In your opinion ... what are the pros and cons of +ve and -ve reinforcement?

[Bay Mare]

Which, in your opinion, *works* best and why?

If you use both together which method do you use the most?

[Wally]

I use negative when a horse would use it. ie when horse invades my space in a rude manner, and my safety might be compromised, or anyone elses for that matter. A barger will get a yelling at, and a slap on whichever part I can reach to send him out of my space in the same way another horse would. Also if anyone threatens me with a bottom, I will kick it before they get the chance to land one on me.

I always use positive reinforcement when I am asking them to do something for me. A horse cannot learn a new skill or "trick", call it what you will, by experiencing negativity, they have to be allowed to do the right thing not the wrong one. Never tell the horse not to do something when training him, allow him to do the right thing.

Set yourself up to succeed.

[ameliet1971]

I use both negative and positive reinforcement.

It all depends on the horse and the situation. If the horse was [say] trying to mug or boss me around, I would use negative reinforcement to make him realise his actions where unacceptable but would immediately relax my body language and reward him with a stroke at the slightest improvement in his actions.

When teaching anything new, I always begin with the softest feel and relaxed body language, if he doesn't understand I then ask for the same thing in a different way. I don't think any horse can learn by being bullied into anything, but neither do I think one can learn by positive reinforcement alone. :)

.

[Sammii]

I tend to use them both, as and when I am quick enough to administer it.

I use positive reinforcement when I am asking.
And negative reinforcement when I'm being told.

[coss]

as with others... i use both
if horse goes to bite me they get an elbow up the nose, goes to kick, grab the front end and pull round, make horse walk backwards (they're legs are longer than mine and would kick before i could kick them... how do you manage wally?)

but when training something you have to make it pleasant for the horse to do the desired task, they'll only rebel and resent you if you try and force anything...

[Welsh_Girl]

As the others said I use negative reinforcement to correct bad uninvited behaviour such as biting, kicking, barging etc. and positive re-inforcement to "reward" any response in the direction I'm aiming for.

It depends on the horse too as one of mine you can't use both in the same session. e.g when leading him into the stable he often barges to which I would scold him with my voice and give a jerk on the lead when he walks on quietly I simply accept this without reward as he gets confused if I then praise him.:rolleyes: and charges ahead again. The next time I lead him in the stable I say goodboy while he's leading properly, before he thinks about barging, and this is enough to stop him barging.

My mare is the opposite in that if you scold her she immediatly corrects her behaviour and then "looks" for the "reward" e.g me saying "goog girl" when she gets it right. If she doesn't get the praise/acknowledgement she keeps trying.

[chev]

Negative when they are being aggressive towards me, positive for training.

Occasionally in training situations... teaching Rhodri to go in a stable is one example. He wouldn't. He was allowed to plant himself and we just praised each step forward. Then he got stroppy, reared and struck out at me. That was not acceptable at all - he got a slap accross the belly. He was allowed to be worried, he was allowed to take his time; but he was not allowed to lose his temper with me. Following the rear he went back to being passive but not co-operative; that was ok. We just waited patiently and rewarded each step forward. And in the end he went in happy.

[chev]

Just realised I didn't really answer the question!

Ok... negative in training situations doesn't give the horse a great idea of what's wanted - just what's not wanted. For me that makes the whole training thing a chore; using positive means the horse starts to look to you for what you want to do next - training becomes something that stimulates and interests them. Most horses I've known who've been trained using positive are happy in work and enjoy learning.

Negative in training... just means they become more wary. I don't think teaching of any sort, be it human or animal, should be based on negatives.

But... there are times when positive isn't going to work. Like Rhodri - I'd have had to have waited til he came down, ignored the fact that he struck out at me, and waited for the next opportunity to praise. Had he just gobe up, I might have done that - but when he struck at me that was definitely not acceptable, and I do think the slap he got was warranted.

[Est]

I use both and think both have a place. I wouldn't say one is/works "better" than the other :) (Though I do love my clicker sessions! :D )

The one thing that people are often not clear on is what "negative" means... it is just removing something, it's not "negative" as in "bad".

So when I touch Merrick's chest to ask him to move back, and take away my hand as he starts the movement (sometimes even the mental movement!) that's negative reinforcement. I'm not "telling him off", I'm not using big body language, I'm not bullying or forcing. All common misconceptions around -ve R :)

Also it's surprising how many people say they are using +ve R, but actually they've asked for something using -ve R (for example, touch or squeeze back of leg to ask horse to lift it for hoof-picking, remove pressure as leg comes up) and then say, "Good boy!" afterwards and believe that they've "only" done +ve R.

Looking at -ve R in that light probably reveals that most of us use a good mix of the two - and undoubtably some +ve / -ve P occasionally though it's far from fashionable to talk about *that* :D

[Roofio]

hmm. so clicker is positive, cause nothing happens till you do what's required and then there's a reward? and anything that is pressure followed by release when the move is completed is negative? that right?

I'm not really sure what i do - am experimenting with parelli at the minute but combining it with clicker also. the biggest issue i have is that they are both feet planters - had a lovely play with both yesterday, walking over an opened out feed bag as brave as anything, all done with clicker and a gentle nudge on a rope halter, but soon as i turn them back to where they're tied up for brushing and all that, they both planted - i can't reward that so use pressure on the halter with release and click/treat for forward movement. in the end had to throw the line at the back end of one (can't remember which :o) to get them to move - is that negative then?

interesting!

