In order to understand canine learning (and therefore training), it’s important to first clarify our terminology. While researching a previous book about animal training, I discovered that many excellent trainers are confused about the terms used in what is called learning theory. Since one of my chief goals in this chapter is to demystify training, it’s crucial that we agree on the meanings of words, especially the technical jargon.

Some helpful definitions

Broadly speaking, any change in the environment that an animal detects by its sensory organs is known as a stimulus. A response is any behavior or physiological event. Animals have innate or instinctive responses to stimuli; for example, newborn pups need little help to find a nipple, and young working dogs seek shade with the same intuition they use to herd ducks.

We say that animals have “learned” when there is a relatively permanent change in their response to a stimulus. When they are repeatedly exposed to a stimulus that reduces responses (including fear), we call this habituation, whereas repetition that increases responses (including fear) is known as sensitization. As well as seemingly simple sorts of learning, such as habituation, the broad definition of learning includes acquiring knowledge, as in knowing that two events tend to go together, knowing where important places are located in a territory, or knowing when important events occur. Dogs show other kinds of learning that may underlie changes in habits or acquiring skills, as in learning how to move a latch so that a door opens or to send a signal to their owner to open the door. Whether any dog ever knows why a particular event happened is debatable.

Not all changes in behavior are due to learning. Motivational factors, physiological changes, or fatigue can all affect behavior. A thirsty dog that drinks despite having refused water five hours earlier has not learned anything in the meantime. It’s just thirsty. Similarly, when a playful pup is temporarily transformed from a bouncing blur of tongue, tail and ears into a snoozing ball of fluff, fatigue rather than learning is the cause.

In my definition of learning offered above, experience is a prerequisite because the definition excludes behavior linked to maturation. For example, when male puppies progress from squatting to leg cocking, they haven’t learned that this new posture elevates the smelly signal they leave for others; they are simply maturing and responding to increased concentrations of testosterone in their bodies.

Changing an animal’s behavior: rewards and punishers

Traditionally, there are two ways to change an animal’s behavior: using a “carrot” and/or a “stick.” Carrots open the way to food, water, sex, play, liberty, sanctuary and companionship. Because they strengthen the responses that lead to these items, they are all effective rewards (primary reinforcers). There are also secondary reinforcers, whose effectiveness depends on their association with primary reinforcers. Meanwhile, the sticks discourage unwelcome behavior, so this defines them as punishers.

CHEW ON THIS

In general, animals prefer things that favor their survival. Less complex animals have a limited range of responses, but all are drawn to attractive stimuli and withdraw from potentially harmful ones. These approaches to and withdrawals from stimuli can often be modified by learning. Invertebrates such as flies, slugs and ants can all learn to modify their behavior. An example of a simple kind of learned behavior in a more complex animal is when a commercial broiler chicken shows a preference for food containing analgesics, presumably because they ease the pain of chronic leg weakness. The reduced pain, though delayed, reinforces their choice of that food over non-medicated alternatives.

Dogs learn despite us

Regardless of our intentions, dogs learn in many different ways. As most vets will confirm, they learn to fear needles very quickly. Indeed, this is why many vets develop injection techniques that spare their patients the sight of the syringe. It’s worth considering why any dog would enjoy a visit to a clinic where the vet might manipulate a sore foot to identify the cause of pain, look down an ear with an auriscope, or perform a rectal examination. And even if the vet mutters, “Just relax,” in the time-honored tradition of human proctologists everywhere, the dog cannot know how to respond. Apart from the occasional friendly face in the waiting room, there is little to make the visit a positive learning experience. Dogs learn to mistrust the smell of the waiting room, while cats often make themselves scarce as soon as they catch sight of the traveling cage.

Opportunistic animals learn to pick up the slightest cues to the possibility of a reward, even cues that humans provide unwittingly. Seagulls follow trawlers and tugs with slavish devotion because there’s a good chance of a free meal. As the arch-opportunists of the animal kingdom, dogs are always on the lookout for food or fun. Some learn that the sound of the car’s indicator means that the park is just around the corner. Indeed, some overexcitable (or under-exercised) dogs might then bark every time the driver makes a turn. Similarly, dogs learn to associate the sound of a can opener with supper and come running when any can is opened. Trainers should remember that their dogs may be alerted by subtle stimuli that are inadvertently included in a training program. For example, obedience-trained dogs may associate the forbidding sound of a choke chain with accurate heelwork and become a little reliant on the threat of pain, performing less well when the chain isn’t worn.

