On my facebook clicker discussion group (Horse Sense and the Clicker) we have been chatting about freeshaping, how to do it, and the problems that trainers face when using it. I had lots of things that I wanted to say, so I decided to take this as a topic for my first blog post, I hope you find it interesting/ useful.

Training through successive approximations. This may sound like some complicated procedure, but in actual fact we use this approach all the time when learning. We rarely learn how to perform a task in one trial. And even if we learn it quickly, there is still room for refinement. So let’s look at how this applies to horse training. The first place to start is with a clear picture of the end goal. We’ll start with the example of a horse standing calmly still in the middle of the arena while we walk to the fence, lift the saddle and return and put it on the (unrestrained) horse. If we want to teach this, no matter what our training approach, we will not start at the end! It is always going to take a number of steps (or successive approximations) to move towards the end goal. The number of steps, and the size, will depend on the individual horse and the individual trainer. This is how we ‘shape’ responses.

In conventional horse training, which mainly utilises negative reinforcement/ positive punishment  (these are scientific terms, so don’t get upset about them, more on that in a future blog post) more commonly known as pressure and release, the process by which this goal will be trained, although it will still involve shaping, will involve the emphasis being on stopping the horse from getting it wrong. So if the horse moves, pressure will be used to return them to position, with the release coming when they are standing. This pressure may be applied via a line, or through body language/ drive to block movement. The shaping steps will involve a gradual reduction in the amount of pressure required to correct the horse. Learning always involves making mistakes, but with this approach, the horse’s focus will be on avoiding the pressure, and avoiding making the mistakes (because they have consequences that they would rather avoid) rather than on the actual task in hand. The end goal is achieved when the horse will stand without mistakes and without requiring pressure or release of pressure.

One of the beauties of training using positive reinforcement is that it clearly draws the horse’s attention to what they are getting right. It certainly isn’t without stress, since all learning involves a degree of stress, but because it utilises a very positive emotional experience (more details on this in a future blog…) then this helps to counterbalance the stress of learning. And it means that we can train without pressure, or with minimal pressure, if we choose.

Using successive approximations to train a response from scratch, without the use of pressure as guidance, is known as ‘freeshaping’. I have heard some people criticise freeshaping because they feel it is more stressful for the horse, requiring the horse to be more creative, more able to puzzle out what is wanted. I would argue that the level of stress is directly related to the skill of the trainer to identify the smallest of tries, and to move forwards through the successive approximations in the smallest of steps. I would also argue that the level of stress is down to the ability of the trainer to read the horse, and to prepare the horse appropriately. For example, if I have a client with a shut down horse, once that doesn’t offer any responses unprompted, then they will struggle with freeshaping that involves offering movement or creative behaviours that don’t involve any kind of external stimuli. So I will first of all work on building a good positive reinforcement history with that horse. I will choose a response that they already give very easily, such as leading, or standing for grooming, and heavily positively reinforce them for that response. In an ideal world, I would also considerably reduce/ remove the use of aversives (negative reinforcement/ positive punishment) in the horse’s training. In order to do this I will choose to work at liberty with the horse where possible, so that there is no temptation to use pressure via the headcollar and leadrope. This tends to lead to an improved relationship, an increase in confidence, and greater interest in the world. This comes about because the use of the positive reinforcement in training stimulates the SEEKING circuitry (if you want to know more about that read ‘Affective Neuroscience’ by J. Panksepp), which is all about curiosity and exploration (as opposed to aversives, which stimulate the FEAR circuitry, which is all about avoidance, escape and withdrawal). Once the horse has started to show me signs that they are growing in confidence, such as a change in posture and expression, a calm interest, nickering, a brighter eye, I will then begin to explore objects. Targetting is a lovely way to do this (this is used extensively in the marine mammal world, and Shawna Karrasch began showing the horseworld the power of targeting in the mid nineties, www.shawnakarrasch.com).

