LATEST THOUGHTS
Why it goes wrong with CT
Why it goes wrong
Perhaps the powerfully motivating nature of positive reinforcement and especially of food rewards means that any unwanted behaviours we inadvertently train are much more obvious to see as they develop quickly and the horse offers them freely in an attempt to be rewarded. In more negative based training methods, horses are usually required to respond to the application of pressure, and if there is no pressure then there is no requirement for them to offer behaviour freely. So unwanted behaviours are only seen on the application of pressure, and labelled avoidance.
With CT the horse is liable to offer behaviours randomly, out of context and without warning in the hope of being rewarded, making CT unwanted behaviours much more visible and likely to be on a variable schedule, as trainers often reward on the “oh, that’s nice, I mark that principle”.
One of the problems clicker training is that people are learning, that is normal, in the process of learning anything we go through the path of unconscious incompetence, conscious incompetence, conscious competence, unconscious competence, and the normal process of learning is to make mistakes and to a learn as you go along. However, the extremely motivating nature of food rewards and the potential of clicker to mark specific behaviours very accurately means as you learn and make mistakes you can mark behaviours which you really don’t want and as a learner, before you know it, the horse as acquired significant unwanted behaviours, which are very hard to remove. CT seems a particularly unforgiving method to learn if you have a horse that is not ideally suited to the method or as a trainer you need more time to learn and comprehend yourself.
Just as with any other method who we are as a person effects who we are as a trainer and this is fact is the biggest single cause in the variances that we experience in the success or failure of CT as a method
People often loose their self awareness when they take up clicker training, they are so focused on the click and trying to spot and reward the desired behaviour they forget about their own body language and behaviour and their effects on the animal and training.
One of the things that seems to disappear when we start clicker training is our feel. We are so focused on finding a behaviour to mark, learning the process of clicker training or concentrating on our shaping plan and rewarding behaviours that actually we lose the ability to stop thinking and just feel right behaviour. When I am picking up a foot without clicker training I feel the moment to put it down, when I will ask horse to step forward and there's pressure on the lead rope, as they move the feel tells me when to release the pressure the same sense of feel has to have to take place during CT training, but often our focus is changed and we think too much. When we think which behaviour to mark, our conscious thinking is not quick enough, and our timing suffers, we miss behaviours we want. We have to grasp a sense of feel and let our subconscious mind choose the behaviours that we see to mark. This is setup by having a very accurate and well thought out shaping plan which allows us to just feel instead of thinking and of course by experience.
Trainers often are not as specific as they think they are being and inadvertently mark to or three behaviours trying to teach just one. E.g, in trying to get the horse to pick up and pass a object, such as a hat, to them, what they often mark is shaking the hat, dropping the hat, tossing the hat, because they are not specific about what they need. In the trainers mind they are trying to transmit a concept, passing the hat to human, so with this concept in mind the trainer marks anything that looks like that. The horse mean time is working on the tiny specifics and actual behaviour, shaking, dropping etc. What is needed is a clearly defined very simple program of teaching the horse to hold the hat still for longer and then gradual succession towards the handler. Horses are very poor at insight learning they rely on specifics. You have to be specific about what you are marking so you have to focus on what you want not what you don’t want.
Trainers often struggle to withhold the click. In their enthusiasm to mark behaviours there is a tendency to click too often for a variety of variable behaviours in the hope that they are shaping behaviour and animal will be able to work out what it is to want them to do. However, too many clicks end up confusing the horse, what the horse needs is very specific, consistent markers that tell them exactly what behaviours are required.
Withholding the click isn't just about waiting longer, it's about knowing which behaviours to reinforce and when to click and when to wait. Withholding the click helps the animal eliminate behaviours that are not successful but our tendency as we grasp firmly the clicker in our hand we want to click anything in the right direction and that silence is often difficult for us to sit with. With good timing and consistently good shaping the silence becomes information and helps the horse to learn. But wait too long and don’t shape well and the training breaks down and frustration and impatience sets in.