[Wally]

(they're legs are longer than mine and would kick before i could kick them... how do you manage wally?)



Mine only have short litte legs! and a bottom at kicking height. Not sure I'd use the same tricks on a 17hh horse whose bum is at my head height.

[Roofio]

Mine only have short litte legs! and a bottom at kicking height. Not sure I'd use the same tricks on a 17hh horse whose bum is at my head height.

there's got to be a thelwell image somewhere of little ponies getting kicked up the backside to get them into shape! :D

[Est]

For the purposes of this thread, I present Est's homegrown definitions of -ve and +ve R (aka encouraging a behaviour), and -ve and +ve P (aka discouraging a behaviour)! :D

Negative reinforcement:
The REMOVAL of a stimulus to INCREASE the likelihood of a behaviour happening again.
Example: squeezing with your calves to ask for trot, then releasing as the horse trots.

Positive reinforcement:
The ADDITION of a stimulus to INCREASE the likelihood of a behaviour happening again.
Example: You're leading your horse inhand, and have to stop suddenly. He politely stops at your side. You say, "Good boy!" and give him a polo ;) :D

Negative punishment:
The REMOVAL of a stimulus to DECREASE the likelihood of a behaviour happening again.
Example: You bring your horse his tea and he starts kicking his stable door as you walk up. You turn around and walk away taking the tea with you until he stops kicking (I am not commenting on the effectiveness of this!)

Positive punishment:
The ADDITION of a stimulus to DECREASE the likelihood of a behaviour happening again.
Example: Your horse goes to bite you and gets a smack (or a kick up the bum!!!) :D

[Kate F.]

I agree with Est. Using those excellent and user friendly definitions, which summarise the scientific uses of the terms positive and negative reinforcement, you're going to have to use both to be effective. You ask the horse to do something, it does it, and you stop asking (negative reinforcement) and stroke, scratch, click, give a treat - what ever you choose as a reward (positive reinforcement). The two go hand in hand. In the initial stages of training there will be really big releases and really big positive reinforcements - but as the behaviour pattern is established, both get more subtle until it becomes almost just a "thought" - but both will still be there. People who claim to use only positive reinforcement are generally overlooking the body language cues and releases they give sub-consciously. Even just the expectation of a response is a type of pressure, and you body language will change when the right response is given, so it's a type of negative reinforcement. You can work with negative reinforcement alone, but it's much more effective to use both.

Punishment is a different matter. Negative punishment, I'm pretty sure, is meaningless to the horse. Using Est's example, the horse is not kicking the door to annoy you, he's kicking the door as an expression of the stress he's feeling. If you then go away with the food, there's no way he can make the connection between your action and him kicking the door - in his eyes, you're just behaving unfairly and weirdly, and will probably kick more. Positive punishment can have it's place, but I prefer to call it a "consequence" rather then punishment. Punishment has emotion from the person implied, a consequence is just something that happens if the horse does a certain undesirable thing. It should be presented without anger or emotion. For example, the horse learns not to touch an electric fence through positive punishment - or a consequence. If it touches the fence, it gets a shock. Then fence has no feelings about it - and the horse doesn't resent or fear the fence, it just knows not to touch it.

Similarly, if the horse does something you don't want it to do - kicking the door of its box, for example, if you can create a consequence for kicking the door, the horse will soon give up. It won't stop being hungry, and it will still be excited, but you can discourage that particular behaviour.

We do it with having someone in the area of the kicker at feed time. Immediately it kicks, that person calls the horse's name sternly and throws a rope up above its eye level so it steps back from the door. This is repeated each time it kicks, and pretty soon it steps back when you call the name. Not long after it gives up kicking the door. However, you have to be really careful that you respond really quickly to the kick - 3 or 4 seconds later is too late, and that it's all done without emotion.

[capalldubh]

Interesting thread!

Not a lot to add to what's above, except two things.

Most of the early examples of "negative reinforcement" are actually positive punishment - something the horse will work to avoid but which comes after the behaviour you don't want, in the hope the behaviour won't happen again.

It's really interesting, people don't like the idea that they're punishing their horse, yet when it's not called punishment, they're OK with it - and people seem to really like the idea of reinforcement - to the extent that they call what's technically punishment, negative reinforcement.

The second thing is that we are all saying we are using negative and positive reinforcement in our training - and we define negative as "pressure" and positive as "a pat or stroke", for example. The difficulty, from an academic point of view, with these definitions is that whether a stimulus is a positive or negative reinforcer is really only determined by the animal's response. So pressure in many circumstances is neither negative or positive - if you place your hand on the horse's side and it doesn't move to remove the pressure, the pressure is not negatively reinforcing. If you stroke your horse's face, and your horse tries to move its head away, the stroke is not positively reinforcing.

