German Shepherds Forum banner

1 - 20 of 31 Posts

·
Registered
Joined
·
97 Posts
Discussion Starter #1
I am getting quite tired of the reward focused training community pointing to studies that show that compulsion training is damaging to the dog and not as effective as reward-only training etc. etc. and then applying this to balanced training. The studies they reference such as ecollar studies show (not surprisingly) that dogs trained to perform new behaviors are more stressed than dogs who are trained initially with positive reinforcement.

I define balanced training (I am aware this def. is a gross oversimplification) as using positive reinforcement to train all new behaviors then introducing corrections in the later stages to proof the behavior. This is completely different from using compulsion to train behaviors from the get-go.

I would hypothesize that balanced training methods implemented properly would produce behaviors that are more resilient to extinction than behaviors that were trained with positive reinforcement only without the fallout (potential residual stress and superstitious behaviors) of compulsion only training.

Does anyone know of any studies on balanced training when compared to either reward based focused training styles or compulsion focused training styles? If not, I would love to see one to shut up the positive-only crowd.
 

·
Super Moderator
Joined
·
4,739 Posts
It would be nice to see. If they are measuring stress hormones, I would expect a rise of those hormones in training that uses corrections. But stress hormones in themselves are not bad. It is a constant high level of stress that is bad. So you would have to see what exactly that study is measuring.

I do know that dogs themselves use corrections. Usually quick and to the point and then they move on (not fighting, just correcting for over the line rough play as an example). Dogs that never get corrections while training will eventually get one. Life is not perfect. When they do will they be able to handle it? Also which dogs they chose makes a difference. Even between my two GSDs, one needs a firmer correction than the other. What is good for one is too much for the other. How big will the sample be? Far too often the sample size is too small to really be significant.
 

·
Registered
Joined
·
975 Posts
Not with dogs, perhaps, but you could certainly wade into research on training paradigms with lab rats from the 1960s forward. There's a LOT out there, much of which has been successfully applied to training paradigms with humans in various lab situations. It's always risky to make cross-species comparisons, but (as I recall) a lot of the basic principles are applicable.
 

·
Registered
Joined
·
2,278 Posts
I once found a study that showed Belgian malinois were less stressed with shock collar corrections than they were with treats and positive reinforcement. SO wish I could find the link for that!!
 

·
Registered
Joined
·
9,919 Posts
This test was conducted with trained adult police dogs who have been been through a thorough selection from the very beginning of their lives. I wonder how representative they can be for the average dog population. It is an interesting article that is food for thought.
As a pet dog trainer most of my client's dogs were raised with over-used treats and no leadership. I prefer to use the clicker for teaching new behavior and the prong or martingale collar and NILIF to reinforce good behavior if the dog decides to disobey. But still in a reasonable way of course and in situations where I know the dog is able to handle it but decides not to comply.
 

·
Premium Member
Joined
·
408 Posts
Yeah it's called IPO, French Ring, Agility, Hunting......

I don't know of any actual dog specific studies that prove what you're asking OP, but it's classic operant conditioning. We know this works, its been born out for along time.

No dog with a decent amount of fight or spirit (aka drive) is ever going to be completely controlled by positive reinforcement alone, at some point it has to be established that doing what I say ISN'T optional. Period, full stop. You don't show that by giving rewards.

And by the way, the people you reference in your first paragraph... they're right! Compulsion training can destroy your bond with a dog and shut down a dog's spirit (drive to work). It might do what you want because it must, but it's never going to reach it's full potential. Ecollars can cause stress especially if you're using them to train a new behavior. Positive reinforcement IS better at training a new behavior.

But the conclusion they reach is the problem, it's classic confirmation bias. They WANT positive only training to work, so they latch onto any evidence they can find and twist it to suit their needs. The problem is because they want things to be nice and don't want to punish their dog, thus they draw some highly suspect logical conclusions.

Stress is merely your bodies physiological response to a challenge, it effects your body in a variety of ways but the simplest way to explain it is it releases adrenaline into your body. I think we've all taken high school biology, so you'll understand that in and of itself, stress isn't good or bad. It can be extremely helpful or if there is too much or it goes on continuously for too long is when it causes damage.

