Positive Reinforcement and Clicker Training - Dispelling the Myths

Horses use punishment on each other but they don't give each other clicks or treats - CT isn't natural

We've all seen horses fighting, sometimes quite viciously. Yet we need to put the herd behaviour of domestic horses into an ethological context. Wild horses live in herds, typically family groups, in large open spaces. Their survival depends on their ability to detect and flee from predators. An injured member of the herd will attract predators - why would they hurt each other?

In the domestic situation we force unfamiliar horses together into small enclosed fields. We restrict their feed and water and we take horses in and out of the herd as it suits us. Is it any wonder why they need to compete with each other and aggression levels increase?

Aggression certainly takes place in the wild just as it does in domestic situations, typically between stallions but sometimes between the remainder of the herd. Often this is actually play behaviour which we mistake for fighting. Other times it is genuine aggression but it has been preceded by the more subtle signs horses give each other to indicate that something is not quite right. They have lots of warnings of the punishment before it occurs. Even in domestic horses the punishments take place relatively rarely, perhaps a handful of times a day at the most.

So for us to make punishment a frequent component of our training methods is not particularly natural. In a typical dressage lesson, or even on a hack, we might punish our horse every few minutes, if not seconds - a very sobering thought. This form of punishment might not be severe but it cannot be called natural.

How about positive reinforcement? Can that be considered any more natural? Horses are peace-loving, sociable, flight animals who want to graze for most of the day. They engage in mutual grooming, they stand guard while others are lying down, they play with each other, mares suckle their foals, they swish their tails to remove flies from each other, they provide companionship for each other. They might not click to each other (this is just a convenient means of bridging the communication gap between our species) but they certainly use considerably more positive reinforcement than punishment. Positive reinforcement is indeed natural.

Clicker training teaches horses to bite

Many people are (understandably!) worried about CT because it entails food rewards. Some horses will indeed attempt to nibble at you and so it may seem counter-intuitive to start food-based training.

Actually it is by using the food-rewards that we can teach the horse not to mug you. By allowing him to explore his environment and find out which behaviour earns him a treat, he will learn the rules of the game. Mugging gets no treat. Touching a target (or whatever) gets a treat. Horses are pretty intelligent - if he wants more treats then he is much more likely to repeat the behaviour that earns him treats. It is your job as the trainer to make sure you never inadvertantly reward the mugging behaviour - if you do then that is what you will get!

Of course, to start with you need to keep yourself safe. You might feel the need to start off with the horse behind a stable door until he has learnt not to mug. You may need to wear a few layers of clothing. When you progress to outside the stable you need to bear in mind that by changing the environment you may get some mugging again. That is a normal part of learning but it will not take long for the horse to learn that the same rules apply, whether inside or outside the stable.

Clicker training is just bribing the horse to do things

Bribery tends not to work. We've all seen someone standing at the top of a trailer ramp with a bowl of feed and the horse standing at the bottom of the ramp, neck stretched out as far as it can reach but not actually moving its feet. And then the owner says "I've tried training with food rewards but it doesn't work". In this example the horse is essentially being rewarded for standing at the bottom of the ramp. So that is what it does!

When training with positive reinforcement the reward comes after (or simultaneously with) the behaviour. That way the horse knows which behaviour to repeat in order to earn the reward. It is thinking about which behaviour to give us. Bribery takes place before the behaviour, trying to entice the desired behaviour. But producing the food early confuses the issue and the horse is more likely to focus on the food than the behaviour.

It's "horses for courses" - just because clicker training works for one horse doesn't mean it works for all horses. Different training methods suit different horses.

I've always felt that "people for courses" is more appropriate, but unfortunately this doesn't rhyme as nicely! Having said that, CT does not suit every horse, owner or combination.

In her book, "Don't Shoot the Dog", Karen Pryor describes the eight ways of changing undesirable behaviour: separation, punishment, negative reinforcement, extinction, changing the motivation, training an incompatible behaviour, putting the behaviour on cue and not giving the cue, rewarding the absence of the behaviour. The majority of training methods depend heavily (although not exclusively) on punishment and negative reinforcement. Training with positive reinforcement, with or without the clicker, gives us various other options. It is not a single method but an approach to life, whether we are dealing with horses, dogs, children, spouses or bosses.

It all sounds very nice but it can't work in practice. Horses, like people, need boundaries and discipline

I think we all agree with this one. But rewarding the desired behaviours and ignoring the unwanted ones doesn't necessarily mean that you end up with an unruly horse running riot. Learning how to set boundaries with positive reinforcement is taking things to a more advanced level but it is an important skill to learn.

