nh vs clicker/reinforcement training

[KP nut]

I've been interested in nh for a while and had various assumptions about 'leadership', 'respect' being the herd alpha (or passive leader)...

But when I watch the groundwork, the horse is being given clear information, and immediate +ve or -ve feedback and so now I'm wondering if actually it's all just different forms of reinforcement training.

I saw clicker as an add-on to nh to address specific issues - ie I viewed it as a particular 'technique' but it seems more like a variant on pressure/release? ('Pressure' is the cue and the ignoring, 'release' is the communication -'yes that's right' and the treat/reward.) If so, what happens to the 'herd hierarchy/leadership' theories?

If successful horse hanlding and training is about how the horse 'sees' you (ie trusts you, respects you) then how come good trainers can work with totally unknown horses in unfamiliar surroundings at demos and get spectacular results almost immediately. Trainers don't say 'I need to spend a month with your horse first so he learns I am a trustworthy and dependable leader'.

 

[stripey]

Most of the available research seems to indicate that dominance heirachies in horses generally arise only where there is competition for resources, such as domestic herds. Researchers such as Dr Marthe Kiley-Worthington seem to suggest that affliliative relationships ie. friendships are by far more important to horses than dominance based ones. 'Horses avoid dominants, not obey' (Lucy Rees)

Once you look at behavioural science and understand the definitions of -re, +re and punishment you can actually look at the 'methods' used by various trainers and begin to decipher what is going on. Using -re and +re we are just communicating with the horse what we would like. Horses are socially co-operative and will generally try to do what they think we want. Obviously experienced trainers are very good at using this communication clearly so the horse understands. There are also nh trainers who do still use domination to train.

Once you understand the science you find that with certain nh trainers there is a big difference between what they tell you they do, and what is actually happening. Whether they themselves don't fully understand or they try and sell it using nice language I don't really know. It's always a good idea to look at what's going on with the sound turned off, then you get a clearer picture, rather than being taken in by the 'language' used.

 

[Francis Burton]

I've been interested in nh for a while and had various assumptions about 'leadership', 'respect' being the herd alpha (or passive leader)...

Don't worry, you're not alone...!

But when I watch the groundwork, the horse is being given clear information, and immediate +ve or -ve feedback and so now I'm wondering if actually it's all just different forms of reinforcement training.

Most of it is, yes.

I saw clicker as an add-on to nh to address specific issues - ie I viewed it as a particular 'technique' but it seems more like a variant on pressure/release? ('Pressure' is the cue and the ignoring, 'release' is the communication -'yes that's right' and the treat/reward.) If so, what happens to the 'herd hierarchy/leadership' theories?

They go right out the window, where they belong in my opinion.

If successful horse hanlding and training is about how the horse 'sees' you (ie trusts you, respects you) then how come good trainers can work with totally unknown horses in unfamiliar surroundings at demos and get spectacular results almost immediately. Trainers don't say 'I need to spend a month with your horse first so he learns I am a trustworthy and dependable leader'.

I think that's because horses are capable of learning very, very quickly. Exactly how quickly depends on how good the trainer is at reading the horse and sending back the right signals in terms of reassuring the horse they are not to be feared, getting the horse's attention, making the horse understand they mean business, etc. in whatever combination is appropriate for the individual horse. A horse can pick up a great deal about a person in the first couple of minutes of interaction, let alone over the course of a 30 min session. The better the trainer, the more this limited time is utilized to good effect. The end result may reasonably be described in terms of leadership, respect or trust, but it is undoubtedly what the trainer does rather that who he/she is that brings about this situation. Moreover, what the trainer does will depend critically on what he/she sees - which is why progress is more rapid when communication goes both ways.

 

[winterwood]

I think that's because horses are capable of learning very, very quickly. Exactly how quickly depends on how good the trainer is at reading the horse and sending back the right signals in terms of reassuring the horse they are not to be feared, getting the horse's attention, making the horse understand they mean business, etc. in whatever combination is appropriate for the individual horse. A horse can pick up a great deal about a person in the first couple of minutes of interaction, let alone over the course of a 30 min session. The better the trainer, the more this limited time is utilized to good effect. The end result may reasonably be described in terms of leadership, respect or trust, but it is undoubtedly what the trainer does rather that who he/she is that brings about this situation. Moreover, what the trainer does will depend critically on what he/she sees - which is why progress is more rapid when communication goes both ways.

