Positive Reinforcement

[Alyth]

To reiterate 'positive' means adding something - not necessarily good!! 'negative' means taking away......so pr means giving rubs or treats for doing something good. But my question is how do you ask them to do it in the first place without pressure!!!!???? NR means taking something away. So in this case you ask the horse to do something with gentle pressure and when they start in the way you want you take the pressure away....like asking with a gentle pressure of legs to ask the horse to move forward, then when he goes to move forward you take your legs off. I have no problem with nr and a lot with pr!!! About the only time I use pr is by having some nice food at the front of the float when I am float (trailer!) training!!

 

[Pale Rider]

You see? Everyone is getting bogged down with the definitions and the horse is still not trained! I knew this was going to happen!

Does it really matter what we call it? "A rose by any other name..." etc.

Would it not be be more productive to look at specific training problems and the various way we solve them?

Thats the thing people do get bogged down in this and even more hung up when you say you are using negative reinforcement in your training. Basically its a package and you have to use it all appropriately to be effective.

 

[Booboos]

With my lecturer's hat on GET THEE TO A LIBRARY!

This is not a topic on which you want to gather views on the internet, it's a scientific topic with existing, clear definitions. Read up on classical conditioning which is the root of these ideas, abnd then instrumental learning which is what you are being asked for.

Start by defining your terms with reference to the literature and it will then become much easier to come up with examples.

A good book to start you off is the Handbook of Applied Dog Behavior and Training Lindsay S.R., Blackwell, 2000. You need volume 1. It's written with the informed general reader in mind but is very comprehensive and has great references. Did your lecturer not provide you with a reading list for this assignment? Did you not get any lectures/seminars on learning theories first?

 

[Tinypony]

You see? Everyone is getting bogged down with the definitions and the horse is still not trained! I knew this was going to happen!

Does it really matter what we call it? "A rose by any other name..." etc.

Would it not be be more productive to look at specific training problems and the various way we solve them?

It might be interesting, but would do nothing to answer Op's original question.

 

[Dry Rot]

So, in negative reinforcement we apply an uncomfortable stimulus and the horse does what is wanted to avoid the discomfort? And if the status quo is uncomfortable and something more comfortable is offered, it's positive reinforcement? What is it called when both stimuli are applied at the same time?

For example, light forward pressure is applied to the head collar (say a rope head collar for the sake of argument) to encourage the horse to move forward and a carrot is offered at the same time? As I understand it, even the psychologists can't agree on the definitions which is why I grumble about confusing things with long words when, in practice, the problem is really quite straight forward.

The OP is apparently asking the question for her response to a question at college. In which case, she should demonstrate that she understands all the facets of the argument, not just what the books say. A university education is meant to teach students to think. I was at university which is why I am so argumentative and difficult to please! We are trained to challenge and question generally accepted beliefs.

 

[Tinypony]

It doesn't have to be "uncomfortable". A light pressure on a headcollar is NR, the trick is not to get hung up on the word "negative". (I think Fburton put it well earlier in the thread).

I agree with those telling Op to go and do her research rather than rely on the confusion that always arises here in these discussions. )

 

[canteron]

With my lecturer's hat on GET THEE TO A LIBRARY!

This is not a topic on which you want to gather views on the internet, it's a scientific topic with existing, clear definitions. Read up on classical conditioning which is the root of these ideas, abnd then instrumental learning which is what you are being asked for.

Start by defining your terms with reference to the literature and it will then become much easier to come up with examples.

A good book to start you off is the Handbook of Applied Dog Behavior and Training Lindsay S.R., Blackwell, 2000. You need volume 1. It's written with the informed general reader in mind but is very comprehensive and has great references. Did your lecturer not provide you with a reading list for this assignment? Did you not get any lectures/seminars on learning theories first?

I think OP is clever. Of course she should go into the library and I bet she has, but trying to get a greater understanding of the situation from the real world, in my book is a good way to go. And, if nothing else, you will get a response from a wide range of experiences in HHO to balance against the theory.

