Commands and Signals Revisited:

Ivan Pavlov, the great Russian physiologist, said that in order to survive, animals must have a way to identify beneficial and harmful events before they actually occur. More simply, he said that if a dog had to wait for the claws of a bear to sink into his flesh before he reacted to the threat, there would be no dogs. Likewise, if animals could not learn that thunder in the distance preceded a flash flood or that the snap of a twig preceded a succulent deer, they would be unable to avoid harmful events or take advantage of available entrees. If these examples sound far to simplistic for great science, think again. Sometimes the very best scientific information helps us build secure foundations for more complex knowledge. In the case of Ivan Pavlov, the diligence of his work pays off big-time for anyone interested in the most fundamental of dog training skills.

Pavlov’s particular field of study was the way the brain learns to make associations from initially meaningless things. His most famous experiment showed that you could pair the sound of a bell to the presentation of food. This type of association can become so strong that the mere sound of the bell can cause virtually all of the reflexes that are associated with eating – salivation, reduction in heart rate, blood pressure, respiration and tail wagging. He discovered many facts that relate to how animals learn to pair the most basic associations. For instance, dogs do not automatically know that meat is food. They must have a small bit of it placed on their tongue before they can connect the smell of meat with eating same. Fortunately, in the process of licking their mother’s mouth in greeting, puppies invariably receive a tiny morsel of meat and forever will know that it is edible. While most people mistakenly think Pavlov discovered a scientific way to teach behaviors such as sit, down and come, he actually researched the foundation for all learning – commands and signals.

What We Do and Don’t Do:
If you take a moment to listen and watch how people control their dogs you will find out some pretty interesting things. Despite the fact that dogs don’t possess the brain-power to speak languages, most humans treat them as if they do. If you doubt this, all you must do is watch someone deal with people who speak another language. An American visitor to Mexico wishes to order a glass of water at a restaurant. In many parts of Mexico, American tourists aren’t the major source of income of the community and many locals don’t speak our language. As the tourist attempts to get some water, he will be frustrated that his perfect English is entirely useless with the locals. When his requests draw blank stares, he is likely to do some interesting things with his speech and voice. Here are the most common ways he’ll change his normal speech pattern when he thinks he hasn’t been understood.

  1. Repeat the word “I would like some water (no response)…water.”
  2. Repeat the word slower with overly precise enunciation. “Waah-Terrrr”
  3. Say the word louder. “WAHTER!”
  4. Modify the word to make it sound more like the dialect in that area. “Yo wanto some water-o.”
  5. Repeat the word in a commanding tone to mildly threaten the listener for being so unresponsive.
  6. All of the above in various combinations.

If you are amused at these typical reactions of silly tourists, don’t be surprised, it’s perfectly natural for humans to do these “verbal behaviors” when they don’t get what they expect from a request – regardless of what nationality or species they are dealing with. If this surprises you, it’s simply the way humans respond generally to similar situations. Speaking louder (magnify the response) is no different than pushing a door harder if it doesn’t open all the way on the first attempt. When a coin operated machine swallows your money, you may knowingly put another coin in after the first. (repeat the response) When speaking to the rather shaky microphone at a fast-food drive through, you speak with extra careful enunciation to make sure the unconcerned, teenage dreamer at the other end gets your order correctly. (offer the response more precisely) All of these reactions are the result of years of interacting with humans and the knowledge that each of these things is likely to improve your result – as long as the listener speaks your language.

Even Trainers Get it Wrong:
Before you assume that animal trainers are immune to this type of misunderstanding, consider this scene that I observed at an agility competition. A woman was running the course with her Australian Shepherd. The woman yelled “over” as the dog was in the air and yelled “right” after the dog hit the ground. The dog blew the command and stood looking for more information. The woman said “Right! Right” in a rough, commanding voice. Then she yelled the word louder and slower with a bit more “growly” roughness. By this time she was standing still in the middle of the course and her dog was slinking toward her, obviously intimidated by the harsh, tones. She started running for the next obstacle and the dog followed the visual signal of his owner running away from him. The next obstacle was the weave-poles. The dog is supposed to enter on the right and then weave his way through a line of vertical poles. As he started to enter from the left, the woman screamed “No! Right” and the dog instantly crouched and started to slink toward her. She had to move all the way to the entrance of the weave poles and use her hand as a visual target to get the dog started correctly. As the dog picked his way through the poles, the woman screamed “WeaveWeaveWeaveWeaveWeave” continuously as the dog dutifully ignored the audible distraction and continued to follow the woman’s hand as a guide. By my counting, the dog failed to follow a single audible command during the entire run – and followed every single visual cue that his handler gave him – whether she knew she was giving cues or not.

Just like the agility handler, many dog owners assume that dogs understand evToyChargeerything we say. This belief usually leads to harmless and endearing fantasies that ultimately reflect our love, affection and respect for our dogs. I have told my dog Petey volumes of things that he doesn’t understand. Not only is there nothing wrong with this, it’s an expected and normal part of being the human side of the human-animal bond. As long as we understand that this is a gentle fantasy, we’re on solid ground. The trouble starts when we want to retain our fantasies while teaching commands that are supposed to work.

