How Horses Learn - Part 2

with Andrew McLean

In this article, I will describe the three forms of learning appropriate to horse training. It is interesting to note that dog trainers in particular, and animal trainers in general, understand in theory, and practice these learning strategies, and use them to make training quick, clear and un-ambiguous.
I guess it is largely the age and classical origins of the equestrian tradition which has shielded it from the modern understanding of animal ethology. It is probably also because very good trainers use these principles, and apply them, without necessarily knowing the theory.
Kyra Kyrklund springs to mind, for if you watch her tapes, you will notice how she talks of repetition, how you, the trainer, should reward every good try, and how horses learn by trial and error. One gets the impression that she has studied learning. The key to keeping the horse's brain receptive is to make as few mistakes as possible, so the horse never learns to switch off. With some understanding of the principles, you will be able to be consistent - whether the horse is a weanling being handled for the first time, or a dressage horse, learning the piaffe.
I must stress that understanding how horses learn will not help you become a better rider, unless your balance is totally unhindering the horse's ability to perform the tasks you endeavour to teach. Balance on the flat, and over fences, comes through correct position, and that is the province of the qualified riding coach.
Habituation is the simplest from of learning because there is no reinforcement. Habituation is 'getting used' to something, for example, living next to a railway line arouses the senses at first but after awhile then you shut down to the noise and can eventually sleep through it. For a sensitive animal like the horse, it is important to have the capacity to learn to desensitise to various stimuli in the environment which in the long run mean nothing, such as the rustling of trees in the wind.
Similarly, the horse habituates to the pressure of the girth and to the image of a human sitting aboard. Imprinting the foal is habituation in the form of desensitisation. It is important to remember as a rider, that the timing of the softening of the aid is critical, or the horse will learn to habituate to your signals. He may eventually generalise and reduce his reaction to all your aids.
This is precisely what a hard mouth is: a generalised partial switching off to pain or discomfort. It can often be remedied but only by changing the stimulus that produces the response. In other words, change the way the rider uses the reins, reduce the anxiety and insecurity, and/or change the bit or the feel of the bit, to resensitize the horse's mouth.
The same habituating phenomenon can be seen with regard to pain.
Psychologists call the condition 'learned helplessness' where the poor animal has had so much pain or discomfort without relief, that it ceases to respond.
Remember, when you are riding, you are training in the sense that the horse is continually adapting and forming new, or extinguishing old, habits. You should keep in mind that the effect of the strength and timing of all the cues, is in your control. Notwithstanding the huge variation in reactivity in horses, laziness on the horse's part is largely up to the rider. We deaden our horses and then blame them for it, by attributing to them the human traits of slackness and non-cooperativeness, or we teach them the fear respons, and they remain continually in a state of flight or running away.
The next learning strategy is Trial and Error learning, a form of association learning. 90% of the performance horse's repertoire is acquired by trial and error learning (also called instrumental learning or operant conditioning).
In the wild, it is by trail and error that the newborn foal finds which end of the mare contains the milk, and once they get it right, (positively reinforced) they score dramatically better each time. Trial and error learning teaches the horse how to move in the physical world in response to all external factors: taking his very first step; developing an effective escape routine during a predatory attack etc. Trial and error learning can be summed up as: if it works, do it again. So in the same way, the horse learns that the softening of an aid (the cessation of mild discomfort, negative reinforcement) means that the action on the horse's part which immediately preceded the softening will be stored and become automatic through repetition.
Try to remember to be quick to reward all the nearly correct responses in the early stages of training with firstly the softening, then stroking the base of the mane, and 'Good Boy'. Compared to classical conditioning, which I will describe next, trial and error learning is the least easily exstinguished from the memory. In other words, once learned , reactions tend to become set rock solid relatively easily with this form of learning.
With some reactions, especially those associated with fear, horses are able to learn from one trial only, but this is the exception rather than the rule, and most tasks require a number of repetitions.