After all, isn't there a lot of stuff of here about college students knowing lots of theory but being let down when it comes to practical application?

 

[Dry Rot]

It doesn't have to be "uncomfortable". A light pressure on a headcollar is NR, the trick is not to get hung up on the word "negative". (I think Fburton put it well earlier in the thread).

I agree with those telling Op to go and do her research rather than rely on the confusion that always arises here in these discussions. )

Well, these are discussion forums! I find it quite interesting when they go off topic. They can end up in some strange places and, unlike college lectures, there is no obligation to read!

If the rope head collar isn't uncomfortable, why should the horse move at all? Uncomfortable does not need to mean pain. The horse is given a choice and the means to turn off the "discomfort". That is the crux of the technique. I regularly get into a more comfortable position in bed by turning over -- and I am not a mascochist!

 

[Pale Rider]

I agree with Tinypony, the pressure doesn't have to cause discomfort, but when you remove the pressure, thats negative reinforcment.

 

[Booboos]

I think OP is clever. Of course she should go into the library and I bet she has, but trying to get a greater understanding of the situation from the real world, in my book is a good way to go. And, if nothing else, you will get a response from a wide range of experiences in HHO to balance against the theory.

After all, isn't there a lot of stuff of here about college students knowing lots of theory but being let down when it comes to practical application?

The understanding and application of these terms is not a matter of opinion. These are scientific terms, used in a particular way. They are actually misundertood and misused in this thread so I don't think OP is getting anything of value, no. Academic inquiry is not a matter of canvasing opinion, but of acquiring knowledge.

If the OP had said "positive reinforcement is X (X= clear definition that includes definition of other relevant concepts such as negative reinforcement, positive punishment and negative punishment), I have A & B examples from my experience, does anyone have other examples?" then it would have made sense to canvas a wider field for practical applications. As it stands all she's getting is misundertandings of the fundamental term.

If I had to grade an assignment showing applications of a concept and the student had misunderstood the concept I would have no choice but to fail them.

 

[Pale Rider]

True that the terms are misused and misunderstood on this thread, but that happens everywhere this theory is discussed. I think Skinner himself lamented that the terms he had given certain behaviours caused confusion and prejudice and he thought it might have been better to call them something else, lol.

Positive reinforcers are favorable events or outcomes that are given to the individual after the desired behavior. This may come in the form of praise, rewards.
Negative reinforcers typically are characterized by the removal of an undesired or unpleasant outcome after the desired behavior. A response is strengthened as something considered negative is removed.

Not going any further than this, you can see that in any training scenario it would be most un natural and almost impossible to train using only positive reinforcers. You could not for example put the slightest pressure on a lead rope as this is in fact negative reinforcment ie. when the pressure is released the unpleasant outcome is removed.

Probably why people like Tom Dorrance said 'Its the release that teaches.'

 

[Brightbay]

Not going any further than this, you can see that in any training scenario it would be most un natural and almost impossible to train using only positive reinforcers. You could not for example put the slightest pressure on a lead rope as this is in fact negative reinforcment ie. when the pressure is released the unpleasant outcome is removed.

Probably why people like Tom Dorrance said 'Its the release that teaches.'

I believe another well known US NH type person is known to say "If you do what you always did, you'll get what you always got".

It takes both a willingness to try something different, plus a little imagination and creativity, and then it is perfectly possible to train a young horse to lead without any pressure on a lead rope (and indeed, without a lead rope, which is introduced later). I used scratches as a reward

 

[Pale Rider]

I believe another well known US NH type person is known to say "If you do what you always did, you'll get what you always got".

It takes both a willingness to try something different, plus a little imagination and creativity, and then it is perfectly possible to train a young horse to lead without any pressure on a lead rope (and indeed, without a lead rope, which is introduced later). I used scratches as a reward

Very true, but as horses communicate through pressure, why try and not use it?

