Clicker training is one of the recent success stories of equestrianism. It makes use of a bridging signal to indicate the moment of the desired behaviour, followed by positive reinforcement. We are told that training with positive reinforcement is more ethical than training with negative reinforcement and/or punishment. We are told that positive reinforcement activates the pleasure circuits of the brain, releasing dopamine in a way totally distinct from the regions activated by techniques involving pressure and release. As clicker trainers we are adept at handling the various erroneous criticisms by sceptics - that horses in the wild do not use positive reinforcement, that hand-fed horses will be encouraged to bite, that understanding behavioural science predisposes us to being unfeeling scientists who can't work with practical behaviour. We have horses who appear to engage in their training enthusiastically, sometimes they even don't want us to end the session. It is just one long string of clicks and treats for us!
So what's the problem?
Firstly there is the perception that clicker training can only be positive. We are giving a horse treats which is better than him having no treats. Therefore it is good. This is a somewhat simplistic view. Skinnerian stimulus-response chains do not take into account anything about the horse's lifestyle and environment. In fact, Skinner seemed even to deny that they were relevant. If a horse pulls faces when you put his saddle on then you can clicker train him to make a happy face instead. If a horse won't stand still in his stable you can target train him to stand motionless while you do things to him. You can train him to adopt dressage postures. You can train him to move at gaits that would require more advanced training if taught conventionally. You can train him not to respond to all manner of scary objects. You can even train him to lie down, permit you to lie down with him and take a great photo for your website. And so much more....
The trouble is that none of these training situations take into account the underlying reasons for the behaviour. The poorly-fitting saddle may be causing pain. The stabled horse may feel worried about a neighbouring horse. He may not have the right musculature to adopt the requested positions or perform advanced movements. He may learn to tolerate the scary objects but what if his fear of them is still greater than the pleasure of the treats? And lying down is all very well if he wants to do it but what about when the ground is hard or there is something in the vicinity which means he'd really rather not?
But horses wouldn't do it if they didn't want to?
This is the age-old question. It has been (and is) said of race-horses, show-jumpers, riding school horses, horses trained with natural horsemanship techniques and even the original process of domestication approximately six thousand years ago. Of course, these forms of horsemanship all include aversive stimuli, both physical and emotional, which provide some level of threat to the horse - "choose to do as I say, or else". So the horse complies, apparently willingly, and the aversive stimulus can remain invisible to all but the most perceptive observer.
Clicker training is different because we are providing something pleasurable for the horse. We are absolved from guilt. Or are we? Domesticated horses have had a lifetime of complying with our wishes and they continue to do so when we pick up a clicker. The rules may have changed and we may be permitting the horse to offer a behaviour before confirming that it is the correct behaviour, but it is still the human who decides whether it is the correct behaviour. We want the horse to choose to offer behaviour spontaneously but it has to be the "right behaviour" - such mixed messages bestow a lot of emotional pressure on an animal who has previously been so well-conditioned to do as intstructed. It is like having "creative thinking" or "independent learning" timetabled at school (as indeed occurs these days), as though autonomy can be switched on and off. Good trainers who understand how to use variable schedules of reinforcement are then able to extract more and more behaviour out of the horse in return for the reward. This "Brave New World" of horse training can often be so blind to what the horse would really choose.
And then we have repetition. Just in case the horse is in any doubt as to who is calling the shots, some trainers seem to feel the need to train a behaviour over and over again. There seems to come a point where any pleasure circuitry triggered in the brain by the treats is more than compensated for by the conflict behaviours seen in the horse - the frustration and aggression, the sexual over-arousal, the boredom, the conditioned suppression, the worry. And the reason for this repetition is typically the perceived need for the horse to respond "less emotionally" or more "cleanly". So our goal has become something coming dangerously close to the shut down automatons of some of the more aversive training methods we have tried to leave behind. What is going on?
The trouble with clicker training is that it is incredibly powerful. The trouble with horses is that the majority are very compliant because they wish to avoid conflict. It is very easy to evolve inadvertently from a novice clicker trainer, who wants to help her horse become more enthusiastic and have a more enriched life, to a more advanced clicker trainer who is looking for perfection and control and has rather forgotten why she started clicker training in the first place. I have never met anyone who actively clicker trains her horse because it is such a good way of exerting her authority. Yet that is so often how it has become. That desire to become a better and more achieving trainer just cannot help getting in the way of what is important to the horse. Yes, with clicker in one hand and treats in the other, we can become over-controlling, aversive stimuli who are actively, albeit inadvertently, working towards reduction of our horses' autonomy and, hence, welfare.
And we haven't even begun to talk about combining clicker training with negative reinforcement and punishment - that was the subject of a previous article http://www.theequineindependent.com/home/?p=234) so I shall spare you that this time.....
So what do I like about it?
Despite all these concerns, I really do rate clicker training very highly and would love to see it taken up by more people. Positive reinforcement (with or without a clicker) allows us to interact with horses in a way to which no other training method even comes close. But in order to tap into this wealth of potential, we really need to change our focus. We need to start again and look at what attracted us to clicker training in the first place.