In the end, we try control our training depending on our own beliefs. I don't subscribe to the belief that it's OK to cause pain or discomfort in training because it's only something another horse would do. I'm supposed to be a human with superior cognitive skills, so should be able to find a more effective way - if we said it was OK for teachers to behave to children as they do to each other in the playground, there would be outcry :D

Animals learn actively - they seek out rewards and avoid negative consequences - so nature trains all of us using all types of reinforcement and punishment - none of them are any better or worse than the other. The main thing for me is that we don't just learn through these things - we also learn through association - if we offer mostly rewards, our presence will be reinforcing, if we offer mostly aversives (negative reinforcement), unless we also provide a lot of very nice dinners, our mere presence may be something the horse wants to avoid (unless we've negatively reinforced them that something bad happens when they leave...)

[coss]

Mine only have short litte legs! and a bottom at kicking height. Not sure I'd use the same tricks on a 17hh horse whose bum is at my head height.

thats ok then... was trying to work out how to kick out at my two but at 14.2hh i think i'll opt out :p

[levi1739]

Negative = inhale

Positive = exhale
-------------------------------------
Negative = send

Positive = draw
-------------------------------------
Negative = pressure

Positive = release

-------------------------------------


For me, it's all about my energy and intent. Regardless of how we humans interpret the "language of words", the horses are always aware of their meanings. Too a sensitive horse (they're all sensitive), every breath has meaning.



Have fun, be safe

Jack

[laceyfreckle]

for me...it tends to be

+ve if asking them to do something eg. pressure on leadrope to back up, pressure stops the instance the thought of backing is there.

-ve if they are 'in my space' trample me, barge me, bite/kick me, you get no second warning lol

[*katie*]

Ummmmm . . .
I should really use both because that way horses will understand my behaviour but I'm a total pushover and tend to use +ve in all of the situations other people have described, and when most people use -ve; i tend to . . . well, not :rolleyes:. I tend to just talk to a horse like (in my best teacher/angry parent/courtroom judge voice :p) 'Erm, excuse me!' or word to that effect. I think both work depending on the situation; i don't think -ve is a nasty way of treating a horse in the proper situation; because a horse isn't a person inside; they think and act differently and in order for the communication not to break down you have to act and treat them like another horse would. I think both are the correct way to treat a horse but only when appropriate (like, it's OK to sit there petting and saying 'good boy' etc if your horse is going out of it's mind with fear but if they're being a rude little sod and misbehaving, you don't sit there stroking their neck :p) but as i said i'm a complete doormat and every horse that's ever met me has themselves firmly wound round my little finger - which really isn't a good thing :D! I'm the perfect example of only using +ve and having horses completely take advantage of me and assume the 'dominant' role :rolleyes:

[ameliet1971]

For me, it's all about my energy and intent. Regardless of how we humans interpret the "language of words", the horses are always aware of their meanings. Too a sensitive horse (they're all sensitive), every breath has meaning.

I couldn't agree more Jack, the human "language of words" is always open to debate and misinterpretation, but horses only understand and respond to one universal language - body and that's the one I strive to perfect. :)

.

[capalldubh]

+ve if asking them to do something eg. pressure on leadrope to back up, pressure stops the instance the thought of backing is there.

-ve if they are 'in my space' trample me, barge me, bite/kick me, you get no second warning lol

the first example is using negative reinforcement, not positive reinforcement.
the second example is using positive punishment (I would guess), not negative reinforcement.

This is the difficulty - people don't actually understand the difference - it's not taught in school, it's defintely not taught in riding schools, and most books on horse training get it wrong. And if you get it wrong, it won't work.

Because for me, there is only one universal language, and that's learning theory. Yes, a horse is quite capable of learning what your body language means, but it learns that it in the same way it learns everything else - through habituation, positive reinforcement, negative reinforcement, positive punishment, negative punishment and classical conditioning. The same way we learn, the same way pretty much every animal learns (it's even been demonstrated - and won a Nobel prize for the man who demonstrated it - in a seasnail called Aplysia).

You can use completely different body language, and provided you get the learning theory right, the horse will learn - it's been demonstrated very nicely by people who have movement difficulties and who can't use the body language that NH practitioner say is innate.

[Francis Burton]

I couldn't agree more Jack, the human "language of words" is always open to debate and misinterpretation, but horses only understand and respond to one universal language - body and that's the one I strive to perfect. :)
A worthy goal indeed!

The trouble is that we are pretty much obliged to use the "language of words" if we want to communicate with each other about body language and training - unless we are right there seeing, and ideally experiencing too, the same things as the other person. That isn't possible in a text-based discussion group like this. (Even photos don't always tell the whole truth.)

I am inclined to think doing is the best way to learn, followed by watching, followed by talking/reading. Talking can help the process or it can confuse it: sometimes people seem to do and say different things!

By using words to describe what we do, are we doomed to failure? Is it worth trying to define meanings of words that at least most of us can agree about?

There's an issue here specifically to do with the commonsense meanings of "negative" and "positive" being at odds with what these words mean in learning theory - equally logical and sensible, but apparently harder to grasp. Likewise, as soon as you use the word "punishment" you trigger all kinds of associations in people's minds, mainly to do with what punishment means to us as humans and our human culture. Some of these may start unhelpful trains of thought. The same can be said of a lot of words and phrases that are used in horsemanship, such as "submission", "respect", "willingness to please" and "leader". What you mean won't necessarily be the same as what other people will take you to mean. I'm not sure what the solution is apart from simply trying to be as clear as possible to explain what we really mean (e.g. by way of example or analogy).

