When is punishment (potentially) justified?

griffinej5

I am aware that some people might be of the belief that it is absolutely never justified. I'm of the belief that there might be rare cases in which in might be justified. Be aware here that I work with human behavior, and my examples and references will be from texts about humans. However, principles of behavior apply to all species, thus I am not worried that these examples do not apply to dogs or other animals. 

  You're wrong about principles of behavior belonging to all species; dogs and other species don't think the way we do and much of their behavior is based on instinct, whereas our behavior is learned. Dogs and other animals are not capable of understanding the concept of punishment, which is why it has been abandoned as a means of training.  I strongly doubt that you will find any respectable trainer who would recommend punishment under any circumstances.

Honestly I think if someone has to sit down and justify every bit of their training, there might be other issues at play.  If someone doesn't have confidence in what they are training and why, that needs to be addressed regardless of what method is used.  Either it works for the dog or it doesn't.  If it's not the right technique for the dog, then don't use it.

I use -R and +P but of the two I use -R more, and I use it for training new behaviors, not proofing.  The things I've used it for lately generally have results within 2 minutes or less and at that point we've already moved on.  Anything I train regardless of what method is always paired with mark, release, reward.  I might use -R to initially shape the behavior but it's always paired with marker training and positive rewards. 

The problem with the word "punishment" is that it seems to apply the dog has done something "bad" and now must pay for it.  I only speak for myself but that's certainly not how I use it.

My most recent use was Pan's retrieve.  I did a forced hold (-R) and did about three reps total which took about 30 seconds. The forced hold was paired with markers, release, reward (and an extended tug session to release the stress).  Immediately following we moved to backchaining the length of the hold using a marker (my "yes";) and food rewards.  Two weeks later the dog is doing a full formal retrieve with a dumbbell, and if you show him a dumbbell he starts jumping around yipping in excitement because he loves the exercise.

Is it "justified"?  Who cares?  There's a dozen ways to train any dog any new behavior.  We have to know what makes the most sense for our dogs.

   Wow Lies; your post is really interesting. I think it would be fascinating to watch you work with your dogs. I know very little about training. We went to an obedience class with Jessie when we adopted her, and that was about twelve years ago. I do remember though that the method used was positive reinforcement, with lots of patience, and we couldn't use the word "no". I loved training Jessie; it was a great bonding experience. 

Liesje

The problem with the word "punishment" is that it seems to apply the dog has done something "bad" and now must pay for it. 

  That's how I interpreted "punishment" in the OP's post. As such, I think it would confuse the dog and may result in the dog not trusting the owner.

In terms of behavior, we would never use the word understand. Why? Because understanding is not a behavior. Understanding is not observable. In order to be a behavior, it must be observable and measurable. Either way, it is not necessary that the organism be able to explain what will happen, or have anything that looks like understanding in order for a consequence to  be effective. The way we know a consequence is effective is by measuring what happens to the behavior. If the behavior increases in frequency (or duration, magnitude, etc) the reinforcer has been effective,  and if it decreases, it has been punished. The ability to speak and explain what has happened is not needed. Punishment works no matter what the species, unless the specific organism in question has some sort of sensory issues (and I might argue that an organism might not be able to live, because if there is a problem with one sense, you could choose an aversive stimulus to apply to another sense, and I do not believe an organism can survive if it is lacking all senses).

There is some debate about what exactly the basic principles of behavior are, but I assure you that they do apply to all species. Based on who you ask, the principles may include only reinforcement and punishment, it may also include motivating operations (I'm going to super simplify, motivating operations are things that change your motivation for a particular reinforcer, like eating a lot of food lessens your motivation to work for food) maybe setting events (again really simplifying, but things in the environment which dictate how you should behave, like whispering because you are in the library). 

Members of all species have both learned and unlearned behaviors. I refer you to the topics of classical and operant conditioning.

Any terms I am using have no emotional interpretation. I am using the definitions which you may find in a Psychology textbook. By punishment, I mean simply, a consequence which reduces the future frequency of the behavior it follows. Reinforcement means a consequence which increases the future frequency of the behavior it follows. Negative means removal, Positive means addition.

Lies, your justifications can be different than mine, and that is perfectly fine. In my professional practice, I am committed to a particular set of ethical standards. If I am prescribing a punishment procedure, I can be required to justify my choice, and I am required to explain to the person giving consent the potential side effects. For the record, I am not absolutely for or against punishment. I have prescribed a punishment procedure for a client in a plan I  wrote, and I work with a client who has  a plan which requires me to apply an aversive consequence. For my part, if I do it with my dog, I am already aware of the potential side effects, and it is my personal choice to do so only if it is justified for one or more of the reasons I previously listed. You're clearly making an informed choice to do so, and you are also aware of the need to combine with other procedures.