When the breaker turns the horse's head to one side for the first time, this is trial and error learning, and most good horse handlers know the immediate timing required to soften the rein. Optimally you have less than a second to soften the aid once the horse gives the correct or near correct response. Remember to keep up the pressure or discomfort of the aid until the horse gives a near response, otherwise you will be rewarding undesireable behaviour. The pressure or discomfort, that is the negative reinforcement, should be either a consistent pressure, an increasing pressure or an evenly spaced one such as tap-tapping of the dressage whip. Inconsistent spaces between taps can serve to give the wrong messages.
As I mentioned earlier, it is important not to demand perfection with most tasks at the outset, but instead you should gradually shape the response. 'Shaping' allows the horse to give the response that he is physically able, and gives you the chance to get the 'ball rolling' and reward the response that most closely resembles the desired one.
Classical conditioning is the third and final category of learning. Embodied in the famous Pavlov's dog experiment, it is the means whereby secondary reinforcements are learned, that is, by association. It is where a previously meaningless stimulus becomes linked to an event. For example, if the sound of a certain motor car engine precedes the advent of food or company, then the sound of that vehicle elicits the hunger and associated motor patterns previously learned. In the wild, classical conditioning evolved to improve the efficiency of interacting with the environment. Any single event which coincides with a reinforced behavioural response will become quickly incorporated as a new cue. So the sound of a certain bird call which precedes the appearance of a predator is swiftly incorporated into the horse's stored associations. Similarly, the sound of twigs crackling in a certain way, just before the predator comes charging from the bushes is quickly learned by this means. So the new cue precedes the old cue, which is the sight of the predator.
Therefore as classical conditioning applies to training, the new cue must always be introduced first, followed by the familiar cue which produces the response. For this type of learning to occur, the horse must already have a reliable response to a primary reinforcement. For example, you are trying to teach your horse to step away from you, using a voice command. Firstly, as with all voice commands, it is important to keep the sound short, distinctive and consistent in tone. Don't add the horse's name to the command as it will only blur the learning process and deaden him to the effect of his name, if in fact, he recognises it. (Also the horse does not have the mental hardware to understand sounds with more than one meaning). The word you choose is 'over', assuming that he already knows to move away from the touch to his sides. So the sequence is:
1. Say the word 'over' once and clearly then immediately
2. Apply the hand or finger to the side of the horse:
3. The horse steps to the side.
4. Remove hand immediately you get the correct response.
5. Then stroke the mane at the nape of the neck and say 'good boy'.
Repeat the process. If he does not comply you will need to repeat the process using more intense knuckle pressure, or resort to trial and error learning by keeping the pressure on his side and increase its intensity or vibrate it, or both, until he steps over, and the moment he does, you remove the pressure. By the rules of trial and error learning, removing the pressure before the horse has given the correct response is teaching him not to comply.
It is well documented that classical conditioning works best when the new cue is before the familiar cue. It works, though less efficiently, when the new cue occurs at the same time as the familiar cue, but it does not work at all when the new cue occurs after the familiar cue.
So saying 'over' after you have already put the pressure on the sides of the horse will not teach him to associate the voice command.
To use another example we would like to teach the young horse who rushes his jumps while at the trotting over poles stage. We need to make him respond to the word 'whoa'. He hurries just a little on landing. So our sequence is say the word 'whoa' (the new cue), followed by the half halt (the old cue) ad the horse slows. This is the most appropriate way to teach the young horse the voice command. Classical conditioning is easily extinguished, so you must be accurate in the timing delivery.
Whatever task you are teaching your horse, it will fall into either trial and error learning or classical conditioning, (unless you are habituating him to something). Teaching a foal to lead, for example, requires trial and error learning - you put pressure on the head collar of the foal (generally your pull is to the side) and the moment he steps toward you - no matter how small the step at first - you soften the pressure entirely. You are, after the first few successes, shaping the response. When the response is reliable and the foal leads well, you can begin to use classical conditioning to improve the response, and to associate other cues (ie the voice) as well. Because your timing, cue presentation and reinforcements are clear, the horse will remain calmer and more attentive simply because your training is designed for his brain, and you haven't just thoughtlessly presumed he should be capable of reasoning his task as if he were human.
As the young horse learns, he also begins to sequence some responses in a chain of responses. This is called chaining. By chaining one event after another, the horse learns repertoires of behaviours which become stored for later use. The storage is in perfect form and may last for years. For example, a 2 year old broken in in say six weeks, might be turned out for two more years, yet it generally only takes a small reminder, if any, to get the response back just as it was when the job was finished two years before. If you punish a horse for, say, shying, he will simply chain the fear response after after the shy response, because of his tendency to chain rather than reason it out through connecting cause and effect.
Everything written above is all very well in a perfect world of compliant and willing horses, but there are some snags. For the horse to be sufficiently motivated to learn, he must not have conflicting drives.
For example, he must not be starving, excessively thirsty, be cold or too hot, be in pain, be in fear, be compromised by sexual factors or be dominant. The last one is the greatest stumbling block for most inexperienced trainers. You must have the horse's respect; he should be placidly obedient. Remember that the horse does not mind you being above him in the hierarchy. In fact he is just as content as he would be if he were the dominant one.
The scariest thing of all for the horse is not to know who is boss. This produces considerable mental conflict for the horse, and leads to all sorts of neurotic behaviours emerging including the fear response, which can be manufactured as an evasion. Natures way of resolving the peck order conflict is to increase the level of rebellion, till somebody does something which finally sorts it out and restores the harmony in the peck-order. Psychological dominance is best achieved by teaching the horse to yield in as many ways as possible. This is the Admiral feature of the Jeffrey method, the Parelli training system and many others, control is achieved without violence-teaching the horse to yield, invading his space, and achieving his 'focus of attention'. for example teaching the horse to lower his head by pressure from the top of his neck will assist in getting the horse's respect through postural submission ( head down is the universal cowering response) and he will become quieter as well. Head down is the universal sign of submission (as in cowering), and this is the powerful effect of riding the horse 'long and low' too. Although trainers continually espouse the muscle effects, the psychological effect is considerable, for it is responsible for the relaxation that comes through submission. The horse which is dominant over you will always look away when you stand in front of him. He is telling you that you are not important to him, yet it instinctively makes him nervous because the hierarchy is in conflict and you are not the leader to be relied upon. Achieving his focus of attention gains his respect without violence and his contentment with the restored hierarchy will be shown by his sometimes quite sudden calmness.
Under saddle, teaching the ebullient horse to go 'long and low' will assist in getting the horse's respect through postural submission (head down is the universal submission posture) and he will become quieter as well. This is the powerful effect of riding the horse 'long and low' too. Although trainers continually espouse the muscle stretching effects, the psychological effect is enormous, for it generates the relaxation of the muscles. Maintaiing consistent flexion is also a means of obtaining submission, as it is often the first aspect of cooperation that is lost as the horse begins to look about.
It is important that fear and learning are never associated, not only because rising levels of adrenaline inhibit learning, but also because the greater the association with the flight response, the more 'trigger-happy' the flight response becomes. Fear, and fear reduction, whether they be real or evasions, are the subject of another article.