Last revised April 20, 2017.
A good mother is a good teacher.1
Horse trainers have strong notions of how horses learn, and how to teach. Some horse trainers get it wrong, though, and while they succeed in changing a horse’s behavior, it often takes much longer than it should have. In this chapter I’ll look at the basics of how horses learn, and how to go about training. In the next chapter, “Carrots and Sticks”, I’ll look at more details about reward and punishment.
Horse training approaches should involve the principles of learning theory (the principles of how all organisms learn), and consider motivation, associative learning, reinforcement, punishment, shaping, and habituation. In addition, horse training should consider equine ethology (the study of the horse’s natural behavior), so that we use stimuli similar to those that horses naturally respond to, and reinforcers that they find reinforcing.2 This seems needlessly complicated, but that’s how the world works. In fact, basing our training on learning theory and ethology simplifies everything. Without understanding these fundamentals, we enter a world of accidental results and mythological explanations of those results. And that is today’s world of equestrian coaching.3
Consider “contact”: the connection of reins to horse’s mouth, legs to horse’s side, and saddle to horse’s back. When contact is increased in the mouth, it has the effect of turning or slowing the horse. When it is increased with your legs, it has the effect of turning, side passing or speeding your horse. “Contact” is not a word that anyone should use, because it is far too imprecise to define what the rider should be doing.4
I hope you will agree to some ground rules if you set out to teach your horse something. Here are mine:
First, do no harm
Trainers using both traditional techniques and natural horsemanship seem to agree that their job is primarily working with humans, “on the grounds that whatever problems the horse/human relationship might have almost certainly originated in equine responses to human behavior.5”
Even if we now know that mild negative reinforcement works best in many training situations, we must remember the word “mild”. We must remember the research showing that making a punishment more severe does not increase its efficacy. Promise that you will never knowingly hurt your horse. Promise this to yourself. Promise it to him. Keep your promises. Remember: First, do no harm.
Everyone thinks they can accept this principle until they must choose between what they want and what the horse wants. Consider what you do when you ask your horse to back up, and he declines. You likely pull back on the reins harder, until your horse is forced, via the bit or pressure on his nose, to yield and move backward. Just a single painful backup is a broken promise to do no harm. If you love your horse, do not hurt him.
Replace your bridle and bit with a halter. Now that you’ve promised to not hurt your horse, throw away that bit. A bit in a horse’s mouth likely feels the same as a bit in your dog’s mouth or child’s mouth or your own mouth. You wouldn’t put a bit in your dog or child’s mouth to control them, would you?
No more spurs. When our horse has become a willing partner, moving forward is by mutual agreement. The willing horse answers the request, and the request need not include pain. Endurance riders do 100 miles without spurs. Certainly you can do a trail ride or loop around the ring without them.
Whether you opt for some form of natural horsemanship or some form of traditional horsemanship, your horse will benefit if you work more respect and kindness into your time with him.
My horse knows about trailers. He is a thoroughbred, and from a line of horses that include War Admiral, Secretariat, Northern Dancer, Storm Cat, and Linkit. From his pastures in Kentucky when he was a baby to the end of his career at the track, G surely rode in a trailer many times. And so it is safe to assume that he knew how to climb into a trailer.
His life at the track was surely made worse by trailers. Every time he had to get in one, it must have meant he would be leaving his stable mates for parts unknown, perhaps never to see or smell or hear them again. So he knew how to get on and off a trailer. And I can’t imagine he looked forward to it.
When I adopted him, I paid someone with a trailer to take him to his new home. They arrived with a stock trailer, we opened the back door. Carlos was ready to exert some serious force to make G get into the trailer, but I asked if I could give it a try. I walked into the trailer, and G followed at my heels. No problem. G was knowledgeable about a trailer, and he seemed to have a good attitude.
When we got to our destination, I got in with G, and asked him to climb out. During the next 5 seconds, Carlos decided that G wasn’t coming off fast enough, and punched him in the face. G, in self-defense, jerked his head up, banging his head on the ceiling and opening a wound.
A few years went by before G needed another trailer ride. When he was given the opportunity to climb in — and Carlos was nowhere around — G would have none of it. G was still knowledgeable about a trailer and knew how to load, but now he had a bad attitude. If you think about it, Carlos taught my horse something. He taught him that the inside of a trailer was a dangerous place, where you could suddenly become very injured. My horse knew how to get into a trailer, how to balance during a long ride, how to get out. And he knew that something very bad could happen to you.
Some years later, I bought a truck and trailer so that I could have some distant adventures with G. In his first weeks with the trailer, he didn’t want to load. Many people would come by and suggest what should be done to train him, to induce him to get on. It took me some time to realize these truths: he was trained. He knew what I wanted. He wasn’t being a brat. I think he dreaded getting on the trailer because he remembered the time Carlos mistreated him. His problem was not training in how to load. It was the training Carlos had given him, which taught him that being in a trailer is dangerous. Horses have good memories, and what G remembered from his most recent training was not good.
Times have changed. G now gets to go with me to some interesting places, and have some big adventures. Now, he scrambles on the trailer when I get the ramp down, whether we are getting ready to leave home, or getting ready to come back. And I am sure that he scrambles because he knows it will be fun. All of his associations with the trailer are now positive. He is able to look out and see the world go by. He has all the hay and water he wants for a snack. And when we get to the destination, he has more adventures or is able to relax with his buddies. He climbs into the trailer the way your dog climbs into the car.