[Teazle]

Positive is always best on a day to day basis, as it gets better results (and is nicer all round!). However, you cannot have one without the other. If you only use positive reinforcement and have never shown the horse (or any animal) where the boundaries are, then you're asking for trouble. Negative reinforcement doesn't have to be hitting though. Things like - tight circles after they buck, or continuously repeating an exercise until they calm down and do it correctly, make doing the wrong thing unpleasant and not worth it, but don't cause pain. Body language also reinforces signals - if the horse is being bolshy and you square up to them, it often makes them think twice. You do have to have the basic respect there first, and that comes mostly through training.

[ameliet1971]

Just a thought, if Negative Reinforcement can be defined as the removal of a stimulus to increase the behaviour happening again, then it can also be interpreted as the release of pressure the instant a desired response occurs. Maybe this is why some psychologists admit that it's not always easy to distinguish between punishment and negative reinforcement.

For instance, in his book 'Considering the Horse', Mark Rashid describes the teaching of ground manners (through pressure and release) to a horse who had little understanding of what to do when pressure was applied.

Part of paragraph:
“As I was leading her the short distance, I stopped to pick up a piece of baling twine that I noticed on the ground. Sugar simply ignored the fact that I was no longer moving and kept right on walking, actually dragging me about three feet before stopping. I took hold of the lead rope and applied light, constant pressure downwards and slightly back, without pulling on her. She responded to leaning into the pressure. Obviously this wasn't what I was looking for, so I met her resistance with with some resistance of my own. Again, I was not pulling on her but rather putting the same amount of pressure on her that she was putting on me. This went on for a short time before she started tossing her head in search of a way to get rid of the pressure. When that didn't work, she finally flexed, giving to the pressure. Immediately I released my hold on the rope, giving her a very clear and noticeable reward for the proper response”.

Obviously, this could be read as positive punishment being used on Sugar, but it could also be defined as the pressure and release applied when using negative reinforcement.

Either way, it sure is a good example of how the language of words can be interpreted and defined by different people.:)

.

[capalldubh]

Just a thought, if Negative Reinforcement can be defined as the removal of a stimulus to increase the behaviour happening again, then it can also be interpreted as the release of pressure the instant a desired response occurs.

But as I said above, only if the horse interprets the pressure as aversive or something they will work to avoid. A touch on a horse's side can be construed as pressure, but if the horse doesn't move away, it's not negative reinforcement. Not all pressure is aversive, it can be a neutral signal as well.

very clear and noticeable reward

Either way, it sure is a good example of how the language of words can be interpreted and defined by different people.


It is interesting because the brain does not interpret the release of (aversive) pressure as a reward. So this was a clear example of negative reinforcement.

Again, I was not pulling on her but rather putting the same amount of pressure on her that she was putting on me. This went on for a short time before she started tossing her head in search of a way to get rid of the pressure.

You could say that since these behaviours stopped or diminished in the face of the pressure, that they were positively punished. And that the behaviour of flexing the neck was negatively reinforced.

The aim of science is to find a common language in which we can all communicate about behaviour, without the personal interpretations that go behind terms like "feel", "respect", and "pressure". Because language is what it is, science doesn't quite get there - but it would be so helpful if the big trainers explained the science behind the terms, so that people from different schools of thought in terms of training were able to communicate more effectively :)

[Francis Burton]

Just a thought, if Negative Reinforcement can be defined as the removal of a stimulus to increase the behaviour happening again, then it can also be interpreted as the release of pressure the instant a desired response occurs. Maybe this is why some psychologists admit that it's not always easy to distinguish between punishment and negative reinforcement.
I would have thought it's harder to distinguish between negative reinforcement and reward than between negative reinforcement and punishment. Some people argue that the relief felt when pressure or something uncomfortable goes away is the same as the pleasure felt when a reward is given. Whether or not this is true, the end result - an increase in the behaviour - is the same in both cases. Negative reinforcement and reward can be combined to achieve the same outcome, and to a large extent they can be traded off against each other. Lucy Rees wrote in The Horse's Mind:

"The more that positive reinforcement - praise and relaxation - is used, the less aversion is necessary; and conversely the less praise is used the more aversion is necessary or the horse never learns. But unpleasantness breeds fear, fear tension, tension inability to respond and thus resistance - and training becomes a desparate, dangerous and unpleasant struggle."

I think the salient point here is that, while the desire for pressure or unpleasantness to stop and the relief felt when it does can be quite effective in changing behaviour, the presence of the 'aversive' (pressure or unpleasantness) may also lead to other unwanted side-effects. The more intense or prolonged the aversive is, the more likely are these side-effects to appear. In contrast, there is no risk of these side-effects when the horse receives 'true' rewards.

Letting the horse relax after working is a bit of a grey area, I admit. If it doesn't come immediately after an action, it probably won't have a direct reinforcing effect on that action, because it happens too late for the horse to make a specific association between the action and the 'reward'. That doesn't mean it isn't worthwhile though, as it can foster generally positive (and relaxed) atmosphere and attitudes. It can also provide a period of relief and recovery after repeated applications of pressure - but that doesn't mean it's negative reinforcement either, for the same reason as above.

