Equine Cognition- The Final Frontier

When I started this blog it was with the caveat that I would stray off piste from time to time.  This post does not deviate too far but it certainly examines matters beyond gravel and hooves!

After many discussions on various forums and a personal journey of discovery and experimentation, I have been feeling moved to write something on a subject that is little researched, acknowledged and talked about-the cognitive abilities of our equine companions.

But this is to start at the end, perhaps a little background would help.  Like many other horse owners, I have long been searching for the best ways to communicate with my horses.  Many years ago, disillusioned with much of what I had witnessed in conventional horse training and seeking a better relationship with my horses, I began what would end up being a winding journey through different methods.  This journey also involved much self discovery, often uncomfortable in nature, and continues today.

I stumbled firstly upon natural horsemanship, attracted by the ideas of interacting with my horses in a language more familiar to them and beginning to understand their body language and communication.  I studied clinicians such as Pat Parelli, Mark Rashid, Tom Dorrance, Monty Roberts and Carolyn Resnick to name but a few.  All of these very experienced trainers have valuable input on different levels and all have their own unique insights into the horse/human relationship.  I was disturbed, however, by the escalation of pressure used by some and the resultant tension I observed in some of the horses being trained.  I also felt that the fundamental flaw with some of these methods was that the human was encouraged to act like a fellow herd member whilst horses were clearly capable of recognising that we are infact human!

Taking away the knowledge I considered to be of value, I began searching for a means of communicating that was more precise and scientifically underpinned.  I have a very good friend who has been clicker training her horse for many years and she kindly invited me to watch a training session she was doing with her horse. I was immediately struck and impressed both by how passive she was able to be as a trainer and also the clarity with which the click and reward was able to communicate her intention to her horse.  Further, her horse seemed engaged in the training interaction and was sufficiently self directed to be empowered to embark upon some problem solving activities.

I decided to research clicker training before giving it a go although the idea of clicking and rewarding behaviours to make them happen more frequently seemed remarkably simple and therefore seductive.  My research, however, indicated that I was wise to wait before experimenting with the horses.  Yes, clicker training is an incredibly effective, clear and motivating communication system but an understanding of the science is really important.  As I read more and attended some clinics, I developed certain concerns.  Would it be possible to addict the horse to being clicked and treated?  Would I really be seeing my horse's 'real' behaviour if our training interactions were based upon positive reinforcement?  Could I cause conflict in my horse if I attempted to ask her to do something with the clicker that she would otherwise be unwilling to do?

I remember attending a workshop with Ben Hart, (http://www.hartshorsemanship.com/), whilst I was in the process of doing this research.  I made a complete nuisance of myself by challenging Ben on all these aspects of clicker training and he answered my concerns with great patience and detailed explanations.  Whilst an advocate of clicker training, it is Ben in his book that actually addressed these potential pitfalls and how to avoid them.  Precautions such as using variable schedules of reinforcement, phasing out the click as soon as possible and intermediate bridges were all discussed.  Did I listen?  NO!

When I finally started clicker training with my horses, I followed the more accepted methodology in equine circles of one treat one click.  I became excited and transfixed by what I could now achieve with my horses. I got great pleasure at seeing that light bulb moment when each horse realised that they could influence the training transaction by their actions.  I loved the feeling of passivity in free shaping and the idea that my horses were stretching their intellectual capabilities by becoming problem solvers.  The only limit it seemed was my imagination.

Despite all of this, some nagging doubts remained.  Some of our herd were becoming rather over stimulated during their clicker sessions.  Some seemed just generally tense.  I also felt that although we were communicating at a different level, there was something rather clinical about the stimulus- response- reward transaction.   I continued clicker training and began expanding my horizons and looking at other trainers, particularly those for whom relaxation and lack of tension was a training priority.  I began paying much more attention to achieving relaxation, not beginning a training session until the horses were calm in the first place and doing much work on clicking for calmness.  I also began being really aware of switching to variable schedules of reinforcement and phasing out the click as soon as possible.

All this helped to an extent, but I still felt that for all the advances clicker training had facilitated, something was missing from our interactions.  I also began to question the more positive aspects of clicker training such as free shaping.  Was it really desirable to be so passive as a trainer, leaving the horses to figure out the answer albeit in small approximations but also to figure out the question?  Would I do this to my kids, put them in a room, not give them any guidance and wait for them to offer behaviour that I could reinforce?  In the human context I would label this inappropriate so why was it ok to be doing this with my horses?  Is trial and error learning really expanding problem solving/intellectual abilities?  I also had times when I couldn't help but feel that the little plastic box that had been such a part of our lives over the last couple of years bore a resemblance to Pandora's Box.  It was now well and truly open and I didn't know how to either close it or mitigate its effects.

