Positive v Negative reinforcement .... our experiences

[tollertwins]

i've seen my NH friend stand very quietly, with hand lightly placed on a horse's side.  she gets very tuned in to their energy, and when she feels any subtle shift in it, or in their stance (even a mere LEAN away from her), she releases that light pressure and rewards.

Yah - the NH guys talk about 'try'....and rewarding for 'try'....'try' can be very, very, very small pieces. And it's not really so different from clicker training in some respects - e.g. you click to 'shape the behavior' in very small steps...a person like shoveltrash mentions takes pressure off to shape behavior - and is generous with praise generally.

I guess that's why I don't want to 'work' my pony for being flinchy...I think he's trying by standing still...even if it's not a CALM still.

[spangal]

Part of being a owner/trainer is that you have to recognize when you use what method.  And different horses need different methods, you really can't say one way is the best because you absolutely must have balance.   

I absolutely agree, what works for one doesn't necessarily work for another and needs adapting.

[kassidy]

Just be aware that the release of the adverse stimuli (i.e. pressure) in negative reinforcement is not the same as reward. It is a common miss conception in the NH community that the release of pressure = reward. They are processed differently in learning and are represented differently in the brain. Their motivational properties are also very different.

Although, I agree that you can shape a behaviour using negative reinforcement using well timed releases for small movements in the right direction and eventually get to the point of aviodence learning where the horse predicts the adverse stimuli and reacts before the apllication of pressure or before the escalation of that pressure.

[spangal]

Spangal....

What is IH, please? And what is the book you are mentioning?

 

Intelligent Horsemanship- Kelly Marks and Monty Roberts particularly From my Hands to yours and his training DVD's.  If you pm your e-mail I'll send the link to his newsletter where you can e-mail him direct and ask for advice

[sandpiper]

But that is the accepted terminology.  It's our interpretation of 'negative' that's wrong, not the terminology.

In that case I use positive, negative and being firm, in that order.

[WendyP]

I just wanted to share my experiences with it.
I bought Pi as a 6 yo, bargy, not good to ride on the flat, good jumper, good hack but Unhappy with a capital U.  Inclined to pull faces, wave legs, nip, shove.  I wondered what I'd bought. He was so badly behaved the vet complained about him at his vetting.   After a year of making only nominal progress on the ground, his flatwork improved but his demeanour did not, he just looked Unhappy under saddle. 

I clicker trained him and in a brief 2 minutes of learning, that he should demonstrate good manners about food and then he got rewarded,  I had a magic moment when a light bulb went on his head. His eyes lit up and he realised if he got something right he got rewarded.  Prior to that I think he'd lived a life of confusion, people asked him to do stuff, he did it (eventually) and people stopped hassling him. He did get rewarded but I don't think he got the link. 

That 2 minutes changed his life, from then on he could see a reason for trying to get things right and he started trying and he got rewarded, he understood when he was getting things right. 

I persisted with the clicker training, training him with the phrase "good boy" rather than a click.  I do use "good boy" a huge amount when riding or handling, he knows that if he has been a "good boy" a lot he's going to get some nice sweeties, if he hasn't he goes and stands in the corner of his stable looking sad.  He's now a pleasure to do, his manners out hacking have improved no end, he is teachable but limited by his bone spavin (which is a whole separate issue).

I'm not an obsessive positive reinforcer in that I do use negative reinforcement but my horse responds much better to positive reinforcement than negative. In the natural way of things in fact you will use both at the same time so if you were teaching say rein back, you ask for the rein back with aids and maybe voice, when the horse did it you'd remove the aid (negative reinforcment) and say "good boy" or pat which is positive reinforcement.   

I prefer the term removal reinforcment rather than negative reinforcement, it sums up better what happens.

[Jane C]

I find that whilst I can work with Bucket with simply pressure and release, sometimes he just isn't that interested - he lights up when we do pressure and release, plus a click and a treat.  Without the clicker he just gets over excited by food so I can't use it by itself, but with it (clicker), he is far more focused on working for a reward and he learns what it is that I am asking him to do so much quicker when he is interested in what we are doing.   