If you are having issues with your horse, consider whether the problem is with his failure to understand what you want, or his lack of desire to do it.
If your horse once loaded fine, but now doesn’t want to load, blame his history and associations with the task, not him. Blame punishment. Blame Carlos. And read the rest of this section.
In the world of horses, everyone seems to confuse training and motivation, and think that if a horse won’t do something, it needs training. Very often, it knows just exactly what to do, but needs motivation. We have too many horse trainers, and not enough horse motivators. Motivation matters.
The interaction with G and Carlos was a case of associative learning. G learned an association between two stimuli: being in the trailer, and being punched in the face and smashed on the top of the head. Now, after some happy trips to happy times and back, G has replace that association with another association: being in the trailer and getting hay, water, and a chance to see and smell the country going by.
Associative learning connects two things, as when G learned to fear trailers, and later learned to love trailers. But associative learning can also connect a stimulus with an outcome.
Pavlov found that dogs salivated when presented with the sight and smell of dinner. When a bell was paired with the presentation of dinner, dogs learned to salivate at the sound of the bell alone. The ringing bell became a conditioned stimulus which could trigger the same response of salivation as the unconditioned stimulus, the sight and smell of the food. Pavlov’s dogs illustrated classical conditioning, sometimes called Pavlovian conditioning. Classical conditioning is a special case of associative learning.
My horse loves peppermint candy, and has learned that he is about to get a treat when he hears me unwrapping the cellophane. His expectation of candy coming after hearing the wrapper is the result of classical conditioning. The sound of crinkling cellophane is another conditioned stimulus.
Classical Conditioning Connects Bridge and Treat
My horse has learned that a click of my clicker is very likely to result in a treat. The clicker serves as a bridge between his behavior and his reward. It could result in the sound of unwrapping cellophane, but that is almost certain to result in a treat. So the click, the unwrapping sound, and the treat are all associated. All we have to do now is add a behavior that we are hoping for, such as picking up my hat when I give him a cue. When he does that, and he gets a click, an unwrapping sound, and a tasty treat, he associates whatever he was doing with all of these good things, especially the treat. These associations form the heart of bridge training (sometimes called clicker training. Bridge training doesn’t require a clicker. A scratch on the withers, a “good boy”, a whistle toot, or a clucking sound from your mouth can also work as bridges.)
Once the click of a clicker is associated with a forthcoming treat, it becomes handy in helping my horse know exactly what will product that treat. I can click the clicker quicker than I can unwrap a peppermint or dig out a slice of carrot, and so can make it even more clear to him just what it is that he has done that has met my approval. The precise timing of my click means that my horse is in a good position to guess what he did that produced the click/praise/treat. Compare this with the friend who, seeing my use my clicker, said “I give my horse a carrot when we get back to the barn, if he’s been a good boy.” Her horse could not possibly guess correctly just what he did to get the carrot.
Associative learning helps improve a horse’s ability to predict what is coming next. When you ask for a turn on a forest trail, your horse can predict the turn from any of the cues that you provide, because he has associated them with each other, and all mean the same thing. Classical conditioning allows us to shift from pressure cues to those not involving pressure, such as a change in seat position.
Bridge Training Works Through Operant Conditioning
Both the use of treats and reducing pressure allow your horse to learn to control his environment, and are examples of operant conditioning — the horse is learning to work as an operator and get what he wants. The chain of events consists of three important parts: stimulus, response, reinforcement. In the case of reining, a tug on the rein applies pressure, the stimulus or cue. The horse turns his head and direction of travel, the response. The pressure is immediately removed: the reinforcement. In the case of the hat trick, you might drop the hat, point to it, and ask him to pick it up — the stimulus. He might then pick it up and hand it to you — the response. You now hand him a carrot slice, the reinforcement. Operant conditioning can be used to teach your horse that if you give him a particular cue, and he then behaves in a particular way, that he will get a particular consequence.
Operant conditioning is good for the horse. It gives him control over his environment.
When you are standing on the ground with your horse, it is easy to use positive reinforcement, and give him a click or hand him a slice of carrot regularly. But when you are on his back, this becomes impractical. We can’t go for a trail ride and click whenever he takes a stride. So most of the time we use pressure — and mild negative reinforcement — whenever we want to increase speed or change direction. But once we have compliance, and our horse is doing what we want, we have no business continuing any form of pressure.
In horse psychology, a reinforcer is anything that occurs after a behavior which makes that behavior more likely. A punishment is anything that occurs after a behavior which makes that behavior less likely.
The formal language of the psychology of learning uses these terms:
- Positive reinforcement: receiving a pleasant stimulus. Your horse gets a treat or praise.
- Positive punishment: receiving an unpleasant or aversive stimulus. Your horse receives a yank on the bit.
- Negative reinforcement: taking away an unpleasant stimulus. You relax the reins, releasing pressure on the bit.
- Negative punishment: taking away a pleasant stimulus. Your cell phone rings, and you stop grooming to answer the call.
