The Horse Forum banner

1 - 20 of 46 Posts

·
Super Moderator
Joined
·
23,204 Posts
Discussion Starter #1
So, basically, as many of you already know, this is a subject I have studied indepth, and am passionate about. I have(after initially falling for the 'experts' claims that 'reward doesn't work on horses') used predominantly positive reinforcement(+R, reward) based training for around 25yrs. I have trained other's horses and worked in the horse industry as farrier(but this can include quite a bit of 'training' too) for around 20yrs now. But as always, there is still so much to learn, and I far from 'know it all' so I'm always happy to have a respectful discussion about the subject.

Particularly, to start the 'ball rolling', I've seen some comments by @untamed equestrian regarding mixing +R & -R in training, 'poison cues'(a term/concept I haven't heard before) and other reasoning that might be behind someone's belief in 'purist' +R training & that you should never use aversives or punishment. I'm curious about this, as I have always(well, bar when I first started learning 'clicker training' & kept it 'pure') used a mixed approach and have found it good. But then, it's been a number of years since I've done any serious study into the subject, so perhaps there is more/different info people can share with me as to the science & reasoning behind it?
 

·
Registered
Joined
·
2,176 Posts
Great thread! I'm excited for discussion as this is a subject I am rather passionate about and trying to learn more about every day!
I stand the same on those points that you mention - I've read up on poison cues after hearing about them here but I'm still not sure I grasp the concept entirely so I'm looking forward to more in-depth discussion on the subject.
 
  • Like
Reactions: loosie

·
Super Moderator
Joined
·
23,204 Posts
Discussion Starter #5 (Edited)
OK, read them. My... perhaps rather messy thoughts on what I read...

Poisoned Cues – Empowered Equines This explains that in the author's eyes, a 'cue' is something that is established solely with +R, as opposed to a 'command' which is established with P or -R. So it can't be a 'cue' if you also use any aversives. Seems to assume a 'cue' is a conditioned reinforcer.

Karen Pryor's article The Poisoned Cue: Positive and Negative Discriminative Stimuli | Karen Pryor Clicker Training explains WHY you may not want to use aversives in conjunction with 'cues' - she also differentiates between 'cues' & 'commands', same as above. If a 'cue' is well established with +R, it can become a conditioned +R in itself, but the '+R value' will of course be 'weakened' if it sometimes paired with an aversive.

So... what I got out of that is, if you think of a cue as a conditioned +R, then yes, you will 'poison' it if you throw aversives into the mix. In the same way you would never want to 'click' and follow up with an aversive.

I don't recall hearing the differentiating 'cue' from 'command' in that way. Cues becoming reinforcers in themselves is also a novel idea for me, tho it makes perfect sense. I have just thought of a cue or command interchangeably, as something that is strongly 'attached' to a behaviour, to the degree that you can then give the signal and the behaviour will follow - it comes to mean 'do the behaviour'. The motivation, the consequence, is separate to that. As I don't use 'cues' as conditioned reinforcers, it doesn't matter to me if they become one, or are 'poisoned' on that front. BUT....

Karen goes on to say; "Behavior tends to break down, interestingly, both preceding and following these ambivalent stimuli: preceding, because the preceding behavior may begin to extinguish due to lack of a positive conditioned reinforcer consisting of the now-aversive stimulus, and following, because the behavior that might be punished tends to be avoided. The shift becomes visible in the learner's attitude, which switches from attentive eagerness to reluctance, often with visible manifestations of stress. Even though successful response to a given discriminative stimulus is still followed by reward, if failure is now followed by punishment, you have made that discriminative stimulus ambiguous in terms of predictable outcome. "

THAT is what I'm interested to learn more about - reasons why you might want to only use +R & 'cues' and never 'commands'(by the above definition). Obviously if you want the cue to be a reinforcer, you don't want to 'weaken' it. The referral to it making the stimulus 'ambivalent' is obvious to me. But what about if you don't care about that(or further, perhaps there are other reasons not stated that mean we should care about that?)?

Just going on the above, seems that IF the 'behaviour preceding' IS reinforced(ie not just with a 'cue'), there's no reason that a 'command' should weaken that behaviour. I would suppose it depends what any 'punishment' may be for, how it's applied, how well it's understood, as to whether following desirable behaviour is weakened.

Before going on, a comment was made in another thread, about differentiating between what happens prior to -R and positive punishment. IOW, they were saying(& I somewhat agreed, until giving it further thought - I have mulled that over a lot in the past, and then some again in past days) the aversive that we are 'releasing' in order to -R is not to be confused with punishment. But to me it seems a very hazy 'line' to draw. It does seem to me that is 'splitting hairs'. It seems from this article, she doesn't really differentiate either - or if she does, it's far from obvious here & she is talking of -R & punishment 'in one breath'. So, to be clear, I'm going with that idea - I am thinking/speaking of the preceding aversive to a behaviour you want to -R as a punisher; you apply a 'command' in the way of a touch, and the failure to respond is punished by an increase in pressure, which is instantly released when the horse responds.

So... when punishment is applied for the purpose of being able to -R, what is being punished is NOT the desired behaviour - yielding softly to pressure. That should NEVER be punished, so, so long as the handler is clear and consistent, the horse should have no fear that yielding will be ever punished.