[Appy2quarter]

Is asking for a reinback or transition when its put on cue still 'negative reinforcement' Kassidy?  By that I mean we still put our legs on to ask for an upward transition and remove them when the horse moves up a gear.  However softly we may use our legs, isn't that still negative reinforcement?

Interesting about the way in which release of pressure vs positive reinforcement is processed in the brain. Can you expand on this as I am intrigued!!! 

[kassidy]

When observing whether negative reinforcement is present with regards to physical cues for riding is really when you have to look at how the learning is taking place. In order for negative reinforcement to be occuring first the horse must be percieving the pressure of the legs or reins to be aversive. If the cue is not aversive the release will not be relieving to the horse and they will not learn from the pressure being released. Different horses will find different amounts of pressure adversive. Negative reinforcement is not necessarily a bad thing as long as the question is asked: how much pressure is too much pressure ? Or how unpleasent a stimuli is too unpleasent? This is a question of welfare. Personally I won't escalate pressure. I feel this is shouting a french person in english. If the horse doesn't understand a light averse stimuli I will find another way to get the horse to move in the direction I want. I feel it is a necessary in training that cues do not cause the horse discomfort or pain.

If the horse is not finding the pressure aversive, it is still possible to teach the horse to move in the correct direction from the pressure, by applying the cue waiting for the horse to move in desired direction (or indeed guiding/luring them in the correct direction) and then rewarding (with positive reinforcement). The horse learns from the positive reinforcement where they are suppose to move to. So the cue is now more a directional cue than negative reinforcement. With incredibly sensitive horses the pressure of the cue would have to be very light not to be aversive but for horses how are unconcerned by light pressure this training can encourage them to become sensitive and light to ride.

Of course negative and positive reinforcement can be combine to good effect, by rewarding (with a petting or food) as you release the cue in time with the correct behaviour being offered by the horse. It may also encourage the horse not to see the cue as averse as it predicts the arrival of reward.

The brain mechanisms underlying the perception of positive and negative reinforcement is a huge subject so here are a few nice papers on the subject. Sorry I am tired and it would be a huge post

Brain imaging of reward -  http://www.jneurosci.org/cgi/content/abstract/24/7/1660
Lots more avaible if you search. The main body of evidence being on the role of dopamine and the straitum/nucleus accumbens in reward.   

Any cue or behaviour that predicts the arrival of reward can inherit by assocaition rewarding properties. This seems to be why positively reinforced animals enjoy there work more. Because the work which predicts reward has some rewarding properties.

http://www.jneurosci.org/cgi/content/abstract/12/12/4595
http://www.sciencedirect.com/science?_ob=ArticleURL&_udi=B6VS3-454797J-7P&_user=10&_rdoc=1&_fmt=&_orig=search&_sort=d&view=c&_acct=C000050221&_version=1&_urlVersion=0&_userid=10&md5=971af369dd11a387f2d0dbf738ef292b
http://books.google.co.uk/books?hl=en&lr=&id=q6RThpQR_aIC&oi=fnd&pg=PA249&dq=neural+negative+reinforcement&ots=zOWu-bEj0q&sig=GPcZTDxcZTWinUYyo7ggsDt6oCY#PPA249,M1

This study is interesting as money is a secondary reinforcer to humans as the click in clicker is to horses - http://www.jneurosci.org/cgi/content/abstract/23/1/303

Differences between positive and negative reinforcement have been found and I haven't seen evidence of the predictors of adversive stimuli firing the reward neurons which I guess seems obvious. http://www.jstor.org/pss/1421126 - looks like an interesting article if you wanted to spend money There seems to be less work on negative reinforcement and aviodence learning than on reward.

Loads on punishment though if you type in punishment into google.co.uk/scholar.

Sorry for cheating but it would have been very long to type out.