In these terms, positive means that you add the consequence to his life, and negative means you remove the consequence. Reinforcement is something that increases the probability of a behavior that it follows; punishment is something that decreases the probability of a behavior that it follows.
A positive reinforcer is rewarding. It is something we want. Negative reinforcement, as defined above, is OK, but we don’t feel any improvement in our well-being if we sit on a tack then jump up. I like positive reinforcement. So does your horse. And nobody likes punishment, whether positive or negative.
|Pleasant Stimulus||Aversive (Unpleasant) Stimulus|
|Adding/Presenting||Positive Reinforcement||Positive Punishment|
|Removing/Taking Away||Negative Punishment||Negative Reinforcement|
Dolphins blazed the way with positive reinforcement training, but sea lions, dogs, and horses quickly followed. Here Jodie Foreman gets a kiss from a sea lion trained by Jenifer Zeligs. In exchange, the sea lion got a fish. Everyone — except the fish — comes out ahead.
Rules for a Good Teacher
The quality of a teacher must be seen in their results. When your student learns, you must be doing something right. When your student fails to learn, the teacher must give themselves an “F”.
To teach well, McGreevy and McLean6 have suggested that we follow seven training principles.
- use learning theory appropriately;
- train easy-to-discriminate signals;
- train and subsequently elicit responses singularly;
- train only one response per signal;
- train all responses to be initiated and subsequently completed within a consistent structure;
- train persistence of current operantly conditioned responses; and
- avoid and disassociate flight responses.
What motivates a horse? We know of several powerful carrots and sticks:
- Reducing discomfort.
- Other Horse Stuff.
A rider pulls on the left rein, putting pressure on the mouth, nose, ear, or right cheek. The horse turns its head left.
How does this magic trick happen? Turns out that any organism that can move has nociceptors — sensory nerve that respond to damaging stimuli by sending signals to the brain. Brains are hard-wired to find these signals unpleasant, and do whatever it takes to avoid these signals. So the horse turns his head left in order to stop the nociceptors on the right side of his face from firing. Never mind all the blather about “horses move into pressure” or “horses move away from pressure.” What we can be sure of is that horses move in such a way as to reduce discomfort. This does not make them unique. You are likely to remove your hand from a pot of boiling water. Cattle, camels, zebras, elephants, horses and mules can all be steered with halters and bridles.
When an aversive stimulus is applied, the horse works to remove it. If you cease pulling on the rein when he turns his head, this release is a reduction in discomfort and an example of negative reinforcement. It is reinforcement because it makes the desired action more likely. It is negative because you have removed the discomfort.
Timing is everything. Negative reinforcement is only effective when the application of a pressure is followed by its immediate removal once the horse performs the desired behavior.7 If his efforts do not remove his discomfort, he may become unresponsive or dangerous.8
Novices learn how to apply pull on the reins long before they learn to release them. If they pull on the reins and make him uncomfortable on the right side of his face, he turns to the left, and the rider keeps the reins taught, then the correct behavior gets no reinforcement. The rider yanks on the reins, the horse jams on the brakes, the rider keeps the reins pulled back, and the horse asks “What am I supposed to do? Backup?”
People who are involved with animals like to think that what they do doesn’t hurt. Fishermen tell me that fish aren’t hurt when they are hooked, even if they swallow the hook and its removal means ripping out part of their digestive system. And horsemen tell me that what they do doesn’t hurt their horse. When we look at how we treat them, is fortunate that animals have no feelings.
Horsemen yank on a rein, and call it “pressure”. And they have a swell theory: horses move away from pressure. Over 845,000 articles on the Internet agree with this claim. But change the search to “horse moves into pressure”, and you’ll find 2,280,000 articles on the Internet. So no matter what we do to a horse, and what it does in response, we have a theory for that, and it involves the horse’s natural tendencies.
When we talk about the pressure of the reins, we are not talking about some broad, distributed pressure. We are talking about intense, focused pressure on a very sensitive area. A horse, donkey, or mule will happily pull into pressure when it is broadly distributed with harness over a well-muscled area. The pressure that causes a horse to yield is different that that. A bit in a horse’s mouth can produce intermittent pressure of up to 2,000 pounds per square inch (PSI) or more.9 Imagine what would happen to your bicycle tires if you inflated them to that pressure. It is no surprise that bits sometimes sever the tongue.10 And beyond bits, we have an array of other medieval devices to apply extreme pressure: curb chains, gags, hackamores, bosals, draw reins, balancing reins, chambons, and, of course, whips and spurs.
There is nothing magical about “pressure”. We could put an ice cube against our horse’s right cheek if we wanted him to turn left. Or try lighting his right ear with a match. With such activities, he will turn his head to avoid our treatment, and if he is moving forward, his body will follow his head. But using the reins is more convenient, and generally doesn’t leave lasting damage.
Horses are certainly capable of working to avoid discomfort from aversive stimuli. And they are capable of learning. So minutes after a horse has learned to go left when the left rein is pulled, he needs nothing but a feather-weight adjustment of the reins to tell him what is wanted. Driving a horse should be different than driving a tractor. Negative reinforcement can be a useful training tool.