I agree.

Also good trainers are good at being effective when working with a horse, and giving clear signals. Their timing is usually pretty good as well

One of the most important factors is getting the horse to feel safe with you. In a prey animal, safety can be a big motivation.

 

[levi1739]

I saw clicker as an add-on to nh to address specific issues - ie I viewed it as a particular 'technique' but it seems more like a variant on pressure/release? ('Pressure' is the cue and the ignoring, 'release' is the communication

I see "clicker training" as "bad timing" unless you consider the click the reward. For me the use of a clicker just interferes with the communication I'm after with the horse.

If successful horse hanlding and training is about how the horse 'sees' you (ie trusts you, respects you) then how come good trainers can work with totally unknown horses in unfamiliar surroundings at demos and get spectacular results almost immediately. Trainers don't say 'I need to spend a month with your horse first so he learns I am a trustworthy and dependable leader'.

Doesn't take a horseman long to establish communications and understanding with a horse. The foundation can happen quite rapidly but refinement can take a lifetime. I know that the horsemen that don't need to spend a month to halter or ride a horse are the ones that I respect.

I think that's because horses are capable of learning very, very quickly.

Learn what? Seem's to me that horses "already know" most things they need.

A horse can pick up a great deal about a person in the first couple of minutes of interaction

Actually it doesn't take that long to confirm what the horse "already knows" about most humans. In minutes horses could write a book about most of us.


Keep on, keepin on

Jack

 

[KP nut]

I see "clicker training" as "bad timing" unless you consider the click the reward. For me the use of a clicker just interferes with the communication I'm after with the horse.

The 'click' becomes rewarding as soon as the horse has made the link between click and treat which is very quick. A horse visibly perks up at the sound of the click.

If the horse is doing something at liberty the click lets them know they've done the right thing so surely it aids communication.

I think your replies are designed to make people think, but please assume I'm a bit dim, and just spell it out!!

How do you 'communicate' and why does the clicker interefere with that communication. And why is the clicker 'bad timing' unless the click is badly timed?

 

[Yann]

I see "clicker training" as "bad timing" unless you consider the click the reward.

That is indeed what happens once you've been doing it for a short while, the horse starts looking for the click rather than the treat once it has associated the two things.

I do wonder about 'pure' clicker fans though, I may be unlucky but none of the ones I've come across seem to do anything particularly demanding with their horses. I would be interested to know how it stands up to situations where the horse and rider are actually put under some pressure.

 

[levi1739]

How do you 'communicate' and why does the clicker interefere with that communication.

I communicate a "correct try" by exhaling. When my timing's good I actually give that "reward before the physical try is visible" but rather at the horses first "thinking" of that correct try.

And why is the clicker 'bad timing' unless the click is badly timed?

once you've been doing it for a short while, the horse starts looking for the click rather than the treat

I don't want my horse looking for either the click or the treat. I want them looking for the "correct try".

Keep on, keepin on

Jack

 

[stripey]

The beauty of the clicker is that, in good hands, it allows you to precisely mark the correct try. Whether or not you can spot that try accurately will depend on your connection/knowledge/horsemanship rather than whether you are a clicker trainer or not. I think if anything working with clicker improves your accuity and enables you to look for that try a bit more. Clicker is a tool, just like a halter, lead rope etc., how it's used and the efficacy depends very much on the hands that are holding it, and the brain controlling those hands. It should be part of a system of positive training and good horsemanship not a substitute.

I'm not sure why people are against the use of food in training, I consider it no different than wages. Although I have a food motivated Cob after a while I've found she is less interested in the treat and more motivated just to try and solve the problem I present. Any good clicker trainer is looking to remove the clicker at the earliest opportunity and get the behaviour on a variable schedule of reinforcement. In basic biological terms the behaviour becomes self rewarding. When the behaviour is performed the click paired with food causes dopamine to be released from the basal ganglia, this dopamine is pleasant for the horse, thus the behaviour becomes self rewarding even when the food is not present. (hope I've remembered the science right)

 

[Yann]

I don't want my horse looking for either the click or the treat. I want them looking for the "correct try".

