Andrew McLean on Punishment

One of the most important concepts Andrew McLean has introduced to the world of the horse is that horses are not ‘naughty’, they are just being given the wrong signals and message. Now Andrew is working in a new direction, the dimension of ‘affection’ between horse and handler, and how that opens up new thoughts on horse training…One of the most piercing examples of the belief that the animal really knows what it ought be doing, and is just being ‘wicked’ by not doing what is required, were the methods Andrew found in Asia for training elephants.

“There is a religious belief that the Hindu god Ganesh gave the people the elephant to serve them and all they have to do is make it submissive. Elephants are far more dangerous than horses, and the death rate of mahouts is far higher than it is for horse people. So they tether the young elephant in various ways and torment it with sticks and spears until it doesn’t fight anymore, and they withhold food, and sometimes water too, so that it’s further weakened. After some days, they then haul it between two large elephants and a rider hops on and gives it all the signals it’s supposed to react to, but he doesn’t think he’s training it, he thinks he’s making it do what it already knows and because it’s been hauled along, it vaguely learns the signals over some days and weeks.  When he puts on the brakes on the hauling elephants, they stop, so the elephant that is being hauled does too, so it inadvertently learns all the cues, but in such a crude way.”

“Sounds crazy doesn’t it? But when you look at sophisticated, civilised horse rider / trainers and their interaction with their horses, sometimes, it doesn’t look all that different… People often believe that the horse has insight into his behaviours, that he knows what he has just done, that he has a concept of right from wrong.”

“Often I think the biggest problem with horse owners is just that they believe the horse does know what he’s meant to do. I think that’s a big mistake, I think to be fair you should apply the precautionary principle: if he doesn’t do it then he doesn’t know what it is that you want him or her to do. And if you are consistent as a trainer, then he will learn to do what you ask, and he will know it and do it when you ask. Tom Roberts was my greatest mentor as a trainer. He was the first to describe learning and reinforcement and put it in easy terms. Whatever the horse is doing will either ‘profit him’ or ‘profit him not’. If you are careful to ensure the horse profits from doing the right response, and you are consistent, the behaviour will become ‘old hat’. Actions transform to habits: simple. ”

“There is a great variation amongst horses and some are always going to be more difficult than others and I think that’s why they all need careful use of pressure / release in early training stages. I believe the main difference between the so-called good horse and the bad one, is a result of what we reward in the early stage of training. The ‘good’ one lucks on the right response and is rewarded for it. The ‘bad’ one offers the wrong response and the handler adds to the problem by rewarding the wrong response. If the horse does something we don’t want, say goes to rear up, and we release the pressure, then we are rewarding the rear. Or if he rushes backwards, and we release the pressure, we are once again, teaching him to do what we don’t want him to do. We must be sure that we only reward the behaviour that we want.”

“For example, the origin of the so-called ‘bad horse’ begins when someone first touches the foal on the rump with a hand, and the foal kicks up and runs off. The foal has learned an important lesson: when humans touch you, you escape and it works, you remove the human. Then when the same horse is being trained to lead or ride and it gives the wrong reaction and kicks out or shoots backwards, and the rider releases the pressure – once again the handler has inadvertently taught the horse the wrong thing. From the horse’s point of view, he’s doing all the right things, he’s learning in exactly the same way as the ‘good’ horse who happened to go forward the first time the rider squeezed him, he’s the horse that’s on a ‘good’ path and everyone will like him while the other horse is the ‘bad’ horse, but really they are both reacting the same way, it is just chance that one is regarded as ‘good’ and one is regarded as ‘bad’.”

“I think that the assumption that the horse knows what he’s meant to do and that therefore you can punish him for not doing it, is a really big mistake, but it’s one of the most common reactions. The good trainer sits back and thinks, ‘what is the training goal, and how can I best achieve it and how would I use my tools of Operant Conditioning – negative reinforcement, positive reinforcement or punishment?’ Even if you’re using Negative Reinforcement – which is just pressure/release – at what moment do you release for what behaviour and how do you shape it by training one variable at a time? I think that’s the most important aspect of the early training and developing a horse to be a sensible horse.”

story continues below the advertisement

“We have to make sure all our interactions are clean. Even our first time we touch the foal, what is he learning when he does a little buck and bucks your hand off and runs off. Does it matter? How formative are first encounters? One of my biggest criticisms is that training books and dressage texts from antiquity rarely mention how to train a response, they all focus on rider position, not the actual tools of training. Position of course is essential in the delivery of clear signals, but it doesn’t train.  Even the German Training scale is doesn’t offer much here. It begins at rhythm/relaxation. At this level the horse already has learned the response, rhythm is just perfecting it and relaxation is an essential characteristic.  Think of piaffe, what you need to know is how to set it up, how to get a basic response, how to get a single step of it from an aid and then how to get a smooth flow of it, which is rhythm.”

“I think the most important thing to recognise is the horse isn’t an animal with high executive mental functions. He’s intelligent at being a horse but he never needed to outwit his dinner, or face the mental challenges of fruit eaters, or use tools.  Before people start writing in to enlighten me about the clever horses that open gates or even let other horses out, yes it’s clever, but it’s Operant Conditioning, not reasoning.  No matter how we rig and re-jig the experiments, horses don’t stack up to the level of dogs or chimps or dolphins. And unless you’re given to fanciful notions or creationist theory, he isn’t a willing-to-please animal either, that would be far too convenient.  At any rate, to be fair in training another species, we should err on the side of caution and be careful not to presume that we can make leaps of faith in training and just do what we’d do in teaching other humans.