So, back to the first part of Karen's quote above, it doesn't follow, in my mind, that using -R will lead to desired behaviours being less likely - it wouldn't be -R if there was nothing to -R, and wouldn't be called that, if it wasn't reinforcing(strengthening) to the desired behaviour.

Re the "learner's attitude, which switches from attentive eagerness to reluctance, often with visible manifestations of stress" I'm not clear about that one, as to whether this happens as a result of lack of clarity, lack of consistency, strong punishment... or my perception of the 'learner's attitude' - which is, IMO VITALLY important, but maybe I'm missing subtleties I should be paying attention to. All I can say on that note is that I do use a 'mixed' approach, and my horses ARE eager to come play with me, they don't seem reluctant(or why would they come of their own volition, to play with me, rather than hanging with their mates in the paddock?) and don't seem stressed.

So... they're my current thoughts. Let's delve! JoBlue, @gottatrot @Kalraii @Avna ... any others who have any respectful contribution to make??

*** Ed to add... It has come to my attention that I may have the wrong idea about tagging (putting @ in front of names), and that my last comment above came across as me wanting this to be an exclusive convo with only those names named. Pardon if it came across that way to others too - that was far & away from my intention & everyone is welcome!
 

·
Registered
Joined
·
2,176 Posts
This is where it gets rather complicated. In my mind, it all depends soo much. It depends on the horse, depends on the handler, depends on their history separate and together, and depends on what they are working on. In some way, this concept makes sense to me. For example: Heidi is currently learning to fetch. She loves it, of course. Her reinforcement for picking up the object (whip in this case) and handing it to me is her desire for a "reward" (praise and treats) and the fact that I withhold those from her until she does what I desire of her. If I were to "punish" her for not picking it up or dropping it before she hands it to me, yes her interest and enjoyment would drop significantly and soon I'd have to really be forcing the behaviour on her. This is nothing I want. I mean who wants that kind of pressure on them to do something perfectly. But I think the confusion arises from the seemingly (correct me if I'm understanding this wrong) interchangeable use of punishment and -R. In my mind they're entirely different things.
Lets take another example - teaching a horse to back up. If you were to apply gentle pressure on their head with the halter and/or tap their chest with your hand or a whip/stick, this would be considered -R, technically, wouldn't it? But this is not punishment. For a horse that knows the concept of this pressure/release strategy (and the concept can be taught very gently and with minimal or no stress) (also I'd like to add in my opinion it's very important for horses to be taught how to respond to pressure anyway, simply for their overall success in life and I find it's reassuring for horses to have an understanding of these things) this is not stressful, not scary, but rather very clear communication. Pressure and release is just that in my mind - another way to communicate with horses. So not a bad thing at all. Any "languages" we can build with our horses are so useful and important. Back to my example: I can see other ways to teach this behaviour, for sure, for people who don't like pressure/release on principle, don't have the practice to do it correctly and clearly, or the few horses who actually can't handle it and get stressed or shut down about it (from past experiences/negative associations with pressure - one of my mares is an example of this. Any form of "pressure" can trigger this response in her). But in the case that it's a regular happy horse (I doubt I can actually make this a category, but you know what I mean) with an understanding handler I see no problem whatsoever. I'm big on rewards and showing our horses how proud we are of them (I love making them feel proud of themselves, and it's so easy to do that I can't understand why it's not more popular. I suppose it's still the old idea that the best reward a horse desires is peace/absence from pressure...), so to imagine ourselves back in this training situation: I would do the tap tap, slight pressure on head, get a response (even just a small one, like shifting weight back) and I would release all pressure instantly and praise cheerfully and treat the horse. I understand all horses are different, but in my Heidi's case this scenario wouldn't be a problem. Yet, this is a "poisoned cue" now, isn't it? Going off the idea that -R and punishment are interchangeable this should make my horse upset and unwilling to learn. But really all that this has been is teaching her a new "trick", and helping her find the correct answer for the first couple of times until she understand what I mean and can do it happily and willingly just with the "cue" for the reward.
Maybe the difference is completely in what we're working on at the time... There's "training" and "trick training". I couldn't "punish" my horse for not giving me a kiss on command or not fetching when I told her to because that would just make her upset, but if I need to walk by her and she doesn't move her body in whatever way I need her to I can back that up "negatively" (forcing, even, if we were to use that word) and it makes perfect sense to her and doesn't make her resentful. Maybe it still has a lot to do with what is natural for them, like having to move their feet, and what is "add-on" tricks. For me to really come to a clear conclusion I would have to work with an untouched horse I think, that I can be certain hasn't had things "drilled" into it by humans yet. (A good time to point out that my experience is very limited; I'm here to learn whatever I can because I'm, probably a bit, obsessed with knowing and understanding more about these matters!) To return to the two distinctions I made earlier, I confess they are, or at least used to be, generally two different things for me in my work. I'm still trying to find the correct balance, actually, but I think the secret to that is just reading your horse every step of the way and listening to what they're telling you! I understand the concepts, just need to work on consistency.
I think the bottom line is first: you all know your horses best, thus can decide what works for you. But overall, "training" and all that jazz is just a matter of communication. Fluid and clear communication between human and horse. Any language that you have well established with your horse, that you both understand and don't get stressed out about, will work.