On the other hand, negative reinforcement and punishment work in opposite directions - one increases a response, the other decreases it. So applying them to the same action/behaviour is liable to cause confusion!

For instance, in his book 'Considering the Horse', Mark Rashid describes the teaching of ground manners (through pressure and release) to a horse who had little understanding of what to do when pressure was applied.

Part of paragraph:
“As I was leading her the short distance, I stopped to pick up a piece of baling twine that I noticed on the ground. Sugar simply ignored the fact that I was no longer moving and kept right on walking, actually dragging me about three feet before stopping. I took hold of the lead rope and applied light, constant pressure downwards and slightly back, without pulling on her. She responded to leaning into the pressure. Obviously this wasn't what I was looking for, so I met her resistance with with some resistance of my own. Again, I was not pulling on her but rather putting the same amount of pressure on her that she was putting on me. This went on for a short time before she started tossing her head in search of a way to get rid of the pressure. When that didn't work, she finally flexed, giving to the pressure. Immediately I released my hold on the rope, giving her a very clear and noticeable reward for the proper response”.

Obviously, this could be read as positive punishment being used on Sugar, but it could also be defined as the pressure and release applied when using negative reinforcement.

I don't read this as punishment (at least, not effective punishment) since the pressure itself didn't stop Sugar pulling. It was the release of pressure, brought about by Sugar's own actions, that actually worked in this case. So I would be inclined to say this was an example of negative reinforcement, with "giving to pressure" being the behaviour that was being increased/taught. However, it is also a good example of what I mentioned earlier: of relief from pressure being equated with reward.

Either way, it sure is a good example of how the language of words can be interpreted and defined by different people.:)

Indeed! :D

[ameliet1971]

But as I said above, only if the horse interprets the pressure as aversive or something they will work to avoid.

capalldubh, purely out of interest - would you interpret this as Positive
Punishment or Negative Reinforcement?
It all depends on the horse and the situation. If the horse was [say] trying to mug or boss me around, I would use negative reinforcement to make him realise his actions where unacceptable but would immediately relax my body language and reward him with a stroke at the slightest improvement in his actions.

When teaching anything new, I always begin with the softest feel and relaxed body language, if he doesn't understand I then ask for the same thing in a different way. I don't think any horse can learn by being bullied into anything, but neither do I think one can learn by positive reinforcement alone. :).
:)

it would be so helpful if the big trainers explained the science behind the terms, so that people from different schools of thought in terms of training were able to communicate more effectively
The problem is, there are probably quite a few people that aren't interested in the science behind the terms. Like me, they'd rather improve their horsemanship by studying horse behaviour/psychology and then putting that knowledge into practice.

Apart from clicker, are there any other trainers or training methods that actually find the need to regulary use these terms when teaching their students? :confused:

.

Sorry Francis Burton, cross posted. Could you also give your views on whether my previous post describes Negative Reinforcement or Positive Punishment.

Thanks

[capalldubh]

studying horse behaviour/psychology

But horse behaviour and psychology involves learning theory! :)

Like me, they'd rather improve their horsemanship by studying horse behaviour/psychology and then putting that knowledge into practice.
so I would say that exactly descibes how I try to learn too :)

I would use negative reinforcement to make him realise his actions where unacceptable but would immediately relax my body language and reward him with a stroke at the slightest improvement in his actions.
If you use something the horse likes to stop them doing something they're already doing, the learning theory description would be positive punishment - punishment always means the behaviour in question stops or becomes less frequent or intense. Reinforcement means that the behaviour in question becomes more frequent or intense - so if there's a behaviour you're waiting for - i.e. the horse is figetting, you correct it (using some form of pressure it doesn't like) so that the fidgetting stops, but you continue to apply the pressure until the horse stands still then release - I would say you have punished the fidgetting and negatively reinforced the standing still.

You could try training an incompatible behaviour - asking the horse to drop its head (hard to fidget with your nose on the ground) and then reward that, or you could ignore the fidgetting and reward the instant the horse stopped moving, which would also work :)

but neither do I think one can learn by positive reinforcement alone. In nature, learning comes about through both types of punishment and both types of reinforcement. We can choose in training how much of each we use - our overall success is likely to be determined how what we choose interacts with the horse's emotional responses to the training as Francis describes above.

Andrew McLean in Australia would be the main trainer who bases his work on a very scientific and behaviourist approach so uses all the language correctly. The only call I would have on his work is that he also occasionally refers to release of pressure as "reward". But different areas of the brain process release from pressure and something that is rewarding - they are not the same. There are lots of trainers who prefer to use predominantly positive reinforcement and very little negative reinforcement or punishment - Alex Kurland would be the one I find most influential.

[Francis Burton]

Apart from clicker, are there any other trainers or training methods that actually find the need to regulary use these terms when teaching their students? :confused:

Not sure about the "big names" - Monty Roberts perhaps? I believe that Charlotte Dennis (Equine Ethology, the offshoot of Parelli in UK) has talked about the importance of distinguishing "positive and negative reinforcement versus punishment and reward", though in general most avoid the word "punishment" (preferring "correction" instead) because of its bad connotations.