If you swat me on my face with your dressage whip, I won’t likely scream in pain, but I will likely step back. I might even reassess our friendship. Your horse works like my face. If something happens that he finds annoying, uncomfortable, unpleasant, or even mildly painful, he will try to avoid that in the future.
If a bee buzzes around his butt, he move forward to avoid it. If it buzzes around his head, he moves back. You and I have learned that bee stings hurt, and we’d do the same. If you tap your horse on his butt with your dressage whip, he is likely to move forward to avoid the taps. If this movement stops the tapping, then the taps served as negative reinforcement. Coupled with positive reinforcement, such as treats, negative reinforcement can be useful in shaping behavior.
Punishment and negative reinforcement are entirely different.
- Negative reinforcement occurs during behavior, and ends with a change in behavior. A horse learns to behave differently with negative reinforcement if it ends the exact moment that the undesirable behavior ends. For instance, suppose you want your horse to move toward you on some cue. You give him a little tap with your whip, he moves toward you, and you stop tapping. This sequence is that of negative reinforcement.
- Punishment occurs after behavior. In contrast, if you were to ask your horse to move toward you, and he did not, and you then punched him in the face for being so disobedient, he would learn to fear you, learn to distrust you, learn to stay away from you. He would not learn to come to you when you asked.
In my sketchy childhood, I was spanked hard and often. Once, on an endless car trip out west, my Dad told me “If you ask one more question, I’m going to pull over the car.” I gave the only possible response: “Why?” I remember being dragged behind a cactus, and learning not to ask questions. But today my wife sometimes calls me “Question Man”, and I suppose it is because my time behind that cactus wasn’t as effective as Dad intended. I don’t think I was a better kid because of punishment. I was just more cautious around my Dad.
By definition, punishment makes a response less likely in the future. Because horses are largely trained by negative reinforcement, they are susceptible to inadvertent punishment. If you apply “pressure” to your horse with bit, reins, spurs, or crop, and he does what you have asked for, if you continue to apply this pressure, you will weaken the connection he has between your pressure and what he is supposed to do. Delays in the release of pressure can make desirable responses less likely and thus punish them.11
There is no place for punishment in our training of horses, children, or anything else. This is not because we are damn tree huggers or softies or liberals. It is because it is not effective. Horses don’t learn what we want from punishment. It only works to help us blow off steam.
And motivating a horse to learn using positive reinforcement has better outcomes beyond training. One study found that “the association of a reward with a learning task in an interactional context induced positive reactions towards humans during training. It also increased contact and interest, not only just after training, but also several months later, despite no further interaction with humans. In addition, this ‘positive memory’ of humans extended to novel persons. Overall, positive reinforcement enhanced learning and memorization of the task itself.12”
Of course, if we swear off punishment, we not only find ourselves in search of substitutes (and we are about to find some that work very, very well), but we find ourselves on the moral high ground. As Black Beauty’s groom once said, “…. there is no religion without love, and people may talk as much as they like about their religion, but if it does not teach them to be good and kind to man and beast, it is all a sham.13…”
Shaping is at the center of all training. It is done with differential reinforcement of successive approximations, changing the rules of what produces a positive reinforcement as our horse gets closer to performing our desired behavior. B.F. Skinner described shaping like this: “We first give the bird food when it turns slightly in the direction of the spot from any part of the cage. This increases the frequency of such behavior. We then withhold reinforcement until a slight movement is made toward the spot. This again alters the general distribution of behavior without producing a new unit. We continue by reinforcing positions successively closer to the spot, then by reinforcing only when the head is moved slightly forward, and finally only when the beak actually makes contact with the spot. … The original probability of the response in its final form is very low; in some cases it may even be zero. In this way we can build complicated operants which would never appear in the repertoire of the organism otherwise. By reinforcing a series of successive approximations, we bring a rare response to a very high probability in a short time. … The total act of turning toward the spot from any point in the box, walking toward it, raising the head, and striking the spot may seem to be a functionally coherent unit of behavior; but it is constructed by a continual process of differential reinforcement from undifferentiated behavior, just as the sculptor shapes his figure from a lump of clay.14”
To shape, we must start with a clear behavioral goal which we can visualize, a goal which includes both what we want and what we don’t want. Example: “On my cue, I want my horse to lower his head to my waist, and hold it there. I want him to stand calmly while doing this.”
Efficient shaping requires that we break a complex task into simple chunks, and train each of them, stringing them together as our horse gets the hang of them. For instance, to get him to lower his head and and hold it there, we might break this into a head lowering chunk and a position maintenance task. We might start with getting him to lower his head farther, and later work on keeping it down. We will only train one criterion at a time, to avoid confusing him.
If our horse happens to lower his head a bit, we want to reward him. But we want him to know just what he did to produce this reward. Fumbling in your pocket for a carrot does not answer his question. He lowered his head, he raised his head, he looked around the barn, and then, for some reason, he was offered a carrot. Did he get the positive reinforcement for the lowering, the raising, the look around the barn, or the entire sequence? We need to pinpoint exactly what he did that is so desirable, so we introduce a clicker.
On the first day of our training, we begin teaching our horse that when he hears a click, he will be getting a treat. The association between click and carrot happens quickly: he learns that a click is followed by a treat. Through classical conditioning, the clicker has become a secondary reinforcer. The treat is a primary reinforcer. If he were one of Pavlov’s dogs, he’d be salivating.