If you get the results you want, why not do what you always do to get that result.

 

[Wagtail]

I have a college question that I would be grateful for any views/experiences/opinions on please: 'Can you think of any ways that positive reinforcement rather than negative reinforcement could be used when training a young horse for riding?'
I understand that clicker training is based on PR but I've never seen it used for starting horses or used for ridden work so I have no experience to draw on with this one! Thanks very much in advance.

Clicker training is excellent for ridden work too. You just tape your clicker to the end of your schooling whip. You need to clicker train the horse on the ground first of all. Once it is ingrained, you do not need to reward after every click. When riding this would not be practical. I used to train my mare (to advanced medium) using clicker. I rewarded every five clicks or so. Worked better and faster than without the clicker, especially for extended trot, which my girl had a bit of a mental block about. Once I used the clicker to reward bigger strides she got the hang of it exceptionally quickly.

 

[fburton]

So, in negative reinforcement we apply an uncomfortable stimulus and the horse does what is wanted to avoid the discomfort?

Relieve the discomfort, yes.

The issue of whether discomfort is necessary for negative reinforcement is where I diverge from Tinypony and Pale Rider, who have said that the stimulus doesn't have to be uncomfortable. If it isn't uncomfortable or unpleasant to some degree -- and it can be a tiny degree with horses because they are so sensitive -- then something must be motivating them to act. If not discomfort, then what? Remember we're talking about teaching a horse to do something it hasn't done before, so it's not an already learned response to a signal or cue.

Right, so if you're deliberately making a horse uncomfortable, isn't that a bad thing? Not necessarily! The discomfort can be very mild and short-lasting, but still be effective. It can be so slight that it has a negligible effect on the horse's state of mind - certainly without any outward signs to indicate "Oh, I didn't like that!". Of course, if the discomfort is too great, or too prolonged, or not removed in a timely fashion, or applied in ways that are "unclear" or inconsistent, then the horse may well become confused and upset and training will suffer. Therefore I believe good trainers constantly strive to find the minimum level of discomfort needed to be effective, and work to reduce it further.

(The same applies to aversives used as punishers - i.e. added when or after a behaviour occurs with the intention of lessening the behaviour. When I am teaching a horse to lead, I will let the horse create discomfort for himself when he turns his head too far away from the front by bracing the leadrope against this movement and hence applying pressure to his head. The pressure is removed as soon as the horse straightens up. Thus swinging the head away from the direction of travel is punished / discouraged. I imagine most people watching wouldn't think I was deliberately inflicting discomfort on the horse - the horse looks perfectly happy, after all. However, the truth is I am!)

And if the status quo is uncomfortable and something more comfortable is offered, it's positive reinforcement?

It's usual to consider the status quo as neutral rather than uncomfortable, and aversives and rewards being perceived as relative to this neutral status quo.

What is it called when both stimuli are applied at the same time?

Depends what you mean by "the same time".

If a reward is added at the same time as the aversive is removed, it would be a mixture of positive and negative reinforcement. The end result would still be reinforcement, a promoting or strengthening of the behaviour.

If a reward is added when a behaviour occurs and at the same time as an aversive is added, the aversive would act as a punisher and tend to lessen the effectiveness of the reward, or override it if it was strong enough. Either way, it would be confusing for the horse and unhelpful in training. It would be like telling the horse "Yes!" and "No!" at the same time.

For example, light forward pressure is applied to the head collar (say a rope head collar for the sake of argument) to encourage the horse to move forward and a carrot is offered at the same time?

Okay, this is different again because you aren't applying the pressure and giving the reward at the same time. Rather the prospect of a reward is being offered - the carrot is being used as a lure rather than a reward. It's only a reward when the horse is given it.

As I understand it, even the psychologists can't agree on the definitions

I would say there is pretty good consensus.

which is why I grumble about confusing things with long words when, in practice, the problem is really quite straight forward.