I'm not positive that I'm getting my tired mind around this correctly... But the thought that punishment is interchangeable with -R isn't quite clear to me or accurate in my mind. Hate to dwell on it again but I think it depends, always. I think we, @loosie, using a "mixed" approach are "good" enough at reading our horses, have worked hard enough to establish that clear communication, that we can balance these two approaches without "souring" the horse with either. I say that simply from looking at the results - neither of us have encountered un-crossable hurdles with this approach and have kept it up because it works and our horses are happy.

Another thought I just had. While using pressure and release (I think the easiest example of -R in horse training?), say for example we're teaching a horse to yield their shoulder away from us. I start with body language and/or a verbal cue (command? Just a "signal" really) then I start tapping the shoulder with my whip/stick. If for some reason the horse is so confused that I don't get a response at all, what I'm not going to do is keep hammering on her, religiously "increasing pressure", and actually hurt her with my training tool. (I can see where, if it gets to this level in any situation, this could be seen more as punishment. In normal life though, aside from this example to try and keep things simple, if a horse is so "dulled" or unresponsive to this language then there is something wrong that you can't fix by just constantly upping pressure and you'll have to work on that, alone, first. Probably the language isn't understood well enough to begin with or it's causing a shutdown) What I would do instead is touch the halter and use that to show her what the answer is and then release, praise and repeat. By helping them out like this you can easily teach them what response you're looking for, if your first "language" wasn't making complete sense to them. Another case of having to be flexible with your training tactics and willing to adapt them as you go along.

I'm not completely against commands (going with the distinction made here). I don't think they necessarily have to make a horse resentful or make them dislike what you're doing. There are instances where I definitely use "commands" - when I tell them to do something that I will reinforce with negativity if I don't get the response I need. But I don't think this is something that necessarily "sours" horses. Their nature is one that understands this and it's common-place in their life. It all just comes down to their handler being clear and fair even when getting firm, and being a figure that the horse can trust and depend on - their leader (not boss, not drill sergeant, but caretaker and the one that's in control (but not the controlling maniac figure either 😉 I just think there are lots of negative connotations that can come with any of these terms...)).

This communication thing, eh? Rather complicated, confusingly adaptable/flexible, and the single most important thing when it comes to working with horses.
I'm tired enough to not be positive if what I've written makes complete sense even to me, now, however this is where my mind is at so I hope this is constructive! :D
 
  • Like
Reactions: gottatrot

·
Super Moderator
Joined
·
23,204 Posts
Discussion Starter #7
*** It has come to my attention that I may have the wrong idea about tagging (putting @ in front of names), and that my last comment above came across as me wanting this to be an exclusive convo with only those names named. Pardon if it came across that way to others too - that was far & away from my intention & everyone is welcome!
 

·
Registered
Joined
·
5,687 Posts
Interesting posts.

My question is about where the individuality of the horse comes in. Meaning, isn't it important to figure out if our reward is truly a reward or a punishment?

It was mentioned that there is the old thinking that letting a horse rest is a reward. But at some point I noticed that for some horses (and dogs), stillness requires focus and movement is more rewarding. Making some horses "rest" can be aversive and allowing them to move rewarding.

Going further, some horses don't care about carrots, and will seem to take one only to please you. A mare I had did not often find food rewarding.

I've heard if a horse is mouthy, rub their head and get in their mouth because "horses don't like that." Try it with a horse that likes that kind of stimulation and you will be rewarding them. My gelding likes having his face fiddled with and his lips pulled on.

It seems important to assess the horse and learn their personality to know if something is positive or negative to them. A smacking sound might cause one horse to avoid a behaviour and another to do it. My gelding likes the whip cracking sound, he finds it fun and has learned to do behaviours to see if he can make me do it.
 

·
Super Moderator
Joined
·
23,204 Posts
Discussion Starter #10
JoBlue, sounds like you and I have extremely similar ideas about stuff!

This is where it gets rather complicated. In my mind, it all depends soo much. It depends on the horse, depends on the handler, depends on their history separate and together, and depends on what they are working on.
Absolutely! I just don't personally(yet maybe?) see a reason to say 'never'. Or for that matter, why using -R assuming it's well applied, combined with adequate +R, etc, etc... is a problem, generally.

Heidi is currently learning to fetch. She loves it, of course. Her reinforcement for picking up the object (whip in this case) and handing it to me is her desire for a "reward" (praise and treats) and the fact that I withhold those from her until she does what I desire of her. If I were to "punish" her for not picking it up or dropping it before she hands it to me, yes her interest and enjoyment would drop significantly
Yep, agree thoroughly too - I do not think punishment or -R is appropriate in that sort of situation. That would be like how they used to teach a dog to 'fetch' by pinching it's ear & making it yelp, so they can shove the ball(or whatever) in it's mouth(WHY??!!).

I think the confusion arises from the seemingly (correct me if I'm understanding this wrong) interchangeable use of punishment and -R. In my mind they're entirely different things.
Yep, they are vastly different, both in theory and effect. It is not correct for them to be used interchangeably. But as discussed in the other thread and here, you can think of +P as a necessary part of -R.