Once the click has acquired this positive value, it can be used to mark exactly what the horse was doing that has so pleased us. By clicking when he dropped his head, the horse learns that dropping his head in response to the cue is going to produce a carrot. Our clicker has become a marker, telling him exactly what will produce a treat.
The click also serves as a bridging stimulus, filling the gap between his good behavior and his treat and connecting them.15 The bridge becomes effective only when our horse links it to both the forthcoming reward and to the behavior that produced it.
Assuming our click is now understood by our horse to mean that a treat is coming, we are ready to move on. We then create a condition where we can elicit the rudiments of that behavior. For instance, we might put him in a halter with a short lead rope, and standing next to him, we gently put some slight downward pressure on the lead rope. In a short time, he is likely to drop his head slightly. That one inch drop is our starting point which we will shape.
Focusing on lowering his head, we repeat and repeat our effort. We might cue for the head drop by providing a slight downward pressure on his lead rope. With our clicker, we tell him he has done the right thing at the moment he does it: click! We then quickly heap on verbal praise, and reach for a treat. The entire package — click, praise, treat — is his positive reinforcement for what he just did. The click happens quickly, allowing him to identify just what he did that produced the package.
As we repeat the exercise, we shift our standards. Where he once got a click/praise/treat for a drop of just an inch, he is now dropping 6 inches, and getting click/praise/treat. Because he can do this consistently, we now try for 8 inches. 6 inches produces nothing, an 8 inch lowering delivers click/praise/treat. Each shift in standards is small, to ensure that he has a good chance of success.
If we have shifted our standards, and he is not succeeding, then we’ve gone too far too fast. We go back. If he won’t drop his head 10 inches, but once dropped it 5 inches, we are headed back to 5 inches. We will reward several 5 inch drops, and then move on to greater distances, but in smaller increments to ensure he is keeping up with our shifting standard.
Once he can meet our goal nearly every single time, we will shift to a variable reinforcement schedule. He’ll get a click/praise/treat only about half the time — or less — to strengthen his connection between the cue and this behavior.
Our original goal was that he lower his head and keep it lowered. Now that we have gotten it lowered properly, we can shift our attention to keeping it lowered. To improve duration, we can count aloud. Initially, he may only reach “A! B!” before he raises his head. So on a subsequent trial if only reaches “A! B!”, don’t click/praise/treat, but if he reaches “A! B! C!”, immediately click/praise/treat.
Alexandra Kurland’s Rules of Shaping16
Alexandra Kurland has written an excellent book on the bridge training of horses. I urge you to buy and read it. In it, she provides a summary of her key rules of shaping. Here they are:
- Train one criterion at a time.
- Get a response, get it consistently, then improve on it.
- Raise your standards in small enough steps that the horse continues to be successful and can be reinforced.
- Once the behavior is established, shift from a fixed to a variable reinforcement schedule to improve response.
- When adding a new criterion, temporarily relax the standards of the old.
- If behavior deteriorates, go back to a previous step in the shaping process.
- If one shaping procedure is not creating progress, find another. There are always many different ways to build the steps to the desired behavior.
Other Topics in Learning
Extinction is the reduction of a learned response to a stimulus when the stimulus is repeatedly presented and no reinforcement for the response occurs. When Pavlov’s dogs were fed after hearing a bell ring, they learned to salivate at the sound of the bell. When Pavlov continued to ring the bell but did not follow the ringing with food, the conditioned response of salivation was extinguished — it came to a stop.
If your horse has been mugging you for a carrot, and you decide you don’t like this, extinction can be your secret weapon: once you decide that mugging will never, never, never produce something he wants, he’ll figure it out and stop. He’s pretty smart. But like a gambler, he plays the odds. If you sometimes yield to his begging, you won’t get extinction. Instead, you’ll strengthen him as a gambler, mugging in desperate hope for a treat.
Can dogs learn by imitation? The answer is yessss! Recently, some have learned that dogs learn just fine by imitating each other or their trainers.17 That’s pretty exciting! But then, I guess humans are pretty dumb after all. How in the world would mama dog teach her puppy anything if puppies needed her to use words and gestures and clickers? How would a baby raccoon or baby elephant or any other baby bird or mammal learn a thing from a parent except through imitation? Parents don’t teach, but babies do learn. They do it through observation, what brilliant researchers have called observational learning.
A variety of studies have documented observational learning in crows,18 keas,19 pigeons,20 parrots,21 (you’ve heard of “parroting”?) chimpanzees,22 orangutans,23 (to “ape” is to imitate behavior or manner.) and dolphins.24
Dogs have a special gift for learning by observation, because they have evolved to become great observers. Dogs are better than wolves or even chimpanzees in tests of sensitivity to human social cues such as pointing, head turning, gazing, and nodding.25 They may even be better than teenagers.
Dogs know how to learn by observation, but most trainers ignore this and persist in “Do as I say” when they could be training by “Do as I do” (DAID). One important study has found that such DAID training is on a par in efficiency with bridge training for simple tasks, but is far superior to bridge training on complex actions and sequences of two actions.26
DAID differs from bridge training in important ways:
- DAID emphasizes a cognitive approach, whereas Bridge Training emphasizes an associative approach.