People start training random behaviours that they would never do if they didn't have the clicker. so not knowing what else to do they start to reward crossing feet over, shaking “hands” and random behaviours that are ill thought out, without a shaping plan that often lead to other behaviours and difficulties. If you didn’t have clicker training you probably wouldn't try to teach your horse to pick up the hat and pass it to you or to touch a target, crossing it front feet or to do any number of other strange activities, which are very far from the horses’ normal and natural behaviours. Without a deep understanding of the science behaviour, shaping, comfort zones stretching, systematic desensitisation its easy to get lost in the enthusiasm of teaching new behaviour. Train what you really need and don’t just teach stuff that seems like a good idea at the time.
What you see is what is being rewarded. Whatever the horse is repeatedly offering is a general indication of what the horse thinks works. This simple observation will tell you what the horse thinks is being rewarded rather than what the trainer thinks is being rewarded.
It is important to observe this, so as you start to see small behaviours such as, the turning of a head or the twisting of a body or offering a foot, indicates that this is what the animal thinks works. We are to busy thinking with our minds rather than the horses. Trainers often miss this simple fact and so fail to make adjustments to what they are marking before the behaviour has become established. Trust the horse what you see is what your marking if it is not what you want change it.
With positive reinforcement the idea is that the animal offers the behaviour themselves in an attempt to receive a reward. This is often with less direct information than might be available say with negative reinforcement. So, when we pull (gently) on the head collar the horse has limited options, if they move forward the pressure is released, there are a more limited number of ways to receive the reinforcement, compared to the vast range of options available to a free thinking unrestrained horse during free shaping, so the free shaped CT horse is likely to have to think more, offer more behaviour and the trainer will have to be better at shaping.
We humans are so used to marking concepts such as move faster, more collected, more relaxed, happy and calm etc. Problem is horses take time to learn concepts and even then the concepts they can seem to learn are ones with solid clear principles, such as choose a shape with a solid centre over a hollow centre to receive a reward. So when the trainer tries to mark faster, slower, calmer, or happy they are marking what to them constitutes these concepts or emotions. This is why some trainers train, happy face i.e. ears forward, or head down, because you can’t mark the concept of happy or calm, so marking ears forward gives the trainer the illusion of “happy.” Marking concepts not physical behaviours can be way new learners get themselves into trouble. For the horse to understand what is required they need more solid physical behaviours, walking faster becomes maintaining the position of the head relative to the trainers shoulder regardless how fast the trainer walks.
We try to mark multiple behaviours with one click, so for instance walking forward faster with your head in the right position while taking longer strides, there are too many things happening for the animal to be able to understand what is getting the reward, unless you get lucky, and you happen to mark the one behaviour that the animal can understand and underpins the others. One click can only reinforce one piece of behaviour. We have to mark one element of each behaviour at a time and when it is learnt add in new elements that mark a different element of the same behaviour. So to start with you can mark walking along side you, but marking walking along side you with their head up and their ears forward is likely to confuse the animal.
We may not make the effort you consciously slow down horses that are frantic and very active and keen through our own slowing of our body language and the controlling of the click. It might be that before we teach behaviours we need to teach the horse how to solve problems and offer behaviours without becoming frantic or overly excited.
Failure to teach patience and not mugging at the start, on the other side of the scale we often see trainers who withhold the reward to long meaning that the animal becomes switched off and lacks motivated because they can't find the right answer to the question in order to unlock the reward. A trainer that sets their standards to high or expects to much is likely to encounter this problem. The trainer that sets their standards to low or rewards to often is likely to meet a frustrated aroused horse who’s behaviour doesn't really improve or change. Understand the importance of with holding the click but rewarding enough to help the horse learn
People don't use a particularly good variable schedule of reinforcement, it is so easy for trainers to get stuck on threes, normally, but it could be 4’s or 2’s. Where the animal is able to predict the behaviour that is likely to be rewarded this affects effort in which they put into offering the behaviour. So if the horse can predict the 3rd try always gets a reward tries 1 and 2 will be quite small and lack motivation. Developing a truly variable reward schedule is a key to maintaining behaviour with minimal arousal.
As I have written this I have come to realise that these errors are of course problems with all methods of training especially if people are learning. So it seems that the numbers of people learning along side the incredibly motivating power of positive reinforcement might mean that it makes our mistakes more visible faster and leads to greater frustrations in the horse when we lack the skills required to clearly help them process the information we are giving them.