Lets take another example - teaching a horse to back up. If you were to apply gentle pressure on their head with the halter and/or tap their chest with your hand or a whip/stick, this would be considered -R, technically, wouldn't it? But this is not punishment.
IF the horse viewed cessation of that stimulus as negatively reinforcing, then the stimulus MUST have been aversive to it, no matter how gentle. So, at the root of it, any 'pressure' which is aversive enough to be useful for the sake of negative reinforcement could be called a 'punisher'. But I think the apparent discrepancy is back to your first comment - It depends. The above may be the technical, literal description(tho as we've heard, depends how you understand terms), but I think there are effective differences in how the horse may see mildly & well understood aversive stimulus. On how strong, how clear, what previous understanding/relationship you have with the horse, etc, etc, etc.


For a horse that knows the concept of this pressure/release strategy (and the concept can be taught very gently and with minimal or no stress) (also I'd like to add in my opinion it's very important for horses to be taught how to respond to pressure anyway, simply for their overall success in life and I find it's reassuring for horses to have an understanding of these things) this is not stressful, not scary, but rather very clear communication.
Again, I agree fully with your opinion here.

Yet, this is a "poisoned cue" now, isn't it? Going off the idea that -R and punishment are interchangeable this should make my horse upset and unwilling to learn. ...
Maybe the difference is completely in what we're working on at the time... ... if I need to walk by her and she doesn't move her body in whatever way I need her to I can back that up "negatively" (forcing, even, if we were to use that word) and it makes perfect sense to her and doesn't make her resentful. Maybe it still has a lot to do with what is natural for them,
Yes. While I too can think of many instances where punishment(however applied) WOULD make the horse unwilling, resentful, stressed... It is also my experience that... it ain't necessarily so. So I'm trying to better analyse what the differences are, and whether, if there are any, how I can better see that my apparently willing & unstressed horses AREN'T as they seem to me.

Also for that matter, whether a little short term, low grade stress is a bad thing anyway. On that note, I don't think it is. While chronic, unremitting, or too frequent stress, regardless of intensity, has absolutely been shown to be a problem, both physically and mentally, the only studies I've seen on short term, minor-medium intensity stress has been on humans, and showed that it wasn't a problem, can even be a good thing, as a motivation for learning & for safety & improving situations, etc (sorry, another one I remember from years ago, but not the source. I think it was from a medical TV show though.

like having to move their feet, and what is "add-on" tricks. For me to really come to a clear conclusion I would have to work with an untouched horse I think,
Hmm, not sure if that would give you any more conclusions. Or, for that matter, whether your conclusions would prove anything - I mean, a certain older gentleman watched wild horses & concluded that he had to run them around in a round yard & that would make them understand and trust & respect him.

I've worked with only a few that were truly untouched - newly caught station horses, and Bee the foal I have now, but her mother was friendly already - but the others were long ago, when my knowledge about both equine behaviour, and behavioural training was... rather lacking. I certainly haven't done anything like experiment & record the differences in horses trained with pure +R v's ones trained with a mixed approach. However, I've worked with a large number of very little handled, fearful and/or 'shut down' horses, and as a general rule, I do reckon that until you've got a good relationship/trust & understanding going, using +R & avoiding any aversives at all, if possible, is best. I actually prefer to 'work' with these horses loose in a paddock. Not only do they feel 'safer', that they can leave/stay away if they feel the need, which seems to speed the process, but I feel it's a good 'test' for me to keep me on track - I know that if the horse doesn't want to hang about, I'm approaching something wrongly.

I think we, @loosie, using a "mixed" approach are "good" enough at reading our horses, have worked hard enough to establish that clear communication, that we can balance these two approaches without "souring" the horse with either.
Yep, that's how I feel, but I'm always willing to entertain the thought that I may have it wrong, have missed something... ;-) So I guess what I'm looking for is information that may make me think twice, go to the 'purist' side.

then I start tapping the shoulder with my whip/stick. If for some reason the horse is so confused that I don't get a response at all, what I'm not going to do is keep hammering on her, religiously "increasing pressure", and actually hurt her with my training tool.
Hmm, I was taught to do this, accepted without question that 'if at first you don't get a response, increase stimulus until you do'. I came to the 'conclusion' that you should 'be as soft as possible, but get as firm as necessary', but that increasing to a level that caused actual pain was not helpful, then to still agreeing with that basic premise, but just seeing 'firm *** necessary' to rarely be very 'firm' at all - it's mostly about clarity & motivation, when a horse doesn't do as you ask 'softly', and I guess that's pretty much where I am now - that I feel so long as you're clear, consistent, fair, provide lots of positive reinforcement, it's extremely seldom necessary to 'up the ante'. And that upping the ante when you're not clear is like... shouting louder at a foreigner because they don't understand your language.

I'm not completely against commands (going with the distinction made here). I don't think they necessarily have to make a horse resentful or make them dislike what you're doing. There are instances where I definitely use "commands" - when I tell them to do something that I will reinforce with negativity if I don't get the response I need. But I don't think this is something that necessarily "sours" horses.
Again, absolutely. When we are, for eg. riding beside a main road & the horse wants to cross, you can bet I'm going to 'command' it not to go there! A lot of our training is a 'conversation' where the horse is allowed to say 'I'd rather not', and I believe it's on me to work out a compromise, or way of getting them saying 'yes, love to' instead. But for safety's sake, if I'm going to ride in public, etc, I need my horses to understand how to follow commands calmly and acceptingly too.

