Wag. Zazie Todd
Читать онлайн книгу.
a sensible response, as it would help the dog to avoid it, but since it is not dangerous it would just be an unnecessary source of stress. Sometimes people assume their dog will simply get used to something, and they accidentally sensitize the dog instead (see chapter 8 for how this can happen with children).
Sometimes it can be hard to know in advance if a dog is going to habituate or sensitize to a stimulus.
Social learning means learning from other dogs or from humans. Stimulus enhancement occurs when the dog’s attention is drawn to something because another dog is manipulating it, and local enhancement means the dog is drawn to a stimulus or location because of the presence of another dog. Social facilitation means the dog will tend to do something, for example joining in running, because other dogs are also running. Dogs’ abilities to imitate have also been investigated, particularly in terms of preferring certain foods, taking detours, or manipulating equipment to gain food.2 Puppies that observe their mom doing narcotics detection work are quicker to pick up drug detection than those who don’t, although it is not clear if this is due to observational learning.3 The Do as I Do dog training method teaches dogs to copy a behavior performed by a human, insofar as canine anatomy allows.4 However, more research is needed to fully understand social learning and whether there are simpler mechanisms underlying it.5
Associative learning
Dogs learn by association with events—that when the car turns in a particular direction it means they are going to the vet, for example. This is called classical conditioning and it affects the dog’s emotions rather than their behavior. For example, if we know that a dog is afraid of strangers, we can make sure that the appearance of strangers predicts us giving the dog delicious food, and over time the dog will learn to like strangers.
Dogs also learn by consequences—if I jump up on you, I get to lick your face; if I sit when you ask, I get a peanut butter cookie. It’s a simple concept (but it’s so easy to accidentally reinforce the dog for doing something you didn’t really want). This is called operant conditioning. When teaching dogs how to behave, we use operant conditioning to reward or punish behaviors.
Extinction happens when the dog learns that the consequences they were expecting no longer happen. Suppose every time your dog barks at the window, you ignore it. Your dog will continue to bark but—if nothing else is reinforcing the behavior—they will eventually stop barking. Before that happens, it’s common to get something called an extinction burst, in which there is even more barking as the dog tries harder and harder to make the behavior work. At this point, people often think that ignoring isn’t working and so they respond to the dog and inadvertently reward the behavior, which undermines the whole attempt at extinction.
Now I have to add a proviso, because dogs bark for many reasons and ignoring the behavior won’t work if something else is reinforcing it (like the person they are barking at going away down the street). We can also accidentally extinguish behaviors we want by removing the reinforcement, like when we teach a dog to come when called using a food reward, and then just stop giving the rewards. The dog will keep coming back for a while, hoping for that cookie, but then they learn it isn’t happening anymore. If other things are more motivating for them, they’ll go and do those things instead.
Operant conditioning is the foundation of most dog training, while classical conditioning is often used to help fearful dogs. Let’s look at these two types of conditioning in more detail.
Classical conditioning: learning from Pavlov
Most people are familiar with the story of Pavlov’s dogs. Ivan Pavlov was a Russian physiologist who learned it was possible to pair a natural reflex like salivation with something totally unrelated (the sound of a bell). Dogs automatically salivate in response to the sight and smell of food, sometimes to the extent that drool dribbles out of their mouth. In technical terms, in classical conditioning we refer to the food as the unconditioned stimulus (US) and the salivation as the unconditioned response (UR). It is an unconditioned association because it happens naturally. Pavlov found that if he rang a bell just before delivering the food, the dogs would salivate in response to the sound of the bell. In this case, we refer to the bell ringing as the conditioned stimulus (CS) and the salivation as the conditioned response (CR). It’s called conditioned because it has to be learned. It’s not normal to salivate in response to the sound of a bell, but the dogs learned it meant food was coming.
Classical conditioning is most often used as counter-conditioning in conjunction with desensitization as a way of helping dogs to overcome fears. Desensitization means presenting the stimulus at a very low level that the dog is happy with, and gradually increasing it so the dog becomes used to it (the opposite of sensitization). In counter-conditioning, every single presentation of the stimulus is followed by something the dog likes (such as chicken or cheese) so the dog learns the stimulus predicts good stuff happening. Note that no behavior is required from the dog in desensitization and counter-conditioning (other than being aware of the stimulus), as the aim is to change the dog’s emotions, not behavior.
Desensitization and counter-conditioning
•The “thing” (CS) happens at a level the dog is happy with—for example, a very quiet recording of fireworks or a stranger standing still in the distance.
•As soon as the dog notices the “thing,” they receive food (US), which the dog likes (UR).
•Over time, the dog learns to like the “thing,” which is the conditioned response (CR).
A great way to do this in real life is to use Jean Donaldson’s Open Bar/Closed Bar technique. As soon as the dog notices the stimulus, start the flow of chicken or cheese (or whatever great treats you are using) as “the bar is open.” Keep the flow going until the stimulus goes away or stops, and then stop the flow of treats (“the bar is now closed”). This technique helps to make the predictive relationship between the stimulus and the food obvious to the dog. All of this should happen while the dog is happy with the level of the stimulus. If you accidentally go “over threshold,” immediately reduce the level of the stimulus (e.g., turn the volume down or put distance between you and the stranger), and then feed as per usual.
Operant conditioning: learning from Skinner
One of my favorite things to teach a dog is a brief sit-stay—especially if the dog is jumpy, bouncy, and mouthy, because it can make such a difference to the ease of interaction with that dog. It is fun too, with early steps that give the dog the chance to earn many rewards in a minute. Some dogs find it really tough to sit still while I dangle a piece of chicken in front of them for just one second; other dogs find the tough part is when I start to move a little and they want to jump up and follow me. Over time, as sitting still gets more and more of a reinforcement history, it happens more often, even when I haven’t asked for it. This is in line with one of the early laws of animal behavior, stated by American psychologist Edward Thorndike as the law of effect: behaviors that get pleasant consequences will be repeated more often, whereas those that have unpleasant consequences will happen less.
B.F. Skinner elaborated on Thorndike’s ideas and did the classic work on operant conditioning. He delineated what dog trainers often refer to as quadrants: positive reinforcement, negative reinforcement, positive punishment, and negative punishment.
Positive reinforcement (R+) means adding something immediately after a behavior occurs to increase the frequency of the behavior. Technically speaking, the term breaks down into two parts. Reinforcement means the behavior continues or becomes more frequent. And positive means something is added. For example, you ask the dog to sit, the dog sits, and you give them a treat (something is added). The dog is more likely to sit next time you ask (the behavior was reinforced). Here, the words positive and negative are not being used as evaluative terms (good and bad), but as neutral descriptions as to whether something was added or taken away.
Punishment means something that reduces the likelihood of a behavior happening again; in other words, the behavior becomes less frequent. So positive punishment (P+) means adding something after the dog does a behavior that decreases the frequency of the behavior. For