The Straight Scoop:

When dog trainers approach training with the same casual attitude they use when speaking conversationally, problems follow. In obedience competitions, giving a command twice twice is an obvious fault. In agility competition, the handler is not judged on how many times she gives a command. Shouting “weaveweaveweaveweave” over and over is common in the agility world. The question is whether it is effective or not. To find out some things that may help you clean up your commands and signals, we need to go back to Uncle Ivan’s science lab and see what he discovered. So get ready to take a trip down memory lane and discover some of the now forgotten wisdom of an early 20th century scientist.

Pavlov 101:
Of all Pavlov’s experiments, there is one that virtually everyone knows about. In his laboratory, dogs were taught to associate various sounds, sights and touches with various consequences. A dog might be taught that the sound of a bell was followed by food. After 20-50 repetitions of this sequence the dog would associate the bell with food and would drool at the sound of the bell, even if food was not present. So much for what you already know.

pavlov_2What you may not know is that Pavlov was a very good scientist and didn’t assume that this was the only way to make an association. He selected a likely, hungry dog and reversed the sequence. In this case, he would feed the dog, then ring the bell. He did that more than 600 times with the same dog. Result? No association. He then took the same dog and went back to his original sequence. He rang the bell first and then fed the dog about 50 times. The dog developed a perfectly normal association between the bell and the food – just like all the other dogs. Conclusion? The sequence of presenting signals and consequences only works one way. If you try to monkey with the sequence, it won’t work, no matter how hard or long you work at it. Remember this if you don’t remember anything else about this article – the sequence is simple, but critical. Attempt to break the rules and you needlessly flirt with failure.

Get with the right sequence:
When people start learning about clicker training, they invariably scratch their heads when it comes to attaching commands to behaviors. In clicker training, the behavior is taught before the command is attached. From the perspective of traditional training, this simply doesn’t make sense. From Uncle Ivan’s viewpoint, it’s right on the money. He would suggest that you can’t connect a signal to nothingness. To put it in plain English, unless the animal already knows how to do a behavior, chanting a word will not give the dog instant knowledge of what you want. A clicker trainer might translate that into “You can’t attach a command to a behavior that doesn’t exist.” For instance, if your dog doesn’t know how to sit, no amount of chanting, yelling, cajoling or begging will cause the behavior to happen.

If your forehead is started to feel warm, don’t panic. In reality, you know of many situations where a behavior developed before you added a command to it. When the woman at the agility contest gave the command “over” after the dog had already jumped, it was entirely too late. The dog jumped, not because she had given a spoken command, but because he saw his owner running along-side the jump. In this case, the dog relies on a visual cue and intentionally ignores the spoken command. In reality, the woman never actually taught her dog the word “over” – but you’d have trouble convincing her of that.

The same is true of dogs that go ballistic when the doorbell rings. No dog owner would admit to actually teaching that behavior, but nevertheless it’s invariably true. Did they go outside and ring the bell over and over so that your dog could learn that the doorbell means someone is there? Nope. The dog’s built-in mechanism for learning the importance of an event did it all by itself. The same is true of the sound of someone crinkling a plastic bag, (potato chips) small objects dropping into a bowl, (kibble) the squeal of brass brake shoes (UPS truck) and a thousand other things virtually every dog knows that their owners didn’t teach. What we as humans need to realize is that the same dog labeled stubborn because we have to yell at him repeatedly to get him to “over”, can instantly react to the sound of a potato chip hitting plush carpet at 20 feet – and will weave perfectly through a myriad of table and chair legs to get there.

Ahhh, now you’re getting the picture. All of these supposedly dense dogs who don’t seem to understand English really don’t understand English. What they understand are connections between things that consistently go together and sequences of events that follow consistent patterns. If the sound of a can opener consistently comes before a taste of liver-flavored dog food, your dog will un-stubbornly learn to race to the kitchen when you open any can. Cats, notoriously un-trainable, learn the sound of the can opener 100% of the time that it leads to food for kitty.

The reality is that connecting a signal to a behavior is as simple as falling off a log. First you give the signal, then the behavior happens and then you give the dog a click & treat. That’s all there is to it. Signal – behavior – information – consequence. These four simple things in the right sequence will dramatically increase your dog’s response to commands. If you do them in the wrong sequence, or add things that aren’t supposed to be there, you will have trouble.

Command and signal connection: Part One
If my list of components for connecting a command to a behavior seems alien, I’ll make it a little more familiar. In nature, animals obviously need to know what to do in any given situation. They do this by connecting important signals in the environment to the behaviors that either help them get what they need or help them to escape danger. Over a series of repetitions with similar circumstances, the animal will form a habitual reaction to that particular context. Soon, there is very little thought involved when the dog senses a pre-learned combination of environmental cues. Dog hears scuffing of shoes on pavement followed by knock on door – dog goes ballistic. Dog sees owner go to pantry and hears crinkling of plastic – dog makes beeline for pantry and possible treats. Dog hears “over” before he starts to jump and is congratulated when he hits the ground.

