definitely has meaning.

  Their language probably has productivity, too, since they can apply the same adjectives to different animals. Dr. Slobodchikoff has also done some interesting experiments to see what calls prairie dogs would make to an object they’d never seen before.

  He built three plywood silhouettes, a skunk, a coyote, and a black oval, and dragged them through the prairie dog colony on a pulley. The prairie dogs gave alarm calls to all three objects, and each prairie dog used the same call for the same plywood object. These calls weren’t invented on the spot, either. At least one of the calls—for the plywood coyote—was a variant of an old call Dr. Slobodchikoff had already recorded them using. That’s more evidence the prairie dogs were combining their old “words” to describe something new.

  Another interesting finding: all three plywood objects were new to the prairie dogs, but the prairie dogs used different calls to identify each one. Dr. Slobodchikoff says that means it’s unlikely the prairie dogs were simply using a rote call meaning “something new is coming.” He also says that the prairie dogs seem to be using transformational rules to create their calls. In human language, a transformational rule allows you to turn words into sentences that make sense. The person listening to you uses the same rules to decode what you’re saying. The prairie dogs seem to have a transformational rule based on speed. Depending on how fast a predator is moving, they speed up their calls or slow them down.