Thinking about behaviour in terms of these basic building blocks really helps, I find, to appreciate why certain things work and, more importantly, understand what's going on when things don't work (and therefore what needs to be changed in order to fix it). If you don't find it helpful, that's fine - no one is saying it's the only correct way.

 

[Dizzy socks]

I though Positive reinforcment was adding something, (Say a pat, or a treat) And Negative reinforcment was taking something away (releasing pressure on a pressure halter, or bit, for exampe)
Is this defination wrong?

 

[Tinypony]

Fburton - if I wanted you to cross the road with me and I took your hand and led you in that direction - it would be NR wouldn't it? But would it involve discomfort?
If I lay my hand on my horse's shoulder to ask it to move away it's NR, but does it involve discomfort?

Dizzy, it's explained further up the thread.

 

[fburton]

Very true, but as horses communicate through pressure, why try and not use it?

Almost everyone who trains horses does use it. It's very hard to avoid using pressure, one way or another! The issue is whether it should be used exclusively, and reward eschewed.

Regardless of how horses communicate, reward is important to horses - both in the survival sense and in the sense of individuals caring about it. So to my mind it doesn't make sense to rule out using it at all. If it is helpful in teaching horses or modifying behaviour (which is essentially the same thing) then I am all for it. Use pressure and release where appropriate and effective; ditto reward; ditto punishment; ditto any other response such as ignoring and not reward a behaviour until extinction occurs.

The argument that reward is 'unnatural' is clearly bogus. Horses modify their own behaviour as a result of reward, so why shouldn't we do so too?

I understand the aspiration of Natural Horsemanship to communicate with the horse in ways that horses communicate with each other. I do think it's important to be mindful of horses' unique characteristics, working with rather than against the equine grain. The important of body language is undisputed. However, I baulk at the notion that everything we do with horses should be as if communicating horse-to-horse using pressure. Why?

First, we aren't horses and normal horses don't tend to regard us as such.
Second, horses are capable of interacting with and learning from other non-horse beings (and things!).
Third, we apply pressure in very unnatural ways that don't resemble anything that horses do to each other.
Fourth, only a fraction of horse communication is achieved via the application and removal of pressure.

 

[fburton]

Probably why people like Tom Dorrance said 'Its the release that teaches.'

I don't think he was claiming that it is only the release that teaches, regardless of his preference for negative reinforcement over positive.

ETA: Surely Dorrance's point was to emphasize the release and its timing over the application of pressure?

 

[fburton]

The understanding and application of these terms is not a matter of opinion. These are scientific terms, used in a particular way. They are actually misundertood and misused in this thread so I don't think OP is getting anything of value, no. Academic inquiry is not a matter of canvasing opinion, but of acquiring knowledge.

You're not saying that all replies to OP show a misunderstanding of the terms, are you? There were some replies that were incorrect/non-standard, but they were corrected by subsequent posts. So I would be surprised if OP were to go away not knowing what was true and what wasn't.

If the OP had said "positive reinforcement is X (X= clear definition that includes definition of other relevant concepts such as negative reinforcement, positive punishment and negative punishment), I have A & B examples from my experience, does anyone have other examples?" then it would have made sense to canvas a wider field for practical applications.

Agreed - this would have been a good and enlightening approach.

As it stands all she's getting is misundertandings of the fundamental term.

With respect, I don't think that is all she is getting.

However, please feel free to say which posts were right (if any)!

 

[Dry Rot]

Some good points and this is getting interesting.

I think many (most?) inexperienced trainers misunderstand how very subtle the signals can be. Sometimes eye contact is enough. Or a very slight body movement.

Sorry, can't be doing with the definitions, though, I'll just continue do what works for me!

I confess I try to break behaviour down into small sections and analyse where things are going wrong, then work on that section. I've a foal here now who resents my approaching from a certain angle. (Long story. His mum was overly protective). So I make a point of approaching at that angle with an offered slice of carrot.