This goes with that

Almost all forms of training depend on the animal associating particular events. As far back as 350 BC, the Greek philosopher Aristotle suggested that the most important principle of association was contiguity (the proximity of two events to each other). He maintained that the more closely together two events occur, the more likely that the thought of one will lead to the thought of the other. The links don’t even have to be logical (as when we memorize poetry or a list of nonsense words that become associated, with each item prompting memory of the next). An association can be based on either spatial or temporal contiguity, or both. For example, a dog may associate his bowl with food by their spatial contiguity and the sound of the door-bell with the arrival of visitors by temporal contiguity.

CHEW ON THIS

The principles of association weren’t studied in a systematic way until towards the end of the nineteenth century. Research began with experiments on human memory and then went on to explore how animals make associations. At almost the same time, but quite independently, an already famous Russian physiologist, Ivan Pavlov, and a then-unknown American student of psychology, Edward Thorndike, carried out the first experiments on learning in animals. Despite their different approaches, their research formed the basis of what became known as classical and instrumental conditioning.

Classical conditioning and Pavlov’s dogs

Conditioning refers to a type of learning in which the timing of events is particularly important. As we’ll see, good timing is almost always critical for dog training to be effective. Studies into conditioning have given us a set of techniques to help modify behavior. In this chapter, we’ll explore these studies, and I’ll explain how their findings underpin the basic principles of dog training. Conditioning is also a powerful way of studying the nature of associative learning.

Pavlov studied dogs because he was primarily interested in physiology, or how bodies work. However, after winning a Nobel Prize in 1904 for his research on digestion, he became interested in what would later be called the neurophysiology of learning. The change was triggered by something Pavlov and his coworkers initially regarded as an obstacle to their work on digestion. Many of the experiments involved measuring stomach “juices” as a reflection of the kind of food a dog was given. When a dog was tested regularly day after day, a complicating factor emerged: Digestive juice would begin to flow out of a surgically constructed hole in the stomach as soon as the research assistant approached, well before the dog was given any food. The physiological reaction seemed to be triggered by an association between the arrival of the assistant and the delivery of food.

Pavlov and his students eventually realized that this “psychic reflex” would allow them to study how associations are formed. Various stimuli were selected to signal the arrival of food: the sound of a metronome, visual signals, and pressure pads on the dog’s body. All of these had the advantage of being easier to control than the appearance of an assistant. Secretions from the salivary glands indicated the strength of the resulting association. The more dribble, the greater the association. Over the next three decades, using this basic procedure, Pavlov laid the foundations for the study of associative learning in animals, becoming more famous for this than for his work on digestion. Although we refer here to the type of learning he studied as classical conditioning, many learning theorists still refer to it as Pavlovian conditioning.

A typical classical conditioning study starts with a neutral stimulus, one that has little effect on the animal, and presents this conditioned stimulus repeatedly, closely followed each time by an unconditioned stimulus, such as food. Eventually, the conditioned stimulus consistently produces a response, the conditioned response, related to the unconditioned stimulus. In Pavlov’s experiments, a buzzer (the conditioned stimulus) that had little effect when first heard except to make the dog prick up its ears, caused the dog to salivate (the conditioned response) after it had been paired many times with meat powder (the unconditioned stimulus). If the buzzer was no longer followed by the meat powder, it became progressively less effective in making the dog salivate. Crucially, in classical conditioning a conditioned stimulus, such as the sound of a buzzer, is followed by an unconditioned stimulus, such as food, regardless of what the animal does when it hears the buzzer. The arrival of food is independent of any response. Thus, classical conditioning lets an animal learn to associate events over which it has no control. Such learning allows your dog to predict events and adapt before they happen.

There are many real-life examples that demonstrate the role of classical conditioning in learning. Some dog breeders make use of a similar effect to ensure reliable performance of stud dogs. They adopt the same routine before taking the dog to the same room prior to every mating to produce conditioned sexual arousal. Even without laying eyes (or perhaps that should be nostrils) on an estrous bitch, the dog is ready for copulation as he enters the room. Similarly, the sound of gravel on the drive tells most dogs that someone is about to turn up at the front door. The sight of a tablet packet spells trouble for some dogs because it is linked with the need to tolerate fumbling fingers down the back of the throat. It is all about associations. So it is the same process that helped Uncle Wolf learn which howl went with which pack member.