So causing stress shouldn't be something you fear exposing your dog to, you just need to be mindful to take things slow and progressive and not go overboard. And of course, if properly structured, challenges (i.e. stressors) can build confidence and reduce stress over time.

And again, yeah e-collars can cause stress it's a challenge, it causes pain (however mild that may be) and thus presents a choice, keep doing what I want and deal with this negative stimuli or give in. That's why it's generally not advised to be used to train a new behavior. It's used to reinforce that an already trained behavior will be done despite the circumstances or to stop a dangerous behavior.

Long and short is that dogs might not be humans, but they're still mammals. They're not some alien species. The same chemicals and brain functions exist in them as they do in us, merely on a more rudimentary and basal level. This topic is severely over analyzed sometimes, and arguments get taken to reductio ad absurdum by their own proponents. Strange stuff, kinda reminds me of politics.

ETA grammer <- (yes, that was a joke LOL)
 

·
Registered
Joined
·
2,408 Posts
This test was conducted with trained adult police dogs who have been been through a thorough selection from the very beginning of their lives. I wonder how representative they can be for the average dog population. It is an interesting article that is food for thought.
As a pet dog trainer most of my client's dogs were raised with over-used treats and no leadership. I prefer to use the clicker for teaching new behavior and the prong or martingale collar and NILIF to reinforce good behavior if the dog decides to disobey. But still in a reasonable way of course and in situations where I know the dog is able to handle it but decides not to comply.
Maybe I'm splitting hairs, and it's been a really long time since my learning theory class in college, but I thought a prong/martingale would be an aversive, aka punishment. And in operant conditioning, punishment is (on a simplistic level) a consequence that decreases the likelihood of the behavior being repeated in the future. Reinforcement, on the other hand would increase the likelihood of the behavior being repeated. Just checking my understanding because I am not a trainer and it's been ages since I took a learning psychology class.
 

·
Super Moderator
Joined
·
32,252 Posts
First you'd need everyone to agree on what the term "balanced" means..... :)

I personally hate it, I've seen too many different things labeled balanced that it's virtually meaningless.
 

·
Registered
Joined
·
975 Posts
Couldn't sleep last night, so I read through the article. I don’t know what anyone could reliably take away from this study. The writing is so problematic in places that it's hard to deduce what they actually did and why --- and I'm accustomed to reading fairly mind numbing studies --- what exactly is a "quitting signal," for example, and was it exactly the same for all dogs, in all conditions? Second, it seems that all dogs wore three collars during the training sessions. What was the purpose of that and was any consideration given to potential crossover effects? What exactly was evidence of learning/training, that the dog didn't make the mistake again (yes/no)? Further, if I understand this correctly, if the dog showed evidence of "learning" (i.e., didn't make the mistake), the assessment portion ended. Apparently, if the dog didn't show evidence of learning, it was given a second chance. If that's the case, then any contrasts simply aren't comparable. Finally, that experts adjusted stim levels means that dogs were differentially primed to respond to the ecollar (vs. the pinch collar or quit signal) before the study began. The authors spent a lot of time in the discussion talking around what is a fairly significant design flaw. These are just a *few* of the issues that I saw.

I think the question is an interesting one, but it’s not answered by this study.
 

·
Registered
Joined
·
9,919 Posts
Maybe I'm splitting hairs, and it's been a really long time since my learning theory class in college, but I thought a prong/martingale would be an aversive, aka punishment. And in operant conditioning, punishment is (on a simplistic level) a consequence that decreases the likelihood of the behavior being repeated in the future. Reinforcement, on the other hand would increase the likelihood of the behavior being repeated. Just checking my understanding because I am not a trainer and it's been ages since I took a learning psychology class.
You are right. A prong, martinggale, choke or e-collar is positive punishment, ('positive' meaning 'adding to' to stop behavior). I always thought that operant conditioning, what they wrongly call "positive training" was the way to go. That worked with all my foster dogs and sight hounds until....I got my first WL GSD. That openend up my mind to be more flexible. No way would I have controlled Deja with a 'quitting signal' when she saw a deer. That took an Ecollar for a few weeks and ow she can see a deer and return to me without any form of correction, reward, yes.
I am too getting tired of the fuzzy, all "positive", and dispensing treats until they literally throw up on the way home (local trainer).
 