I used R- in the process of teaching Luke's box turn. His most preferred reinforcer is the ball. He prefers to take the ball, run into my room, or onto his bed, and attempt to squish it.  He will not work on it for food, even if it is his most preferred food, even if he hasn't eaten all day. Once I started putting the ball on, if he did it correctly, I let him take the ball and run away with it while I set up the next trial. If he doesn't get all 4 feet on, I grab the ball from him as quickly as possible. He gets nothing while I set up the next trial. It's negative reinforcement since  I removed the ball, and the frequency of getting 4 feet on the box increased. I'd probably choose to do that again with another dog for whom nothing competes with the ball, but clearly that wouldn't work an animal who doesn't care about the ball.

Edited to add... I also meant this as a discussion for those who believe punishment is sometimes justified, or people who might wonder when it might be justified.  If you are truly convinced that it is absolutely never justified, I could regale you with horror stories of things that have occurred under the care of people with such a belief, just as you could cite examples of people abusing punishment. I'm probably not the best person to convince you otherwise. I'd have to refer you to some literature, and you might just be that set in your ways that you won't even consider what you are reading.

The more I learn, the more I am convinced that it's unnecessary.  Dogs repeat behavior that works (for them).  One of the key reasons that dogs continue to exhibit behaviors that humans don't like is that we either insufficiently reinforce and proof the correct or incompatible behaviors, or that the unwanted behaviors are inadvertently reinforced.  In the 1970's, before I was a dog trainer, one of my dogs was killed because he would not recall when chasing something that was more interesting than me (in this case, another dog).  I vowed, from that day on, that my dogs would always have a rock solid recall.  Guess what?  They all have.  It's because I train it religiously according to the principles of science, with an extremely high rate of reinforcement until the behavior is so conditioned that the recall signal elicits some primal "I must come" chemical in their little brains that overrides all else in the universe, and then I proof it in different locations, and THEN I add distractions, and THEN I add more significant distractions.  I don't just teach the recall in the yard and expect my dog to come when he's at the dog park or the beach.  I also never use the word "come" when it will result in all the fun stopping, or in getting one's nails clipped.  I must admit that I don't care for threads in which someone accuses others of not being able to come to a decision that is different from theirs, especially since, at least in my case, I used to believe as you do, and I have since changed my views on the subject.  When a dog doesn't learn, it's nearly always because of a mishandling of the method, not the method.  Science doesn't lie.  Punishment works (it has significant fallout, which is why I don't use it, but it works.)  By the same token, reinforcement works, too.  The only real difference is in ethics and how you wish to conduct the training process.  If you think you need punishment, I would suggest that you don't make that a final decision until you seriously study learning, and attempt to train a dog, either a newly rescued one, or a puppy, from the ground up so that you can truly see the difference. 

 Anne, I think the link you posted on the other thread says it best. Some extremely skilled trainers may be able to use it (I don't recall exactly what the statement there was). That particular type of punishment is one that I personally believe, whether the subject be a human or a canine, should only be applied by extremely skilled trainers, under extremely limited circumstances. That particular punishment is not one I will personally implement, because I lack the appropriate skills and training to do so.

I agree too that when the subject fails to learn, it is not because reinforcement (or punishment) doesn't work. By their very definition, they work. What has probably failed is that particular reinforcer or punisher for that particular organism, or something else in the situation (ie- lacked appropriate pre-requisite skills).Your reinforcer may not have been more valuable than whatever else was available. I'd argue of course that if you offered a milk bone to your dog to call him off deer, you'd quickly not have a dog. However, you may have a dog for whom no other stimulus is as reinforcing as chasing a deer (I'd control it by not letting him off leash, and I am not arguing that as a time when you should use shock, but only as an example of your having no reinforcer available which competes with that which is available from another behavior). Or, you may be dealing with a behavior which is automatically reinforced (the reinforcement results from engaging in the behavior, thus you may have no means of controlling access to it unless you can completely prevent the behavior). 

Then, you may have a case in which you need to reduce a behavior more quickly than you may be able to by reinforcing an alternate behavior, or reinforcing the absence of the behavior (and there are some positive only people who would be opposed to this procedure as well). If you have a person who is injuring himself to the point of causing permanent damage to himself, severe injuries to others, or who has a deadly behavior, I might argue that you would be justified to use punishment in addition to a reinforcement procedure, because it is absolutely critical that you eliminate that behavior very quickly. If we're talking about a dog, what if you have a dog who barks frequently? What  if, as a result of his barking, you will be forced to return him to the shelter, where he will die?