Teaching targeting with positive reinforcement involves utilising freeshaping. So, when the target is first presented, the skilled trainer will do it in such a way that the most likely response the horse is going to make will be explore the target with their muzzle. For a confident, open minded horse, this is generally their first response anyway, however, with the more withdrawn, shutdown or anxious horse, they may be very reluctant to explore any novel object. With a less skilled trainer, and a horse that has had enough experience of clicker training to want to seek out the click, poor shaping in this kind of situation is going to be a stressful business. Imagine you are prone to getting static shocks when you touch the car door when you get out of the car. So you have developed a way of dealing with this which means that you close the door with your foot in order to avoid the shock. Then you take part in an experiment where you are told you will be given a small reward for every correct response you give, and that will be related to how you get out of the car, but you don’t know what the correct responses are. Are you likely to touch the car door with your hand? Probably not. So you will never find out if that is one of the responses that might earn you a reward. With a poor trainer, you might not earn any rewards at all. Are you likely to want to take part in the experiment again? Probably not. You may just give up or walk out, or you may become very frustrated at your inability to earn the rewards that you have been told about. However, with a good trainer, they will first of all ensure that you won’t infact get a static shock. They will then reward any movement towards touching the door, even if this is only a slight movement of the hand. It may take many approximations, spread out over many sessions, to get you to the point that you actually touch the door, but in the process of that training you will have received lots of rewards. You will have begun to trust your trainer, you will be willing to try things for them, and the earning of the rewards will have become more salient (meaningful) to you than any potentially aversive experience (the fear of the static shock). When you do actually touch the door with your hand, they will give you a huge jackpot. What will happen if you are then asked to take part in a different, yet similar type of experiment? Will you want to do it again? Most likely yes, and with much more enthusiasm than when you agreed to the first one!

Freeshaping does not mean standing back and just waiting for your horse to pick up their foot and hold it on the spot so you can reinforce it. It does mean spotting the smallest tendency towards your end behaviour (which means you have to have an understanding of how the response breaks down), and being quick to reinforce that (preferably within a very short time of starting the session). For foot lifting, this could be the tiniest shift of weight. If you haven’t spotted something within the first minute that you can reinforce, then you either haven’t broken the steps down into small enough pieces, or you have picked the wrong start point, or you have a horse that is not confident enough for the behaviour you have chosen. Freeshaping is a skill that some horses need to learn. If they have come from traditional training backgrounds, they will most likely have been punished for creativity, so they are unlikely initially to offer any kind of responses unless they have been cued. If this is your horse, then you certainly want to begin your freeshaping with something simple like standing still. This doesn’t require any kind of creativity from the horse, but they do have to use their brain to work out what you want. Then you can progress to simple targeting of a safe and familiar object. You may even have to begin your shaping by reinforcing them for simply looking at the object, then, through your successive approximations, you can gradually stretch them to the point where they will touch the object. Over time you can turn this into object play, such as kicking a ball or barrel along, or nosing something, or picking up an object, or doing a retrieve. These are all fantastic ways to teach a horse confidence and creativity around objects. The knock on effect of this is a horse that is not fearful of novelty (which means less spookiness!), and one that is willing to try new things. You can then progress to freeshaping of behaviours and movements that don’t involve objects, because by this stage your horse will be more creative and confident about offering responses. One thing that you do need to remember though, is that if you are freeshaping a response that doesn’t involve an object, then the horse will be very quick to learn the environmental stimuli that are clues that they will get reinforced for performing that response. This could be the way that you are standing, or the particular area that you are working in. This is why if you are freeshaping, you need to be aware of what these clues might be, and when they are present, only work on the response you have been shaping, until you get it to the point where you can introduce a clear cue that you and your horse both know and understand. If you don’t train to this point, and you work on multiple responses, you will end up with a tense horse that is continually offering a variety of responses in the hope of earning reinforcement. This is why the first (and most important) response that any clicker trained horse can learn is to stand calmly and quietly. This is the behaviour that I teach when I am first conditioning the clicker, and the cue for it is simply me being there and the absence of any other cue. This reduces the chance of the horse offering other unwanted responses when I am standing about doing nothing!

This just skims the surface of so many issues, I find it hard to limit myself to just talking about freeshaping, when I know that there are so many other related points that I think people need to understand! However, I am writing about these and other issues in more detail in my book, which I hope to have finished soon. I hope you enjoyed this, my first blog post!

Over the next few weeks I intend to film some short videos of freeshaping in action that I can share with you…..