I'm interested in whether there is reason to think of this as a 'necessary evil', or something that is, as I suspect, fine to do, assuming it's well applied, balanced with understanding, positive reinforcement, good, trusting relationship, etc.
 

·
Super Moderator
Joined
·
23,204 Posts
Discussion Starter #11
My question is about where the individuality of the horse comes in. Meaning, isn't it important to figure out if our reward is truly a reward or a punishment?
Absolutely! A 'reward'(using that term to mean +R) or punishment is extremely subjective, not only when looking at different animals, but the same one at different times. Eg. carrots may be VERY rewarding to a certain horse, but not if they're afraid - and if a reward (or cue or anything else) is given when the horse is afraid or such, it can, by association, become a punisher. If the horse has been cooped up in a stall, 'rest' may be a punishment, while that same horse may indeed feel negatively reinforced with a rest, at the end of a hard workout. A horse may understand and be comfortable with 'pressure/release' tactics generally, but if it's in a situation where it's afraid, any 'pressure' whatsoever might be 'over-reacted' to and cause the horse significant stress and confusion. So we need to be aware enough to know what is appropriate or not for the horse at that time & place too.
 

·
Registered
Joined
·
2,176 Posts
Again, absolutely. When we are, for eg. riding beside a main road & the horse wants to cross, you can bet I'm going to 'command' it not to go there! A lot of our training is a 'conversation' where the horse is allowed to say 'I'd rather not', and I believe it's on me to work out a compromise, or way of getting them saying 'yes, love to' instead. But for safety's sake, if I'm going to ride in public, etc, I need my horses to understand how to follow commands calmly and acceptingly too.
I think that the important thing is for communication to be fluid, "give n take", if you know what I mean. A horse will be more likely to listen to you "commanding" them in these instances because they know that they are always listened to and they do usually get to have a say in matters. This creates enough trust in their handler that every now and again they can willingly agree to let the human "take the reins" completely (literally lol). Even in these instances they know you understand them yet have decided to insist anyway - generally as long as horses don't feel that their "words" are falling on deaf ears, they're very willing to "listen" and believe in their handler.
I always come back to thinking of this relationship between horse and handler as similar to a parent and child. I think I'm used to this approach/mindset from working with and learning to train birds, but I think it really explains matters with horses a lot. This kind of relationship also needs that "give and take", communication, making sure both sides are heard and understood but the parent is still in charge and can still make decisions for the child. They listen to their child and understand their point of view (I'm saying this generalizing) but in the end it's still up to them to make the final decision. It's tricky to explain this "balance" in words but this comparison makes it easier for me, personally, to understand but also explain what I mean. The same thing goes for boundaries/the behavioural expectations that we have from horses. There are the people who don't like to "correct" horses and would prefer if this isn't necessary at all. The problem is just that it's such a delicate balance and everyone needs to reach/understand it themselves. You can and should expect good, polite behaviour from your horse by setting boundaries for them without being cruel or making the horse resentful. Horses thrive with boundaries, just like children do, which I believe has been proven often. Even though boundaries obviously have to be backed up negatively sometimes, it's still a positive thing. And I think it's not doing them any favours at all when their people refuse to set boundaries.
It's just a careful balance. Not a kind of "one size fits all" or a specific "code", "method"... everyone needs to work it out themselves with their own horses.

Hmm, I was taught to do this, accepted without question that 'if at first you don't get a response, increase stimulus until you do'. I came to the 'conclusion' that you should 'be as soft as possible, but get as firm as necessary', but that increasing to a level that caused actual pain was not helpful, then to still agreeing with that basic premise, but just seeing 'firm *** necessary' to rarely be very 'firm' at all - it's mostly about clarity & motivation, when a horse doesn't do as you ask 'softly', and I guess that's pretty much where I am now - that I feel so long as you're clear, consistent, fair, provide lots of positive reinforcement, it's extremely seldom necessary to 'up the ante'. And that upping the ante when you're not clear is like... shouting louder at a foreigner because they don't understand your language.
Absolutely, and I agree. My example was more an attempt to show that I don't believe in sticking religiously to "methods" and that I'm used to adapting and changing flexibly as I go along, to fit the situation and the horse. This is a problem that I think many people encounter - that they feel they need to really stick with a certain "method" they've been taught, whereas in reality it's much more productive if you can change and adapt as you go along (by reading your horse and their responses).

Yep, that's how I feel, but I'm always willing to entertain the thought that I may have it wrong, have missed something... ;-) So I guess what I'm looking for is information that may make me think twice, go to the 'purist' side.
For sure. I'm also interested in seeing other's perspectives and potentially applying more of these concepts to working with my horses.

Hmm, not sure if that would give you any more conclusions. Or, for that matter, whether your conclusions would prove anything - I mean, a certain older gentleman watched wild horses & concluded that he had to run them around in a round yard & that would make them understand and trust & respect him.
A very valid point... This was evidently not real "watching". But all I'm trying to say is to take a horse that I can be certain hasn't had things drilled into it before, so that I know that my "results" aren't skewy from their past experiences. I really just want to work with the horse itself to see what actually works and what responses are actually honest; would rather not work with what information other people have imprinted onto the horse's mind previously.