Sorry Francis Burton, cross posted. Could you also give your views on whether my previous post describes Negative Reinforcement or Positive Punishment.
I thought I had! Do you mean Mark Rashid's description of Sugar's pulling? I think that was negative reinforcement because Mark allowed the horse to discover the alternative to pulling, i.e. giving to pressure, whereupon the pressure was released. If he had made her stop by e.g. delivering a sufficiently strong/alarming/unpleasant jolt through the rope rather than maintaining an equal pressure, then I would call that positive punishment. In that case the behaviour would have been ended by Mark's action. As it happened, however, the behaviour was ended by her own yielding action. Does that make sense?

[Kate F.]

But different areas of the brain process release from pressure and something that is rewarding - they are not the same. There are lots of trainers who prefer to use predominantly positive reinforcement and very little negative reinforcement or punishment - Alex Kurland would be the one I find most influential.

I would imagine that different areas of the brain may well also process different types of reward... I don't know, but I would think it's quite likely that it might depend on the sense involved - eg food rewards would be processed differently from tactile rewards?

I think it's REALLY important to distinguish negative reinforcement (removal of the stimulus) from punishment (a negative consequence of an action) - they are opposites. I am not at all convinced that you can train with no negative reinforcement.... or rather you can, but then it's called desensitizing! If you don't, for example, take your leg off when the horse moves forwards, it will very soon stop responding to the leg. It's an old addage - but very true.... the horse learns when you STOP asking! Stopping asking = negative reinforcement.

There are the people who have the patience to attempt free shaping, I think its called. You wait for the horse to do what you want by accident, then reward it. Eventually you build in a cue, so the horse does it on cue. However, it's a very long way around and not terribly efficient or practical for most people. It's much quicker, simpler and easier for all to give the horse a signal (squeeze of the leg, shake of the rope, whistle, wave - whatever you want) that then STOPS when it does the right thing. It's the way they communicate with eachother, and seems to be the way they learn most easily. If you reinforce the negative reinforcement with positive reinforcement, the process goes quicker again.

Even the clickers who SAY they are using only positive reinforcement actually aren't because our own sub-conscious body language creates a pressure/tension and a release just by the fact that we are looking for a particular response - and you have to look for the response to be able to click at the right time! :D

[capalldubh]

I would imagine that different areas of the brain may well also process different types of reward... I don't know, but I would think it's quite likely that it might depend on the sense involved - eg food rewards would be processed differently from tactile rewards?

It is certainly possible, but for me the main point is that you can demonstrate firing in reward pathways on presentation of a positive reinforcement, but not on the removal of a negative reinforcer. Removal of pressure does not equal reward. That doesn't mean removal of pressure does not equal learning - it does, but in a context where the avoidance part of the approach/avoidance system is activated.

It's much quicker, simpler and easier for all to give the horse a signal (squeeze of the leg, shake of the rope, whistle, wave - whatever you want) that then STOPS when it does the right thing.

It is certainly fast. When you want to increase behaviour, both positive and negative reinforcement act at the same speed. The difference is that learning does not occur in a vacuum, and as Francis pointed out above, if you have trained by presenting something the horse does not like and then removing it, the horse learns to produce the behaviour in the when the stimulus is presented, but also learns that unpleasant stimuli tend to occur when you work with them. That's absolutely fine if you have no problem catching your horse and always work with them on a line, and are a good enough trainer to know you also have to train your horse that there are negative consequences to leaving when you want them to stay as well as staying when you want them to leave.

Training with mainly positive reinforcement has the opposite problem - very easy to get your horse to stick to you like glue, but more of a training challenge to persuade them that there are also good consequences to walking away :)

And I would emphasise again that something can be a cue without being negative reinforcement...
Even the clickers who SAY they are using only positive reinforcement actually aren't because our own sub-conscious body language creates a pressure/tension and a release just by the fact that we are looking for a particular response I know of only one person who says they use only positive reinforcement... however again, the example you give of a demand cue isn't necessarily negative reinforcement - by definition, to be negative reinforcement your demand would have to be an aversive - something the horse will work to avoid, and it's not clear that this is in this case.

I trained my horse to move off the leg by teaching him (at liberty) to match my energy on the ground - i.e. walk when I walk, trot when I trot, then added a touch on his side every time we changed gait. When I started riding, the touch on his side was his cue to change gait. It was not negatively reinforced, it was a neutral stimulus that came to have meaning through positive reinforcement. As I said above, nature uses both types of reinforcement and both types of punishment, and so do we in training - but we have the option to choose which we use most frequently (provided we know what they are...) :)

[Kate F.]

Hmm. I think it's a bit of a broad assumption to imply that negative reinforcement necessarily means the stimulus is unpleasant. All it means is that is stops when the horse does the right thing - ie there is a change. It doesn't have to be unpleasant.

[ameliet1971]

I thought I had! Do you mean Mark Rashid's description of Sugar's pulling? I think that was negative reinforcement because Mark allowed the horse to discover the alternative to pulling, i.e. giving to pressure, whereupon the pressure was released. If he had made her stop by e.g. delivering a sufficiently strong/alarming/unpleasant jolt through the rope rather than maintaining an equal pressure, then I would call that positive punishment. In that case the behaviour would have been ended by Mark's action. As it happened, however, the behaviour was ended by her own yielding action. Does that make sense?
I was refering to one of my posts that I asked capalldubh to give her views on (post 3 on page one). :)

Your explanation makes perfect sense Francis and I agree with your opinion that Mark Rashid applied negative reinforcement to Sugar, rather than positive punishment in his description.