- DAID takes into account the role of attention and the knowledge of the learner; bridge training is an associative training approach that connects two events (unconditioned and conditioned stimuli) and/or an association between a discriminative stimulus and an operant behavior.
The effectiveness of DAID is not limited to dogs. It was originally demonstrated with a chimpanzee learning from a human.
But since this is a book about horses, we want to know if horses can somehow learn by observation, through imitation. Us mortals sometimes believe that our horses learn by imitation. One recent study found that about half of all horse owners believe that cribbing or crib-biting is learned by observation, but only 1.0% of horses took up cribbing after being exposed to a cribber.27 Our notion of a horse’s ability to learn by observation may not be correct.
In fact, there are many studies that seem to establish that horses cannot learn from humans by imitation.28 It is possible that some horses can learn by imitation, and others can’t. For instance in one study, warmbloods appeared to be slower at learning a bin opening task than non-warmbloods, possibly due to higher anxiety levels in the study.29 A recent study30 found that young, low-ranking and more exploratory horses learned by observing older members of their own group. The older the horse, the more slowly it appeared to learn by imitation.
A similar study found that horses were more likely to follow humans if dominant horses from their social group were following that human, but not more likely to follow if subordinate horses from their group, or horses from a different social group were following. So we can conclude that horses are more willing to imitate other horses if those horses are dominant.
These studies indicate that imitation will occur in horses if a) the horse that imitates is young; b) the model for the imitation is a horse; c) the model is a high ranking horse from the young horse’s herd. If these conditions are not all met in a research study, imitation may not occur.
Will your horse imitate you? Try a trail ride which includes a scary bridge. Your horse might balk, but if others in your group go across the bridge first, or you dismount and lead him across, he is very likely to follow. Some of this form of imitation might depend on your horse’s respect for the others crossing the bridge. If he will follow you, consider yourself in the same class as dominant horses from his herd.
If he follows you over the bridge successfully a few times, and you remount, is he now willing to tackle the bridge? If so, he has learned through imitation. He has learned that the bridge crossing is safe. He didn’t not learn anything that requires a new motor behavior, but he has learned about the appropriate emotions that determine whether he can show that motor behavior.
Imitation has been studied in cows: cows that have been handled gently stood close the gentle handler who hand fed them hay than they stood to an aversive handler who slapped them in the face. Observer cows also stood closer to the gentle handler than to the aversive handler.31 And as we think about it, horses do seem to adopt the calm or fearful feelings of others around them.
None of this blather comes down on the side of DAID: if you ask your horse to watch you as you do a figure 8 in the round pen, then tell him to “Do It!”, I will be surprised if he does it. I will also be surprised if he learns to play the ukelele by watching you.
This section ends with a whimper, not a bang. It is an area that needs much more research. For now, we know that bridge training and extinction definitely works. We know that negative feedback shapes learning fastest, while positive feedback best helps with retention. But we aren’t so sure about imitation. If you know of a situation where a horse has learned by imitation, please let me know.
Is Learning Fun?
I went to college for a total of 9 years. Somewhere along the way, I asked myself whether I found learning to be fun, and decided that the correct answer was no. For me, learning was not fun, but not knowing was more unpleasant. Learning removed the “not knowing”, and that removed the unpleasantness.
Cockatiels have a “word” for “Ow!”. They use it in a variety of situations, and its exact pronunciation depends on the source and severity of the discomfort they experience. We once had a wonderful cockatiel named Stumpy. He loved to have his head rubbed, and would stand on my chest, head bent over, so I could get his neck, his ears, the top of his head. Sooner or later I’d have to stop, at least briefly, to return to what I was doing before he had landed on my chest. He’d say “Ow!”, but with a pronunciation that he only used when the pleasure stopped. Stumpy wasn’t in pain when the pleasure stopped, but he was clear that the cessation of pleasure is unpleasant.
Physical pleasure and physical pain arise from the firing of very different nerves. When either of these types of nerves fire, we interpret the result by our understanding of the situation. That interpretation — our emotional state — falls on a continuum ranging from joy to misery. Stumpy was correct. The removal of pleasure is something like a modest pain.
Is your horse having fun with your training? There are some standard ways that horses try to tell us they are not happy, including avoidance (horse ignores you), nostrils widened, eyes widened, snapping tail, pinned ears. Keep your eyes open for signs that he’s not having as much fun as you think, and change your approach if you find any of these clues.
It is intuitive that getting a reward is a positive experience. So it should be intuitive that not getting a reward is not a positive experience. Not getting a reward when we expected one is probably a very slightly negative experience. If this is so, then surely training using positive reinforcement is not all positive. As the animal learns, it encounters a mix of positive (when it gets it right) and negative (when it gets it wrong.) The degree of negative does not approach the negative effect of bit or spur or whip, but still cannot be denied. I am not arguing that training through positive reinforcement is cruel or undesirable, but only that it is not quite as purely positive as some of its advocates claim.
If you get angry at your horse, other bad things can happen. If you punish him, and he fails to understand what it is that he has done wrong, you will likely get even angrier. You may punish again. He may again fail to understand… Emotions beget emotions.