I certainly haven't done anything like experiment & record the differences in horses trained with pure +R v's ones trained with a mixed approach.
Neither have I, but this is something I'd be very interested to do. "Experiment" and record my results.
However, I've worked with a large number of very little handled, fearful and/or 'shut down' horses, and as a general rule, I do reckon that until you've got a good relationship/trust & understanding going, using +R & avoiding any aversives at all, if possible, is best.
This is a very good point. Until you have some actual trust in place, any "telling" can be taken badly. It makes sense though; who wants to take orders from someone they don't know or trust. Certainly no one who values their life/safety!
Shut down horses are also a point of "interest" for me. I've only worked with one really shut down one (though it's easy to see those types of responses even from normal horses and it's important to recognize them - stress indicators, WS calls them I think) but for multiple reasons (including health) she was really far "gone". So much so that I, still on a completely different track at the time, had to completely exhaust the "normal" tactics and methods that I had "learned" should "break her out of it", realize that I'd reached the end of the road there, realize how shot my logic had been in the first place, and then start on a completely new track (which, believe you me, completely blew my mind). And now after about a year of not forcing any of the little things (only very rarely when there was no other way, and it always set us back a little) she's an alive horse again with a personality, with thoughts, with things to tell me (this still only goes for out in her pasture, working at her speed. We haven't "trained" or gone out or anything in all this time because anything like that would "trigger" her and she'd think that she had to go back to being "walking dead". So yes, it's a slow process and one has to completely drop one's own expectations - there's simply no other way). Having her show her mind and feelings again has been an amazing thing which still gets me every time, but now everything else is starting to fall into place and she's more willing to do things - and since I'd dropped all my expectations of this, every time it's another shock. It's seriously been a mindblowing experience for me. I took a picture of her the other day - she was right up in my space ears relaxed to either side and both eyes on me (waiting for a treat, no doubt :p). No special picture by most standards but it makes me so happy - she used to not be able to give me two eyes (at least never when I tried to force it and my mistaken perception once was that this was "disrespect"). But now she lets me into her mind. Horses are so forgiving. It's absolutely insane.
I think I always come back to this; it's really been a turning point for me and it's like a whole new world was opened to me. A lot of new information to understand, work with, and balance.
I actually prefer to 'work' with these horses loose in a paddock. Not only do they feel 'safer', that they can leave/stay away if they feel the need, which seems to speed the process, but I feel it's a good 'test' for me to keep me on track - I know that if the horse doesn't want to hang about, I'm approaching something wrongly.
Agreed, 100%. I used to really think that I needed a round pen to accomplish anything with my horses, cuz hey that's how all the celeb trainers show it. But that entire mindset through me off, you know? The idea that the horse being a horse wasn't what I needed to start with but that it was some different "level" that we needed to get anywhere.
Like you say, it's a brilliant "test". It keeps us honest. :p It's so easy, for me at least, to be in denial if I've skipped steps, cut corners, or done anything else that's getting in the way of learning and has made my horse unwilling to keep going. One of the things I need practice in. So to work outside at (true) liberty, you just have to stick to your principles. Like you say, they also feel safer and this opens their minds more.

Yes. While I too can think of many instances where punishment(however applied) WOULD make the horse unwilling, resentful, stressed... It is also my experience that... it ain't necessarily so. So I'm trying to better analyse what the differences are, and whether, if there are any, how I can better see that my apparently willing & unstressed horses AREN'T as they seem to me.
100%. This is where I'm at too.

Also for that matter, whether a little short term, low grade stress is a bad thing anyway. On that note, I don't think it is. While chronic, unremitting, or too frequent stress, regardless of intensity, has absolutely been shown to be a problem, both physically and mentally, the only studies I've seen on short term, minor-medium intensity stress has been on humans, and showed that it wasn't a problem, can even be a good thing, as a motivation for learning & for safety & improving situations, etc (sorry, another one I remember from years ago, but not the source. I think it was from a medical TV show though.
Yes, agree with this. But then again it's a matter of careful balance, and when it comes down to it I think it's better to work with too little stress than too much. One of the fundamental ideas that I try to work with is that everything must start with relaxation.

Yep, they are vastly different, both in theory and effect. It is not correct for them to be used interchangeably. But as discussed in the other thread and here, you can think of +P as a necessary part of -R.
IF the horse viewed cessation of that stimulus as negatively reinforcing, then the stimulus MUST have been aversive to it, no matter how gentle. So, at the root of it, any 'pressure' which is aversive enough to be useful for the sake of negative reinforcement could be called a 'punisher'. But I think the apparent discrepancy is back to your first comment - It depends. The above may be the technical, literal description(tho as we've heard, depends how you understand terms), but I think there are effective differences in how the horse may see mildly & well understood aversive stimulus. On how strong, how clear, what previous understanding/relationship you have with the horse, etc, etc, etc.
Yes, for sure. But maybe the difference is that "punishment" often is an (in some instances) unnecessary negative experience that is just a "response" from you to an action from the horse, whereas -R training has a purpose and when used correctly is a clear form of communication to the horse. So even if it qualifies as this category in technical terms, it can still work just fine for some horses.

Yep, agree thoroughly too - I do not think punishment or -R is appropriate in that sort of situation. That would be like how they used to teach a dog to 'fetch' by pinching it's ear & making it yelp, so they can shove the ball(or whatever) in it's mouth(WHY??!!).
Yes exactly. That would just be ridiculous. The fact that training methods like what you mention used to be normal doesn't tell good things about the brightness of humanity...