Turning a negative into a positive

A particularly useful variant of classical conditioning is called counterconditioning . This is when an unpleasant stimulus is changed into one that is positive for the animal. The first known example came from Pavlov’s lab. He used a mild electric shock, which initially elicited signs of pain, as a conditioned stimulus. After the shock had been paired repeatedly with food, it started to encourage salivation, and there was no sign that it was still painful. A more familiar example is the humble leather collar and leash. A pup might initially scratch and resist the collar and leash, but very soon it will associate them with the adventure of a walk.

Rewarding behaviors that are mutually exclusive to unwanted responses—for example, in a dog that races after cars, sitting calmly with him near a busy road—is called counterconditioning. It can be very useful in behavior therapy and in getting dogs to accept painful therapeutic procedures.

The importance of timing

Some of Pavlov’s early experiments confirmed the importance of Aristotle’s principle of temporal contiguity. Conditioned salivation in response to a buzzer developed much faster when the food arrived within a few seconds of the buzzer sounding than when the gap between the two events, the interstimulus interval, was longer. However, for training a conditioned response, the optimal length of the gap depends on the hoped-for response. At one extreme is eye-blink conditioning, which can be produced by sounding a beep before delivering a puff of air to the eye. The best interval for this is about half a second. With a delay between the beep and the discomforting puff of more than a second or so, it becomes difficult to get a conditioned blink to the beep. However, the arrival of food to a hungry dog or water to a thirsty one still allows strong conditioning to a light or sound even after a gap of several seconds.

Associations between two events are achieved more quickly if the events are novel. If a dog is exposed to a conditioned stimulus (such as a verbal command) on a number of occasions before the conditioning procedure begins (before it is paired with any training), it will take a while for it to learn the conditioned response. The dog may simply learn to ignore the stimulus because it has no important consequences. This is why some trainers feel it is better to add the cue after the responses it has initiated. So, by way of an example, to train a dog to spin 180 degrees (for freestyle heelwork), I will first lure it with a treat; then give the treat only after it has followed my hand as it describes a circle; then wean it onto a circular finger movement, and finally, when it is spinning merrily with this minor visual cue, I will replace it with a verbal command such as “Round!”

The importance of consistency

Pavlov found that when a conditioned stimulus (the sound of a metronome) was paired with food as an unconditioned stimulus, it continued to make the dog salivate just as long as the conditioned stimulus was followed by food. If the metronome sounded again and again but no food arrived, the dog stopped salivating to the tone, an outcome that is called extinction. Extinction applies to all the examples of classical conditioning given here. The dog will eventually stop blinking to the beep if the air puff no longer follows. The stud dog won’t get aroused by trips to the breeding barn if none of his sexual encounters ever occur there. Anything that weakens desirable associations can compromise training programs. If I interchange “Round!” with “Go round!,” “Now round!” or complicate matters by sometimes adding the dog’s name (e.g., “Tinker, Round!”), I should expect a diminished quality of responsiveness, with the dog spinning sometimes but not others. Put simply, inconsistency impedes training. The best life-coaches put great effort into being as consistent as possible.

CHOICE CUTS

Principles of training apply to all species over a very wide range of conditions.

Food, water, sex, play, liberty, sanctuary and companionship are all effective primary positive reinforcers because they make the responses that gave rise to them stronger.

Dogs learn to pick up the slightest cues when there’s a possibility of a reward.

The more closely together two events occur, the more likely will the thought of one lead to the thought of the other.

Classical conditioning is the same as Pavlovian conditioning.

In classical conditioning, a conditioned stimulus is followed by an unconditioned stimulus, regardless of what the dog does when it detects either.

Associations between two events are acquired more rapidly if the events are novel.

Good timing defines good training.

Inconsistency impedes training.

Ivan Pavlov studying one of his laboratory dogs. The disc on the side of the dog’s face is a fistula used to collect saliva as a measure of the association dogs had between food and various conditioned stimuli, such as noises.