Fate, as it has a habit of doing, then intervened.  I had bought a training manual pre clicker training on SATS.  I had given it a rather cursory glance and then dismissed it as complicated and scientifically unfounded.  Whilst wallowing in a state of uncertainty, I noticed in a friend's blog that having experienced similar sentiments, she was exploring SATS.  I dusted off the manual I had bought a few years previously and began to read, not expecting to have much light thrown into the situation.  

What I found, however, was many answers to the questions I had been asking and a way of communicating that would take the best of operant conditioning but also harnessing the cognitive abilities of animals.  SATS, stands for Synergistic Alliances and more information can be found at http://synalia.com/  SATS is not so much a system or a methodology but a philosophy.  In it's most basic form, it can be stated as a way of working with animals that utilises our understanding of learning and behavioural theory by using bridges and targets but also acknowledges the cognitive capabilities of animals.  Perhaps a practical example would help!

It we wanted to teach a horse to be calm, with a clicker we would ideally only start a session when the horse was relatively calm or at least not overstimulated.  We would then be looking for small signs of relaxation and reinforce these signs such as a relaxed muzzle or ears or lowered head.  Our own relaxed stance would gradually become a cue for relaxation and thus over time we would aim to have a relaxed horse.

With SATS, recognising that the horse is actually capable of understanding what relaxation is and feels like, we would seek to provide them with the information necessary for them to achieve this state.  We would firstly therefore name for them all their body parts such as ears, muzzle, nose, shoulders etc, creating a mental map of their body.  We would then help the horse to achieve a relaxed state by massaging them or for touch sensitive horses, other means that would be likely to induce relaxation.  As we are doing this, we would tell the horse that we want them to be calm and as we see signs that they are beginning to be calm would give the intermediate and then terminal bridges.  Further, we would talk them through when their individual body parts look calm such as 'good calm ears, muzzle etc'.  If the same ears became alert, this is reported back to the horse, 'those are alert ears, can you give me calm ears'?  By undergoing this process, the horse begins to 'own' the state of calmness very rapidly and to consequently be able to reproduce it.  No food is used throughout the process in order to avoid over stimulation, the bridges being sufficiently reinforcing.

What is the difference between the two examples and does it matter?  In the clicker training example, the horse is shaped over time to be ostensibly relaxed.  I use the word 'ostensibly' deliberately.  Does the horse really understand that what we are training is relaxation or merely standing still for a prolonged period?  Can the horse generalise relaxation easily so that it is on cue in a variety of contexts?  I am not sure.  In the latter example, I believe that the horse is indeed equipped to understand the concept of relaxation because they are given an awareness of what different parts of their body feel like when given the cue to relax.  Relaxation means something to them.  As they are also taught the contrasting state of 'alert' at  the same time, relaxation is put into context for them as well.  

If we accept that horses, (and many other animals), possess the ability to understand many of our words and concepts if only we talk to them, this dramatically changes the training landscape.  Of course we still want to use the principles of operant conditioning as part of our communication, (it would both inefficient and naive to think that we could or should exclude them).  However, we can greatly enhance the learning experience by mapping out the animal's world for them in our language and then using this knowledge when interacting with them.  

Of course it is perfectly possible to teach words and concepts by using operant conditioning alone but generally labels are only attached once actions are learned.  This artificial distinction between words and actions can result in misunderstandings and a lack of appreciation of meaning.  If the animal is taught words contextually, I believe it makes much more sense to them.  This claim is made several times by Irene Pepperberg amongst others in her books The Alex Studies and Alex and Me.  She used the rival/model method to teach Alex his vast vocabulary and understanding of concepts.  This essentially involves two trainers demonstrating infront of the animal the task that they intend the animal to learn.  The animal then gains an understanding of the context with which the words are used and thus their meaning.  Although it is claimed that equines lack the ability to learn by imitation this seems strange for a social species and may be down to the design of the studies undertaken to date rather than a comment on this aspect of equine cognition.

To conclude, this post is not intended as a critique of clicker training but rather an observation of the issues that can occur if it is not used with care.  It is also a recognition of the cognitive abilities of horses and an encouragement to other owners and trainers to experiment beyond the parameters of operant conditioning.  We have become so concerned to do the right thing for our horses, to be ethical and fair it is as if we are now afraid to interact with them on a  more instinctive, energetic level.  Whilst I certainly would not advocate abandoning those principles that are effective and inform our relationships with our horses, I also feel that it is time to abandon the more clinical aspects of behaviorism for a little cognitive revolution of our own.