Again, we turn to the eight ways of changing behaviour. We have many cards up our sleeve. For example a horse that barges can be rewarded for standing nicely (incompatible behaviour) so that he gradually learns that walking nicely (reward the absence of barging) is more beneficial to him than barging (changed motivation). We need to start using our imagination in our training so that we don't need to resort to punishment the minute things don't go to plan. Some people might ask their barging horse to back-up a few strides each time he barges, they might have been doing this for months or even years - the punishment doesn't seem to be having the desired effect then! Has the horse really been disciplined?

Riding uses negative reinforcement and punishment so what's the point in using positive reinforcement on the ground?

Traditionally riding has used negative reinforcement and punishment. There is no reason why positive reinforcement and CT cannot be brought to the saddle, just as it is used on the ground. It is all about using your imagination.

Clicker training is just pointless tricks

We seem to have become so used to riding horses, lungeing horses, tying horses up, shutting horses in stables that they are deemed acceptable; yet rewarding a horse for (for example) touching a target or (God forbid!) bowing is considered demeaning, degrading or pointless.

A CT horse has the choice of performing a behaviour and earning a treat or not doing it. If it were to find the behaviour degrading then it could choose not to participate. You could argue that the desire to have the treats over-rides its desire not to be demeaned (assuming of course that the concept is not just purely anthropomorphic!). In which case we would need to consider what a horse might find degrading.

Horses are sociable flight animals who have evolved over 60 million years into athletes. Yet we regularly shut them up in small boxes, we keep them by themselves, we feed them inappropriate diets, we put concussive metal shoes on their feet, we cause them to suffer with pathologies they would not have in the wild (e.g. COPD, navicular), we clip them, we trim their whiskers, the list is endless. Is not touching a target or learning to bow in return for a food reward infinitely preferable from the point of view of the horse?

Of course there are some behaviours that would be unethical to request, particularly in the case of a horse who would do anything for treats - CT can be done well or it can be done badly and the physical and emotional well-being of the horse should always be paramount.

I would argue that it is not pointless either. These basic CT exercises are excellent for teaching the horse that click means treat. They are excellent for encouraging the horse to think for himself and make decisions, rather than becoming a robot who responds only to cues. They encourage a horse to explore its environment and offer behaviours, not have them shut down. They increase the horse's confidence because he does not fear the consequences of his actions. How could this be considered pointless?

Will clicker training make my horse trust me?

Nope! No training method will make a horse trust you. Your horse will trust you if you prove yourself to be a trusting and trustworthy person, consistently and over time - CT can be a part of this. Trust won't come instantly, it won't come through sending your horse away to a professional and it won't come through dominating your horse. If you think your horse doesn't trust you then it is highly likely that it is not the horse who needs the training......

CT seems too mechanical - I prefer to work with "feel"

This is a frequent criticism of CT - who wants to be learning the science of operant conditioning when horses are thinking, feeling animals who live outside the laboratory?

Unfortunately, like it or not, operant conditioning is all around us. A horse will forage for his food (positive reinforcement), find some shade when he is too hot (negative reinforcement) or make grouchy faces at a horse invading his personal space (punishment). They use variable schedules of reinforcement, they display extinction bursts and spontaneous recovery. They habituate to stimuli and succumb to conditioned suppression through inappropriate training. If we understand this behaviour then it can be invaluable to us. It helps us understand one of the most important factors in our training - the reason why. Attempting to change our horses' behaviour without knowing the reason behind the behaviour can be very dangerous, either physically or emotionally. We need to understand why they behave as they do.

Of course, horses don't read the text-books and there are exceptions to every rule. And this is where the feel comes in. The feel element is what still remains after we have accounted for all the bits of science. It is where training from the brain meets training from the heart. It is harder to describe, there are no guidelines and it only comes with experience (with the rare exception of those who just have that extra "something" naturally). But without the foundations of behaviour in place it is too easy to confuse "feel" with anthropomorphism, misread behaviours, miss the reasons for behaviours. "Feel" so often becomes an excuse to dominate and such trainers can often be seen doing something very different from what they claim to be doing. (A useful exercise is to watch videos of different trainers but with the sound turned down - do the horses look happy or confused?).

I always find music is a useful analogy. If you can't understand the music then you will struggle to play a piece of Mozart, however much you want to play with feeling. Yet if you persevere with the music theory and you work out where the notes are, then you will be able to play the basic tune. As you become more experienced the feel will take over, you will no longer need to concentrate on the music theory and there will be beauty in what you are playing. The music will sound as though it comes from the heart.

Whether we are dealing with equine behaviour or music there is theory all around us, it is unavoidable. But when we build on those foundations and find the balance between science and feel then we will truly be able to work with our horses rather than against them.

Copyright Catherine Bell 2004