·
Super Moderator
Joined
·
5,635 Posts
I have a friend ask me one time why all the GSDs she saw in obedience trials were mentally shut down. They respond very slowly to commands, and have zero enthusiasm in the ring. I told her it's because they were all trained using compulsion. I can't tell you how many trainers I've come across that teach a down by stepping on the leash next to the dogs collar, watch the dog flop around like a fish, until it stops fighting. It's so stupid and unnecessary. Why would you want to scare the crap out of your puppy to teach it to down? I was at a trial one time, and a woman had 2 GSDs entered. Both of her dogs when they were in the ring, hit the deck when she made a sudden movement. She was visibly embarrassed by it. Her heavy handedness came back to bite her in public.

I'd love to have this new dog in classes but I can't find anything around me that isn't "all positive" or old school compulsion. If I could find someone that had a bunch of tricks in their bag, that would be lovely. Clicker, prong, e-collar, whatever. So instead I'm just winging it on my own. And, for the first time ever, I have a dog that works better for praise and toys than food. Oh she loves her snacks, but a cat toy really gets her going. If I throw a party with my voice, I get snappy results. I love the clicker, and have trained a lot of IGs and Dobes with it, but Scarlet is a no go. She just looks at me like I'm stupid, and waltzes off. I bring out that silly IKEA mouse and she's happy to train.
 

·
Registered
Joined
·
97 Posts
Discussion Starter #13
Aly,

IMO the gist of the study is that negative punishment can cause more stress (higher cortisol levels) and inhibit learning more than physical corrections whether prong or ecollar in highly motivated dogs. It's basically saying that withholding a cookie or toy from a highly motivated dog is more stressful than a physical correction. It makes sense if you think about it. If someone is a crack/heroin addict, withholding the crack/heroin is more punishing to them psychologically than the physical pain of them getting beaten.
 

·
Registered
Joined
·
9,919 Posts
There is just no 'fit for all'. I train the dog the way that works best for him/her, once I figure him/her out, which is usually pretty quick as they show you who they are right away. The hardest thing is to teach the owners that his /her fur baby is actually a domesticated predator :)
 

·
Super Moderator
Joined
·
4,739 Posts
Aly,

IMO the gist of the study is that negative punishment can cause more stress (higher cortisol levels) and inhibit learning more than physical corrections whether prong or ecollar in highly motivated dogs. It's basically saying that withholding a cookie or toy from a highly motivated dog is more stressful than a physical correction. It makes sense if you think about it. If someone is a crack/heroin addict, withholding the crack/heroin is more punishing to them psychologically than the physical pain of them getting beaten.
reminds me of a boy who told his dad...I'd rather be spanked than have to sit here and listen to your lecture...please.
 

·
Registered
Joined
·
975 Posts
IMO the gist of the study is that negative punishment can cause more stress (higher cortisol levels) and inhibit learning more than physical corrections whether prong or ecollar in highly motivated dogs. It's basically saying that withholding a cookie or toy from a highly motivated dog is more stressful than a physical correction. It makes sense if you think about it. If someone is a crack/heroin addict, withholding the crack/heroin is more punishing to them psychologically than the physical pain of them getting beaten.

Hi @Wolfhund,

Oh, I understood what the purpose of the study was and what the authors claimed to have demonstrated. I just don't think that their claims (as shown in the Discussion) are supported by this study. Perhaps I was unclear (wuz trying to not to be too much of a nerd ;)), but the problematic design aspects (only some of which I noted previously) make it difficult to know what they did demonstrate.

Assuming that there is a viable "take away" to be had with this study, my interpretation of the findings would still likely differ. I suspect that a better designed study would demonstrate the greater efficacy of e and pinch collars (vis a vis quitting signals) in disrupting cortisol flow and, thereby reducing the frequency of unwanted behavior (here defined as 'learning'). So, it's less that quitting signals fair poorly in interrupting unwanted behavior and more that e and pinch collars (in that order) are more effective in doing so. This is a different emphasis than what I understand you to have suggested.

A final thought (and thanks for the conversation, everyone, this has been fun to think about): I think that this study, whatever its limitations, points to the need for the handler to have different skills to manage/train different behaviors --- even in the same dog. (Not a novel concept, I know ;)). I think that comparing the relative efficacy of positive only vs. compulsion only approaches in teaching new behaviors vs. extinguishing existing, undesirable behaviors would be fascinating, and may point to what "balanced" training approaches should look like.

