Most everyone knows that it doesn’t do any good to beat a dead horse. No matter how hard you beat, you’re not going to be able to convince him to load up into the trailer, jump that 2 foot oxer, or perform any other sort of task.
(Most) people these days also realize that it doesn’t do a whole lot of good to beat a living horse. Yes, if you scare him badly enough you might be able to force him into that trailer. (This is mainly because at some point the “scary” trailer starts looking a whole lot more comforting than the person using force).
So, beating a horse doesn’t work. What should we do instead? How about using small amounts of pressure and release of pressure coupled with other motivators (such as comfort). Wouldn’t this be the better solution? Intuitively, we think yes. However, what if there’s an even better solution?
A recent study by several Welsh researchers evaluated the use of positive and negative training methods in retraining simple tasks (such as leading, standing for grooming, trailer loading) to rescued horses. (The original article was published in the August 2008 issue of Applied Animal Behavior Science. If you have access to journal databases through your school, you can find it online. Otherwise, you can at least read the abstract here, if you are interested.)
Here’s what they did:
Horse training methods often rely on positive reinforcement, negative reinforcement or a combination of both. (In this context, positive reinforcement refers to something added after the behavior that increases the chance of the behavior occurring again, such as a treat or a scratch behind the ears. Negative reinforcement refers to a stimulus applied that the horse works to avoid, such as pressure. When we pull on the reins, the horse stops because of negative reinforcement, it is seeking to avoid the pressure.)
16 rescued Welsh ponies were trained for seven weeks (4 times a week for 30 minutes) by experienced staff to perform 5 tasks (leading, standing to be groomed, picking up feet, trailer loading, and walking an obstacle course) using either positive reinforcement or negative reinforcement. The 8 ponies in the positive reinforcement group were trained using clicker training. The 8 ponies in the other group were trained using a whip/training stick. The stick was applied with a very light touch at first, and then with increasing pressure if the animal did not respond. As soon as the animal responded, the tapping was stopped.
Weekly, a novel object test was given where the ponies were individually released into an arena with a novel object (a hung umbrella). After about week 4 of training, the ponies being trained with positive reinforcement (+R) were significantly more likely to approach the novel object and the observer. This group was also more likely to approach the door to the stable while turned out in the pasture. During training, ponies in the negative reinforcement (-R) group were more likely to exhibit pawing behavior, which is often a sign of frustration. Although both methods were successful for training the ponies, the ponies in the +R group were more likely to exhibit explorative behaviors and were more willing to participate in training. Also, more of the ponies in the +R group met the criteria for re-homing at the end of the study.
Some of my thoughts:
These were all rescue ponies (all had little previous contact with humans, and most past interaction with humans had been aversive). So, they could have been more sensitive to -R techniques than regular horses. This has important implications for rescue groups, who might want to incorporate more +R into their training programs. Pressure (and -R) can be a very effective for training horses. However, as was seen in this study, the horses trained in this method were less likely to exhibit curiosity or to exhibit correct behaviors on their own. So, if we want correct behaviors and control, either training method will do. But, if we want horses that are braver and more curious and that (potentially) can be rehomed faster, +R methods (such as shaping, targeting or luring) might be a better approach. It would be really interesting to see this study redone with “normal” horses that didn’t have a lot of previous stress and baggage.
(And for all you skeptics about whether or not horses can be trained with mainly positive reinforcement, go read the clickRyder blog or go watch some youtube videos. Like this one, or this one, or this one.)
9 responses so far ↓
1 Elizabeth P // Dec 2, 2008 at 11:01 am
Is +R more difficult than -R? Does it require more imagination and patience on the part of the trainer?
This post makes me think of parenting, and trying to train behavior in children. I see parents smack and yell at their kids all the time, and it’s easy for me to say that there has to be a better way of doing it, but it sounds very difficult.
2 Simon Mulholland // Dec 2, 2008 at 12:00 pm
My instinctive response is that in the interests of “scientific” data, they used the Negative reinforcement in a way that few trainers would use. I don’t use clicker training and would describe myself as “negative reinforcement” but if you watch a link of me training a pony in a new vehicle, you will see that negative can be very gentle. I certainly wouldn’t use a stick to signal.
Try this link
http://picasaweb.google.co.uk/saddlechariots/Guido421465#5273819578001572706
3 ozvet // Dec 2, 2008 at 1:34 pm
Hi Simon,
Thank you for your comment (and video link). It was obvious from watching that you are taking the right amount of time and making sure the cart training is a pleasant experience for the pony.
In my experience, negative reinforcement can be extremely gentle and can produce phenomenal results when the welfare and dignity of the horse are taken into full consideration.
I plan to e-mail the researchers, as I’m very interested in learning more about the precise methods they used. If I find out anything interesting, I’ll be sure to post it here.