In the above examples, it wasn’t necessary for a human being to point out to the dog the association between one thing and another. The dog is completely capable of putting any number of repetitive and predictable steps into a line, like dominoes falling over. Once the first domino falls the dog anticipates the sequence. This is equally true whether the sequence has one step or a dozen. It is so painfully clear that every dog is capable of this mental feat that we humans should rightly be embarrassed when we give a command twice twice, or yell it louder in order to get the behavior to happen. Potato chips, UPS trucks and marauding cats are instantly identified by even those dogs considered brain dead by their owners – they don’t need us to constantly remind them.

Qualities of a Command:
After you learn the all-critical sequence of connecting a command, the next thing to know is what makes a good command. A command can be anything your dog can see, hear, smell or feel. Contrary to popular belief, there isn’t anything extra complex or special about a hand-signal. Though some dogs are more visually adept than others, unless they are blind they can tell that you waved your hand just as well as they can hear you say “come.” Our Australian Shepherd friend used the sight of his owner to help him solve a direction puzzle that no audible command could decipher. Generally, a dog isn’t too concerned about whether he sees a rabbit before he smells or hears it – and if obeying you leads to food treats, he won’t care if you wave your hands over your head or shout to the heavens. Likewise, he’ll be very focused if your command is the mere movement of a finger or a spoken whisper – as long as it leads to something he wants.

Sound signals: There are several traditional sound signals that work well with dogs. The most obvious is a spoken command. Spoken commands should be taught softly, up close and then gradually at a greater distance. The reason to teach spoken commands softly is that at a distance, air resistance will mute your voice. If you have taught your dog that you are going to bark the commands at 100 decibels, up close, then the dog may well not react to your screeching at 50 yards when your voice is half as loud. One simple reason for this is that it is unlikely that you have established consequences for mistakes done at 50 yards. All animals eventually ignore signals that don’t lead to tangible consequences. Read that twice twice – it’s the number one reason that dogs fail to listen to our commands.

Another type of common sound signal is a whistle. A medium length blast on a whistle is good for distance work – which is why virtually every good gun-dog is trained to whistles. You can make the sound mean one thing in particular, or you can teach that a series of long and short tones have different meanings, like Morse Code. Just remember to keep it simple. For instance, one long blast means “come,” and two short blasts can mean “Stay.”

Sight signals:
Every culture on the planet uses a hand held over the head to signal location and to invite someone to come, stay or go away. The last three behaviors are controlled by commonly accepted movements of the arm. For pet owners, a raised hand fully above your head and waving it side-to-side makes the perfect signal for “come” or “stay.” It’s your choice. You will never be more visible than when you are at your full height, waving your hand over your head. (Note: The reason that obedience trainers don’t use this signal for “come” is that they have more difficulty with the “drop on recall” where the dog must stop and lie down at a distance. Obedience people save the most visible signal for that behavior, but it is not written in stone. ) If you think that “come” is the most important behavior your dog should know, you should connect the most visible signal to it.

General tips for hand signals: Dogs are usually shorter than we are. When you give a hand signal, it is invariably above the dog’s head. That means that the dog is looking upward and it’s a 50/50 chance that you are silhouetted against the sky or a light in the room. If that is the case, you will appear to be a big, dark object to your dog. That’s why giving hand signals in front of your body is not a good idea. To find out what works best, stand outside with your back to the sun or inside with your back to a strong light. Look at your shadow and make sure that your hand signal is plainly visible. Including some movement is also a great idea. Dogs see movement very well, even in low-light conditions. Here are the Beatles spelling out the name of their album in semaphore – HELP. You’ll note that on a white background (back lighting) their postures are plainly unique. Each of these poses could become an easily recognizable and distinct command.

Practical Applications of Pavlov’s discoveries:
To put this scientific stuff in perspective, Pavlov confirmed that associating an initially unknown signal to a particular event is very simple. First you give the signal, then the behavior happens and you add a consequence. About 20-50 repetitions and you have laid the foundation that connects the signal with the behavior. Changing the signal doesn’t make the process automatically better, but often makes it worse. You don’t need to give a signal twice twice, louder or with some kind of “command tone” in order to get a dog to pay attention. The tiniest sound of a twig breaking can cause a hunting dog to instantly change direction – even though the twig doesn’t use a command tone, lengthen its own sound or enunciate the snap sound. If you get in the habit of making very simple associations in very precise ways, your dog will be far more likely to respond to your requests. The next time you give a command to your dog, there’ll be a good chance that it might just ring a bell.


4 thoughts on “Commands and Signals Revisited:

  1. Hey GW,
    I need to reread this article on a regular basis. I try to get all of my clients to check out your training articles. There is no one out there who can explain behavior as well as you do. Thanks for all of your efforts.

Leave a Reply

Your email address will not be published. Required fields are marked *