.

[Francis Burton]

Hmm. I think it's a bit of a broad assumption to imply that negative reinforcement necessarily means the stimulus is unpleasant. All it means is that is stops when the horse does the right thing - ie there is a change. It doesn't have to be unpleasant.

Isn't an aversive stimulus part of the definition of negative reinforcement?

As I wrote on another board...

For it to work, I think negative reinforcement has to involve some discomfort (physical or mental) at least to begin with. Otherwise what motivates the horse to move in the first place? And how does he know what the cue means? That said, the discomfort might be very small in some cases, horses being the sensitive and reactive animals they are. As the horse learns whatever action you are teaching him and associates the stimulus with doing the action, the stimulus can be reduced to the level of a cue where it no longer has any discomforting or aversive quality. (Or you can pair a pure cue, like a voice aid, with the negative reinforcer initially, and then phase out the -R.)

To which I should add...

Or you can pair a pure cue (neutral stimulus) with reward, and then phase out the +R - just as Capalldubh did!

Here's the original thread:
http://ihdg.proboards91.com/index.cgi?board=talk2&action=display&thread=1201129382

(I think CatherineB's post is particularly illuminating, and Emmal comments on having to wait for behaviours with positive reinforcement.)

[Francis Burton]

I was refering to one of my posts that I asked capalldubh to give her views on (post 3 on page one). :)
Ah, I misunderstood - sorry. :o

I use both negative and positive reinforcement.

It all depends on the horse and the situation. If the horse was [say] trying to mug or boss me around, I would use negative reinforcement to make him realise his actions where unacceptable but would immediately relax my body language and reward him with a stroke at the slightest improvement in his actions.
When I first read this, I thought you were really talking about punishment. But as you clearly are aware of the difference between punishment and negative reinforcement, I'm slightly confused now because I am finding it hard to picture what you actually do with the horse when he tries to mug or boss you. (The bit about relaxing body language and rewarding slightest improvements is fine though!)

When teaching anything new, I always begin with the softest feel and relaxed body language, if he doesn't understand I then ask for the same thing in a different way. I don't think any horse can learn by being bullied into anything, but neither do I think one can learn by positive reinforcement alone. :)
It's great to hear people espousing the gentle approach - it really is! As for learning by positive reinforcement alone, I think I do agree with you in practice - you might be interested to read the thread I referred to above.

Your explanation makes perfect sense Francis and I agree with your opinion that Mark Rashid applied negative reinforcement to Sugar, rather than positive punishment in his description.
And yet it is quite a subtle difference. Thank you very much for quoting this example - it's a useful one to ponder.

[ameliet1971]

When I first read this, I thought you were really talking about punishment. But as you clearly are aware of the difference between punishment and negative reinforcement, I'm slightly confused now because I am finding it hard to picture what you actually do with the horse when he tries to mug or boss you. (The bit about relaxing body language and rewarding slightest improvements is fine though!
If I was leading a horse that tried to boss or mug me, I'd immediately raise my energy and turn him in tight circles.

As mentioned above, at the first signs of any improvement in his actions I'd relax my body language and praise/stroke him. I would repeat the same procedure each time he attempted these antics until he would stand calm and relaxed at my side.

Hope I've explained things clearly. :)

[Kate F.]

Isn't an aversive stimulus part of the definition of negative reinforcement?



I suppose it depends how strictly or broadly you define negative reinforcement. If it's defined merely as the removal of a stimulus - no the stimulus doesn't have to be aversive.

I think there is a lot of human emotion wrapped up in the word "negative". If we called it "positive release" or something, I think people would be more willing to call it that! :D

There's also the very fine and hazy line between what is a cue and what is an aversive stimulus. When the phone rings, is it a cue that someone wants to talk to you, or an unpleasant or irritating noise? If you pick the phone up and there's no one there and it doesn't stop ringing, you'll pretty quickly look for a way to stop it. If you pick it up and it stops ringing and you hear your best friend with some good news, you don't mind at all that the phone rang, and will look forward to it ringing again.

What I mean is, you can't really separate the cue/stimulus from the things that go on around it. Whether the horse finds it uncomfortable and to be avoided or pleasant and motivating (presuming we're not talking about physical pain, just touches, gestures etc.) will depend on the surrounding feelings and events - and particularly the mood/feel of the trainer.

I think, while it's good to analyse the details of our stragegies to some extent, we also need to keep sight of the bigger picture and see the training situation as a whole communication, rather than a series of individual "words" so to speak. It ain't the words, its the way you say them!! :D:D

[Lili & Morgan]

Superb thread :cool:

Bay Mare : Are you refering to what Becky from EE does for training horses, including your mare?

The experts here will correct me, but I guess clicker training : click and used food as a reinforcer is positive reinforcement training.

If you are refering to this training method, I am not convinced it works on its own. According the state of mind of a horse, the horse will or will not eat the reward.
A frighten horse or playful horse won't care less about a click and treat. Others horses are very food oriented ...