The most important things to do, even with a horse who has had a bad early experience, are:
– to get him very clear on the ground so it doesn’t infect under-saddle work
– to make sure we’re aware of the signals we want to teach him
– to make sure all the signals are different to each other so he can discriminate between them
– to do our best to make sure they’re not overlapping, because some aids will be a little bit close to another kind of aid e.g. to make the horse come ‘round’ from the hands or doing one-rein stops can clash with the earliest stop and turn responses respectively, so we need to rethink these.
– to make sure there are no exceptions to the rule, because we can’t expect animals that don’t have higher mental abilities to handle exceptions.

“One big exception to the rule is – and I’ve learned this only more recently – when you lead horses don’t let them learn to follow your feet. I know it is standard in so much of the horse industry, it all looks rather nice and impressive when the horse does follows the handler’s feet, but the trouble is if the horse sees your feet as a signal, and you tie the horse up, or put it into a float and you walk off, in those situations the horse is going to want to go with you. That’s a big part of the origin of fidgety, pawing behaviour in horses and ones that have problems in the trailer. It’s not true for all horses, some cope with the exception to the rule, but even the ones that cope it is not that they have necessarily learned the exception – they just lower the threshold of responding to humans in all areas or they engage in out-of-the-blue, unreliable behaviours.”

“Horses are really fundamentally clear on how they relate to each other, and I think if we made ourselves as clear, by for example, giving very clear leading signals, they would be as alert to us as they are to other horses.”

“I teach people to press the horse forward with the lead rein before they step forward, and then if they want to add another cue, for example ‘walk on’ or Georgia Bruce uses an arm raise, any of those cues are good idea, but we must make sure we don’t step first or stop first with our feet or else we’re embedding a small but solid brick in the wall of confusion. When horses are on clear lead signals that never deceive, including teaching them to stand still, they’re just so much calmer. They are not worried what your feet are doing. I’ve transformed many nervous horses by just being clear. ”

“I’ve since used that technique with the elephants as well. It’s really valuable in getting a sensible animal from one that was previously confused. These are really important principles to begin with in whatever system of training you follow, to make sure it is clear and consistent and the cues aren’t overlapping. They’re distinct and that there are no exceptions to the rule.”

story continues below advertisement

And one rule to which there are really no exceptions is that punishing a horse in anger is a complete no no…

“One thing we know is that violence is fairly reinforcing for humans. It somehow makes you feel like you’re solving the problem when you use a bit of violence. When you have a hammer in your hand, all your problems start looking like nails. And one of the things we know about punishment is that it can create fairly insecure relationships between the punisher and the punished, so that’s a problem for the future when you think of the horse’s fantastic memory, especially for fear. You really should avoid punishment and fear behaviours as much as possible when dealing with horses. Punishment lowers an animal’s tendency to try new learned responses, so in difficult situations when you’re training something a little more complex, the horse is less likely to try and more likely to stress. Just like a child in a classroom who’s been punished or humiliated, he’s not likely to put his hand up even if he thinks he might know the right answer.”

“The difference between Punishment and Negative Reinforcement can seem to be a blurry line to the amateur behaviourist. The difference is that Negative Reinforcement promotes a behaviour, so the pressure is released when you get the behaviour you want. You try and engineer the correct response, whereas punishment is designed to delete a behaviour or to remove one, so if the horse kicks you whack it… it’s easy to overdo it. I’d rather find out why it kicks or bites – they’re almost always associated with poor/heavy stop rein responses (they can’t trot/halt from a light aid in 4 steps of the forelegs, or step back from a light rein aid) and sometimes a poor go response.”

“I have a good diagram that describes Operant Conditioning.”

“With this diagram it’s really just showing how the four forms of operant conditioning can be understood i.e. Negative Reinforcement, Positive Reinforcement, Positive Punishment and Negative Punishment. If you look at the horizontal axis, you have on one side something the horse doesn’t like and on the other side something he does like – attractive and unattractive stimuli. Unattractive stimuli could be leg or rein pressure and attractive stimuli could be food or caress.”

“On the vertical axis we have a response we want to make more likely (top) or less likely (bottom). Positive means adding, negative means subtracting. So now we have four quadrants. Positive Reinforcement is using attractive stimuli to make a response more likely (e.g. using clicker training to train piaffe); Negative Reinforcement is using unattractive stimuli to also make a response more likely (e.g. softening the leg aid when the horse goes forward); Positive Punishment adding an aversive to delete a behaviour (e.g. the horse gets whacked with the whip for kicking) and Negative Punishment is removing something attractive to delete a response – we walk away from the float when the horse paws.”

“We’ve also superimposed over this diagram our latest research areas. At the University of Sydney, Melissa Starling, Prof Paul McGreevy’s PhD student is researching how arousal (alertness) states and affective (emotional) states interfere with learning. For example, it would be futile to attempt to train a horse when it has been cooped up all night in the stable, so hyper-aroused and all it’s motivated for is exercise. Similarly it may be very difficult to get a reliable dressage test from a horse that has had a fearful experience at a certain venue due to its high arousal level and negative emotional state.

Of course, trainers have always known that arousal levels and emotional have an effect on learning. The most interesting thing about this research though is that we are on the brink of measuring it and making accurate predictions about the usefulness of a certain type of training such as negative reinforcement or positive reinforcement or punishment which all have limitations as a result of inhibitions via arousal and affective states. My interest has been in the area of Attachment theory which is the other big brick in the wall. It’s about drilling down into the bond between human and horse.

This article first appeared in the October 2013 issue of THM.

Want more articles with Andrew McLean? Here they are:

McLean, Andrew