I was once a jump judge at a Pony Club Cross Country event. In the distance I could see a young woman charging along, fighting with her horse. At the previous jump, her horse had refused, and she had banged him again and again with her crop. Now she was headed to my water jump, where I was sure her horse would refuse since things were going so badly for them. He stopped short of the water, and she beat him again. Then she came around to try the water again. But her approach was far too short for her horse to size it up and consider it. He refused again, she beat again. I reported her. She was disqualified. Her trainer came and took her horse away from her. In the end, the rider who punished was punished.
At a mulemanship clinic, a woman and her mule got into difficulty. Her mule didn’t do what she wanted, and she tried to force it. Mules can’t be made to do anything, it turns out. Her efforts resulted in her mule kicking her pretty hard. None of the humans who were watching seemed to approve of this kick, but I bet some of the mule observers approved. A bit later I got a chance to spend time in the round pen with the woman and her mule. Her mule was in a hard sweat, totally stressed out from his session with his owner. I explained that we couldn’t ask anything of her mule until we got him calmed down. I proceeded to do that, letting him run, rubbing him, and talking quietly. Finally, his coat was dry and he was standing still. We again tried whatever exercise this mule previously could not do. He did it just fine. And he did everything else he was supposed to do that day in that clinic. Of course, the stupid mule was not stupid at all. But he had not understood the initial instructions, and didn’t appreciate the pressure she tried putting on him. As he stressed out, his heart rate increased, his ability to understand her instructions decreased, and his willingness to obey her dropped off. All of this helped her stress out, causing her heart rate to increase32… Her relationship with her mule was damaged by his kick, but also damaged by her failure to appreciate his emotional state. In the morning, the mule owner and her mule both had similar emotions, feeling stressed and angry. In the afternoon, both the owner and her mule seemed happy. Emotions beget emotions. But negative emotions don’t have any business in your relationship with your horse.
1 Image source: https://upload.wikimedia.org/wikipedia/commons/5/59/Finnhorse_mare_with_foal.jpg
2 McGreevy, Paul D., and Andrew N. McLean. “Roles of learning theory and ethology in equitation.” Journal of Veterinary Behavior: Clinical Applications and Research 2, no. 4 (2007): 108-118.
3 Warren-Smith, A.K., McGreevy, P.D., 2006. An audit of the application of the principles of equitation science by qualified equestrian instructors in Australia. Minero M., Canali E., Warren-Smith A., McLean A., Goodwin D., Zetterqvist M., Waran N., McGreevy P. (Eds.). Proceedings of the 2nd International Equitation Science Symposium, Milan, Italy, September 2006.
4 McGreevy, Paul D., and Andrew N. McLean. “Roles of learning theory and ethology in equitation.” Journal of Veterinary Behavior: Clinical Applications and Research 2, no. 4 (2007): 108-118.
5 Birke, Lynda. “Learning to speak horse”: The culture of Natural Horsemanship.” Society & Animals 15, no. 3 (2007): 217-239.
6 McGreevy, Paul D., and Andrew N. McLean. “Roles of learning theory and ethology in equitation.” Journal of Veterinary Behavior: Clinical Applications and Research 2, no. 4 (2007): 108-118.
7 McLean, A., 2003. The truth about horses. David & Charles, Devon, UK.
8 McGreevy, P., 2006. Review: the advent of equitation science. Vet. J. 174, 492-500.
9 “Peak rein tensions recorded from riders of different levels of ability are reported to be most frequently in the range of three to ten pounds. In horses that are allowed to lean on the bit, peaks of 15 to 20 pounds are not uncommon. When a horse (or rider) snatches on the reins, peaks exceed 30 pounds.15 If one translates these figures into pounds per square inch at the level of the mouth from, say, a snaffle bit measuring 3/8” diameter by 4” long, a 3 to 30 lb peak translates to a pressure on the tongue ranging between 2 to 20 psi. If, on the other hand, the tongue is retracted and the same pressure falls on the knife edge of the mandibular diastema, the area of contact is infinitely smaller and the pressures would range from c.200 to 2000 psi or greater. It hardly bears imagining what the pressures might amount to if these same rein tensions are applied to a leverage bit that has the effect of multiplying the psi by a factor of three or four.“ From Cook, W. R. (2002). The Horse’s Bit: A Bronze Age anachronism and cause of many an idiopathic problem. In Unpublished article submitted to the Program Committee of the American Association of Equine Practitioners for the 2002 Convention, Orlando, FL. Retrieved from http://www. bitlessbridle. com/AAEP_UM_THE_BIT. pdf. Also found here: http://citeseerx.ist.psu.edu/viewdoc/download?doi=10.1.1.492.2507&rep=rep1&type=pdf
10 Rollin, B.E., 2000. Equine welfare and emerging social ethics. Animal Welfare Forum: Equine Welfare. Journal of the American Veterinary Medical Association 216, 1234–1237
11 McGreevy, Paul D., and Andrew N. McLean. “Punishment in horse-training and the concept of ethical equitation.” Journal of Veterinary Behavior: Clinical Applications and Research 4.5 (2009): 193-197.