When starting clicker training we tend to offer a neutral target; either through natural curiosity or by accident the horse touches it. He hears a click and receives a reward. After a few repetitions we see that incredible "light-bulb moment" as the horse works out what is happening. The horse realises that he can turn the human into a vending machine - it is the moment of a surge of self-confidence, empowerment and autonomy. As horse-loving owners/trainers we are hooked from this moment onwards. It is why we wanted to clicker train, we liked seeing our horses so happy and expressive. We liked the moment of being able to read our horses' minds. I like clicker training when we stay in this place, when we don't move out into the world of training behaviours just because we can, or over-training, or worrying about excessive stimulus control or trying constantly to deal with so-called behavioural problems.
When engaged in a simple free-shaping session, such as this, we are conveying a very powerful message to the horse. We are saying that he can choose to participate or not (even better if the session is in the field so grass is always available as an alternative to training). We are saying that he can earn rewards or opt not to earn rewards and nothing bad will happen, whichever option he chooses. We are saying that we will respect the decisions he makes, rather than trying to find alternative ways of obtaining compliance. The horse choosing to say "no" is not a slur on our training or on our relationship. It can be a sign that he is in good psychological health and feels sufficiently secure in his relationship with the owner that he can say "no". After previous years of being conditioned to do as he is told, learning that he can opt to do or not to do something is incredibly liberating. When we turn clicker training into something bordering on authoritarian, we lose the most enlightened element of it - the opportunity to reinstate the horse's autonomy. This is where clicker training has advantages in its ability to increase welfare; any technique using pressure and release cannot increase a sense of autonomy.
Despite being a strong advocate of positive reinforcement, often to the point of being misquoted as attempting a route of pure positive reinforcement, I have come to believe that autonomy is perhaps the most beneficial gift we can incorporate into our training. When positive reinforcement training is controlling and manipulative it erodes autonomy and diminishes the value of the rewards - it becomes a poisoned cue in itself. Horses have evolved to make many decisions for themselves - the erroneous idea that the majority of horses just blindly follow a leader is outdated - and there is no reason for this to have changed over the relatively brief period of domestication. Yet the vast majority of domesticated horses have no say in what they do when, are fed a prescribed diet at specific times and have no choice as to their companions. Indeed, the manner in which most horses are managed is contrary to even the most basic ethological time-budgets.
I do not pretend to use positive reinforcement all the time, but I reserve it for when I want to encourage my horse's autonomy, alongside careful consideration of his evolutionary needs. I will use discrete and well-defined free-shaping sessions to reinforce the message that I will listen to my horse's opinions. This is not to say that I will never over-ride my horse's opinions because sometimes I do - afterall, none of us has autonomy 24/7 - but within a free-shaping session it is all his choice. The balance needs to be found where the horse has the self-confidence and trust in the owner that he can offer opinions confidently without feeling "shut down" if the opinions are over-ruled. I don't use clicker training to train away problems or to train behaviours I actually care about training. I use clicker training to build a sufficiently strong relationship from which I can later use mild negative reinforcement when I feel it is appropriate. Obviously it depends very much on the horse as to how much of a balance must be struck between the need for free-shaping sessions and the appropriateness of incorporating mild pressure. In the early days of working with a new horse it may be that every interaction needs to be the horse's decision. The long-term shaping plan will include being able to cope with direction from the human.
Free shaping allows the horse to behave in the most open and honest way, rather than just trying to avoid pressure whichever way he can. It is a means of communication, two-way communication as opposed to formal training. As a result, we are provided with the closest insight as to how a horse might be thinking. We can use this information to improve the life of the horse - we can learn about his learning style, what he likes and dislikes, how he values things, what he feels scared about. We can apply this information to any form of equestrianism in which we wish to participate - not to exploit and manipulate but to add value and reduce conflict.
I strongly believe that this approach to horsemanship is analogous to some of the methods used in human psychotherapy, most notably, the person-centred style of therapy pioneered by Carl Rogers (e.g. On Becoming a Person). There is also a beautiful description of such therapy applied to a six year old boy, thought to be mentally deficient but given the opportunity to develop a positive relationship with play therapist, Virgina Axline, and transform into the highly intelligent and advanced boy he was (Dibs: In Search of Self). This book shows the power of free-shaping in action and is remarkable for so many reasons, not least because the therapy took place for only one hour a week with the boy returned to a fairly aversive home life in between. Rogers believed that a therapeutic relationship hinged on three key factors - empathic understanding, genuineness and unconditional positive regard. While his earlier work studied the relationship between therapist and client, he later extended it to just about all relationships. I see no reason why this should not apply to horse-human relationships as well. Working with a troubled horse requires these same three attributes - an understanding of how that horse might be feeling, the patience to allow that horse to behave how he needs to behave without trying to manipulate or creating an agenda and respect and appreciation for every try that the horse makes. I think it's fair to say that no equestrian discipline has these core points at the heart of the horse-human relationship. Yet.....
Copyright Catherine Bell 2013