Aly
 

·
Registered
Joined
·
4,432 Posts
I think that comparing the relative efficacy of positive only vs. compulsion only approaches in teaching new behaviors vs. extinguishing existing, undesirable behaviors would be fascinating, and may point to what "balanced" training approaches should look like.

Aly


I can't think of a way to extinguish an existing undesirable behavior using positive only...

Unless one was somehow teaching a new desirable behavior as a replacement strategy...but even that seems impossible without some form of compulsion initially.

Interesting idea though...let me know when you publish your findings ☺
 

·
Registered
Joined
·
5,452 Posts
I have a friend ask me one time why all the GSDs she saw in obedience trials were mentally shut down. They respond very slowly to commands, and have zero enthusiasm in the ring. I told her it's because they were all trained using compulsion. I can't tell you how many trainers I've come across that teach a down by stepping on the leash next to the dogs collar, watch the dog flop around like a fish, until it stops fighting. It's so stupid and unnecessary. Why would you want to scare the crap out of your puppy to teach it to down? I was at a trial one time, and a woman had 2 GSDs entered. Both of her dogs when they were in the ring, hit the deck when she made a sudden movement. She was visibly embarrassed by it. Her heavy handedness came back to bite her in public.

I'd love to have this new dog in classes but I can't find anything around me that isn't "all positive" or old school compulsion. If I could find someone that had a bunch of tricks in their bag, that would be lovely. Clicker, prong, e-collar, whatever. So instead I'm just winging it on my own. And, for the first time ever, I have a dog that works better for praise and toys than food. Oh she loves her snacks, but a cat toy really gets her going. If I throw a party with my voice, I get snappy results. I love the clicker, and have trained a lot of IGs and Dobes with it, but Scarlet is a no go. She just looks at me like I'm stupid, and waltzes off. I bring out that silly IKEA mouse and she's happy to train.
I believe the reason dogs in obedience trials look shut down is because,imo, they are asked to perform exercises not in drive. Actually, I have heard judges and trainers of AKC frown on seeing a dog perform in drive. The nature of CD, CDX, and UD has always been slow execution of desired exercises. I have heard obedience judges say that they don't like that drivey obedience no matter how correct it is....someone explain that to me.
I am member of GS AKC club that has many obedience folks in it that train with food, praise, positive only, balls, etc., they all look dead in the ring because it appeals to the judges and it is the norm. I don't think it's the training methodology....but I could be wrong?.
 

·
Super Moderator
Joined
·
5,635 Posts
There is a girl that has recently moved here from a different part of the country, who teaches obedience and has GSDs. Her dogs aren't shut down. I think I'll bring this up with her, about the judges. I've only watched obedience trials around my area, so I don't have other places to compare. The obedience club has been here since the early 50's, and they have a one size fits all mentality. When I had a young Dobe enrolled in a novice class there, the two of us were bored out of our minds. They didn't even touch on engagement in class. At all. Why would judges want to see dogs that lag when heeling, that take 15 seconds to finally sit? Because that's the norm with the GSDs around here.
 

·
Registered
Joined
·
2,278 Posts
I believe the reason dogs in obedience trials look shut down is because,imo, they are asked to perform exercises not in drive. Actually, I have heard judges and trainers of AKC frown on seeing a dog perform in drive. The nature of CD, CDX, and UD has always been slow execution of desired exercises. I have heard obedience judges say that they don't like that drivey obedience no matter how correct it is....someone explain that to me.
I am member of GS AKC club that has many obedience folks in it that train with food, praise, positive only, balls, etc., they all look dead in the ring because it appeals to the judges and it is the norm. I don't think it's the training methodology....but I could be wrong?.
:surprise: Dear God, Cliff, I wish you were joking... I have toyed with the idea of doing AKC obedience, but after reading with this, I think I'll stick with IP0, thank you very much! I am not a huge fan of dogs being so focused that they can't watch where they are going, but at least the best of them look HAPPY while they are doing it, and some are even wagging their tails.
 
1 - 20 of 31 Posts
Top