12 Sankey, Carol, Marie-Annick Richard-Yris, Helene Leroy, Severine Henry, and Martine Hausberger. “Positive interactions lead to lasting positive memories in horses, Equus caballus.” Animal Behaviour 79, no. 4 (2010): 869-875,
13 Black Beauty, Chapter 13, last paragraph.
14 Skinner, B.F. (1953). Science and human behavior. pp. 92–3. Oxford, England: Macmillan.
15 Pryor, K. 1999 Don’t Shoot the Dog! The New Art of Teaching and Training. Ed. Bantam Books, New York; Pryor, K., 2005 Getting Started: Clicker Training for Dogs, Rev. ed. Sunshine Books, Inc., Waltham, MA, 3–7; Williams, B.A., 1994. Conditioned reinforcement: experimental and theoretical issues. Behav Analyst 17, 261–285
16 Taken verbatim from Kurland, Alexandra Clicker Training for Your Horse. Sunshine Books, Inc. 2007. 175 pp. Page 27.
17 Kubinyi, Enikő, Péter Pongrácz, and Ádám Miklósi. “Dog as a model for studying conspecific and heterospecific social learning.” Journal of Veterinary Behavior: Clinical Applications and Research 4.1 (2009): 31-41.
18 Hunt, Gavin R., and Russell D. Gray. “Diversification and cumulative evolution in New Caledonian crow tool manufacture.” Proceedings of the Royal Society of London B: Biological Sciences 270.1517 (2003): 867-874.
19 Huber, Ludwig, Sabine Rechberger, and Michael Taborsky. “Social learning affects object exploration and manipulation in keas, Nestor notabilis.” Animal Behaviour 62.5 (2001): 945-954.
20 Klein, Emily D., and Thomas R. Zentall. “Imitation and affordance learning by pigeons (Columba livia).” Journal of Comparative Psychology 117.4 (2003): 414.
21 Moore, Bruce R. “Avian movement imitation and a new form of mimicry: tracing the evolution of a complex form of learning.” Behaviour 122.3 (1992): 231-263.
22 Hayes, Keith J., and Catherine Hayes. “Imitation in a home-raised chimpanzee.” Journal of comparative and physiological psychology 45.5 (1952): 450.
23 Call, Josep. “Body imitation in an enculturated orangutan (Pongo pygmaeus).” Cybernetics & Systems 32.1-2 (2001): 97-119.
24 Herman, Louis M. “Vocal, social, and self-imitation by bottlenosed dolphins.” (2002).
25 Brauer, J., Kaminski, J., Riedel, J., Call, J. & Tomasello, M. (2006). Making inferences about the location of hidden food: social dog, causal ape. Journal of Comparative Psychology 120, 38–47.; Hare, B. & Tomasello, M. (1999). Domestic dogs (Canis familiaris) use human and conspecific social cues to locate hidden food. Journal of Comparative Psychology 113, 173–177.
26 Fugazza, Claudia, and Ádám Miklósi. “Should old dog trainers learn new tricks? The efficiency of the Do as I do method and shaping/clicker training method to train dogs.” Applied Animal Behaviour Science 153 (2014): 53-61.
27 Albright, J. D., Mohammed, H. O., Heleski, C. R., Wickens, C. L., & Houpt, K. A. (2009). Crib‐biting in US horses: breed predispositions and owner perceptions of aetiology. Equine veterinary journal, 41(5), 455-458.
28 Ahrendt, Line Peerstrup, Janne Winther Christensen, and Jan Ladewig. “The ability of horses to learn an instrumental task through social observation.” Applied Animal Behaviour Science 139.1 (2012): 105-113.; Baer, K. L., Potter, G. D., Friend, T. H., & Beaver, B. V. (1983). Observation effects on learning in horses. Applied Animal Ethology, 11(2), 123-129.; Baker, Ann Eileen Miller, and Ben H. Crawford. “Observational learning in horses.” Applied Animal Behaviour Science 15.1 (1986): 7-13.; Clarke, J. V., Nicol, C. J., Jones, R., & McGreevy, P. D. (1996). Effects of observational learning on food selection in horses. Applied Animal Behaviour Science, 50(2), 177-184.; Krueger, Konstanze, and Birgit Flauger. “Commentary Social learning in horses from a novel perspective.” Behavioural processes 76 (2007): 37-39.; Lindberg, A. C., Kelland, A., & Nicol, C. J. (1999). Effects of observational learning on acquisition of an operant response in horses. Applied Animal Behaviour Science, 61(3), 187-199.
29 Lindberg, A. C., A. Kelland, and C. J. Nicol. “Effects of observational learning on acquisition of an operant response in horses.” Applied Animal Behaviour Science 61.3 (1999): 187-199.
30 Krueger, Konstanze, Kate Farmer, and Jürgen Heinze. “The effects of age, rank and neophobia on social learning in horses.” Animal cognition 17.3 (2014): 645-655.
31 Munksgaard, L., DePassillé, A. M., Rushen, J., Herskin, M. S., & Kristensen, A. M. (2001). Dairy cows’ fear of people: social learning, milk yield and behaviour at milking. Applied Animal Behaviour Science, 73(1), 15-26.
32 Studies have found a close connection between the heart rate of the horse and handler. For example, Keeling, L.J., Jonare, L., Lanneborn, L., 2009. Investigating horse-human interactions: the effect of a nervous human. Vet. J. 181, 70-71.