I think like Capalldubh says, it makes more sense to use it "all" : positive, negative reinforcement, positive and negative punishment.

Also clicker training was developped with predators animals : marine mammals and dogs etc ...
I am not convinced that horses are as sensitive/repsonsive to this type of training ... :confused:

[ameliet1971]

I think, while it's good to analyse the details of our stragegies to some extent, we also need to keep sight of the bigger picture and see the training situation as a whole communication, rather than a series of individual "words" so to speak. It ain't the words, its the way you say them!! :D:D

I'm in total agreement with you Kate F :)

[molly34]

Interesting thread - but the biggest thing that it shows me is how misused and misunderstood the different techniques and methods can be. Congrats to Est and Capadullah for some very good explanations! :)

Lil & Morgan, yes, clicker training is positive reinforcement. Its an extension of the most basic behaviourist technique (remember the old experiment with Pavlov's dogs and bells?). Fix on a positive reinforcer (eg. a food treat) to reward the desired behaviour, and accompany it with a particular noise (the clicker). The animal will associate the click first with the food and then with the reinforcer/reward. Use it often enough and the click itself becomes the reinforcer. Used correctly, with tasks broken down into small enough stages, positive reinforcement will work with pretty much any animal - my dad even managed to train a cat!! The key is to get to a point where your horse can understand what you want, and what you're rewarding.

To take it further, behaviourists use different degrees of positive reinforcement - at its simplest form you reward everything. This yeilds very fast results (obviously!) but risks producing a horse who is constantly looking for treats/reinforcers and won't do anything without.

'Fixed interval' reinforcement is just as it sounds - you reward the horse every (eg) 3rd time it does the 'right' behaviour. It encourages them to start thinking for themselves a bit more - they don't take the reward for granted but do the right thing because they MIGHT get it.

'Varied interval' - as fixed, but change the gaps, so maybe reward the 3rd, 5th, 9th, 10th, 15th etc etc.... encourages the animal to strive for the reward without knowing when to expect it, so start thinking for themselves a bit more. It also sets you up to stop rewarding the behaviour - just let the intervals get longer and longer until one day they stop. This one works really well with dogs, I've never tried it with horses.

I'm a big fan of negative reinforcement too, the most obvious case with horses being pressure/release. I think training horses with punishment is a big no no though. (But a slap or strategically placed elbow if I'm barged is a totally different thing, lol!)

[Kate F.]

Also clicker training was developped with predators animals : marine mammals and dogs etc ...
I am not convinced that horses are as sensitive/repsonsive to this type of training ... :confused:

I know what you mean - and it's not really my thing either - though there's no doubt it does work, and can be effective for particular things. I was very indifferent to clicker and thought it a bit of a waste of time and missing the point, to be honest, until I saw what some of my students in Russia were doing with it. They were using it for remote situations - rather as you would with dogs. One girl had used clicker to teach her horse to hold one end of the bandages while she rolled them up - cute - and fun. To teach this, she wanted to be able to reward the horse while she was at a distance and holding the bandage herself - so the clicker was a good way. They both enjoyed it - so why not?
Another girl has a horse with very bad laminitis - and she was using clicker to teach various tricks that kept the horse interested and motivated when it couldn't walk very far. Another good use.

For myself, I don't fancy clicker for every day stuff - defining personal space, controlling movement etc. There I think body language is better, and rewarding by touch - scratching and stroking etc. is more appropriate than food/clicker, especially for overcoming the horse's anxieties etc. However, for the more "circus" type things clicker is probably as effective as anything else - and although it uses food at the beginning, is probably preferable to using food treats throughout! It's not the only way, but it is A way, and it does work! ;):D

[Skib]

We have had threads like this before. Intelligent as I am, I don't find the definitions useful - and in the context of my riding they sometimes seem weird. Particularly if one begins to classify all cues to the horse (leg aids etc) as negative.
I'd prefer to look on interaction with the horse as a communication system. And you use it differently when teaching the horse something new, or refining something, than you do once the horse has learned?
How one applies that system Bay Mare may vary according to what you want to do with your riding.

I'm finding classical dressage is different. Is that what you are asking?

I was (am still) profoundly influenced by Mark Rashid - emphasis on the release.
Using his approach one can teach a horse to halt just on an exhale. It gets so there is nothing to release. So I might let go the rein.
But I have an RI who does classical dressage. No letting go of the rein she says. Why do you want to let go the contact, she asked me.
To tell the horse she has done right. To say thank you.
The horse isn't a fool, says my RI. She knows she has done right. You don't have to tell her. Or if you must, just soften your fingers.
But if I soften my fingers to "thank" her for halt, I cant soften my fingers as a signal that I want her to walk forward again.

The negative or positive stuff - linked to negative and positive emotions seems to me to get in the way. I am having to train myself not to insist all the time on release.
You can talk to a horse nicely and politely and intelligently and in the majority of cases the horse is nice and polite and intelligent in return.

I'd like to rephrase the discussion.
Do horse like being signaled too? Or is every cue negative?
I think the horses I ride all like interaction with me the rider. In moderation. Like we might enjoy doing a crossword but wouldn't want to do it all day.

And when Saffy does Spanish walk, or gives a first riding lesson, you don't feel she is doing it under duress.