Anne, I totally agree with you, I just come to a slightly different conclusion. It is something I think that can be used, by a limited few people, in a super limited set of situations, combined with other procedures, etc. 

It is absolutely never silly to ask. It's silly to assume you know, get it wrong, and make a fool of yourself. I'm also silly to shorten it when not engaged in conversation with the public...

Anyway, R= reinforcement, P= punishment, += positive, - = negative.

LadyFreckles

I know it seems silly to ask this... but what is -R and +P? Is it like negative reinforcement and positive reinforcement? I studied both in school, I've just never heard the terms shortened like that before.

-R is negative reinforcement also known as "escape training".  You apply a negative stimulus until the dog does what you want and that shuts of the stimulus.  +P is positive punishment also most commonly known as a "correction".  Like your dog starts barking at another dog and you pop his collar to get him to stop.

Because I'm open about my training methods, and how rarely I use punishment in training, I'll provide an example in which I thought using punishment was appropriate, as it was actually quite recently. 

Situation:  Mouse had gotten into our propane fireplace casing/mantle. 
Problem Behaviour:  Gaci fixated on mantle/fireplace, and was digging inappropriately at the sandstone mantle and fireplace casing.

First tried:
- using puzzle filled toys to distract her.  Worked temporarily, but did not work long-term. Once finished, she fixated again on the fireplace.
- used "Leave It".  Worked again, temporarily, but had to ask her to leave it several times over several days.
- using a body block to get between her and the fireplace to redirect her to another task.
- using a "time out" - removal from room as consequence for digging at fireplace. Worked so-so, but motivation for mouse trumped motivation to avoid time-out.
- asked her to Go To Mat for a time, but when release would immediately go back to fireplace when released.

Other factors:
- My ground floor is an open layout, so using gates was not a realistic goal in this situation, to prevent her access to the fireplace.
- Extinction would not have worked well in this situation - risked destruction to expensive fireplace sandstone material, as well as high motivation to reach the mouse that was hiding (Gaci is a very good hunter) to maintain behaviour (hunting was self-reinforcing).

Because of all of that, and because Gaci visually got more stressed the more she practiced digging and fixating, I decided in that case to use a mild aversive to eliminate the behaviour.  The aversive of choice was a spray of water, set up such that she did not know I was spraying. I also did not spray her directly, I sprayed to a spot just nearby her.
 
Effectiveness:  Very effective. First night I used  two sprays, approx. 2.5 hours apart.  Second night I used one spray.  None used since, and behaviour was completely eliminated. 

Fallout: None observed. She is not afraid of the bottle (for instance when I take out spray bottles to clean windows, puppy pees, etc).  She was not even afraid of the water, it was a reaction of disgust rather than fear (the "Blech" factor).

Now, lest anyone read this and think it's a great tool for their dog - realize that this is the first time I have chosen to use a spray of water as a punisher, and I have been training dogs for about ten years.  It wasn't an "I'm angry so I'm going to take it out on the dog" - it was planned, thought out, and executed carefully.  There were other things tried FIRST, to see if we could change the behaviour without using a positive punishment.  In this case, the effectiveness of the punisher was definitely less stressful than allowing her to obsess on the mouse.  The aversive fit the behaviour, and it worked out as it should have. This is the first time Gaci has experienced a positive punisher used to change behaviour. She's eight years old. It's not commonplace, nor should it be.  To date I have never recommended this to any of my clients, because it was not needed. 

In the reality of my life as a trainer, any time I have needed to use an aversive - rarely was it a punisher. What I mean by that is that the dog likely did not learn anything from the aversive, rather the aversive was used to interrupt a very dangerous behaviour. But it wouldn't have had any impact on the future probability of the behaviour occurring. So using a loud "HEY" when an owner dropped the leash and the dog was running to the road. Or using a type of spray to interrupt a dog fight.  Once the crisis was averted, and the danger removed, then we use other means to teach what it is we are looking to teach.

My goals in working with dogs are to remove aversives from training.  It doesn't mean I will never use one, as I gave an example above (as well as information on how rarely I have used it - once, ever), but aversives should be a very rare thing, if used at all, and it should be done carefully, to prevent fallout (or side effects) as much as is possible.  They should not be a first choice, or a choice used "just because someone told me to".  There are ways to change behaviour without using fear, force, or pain, as well there are punishments that don't include these things, and it's about bringing awareness regarding humane punishments (and ain't THAT a loaded word!, but generally talking about using P-, or conditioned negative punishment markers) that change behaviour without harming the dog's emotional wellbeing or fracturing your dog's trust in you.