It's working but if I spent my time trying to work out whether it was negative or positive reinforcement, he'd have been in a can by now! I offer the carrot with my left hand and put my right hand where he just views it with suspicion. Eventually, he will give up being suspicious and will accept it as part of the price he must pay to get the carrot, then finally ignore what he now considers threatening completely. The last step will be him offering the head collar for me to take hold of so he gets the reward. I've used a similar technique to teach a head shy pony to actually put his head into the head collar.

 

[Pale Rider]

Almost everyone who trains horses does use it. It's very hard to avoid using pressure, one way or another! The issue is whether it should be used exclusively, and reward eschewed.

Regardless of how horses communicate, reward is important to horses - both in the survival sense and in the sense of individuals caring about it. So to my mind it doesn't make sense to rule out using it at all. If it is helpful in teaching horses or modifying behaviour (which is essentially the same thing) then I am all for it. Use pressure and release where appropriate and effective; ditto reward; ditto punishment; ditto any other response such as ignoring and not reward a behaviour until extinction occurs.

The argument that reward is 'unnatural' is clearly bogus. Horses modify their own behaviour as a result of reward, so why shouldn't we do so too?

I understand the aspiration of Natural Horsemanship to communicate with the horse in ways that horses communicate with each other. I do think it's important to be mindful of horses' unique characteristics, working with rather than against the equine grain. The important of body language is undisputed. However, I baulk at the notion that everything we do with horses should be as if communicating horse-to-horse using pressure. Why?

First, we aren't horses and normal horses don't tend to regard us as such.
Second, horses are capable of interacting with and learning from other non-horse beings (and things!).
Third, we apply pressure in very unnatural ways that don't resemble anything that horses do to each other.
Fourth, only a fraction of horse communication is achieved via the application and removal of pressure.

I'm not saying that anything should be used exclusively or not used in training. The whole gambit should be employed, why not use pressure, why not reward?

I'm sure Tom Dorrance's point was to emphasize the release and its timing over the application of pressure, it's still negative reinforcement.

 

[Pale Rider]

When we talk about horses using pressure to communicate I believe we only see the more obvious signs. I'm sure we miss the pressure a horse exerts with a look for example, or an expression. Other horses don't miss it.

When someone gets kicked, what have they missed before the the horse resorts to a kick.

 

[fburton]

Fburton - if I wanted you to cross the road with me and I took your hand and led you in that direction - it would be NR wouldn't it? But would it involve discomfort?
If I lay my hand on my horse's shoulder to ask it to move away it's NR, but does it involve discomfort?

I think it would depend on whether the leading or moving away was already learned behaviour.

I probably wouldn't experience any discomfort if you were my mother and you had previously taught me to follow you when you took my hand. However, if a stranger attempted to do that with me, I would almost certainly feel discomfort - if not physical, then definitely mental/social.

Has your horse already learned to yield to pressure on his shoulder by moving away? If so, there should be no discomfort in simply responding to a cue - and it would not be considered NR as the behaviour has already been established. If not, has he/she learned to yield to pressure in general but not specifically at the shoulder? Then your horse may or may not be able to generalize and apply this existing 'knowledge' by offering the desired response. If this is a completely new thing you are asking him/her to do, then yes I think that a degree of discomfort must be present, if only transiently.

 

[fburton]

If you get the results you want, why not do what you always do to get that result.

Because there might just be a better way? (Not always, not necessarily, but the possibility is there.)

 

[fburton]

When we talk about horses using pressure to communicate I believe we only see the more obvious signs. I'm sure we miss the pressure a horse exerts with a look for example, or an expression. Other horses don't miss it.

When someone gets kicked, what have they missed before the the horse resorts to a kick.

Sure, I take your point. Nevertheless, even if you count the pressuring looks and expressions there is still much more to equine communication. Concentrating on the agonist (hostile) signals to the exclusion of the rest skews the mindset. Then almost everything becomes an issue of who is further up the pecking order. It is true that if you want a horse to go away from you, threat is the most natural signal to use. But what if you want the horse to follow you, for example? Then, in this paradigm, you have to resort to tricks like punishing not following by chasing away. If your only tool is a hammer, every challenge looks like a nail.