Absolutely! I just don't personally(yet maybe?) see a reason to say 'never'. Or for that matter, why using -R assuming it's well applied, combined with adequate +R, etc, etc... is a problem, generally.
Yes. I'm very interested in learning more about arguments for this as I would genuinely like to understand better. But I'm inclined to lean towards thinking that there is no need to say "never", and that -R if used correctly can be very good and effective. I suppose part of the problem might be again that it is a bit of a personal "balance" that everyone has to reach, and this makes it difficult to teach (to people). It's always easier to stick to clearer methods and principles when starting out.

JoBlue, sounds like you and I have extremely similar ideas about stuff!
Haha this is exactly what I thought after returning to the forum, after my last longer break, with a different "world view". It's funny cuz I never expect people to come at these things from the same angle I do, now, because it's just not the normal in the horse world. Very refreshing to actually get to discuss and brainstorm!
 

·
Registered
Joined
·
2,176 Posts
Interesting posts.

My question is about where the individuality of the horse comes in. Meaning, isn't it important to figure out if our reward is truly a reward or a punishment?

It was mentioned that there is the old thinking that letting a horse rest is a reward. But at some point I noticed that for some horses (and dogs), stillness requires focus and movement is more rewarding. Making some horses "rest" can be aversive and allowing them to move rewarding.

Going further, some horses don't care about carrots, and will seem to take one only to please you. A mare I had did not often find food rewarding.

I've heard if a horse is mouthy, rub their head and get in their mouth because "horses don't like that." Try it with a horse that likes that kind of stimulation and you will be rewarding them. My gelding likes having his face fiddled with and his lips pulled on.

It seems important to assess the horse and learn their personality to know if something is positive or negative to them. A smacking sound might cause one horse to avoid a behaviour and another to do it. My gelding likes the whip cracking sound, he finds it fun and has learned to do behaviours to see if he can make me do it.
For sure, this is a very good point. Two of my mares have been at a point in flightiness where my normal cheerful praise (which Heidi practically feeds on) made them go "whoooo whats that??", thus completely destroying the point of trying to reward for a behaviour that in their stress they have now forgotten... Some horses also don't have the understanding of treats being a reward and forget about "doing stuff" and go searching for more treats instead, if they've had inconsistent experiences with treats (this I think can be worked on easily, though generally people opt to simply not use treats). It's really important for people to know and understand their horses.
Your gelding sounds like a hoot; that's awesome!
I've heard if a horse is mouthy, rub their head and get in their mouth because "horses don't like that." Try it with a horse that likes that kind of stimulation and you will be rewarding them. My gelding likes having his face fiddled with and his lips pulled on.
Just to respond to this, though it's probably off-topic a bit: there is also the train of thought that you get in their mouth and rub on their face to satisfy this need which some horses have. I think this likely only works for very specific individual horses though.

Absolutely! A 'reward'(using that term to mean +R) or punishment is extremely subjective, not only when looking at different animals, but the same one at different times. Eg. carrots may be VERY rewarding to a certain horse, but not if they're afraid - and if a reward (or cue or anything else) is given when the horse is afraid or such, it can, by association, become a punisher. If the horse has been cooped up in a stall, 'rest' may be a punishment, while that same horse may indeed feel negatively reinforced with a rest, at the end of a hard workout. A horse may understand and be comfortable with 'pressure/release' tactics generally, but if it's in a situation where it's afraid, any 'pressure' whatsoever might be 'over-reacted' to and cause the horse significant stress and confusion. So we need to be aware enough to know what is appropriate or not for the horse at that time & place too.
Agree completely - you've always got to be able to adapt to the horse you have "right now" (meaning, they can change completely from one day or minute to the next). This reminds me too of the fact that with some horses you can use this train of thought to test where their minds are at, if it's tricky to read from other signs. My mare, if she's not taking treats anymore I know that she's been "triggered" by something and has gone back into her worried mind. I know then to forget anything we were trying to do and just let her do whatever she needs to calm herself and feel safe again. It's just something that makes judging her mindset easy.
 

·
Registered
Joined
·
117 Posts
Sorry about replying to this thread late! I haven't read al the replies, but these are my thoughts concerning mixing. With the poison cue theory (which is still a small theory) the animal is trained with both aversive and +R, which can lead to uncertainty and anxiety in the animal performing the behavior, as he never knows for certain if he will get rewarded or not, or if his behavior will be followed (or shaped) by and aversive. Here's a link to Mary Hunter's article on it:
thoughts-poisoned-cues
I personally do not believe in ever using +P (and avoiding -P as much as possible as well). I also do not agree with dominance theory. Punishment really doesn't solve the problem or answer why the animal was performing the undesired behavior. For example, your horse has ulcers and as you go to do up his girth he swings his head around to bite you. You smack him, and the horse either stops trying to bite you and suffers in silence since he knows he will be hit if he tries to tell you that he is in pain, or the pain from the ulcers are enough that he will try and bite you again the next time you put the girth on. I the end, punishment doesn't solve problems and can really only act as a bandage.
 

·
Super Moderator
Joined
·
23,204 Posts
Discussion Starter #16
Thanks for replying untamed. I read that other article, linked above, by that Jesus guy & thought 'meh' but the way this Mary has put it, makes much more sense to me, why you'd avoid/minimise doing that as much as possible. Seems lack of clarity of what is taught is probably the biggest single factor.
 