It's a terribly easy and effective way of getting them to do that.

 

[Bay Mare]

I don't want my horse looking for either the click or the treat. I want them looking for the "correct try".

Keep on, keepin on

Jack

They do look for the correct try. Clicker encourages a horse to try a behaviour because they know that they will not be punished and may be rewarded for their efforts. The click/treat isn't used once the behaviour is established anyway, it's for teaching new things. Once the behaviour is established you move onto a schedule of reinforcement anyway so that you aren't clicking/treating ever time. They don't become resentful or stop working because the behaviour itself becomes rewarding because of the positive connotations.

In basic biological terms the behaviour becomes self rewarding. When the behaviour is performed the click paired with food causes dopamine to be released from the basal ganglia, this dopamine is pleasant for the horse, thus the behaviour becomes self rewarding even when the food is not present. (hope I've remembered the science right)

It's definitely something like that, I didn't even attempt the science bit


I really don't understand how clicker training is 'bad timing'. The basis of clicker is GOOD timing otherwise it doesn't work.


Most 'nh' trainers use predominantly -ve reinforcement in their work with varying levels of +ve reinforcement.

By definition the stimulus in -ve reinforcement must be aversive, it doesn't mean that it hurts but there must be at least a small amount of discomfort for them to want to move away or escape it. The release of pressure, for example, isn't a 'reward' as nh trainers would have us believe but more accurately, relief.

-ve reinforcement does work because the horse will seek to avoid the pressure etc in the future but is that really the best way of learning?


I, personally, don't necessarily see clicker as an add on to nh as I'd rather use predominantly +ve reinforcement with as little -ve reinforcement as possible. I do think that it can (and should) be used as part of nh but then there are some nh trainers who don't agree with giving horse's treats and although there are other ways of rewarding I'd prefer to deal with potential mugging rather than avoid the issue.

But when I watch the groundwork, the horse is being given clear information, and immediate +ve or -ve feedback and so now I'm wondering if actually it's all just different forms of reinforcement training.

It's not always clear communication but that's another story but, yes, it is mostly reinforcement and, in some cases, punishment. I don't buy the 'alpha human' story at all, as I've said many times before, I'm a human and horses don't relate to me as another horse. I don't have the subtleties of body language that a horse does so there's no way that I can accurately mimic their interactions no matter how skilled I am.

 

[Yann]

-ve reinforcement does work because the horse will seek to avoid the pressure etc in the future but is that really the best way of learning?

It's certainly something horses understand innately and use amongst themselves, so there's probably not a lot wrong with it, and there's absolutely no doubt that it works, particularly done well with good feel and timing. As I've said before I think this desire to avoid aversives as much as possible is more for our benefit than the horse's.

 

[levi1739]

It's a terribly easy and effective way of getting them to do that.

Yann can you clarify a bit. I'm not certain if you mean it's effective in getting the horse to search for the click, or do you mean it's effective in searching for the "try"? (Or maybe you meant it's effective at both?)

Bay Mare

They do look for the correct try. Clicker encourages a horse to try a behaviour because they know that they will not be punished and may be rewarded for their efforts.

How does a clicker encourage a horse to try? It may become a signal for something but I don't see how it "encourages" anything. When I get a correct response from my horse I exhale instantly as a signal. You suggest that the clicker is somehow different. How so?

Also, the word "punished" is not a concept that I use when considering horses. Punishment is a very long way from the things I speak of and has no place in my horsemanship. "Being punished" is not a choice I would offer any horse myself.


Keep on, keepin on

Jack

 

[KP nut]

When I get a correct response from my horse I exhale instantly as a signal.

How does the horse know what your exhalation signifies?

 

[levi1739]

The "timing". I "exaggerate" the exhale that happens when the horse responds correctly. This "exaggeration" results in my body language becoming relaxed. These are the same "signals" that a horse learns from it's mother and they can be very subtle.