 

[fburton]

Some good points and this is getting interesting.

Yeah, I think so too (though I've probably posted too much and will shortly shut up!).

I think many (most?) inexperienced trainers misunderstand how very subtle the signals can be. Sometimes eye contact is enough. Or a very slight body movement.

I think you're right there. The first step is to see it - perhaps by being shown by someone else? - and then to experience it for oneself.

Sorry, can't be doing with the definitions, though, I'll just continue do what works for me!

No one is forcing you to use definitions! Some will find the ideas helpful and others won't. Vive la difference! One argument which I do think is valid, however, is that if people choose to explore these ideas and to discuss them, it helps if a consistent set of words is used to express them.

I confess I try to break behaviour down into small sections and analyse where things are going wrong, then work on that section.

A very logical and sensible approach.

It's working but if I spent my time trying to work out whether it was negative or positive reinforcement, he'd have been in a can by now!

Well, it does get easier with practice.

I offer the carrot with my left hand and put my right hand where he just views it with suspicion. Eventually, he will give up being suspicious and will accept it as part of the price he must pay to get the carrot, then finally ignore what he now considers threatening completely. The last step will be him offering the head collar for me to take hold of so he gets the reward. I've used a similar technique to teach a head shy pony to actually put his head into the head collar.

A very nice description of a process we can all appreciate, even if we aren't inclined to reduce it to 'learning theory' building blocks.

 

[Tonks]

Depends what you mean by "the same time".

If a reward is added at the same time as the aversive is removed, it would be a mixture of positive and negative reinforcement. The end result would still be reinforcement, a promoting or strengthening of the behaviour.

If a reward is added when a behaviour occurs and at the same time as an aversive is added, the aversive would act as a punisher and tend to lessen the effectiveness of the reward, or override it if it was strong enough. Either way, it would be confusing for the horse and unhelpful in training. It would be like telling the horse "Yes!" and "No!" at the same time.

Just wanted to clarify this and it's not entirely correct.

If negative and positive reinforcers are used stimutaneously, one may OVERSHADOW the other depending upon their relative salience.

So, for example, if low levels of negative reinforcement concur with higher levels of positive reinforcement, the behaviour that is positively reinforced is more likely to reoccur. However, if pressure increases, positive reinforcment may become less salient and the behaviour that is modified by negative reinforcement will become reinforced.

To answer the OP - I think that the nature of horse riding means that we have to primarily train through negative rather than positive reinforcement. A lot of other animals trainers do use positive reinforcement as they can AFFORD to ignore errors made by the animal and thus use positive reinforcement to modify and shape wanted/desired behaviours.

However, as it is dangerous to allow the horse to take flight with a rider on board he cannot wait for the horse to offer up the correct behaviour and reinforce it. Rather, the rider MUST contrive the desired behaviour [throught the use of pressure and release] and thus reinforce it - hence the use of negative reinforcement.

However, this does not mean that PR does not play a role in horse training. It does and it should. However, it is easier to use it in groundwork where it is EASIER to apply to avoid danger.

 

[AengusOg]

...

 

[AengusOg]

A light pressure on a headcollar is NR,

No, the NR is the removal of the pressure on the headcollar.

Fburton - if I wanted you to cross the road with me and I took your hand and led you in that direction - it would be NR wouldn't it?

Again, this is not entirely accurate. The taking of the hand would equate to pressure, whereas the NR would equate to the release of the hand once FB had understood your request and had begun to cross the road voluntarily.

If I lay my hand on my horse's shoulder to ask it to move away it's NR,

Laying a hand on your horse's shoulder may or may not signal to the horse that a response is required. Nr is the cessation of the contact on the shoulder, usually in the case where the horse offers the desired response.