·
Registered
Joined
·
223 Posts
FWIW, in my mind, the basic point of +R training is for the horse to have the seeking system turned on which increases the ability to learn, along with an association being formed that what they are actually doing at the time is fun, with the fun becoming an intrinsic motivation in time.

When an aversive is introduced that causes the horse to go over some level of threshold, the seeking system shuts down. Endorphins and all the good stuff shut down with the fear system activating and cortisol being produced. Most or all learning stops or is significantly reduced.

Emphasizing an aversive that throws the horse over some level of threshold. The aversive otherwise does no harm. In fact mild aversive can bring them into an area of more efficient learning if their attention stays below threshold.

I think then that it is important to talk about the level of aversive when talking about using aversives with +R.

But to use an aversive as punishment, the aversive must be strong enough to reduce the chance of the punished behavior reoccurring. It has to be strong enough to cause avoidance. If a horse is worried about the need to avoid punishment, the horse is not in the happy seeking system and is not learning efficiently.

Poisoned Cues: In +R lingo, a cue is not an ask or request as used in -R. A tactile cue in +R lingo just touches the horse in a place or way that has been associated with an activity that has become to be regarded as fun, intrinsically, and that good things happen when that activity is entered into. The good things become the endorphins etc when the activity becomes intrinsically motivated.

The click or similar is the first thing associated with the behavior which means food is coming, eventually, and finally the click is replaced with the tactile cue, or non-tactile cue, which indicates good things are coming, again food at first, until the intrinsic motivation is developed at which time it's very close to just tactile cue and go.

Shawna Karrasch has a 28 YO she started as a 2 YO with pure +R that many people that have only ridden -R horses but have ridden hers and did not observe any difference in the +R trained horse. But great care was taken to teach tactile cues that would resemble a light -R cue by a rider with sensitive hands. I doubt she allowed any rough riders to ride her horse.

Appetitive: Yes, it has to be something the horse enjoys. Food that the like or scratches they like. Keno would just as soon eat dead grass than timothy pellets. Can't figure that out but that's him. And a good scratch in one place can be aversive to another horse. Plus an appetitive with too high of value can become problematic also. Depends on the horse. A shut down horse that just wants to be left alone until he dies may need something with lots of molasses in it at first providing he's not IR.

And that's the extent of my 1 1/4 cents.
 

·
Registered
Joined
·
223 Posts
The trainer in the video below working with a wild horse at first places a pan of food in the middle of the pen as a reward just for the horse looking at her. Then she leaves.

In the notes at the bottom she says both +R and -R is being used in this case, with the food being +R and her leaving being -R.

That is similar to a technique called approach/retreat which is not technically -R as the pressure from the approach stays below threshold since the astute trainer removes the pressure before the horse retreats which would be over threshold.

So I believe this trainer is not using -R as described in operant conditioning.

That said, the video is, to me, great and very useful.

 

·
Registered
Joined
·
223 Posts
And now for the intricasies:

The Premack Principle:

This is the idea that a favored action or behavior can be a reinforcer.

When I first ran across this quite a while back it was associated with something called Grandma's Rule where if you ate your broccoli you could have some chocolate cake. Seemed pretty clear and simple. I did not investigate further.

Then just yesterday I happened upon a YouTube video of an interview with Dr. Patricia Barlow Irick, biologist, animal behaviorist, and mustang trainer. Founder of Explore Mustang Camp

The topic of The Premack Principle came up. The Dr. also trains dogs and she related a situation for explanation of the principle along with the dangers of redirecting an unwanted behavior to a desired behavior. The dog in question enjoyed the behavior of sitting. The dog also enjoyed the behavior of the owner petting the dog.

The dog learned that if s/he jumped up on the owner, the owner would say "sit" which the dog would do and the owner would pet the dog for being such a good dog. The owner believed in reward training and just couldn't understand why it wasn't working to stop the dog from jumping up.

Since the dog can't speak, (in english anyhow), the Dr. spoke for the dog. "I know how to get my owner to let me sit and also to get petted. I just jump up on her. See?"

Quite funny but it certainly does add a dimension to the Premack principle that I shall now be investigating further.

Here's the video but Premack was only discussed briefly.

 

·
Registered
Joined
·
117 Posts
Seems lack of clarity of what is taught is probably the biggest single factor.
The poison cue is still a smaller theory being fleshed out, so yeah, it can be confusing. I like how Mary Hunter explains alot of principles of +R, I find she does a good job.

The dog learned that if s/he jumped up on the owner, the owner would say "sit" which the dog would do and the owner would pet the dog for being such a good dog. The owner believed in reward training and just couldn't understand why it wasn't working to stop the dog from jumping up.
Interesting, I’ve never heard of the Premack Principle before. What you’re describing with the dog reminds me a lot of loopy training. The loop for the dog has become jump on the owner to be asked to sit, get the reinforcer, and then reposition himself to be cued again. (Jumping on the owner) It has become a loop. If reinforced enough behaviors can become rewarding (which can lead to developing behavior chains), and I find that they are certain behaviors that my horse enjoys doing. Here’s a link to it:
Knowing this and keeping loopy training in mind when training has greatly improved how I work with my horse.
 
1 - 20 of 46 Posts
Top