Keep on, keepin on

Jack

 

[old_woman]

In my non-scientific non-behaviouralist experience of clicker training, which is something I use as an adjunct to other methods of 'training', i would prefer to refer to clicker training as clicker learning.

Again in my experience, it is an extremely effective way of encouraging a horse to learn.

Someone - anyone! horse or human - who is eager to learn is far, far easier to teach,and finds it more enjoyable to be taught, than one which does not want to learn.

I suppose, though, that it depends how you use clicker.

 

[alwaysfallingof]

I don't see what the difference is between a horse trying to find the 'correct try' to achieve an exhalation and a horse trying to find the 'correct try' to achieve a click and subsequent treat. I can see it's about timing. You time the exhalation. You time the click. Horse gets reward.
Ok...you can be more subtle with an exhalation but I'm sure you wouldn't get the same degree of 'try' with my welshie if all you were offering was to be slightly more relaxed.
Surely if you want to motivate the horse to try as much as possible, you use the thing he wants most.

 

[Yann]

Yann can you clarify a bit. I'm not certain if you mean it's effective in getting the horse to search for the click, or do you mean it's effective in searching for the "try"? (Or maybe you meant it's effective at both?)

Yes, both, since one leads to the other. Your sigh, or release of tension is positive reinforcement too, you're presumably rewarding the horse with a rest when it's got something right. All the clicker adds to the equation is a precise marker so the horse knows much more clearly and to the split second what the desirable action is.

 

[levi1739]

i would prefer to refer to clicker training as clicker learning.

Again in my experience, it is an extremely effective way of encouraging a horse to learn.

All the clicker adds to the equation is a precise marker so the horse knows much more clearly and to the split second what the desirable action is

.

I agree that “clicker training” is a learning tool but I think it “marks” the timing for the human rather than the horse. It seems to me that once the human develops their own timing they no longer have need for a clicker as is indicated here.

The click/treat isn't used once the behaviour is established anyway, it's for teaching new things.

I always have trouble with the idea of horses “learning” or humans “teaching” things to animals. I believe that horses “already know” how and why they do the things they do, and that the “learning” is actually needed by the human to accomplish these things “with the horse”. It’s hard to put into words but it seems to me the horse teaches us how we can become more refined with it’s movements, not the other way around.

As I listen to different folks talk about “clicker training” I often find myself pondering their own physical actions while they manipulate the clicker. Do you suppose that most people “inhale” when they click, or rather “exhale”? I believe that many people would do better to become more aware of themselves rather than the horse. Believe me, the horse knows when you inhale or exhale, and everything else too.

I just don’t get the difference between “pressure/release” and “clicker training” other than the timing and manner of signaling the release.
The implication that somehow by using “clicker” we are using only “positive re-enforcement” just doesn’t ring true for me. I would encourage people to discard the clicker and spend their time finding some other ways to “mark” a horses correct response to our requests.

I've heard Chris Cox, a well known western clinician, say this often, "the best reward you can give a horse is to allow it to relax and do nothing". I agree with the thought. My horses enjoy a treat as well as any but they don't have the same appreciation for them as those "instant and constant" rewards that they earn when "trying" to do thier jobs correctly. It's another thing I find troublesome about "clicker training" undersaddle. I strive to "reward" each foots correct placement when I ride and I don't see how that's possible using a clicker or treat.

I know clicker works for many but I personally have found no need to try it. Since the thread is titled "NH vs Clicker" I thought an opposing view might be interesting for some folks to consider.


Keep on, keepin on

Jack

 

[KP nut]

This is really interesting!

I'm loving the thought of 'effective communication' being the main point of all successful training systems. When I use clicker I can almost see the lights go on when the horse has understood something. And I love that feeling. But today I tried to communicate while riding in a different way - experimenting with adjusting my weight/seat to speed up/slow down. He was just as responsive to that as he had been to clicker - his ears were half back as if 'listening' and he became more and more responsive to slight changes. It almost felt telepathic at times. I'd think 'HALT' and he'd stop dead. That felt pretty good too.

Imagine the confused garbled nonsense I give out most ot the time - when I'm not even aware that I'm communicating anything, as my mind is a million miles from my horse.