Not a whole lot of work of this kind has been done with horses, and the fact that these horses came from a bad situation could have played a large role in how they responded to even very mild negative pressure (even such as training to trailer load by a gentle tug on the lead rope instead of something like targeting).
I think what’s especially interesting is that the ponies trained exclusively with positive reinforcement were more brave/curious than the other group.
I know work with dogs has shown that dogs trained with even a tiny bit of coercion or correction are less likely to exhibit curiosity and more likely to exhibit stress related behaviors (even when positive reinforcement was used in addition to the coercion/correction).
4 ozvet // Dec 2, 2008 at 1:42 pm
Interesting comment, Elizabeth.
I think +R does take a lot more imagination, patience and planning from the trainer. The trainer has to have a plan, has to be able to modify that plan on the fly, and has to understand how to set the pupil up so they can succeed most of the time.
I think parents often resort to yelling at and smacking kids out of frustration. While this does produce results, it also creates stress and other negative emotions. However, because it works some of the time, they often keep doing it.
Karen Pryor’s “Don’t Shoot the Dog” is an excellent introduction for getting results with animals (and people) using operant conditioning and mostly positive reinforcement. (It also details why positive and negative punishment, and to some extent, negative reinforcement, should be avoided).
5 Simon Mulholland // Dec 2, 2008 at 5:28 pm
Keep me posted if you get any details of the methods used. I just feel it’s skewed. Clicker training versus whip. Why a whip, that isn’t pressure release first choice tool. A rope halter yes. But you get the pony to come to you and give it a scratch, and talk to it.
Now if they had done clicker versus pressure release and allowed the pressure release types to do what they normally do, I think you would actually see a vastly greater middle ground with both parties doing pretty much the same thing.
It is the problem with “science” that it wants to categorise, and some things don’t categorise that easily.
Pressure release can be really gentle and parents smacking children may not even be applying pressure release, they may just be applying pressure and letting fear do the rest, a very different concept.
Keep looking, but I suspect the answers will be multiple, and won’t fit into a neat graph because the concepts are too complex to divide into + and -. But keep looking. I keep trying new methods, the one thing I have learned is that the British Equestrian Establishment have no desire to look or to learn. And that is depressingly easy to prove scientifically.
Simon
6 Simon Mulholland // Dec 2, 2008 at 6:42 pm
Sorry just realised this looked anti clicker. I have seen plenty of really well trained ponies whose owners use clicker, and I have met Alexandra Kurland who I think is super.
My concern is that the “alternative” movement can spend so much energy fighting each other, when there are people out there using “pressure pressure and more pressure, OK we obviously needed bigger spurs and a harder whip” school.
7 ozvet // Dec 2, 2008 at 7:11 pm
I agree that most of these concepts are probably way to complex to fit into a neat +R and -R spectrum. Even when using all +R, something as simple as moving away from the horse can release pressure if the horse is stressed (which means you are also using -R). And I don’t think anyone uses exclusively -R. So, I think it’s probably impossible to use all one or all the other, which of course makes designing research protocols more difficult.
It would be interesting to have a study that looked at different variations of +R and -R. Such as combining the use of a whip/training stick with treats or scratches, as opposed to either method alone, or how stress and learning related to different tools, such as whips, training sticks, rope halters, etc.
I agree completely with you about alternative movements spending too much time fighting each other. All this does is just make everyone look bad and defensive.
As for parents smacking children, I think this almost certainly always falls in positive punishment, which is different than negative reinforcement.
8 Simon Mulholland // Dec 3, 2008 at 10:46 am
The thing I have learned is to use less and less equipment, and basically I am looking for signs of stress. Basically I don’t want to put a vehicle on a stressed animal, after all I am going to be on the vehicle. So I just muck about and see what I can do, taking it gradually. But as you say, any good trainer is responding to the actions of the animals they are training, and by that response they are going +- pretty much at random.
Look I am all for the scientific method, but I can see its limitations. You can’t run double blind controls on people because the law won’t let you, and no horse lover is going to let their animal be used in an experiment where it might not get the best treatment, so you either get experiments like the welsh one, which is useful, but doesn’t prove anything.
We need to go to the model used by psychoanalysts, extract huge sums of money from the patient for doing nothing, sorry getting cynical, but we need to try and build a theory of human horse communication and then test the viability of the theory.
Can I relax a random group of ponies and get them in a horse drawn vehicle in under two hours? Yes, just done it for a German Magazine, Cavallo. Exactly how I do it varies as each animal varies, but the principle holds good.
Try getting away from the idea of equipment and look at this article http://naturaldriving.co.uk/content_chivalry.php which looks at the difference between horse and dog training in the UK.
Simon
9 sex ??? // Nov 23, 2009 at 7:59 am
eh. bookmarked post :))
Leave a Comment