Using punishment is always a slippery slope. We can't always avoid it, but we should always be thinking about it heavily before using it.

Kim_MacMillan

Because I'm open about my training methods, and how rarely I use punishment in training, I'll provide an example in which I thought using punishment was appropriate, as it was actually quite recently. 

Situation:  Mouse had gotten into our propane fireplace casing/mantle. 
Problem Behaviour:  Gaci fixated on mantle/fireplace, and was digging inappropriately at the sandstone mantle and fireplace casing.

First tried:
- using puzzle filled toys to distract her.  Worked temporarily, but did not work long-term. Once finished, she fixated again on the fireplace.
- used "Leave It".  Worked again, temporarily, but had to ask her to leave it several times over several days.
- using a body block to get between her and the fireplace to redirect her to another task.
- using a "time out" - removal from room as consequence for digging at fireplace. Worked so-so, but motivation for mouse trumped motivation to avoid time-out.
- asked her to Go To Mat for a time, but when release would immediately go back to fireplace when released.

Other factors:
- My ground floor is an open layout, so using gates was not a realistic goal in this situation, to prevent her access to the fireplace.
- Extinction would not have worked well in this situation - risked destruction to expensive fireplace sandstone material, as well as high motivation to reach the mouse that was hiding (Gaci is a very good hunter) to maintain behaviour (hunting was self-reinforcing).

Because of all of that, and because Gaci visually got more stressed the more she practiced digging and fixating, I decided in that case to use a mild aversive to eliminate the behaviour.  The aversive of choice was a spray of water, set up such that she did not know I was spraying. I also did not spray her directly, I sprayed to a spot just nearby her.
 
Effectiveness:  Very effective. First night I used  two sprays, approx. 2.5 hours apart.  Second night I used one spray.  None used since, and behaviour was completely eliminated. 

Fallout: None observed. She is not afraid of the bottle (for instance when I take out spray bottles to clean windows, puppy pees, etc).  She was not even afraid of the water, it was a reaction of disgust rather than fear (the "Blech" factor).

Now, lest anyone read this and think it's a great tool for their dog - realize that this is the first time I have chosen to use a spray of water as a punisher, and I have been training dogs for about ten years.  It wasn't an "I'm angry so I'm going to take it out on the dog" - it was planned, thought out, and executed carefully.  There were other things tried FIRST, to see if we could change the behaviour without using a positive punishment.  In this case, the effectiveness of the punisher was definitely less stressful than allowing her to obsess on the mouse.  The aversive fit the behaviour, and it worked out as it should have. This is the first time Gaci has experienced a positive punisher used to change behaviour. She's eight years old. It's not commonplace, nor should it be.  To date I have never recommended this to any of my clients, because it was not needed. 

In the reality of my life as a trainer, any time I have needed to use an aversive - rarely was it a punisher. What I mean by that is that the dog likely did not learn anything from the aversive, rather the aversive was used to interrupt a very dangerous behaviour. But it wouldn't have had any impact on the future probability of the behaviour occurring. So using a loud "HEY" when an owner dropped the leash and the dog was running to the road. Or using a type of spray to interrupt a dog fight.  Once the crisis was averted, and the danger removed, then we use other means to teach what it is we are looking to teach.

My goals in working with dogs are to remove aversives from training.  It doesn't mean I will never use one, as I gave an example above (as well as information on how rarely I have used it - once, ever), but aversives should be a very rare thing, if used at all, and it should be done carefully, to prevent fallout (or side effects) as much as is possible.  They should not be a first choice, or a choice used "just because someone told me to".  There are ways to change behaviour without using fear, force, or pain, as well there are punishments that don't include these things, and it's about bringing awareness regarding humane punishments (and ain't THAT a loaded word!, but generally talking about using P-, or conditioned negative punishment markers) that change behaviour without harming the dog's emotional wellbeing or fracturing your dog's trust in you.

Using punishment is always a slippery slope. We can't always avoid it, but we should always be thinking about it heavily before using it.

All of it very well said Kim. Though, I would think that even when you are using an aversive to interrupt and teach an appropriate behavior, you're probably punishing that inappropriate behavior. 

I've only ever prescribed the use of a punisher twice with human clients. In both cases, we were talking about serious behaviors which posed a health risk to themselves and others. For both kids, it limited their access to social reinforcement, because basically, nobody would want to be near someone who engaged in these behaviors. For both, the behaviors had gone on for years, and numerous reinforcement only treatments had failed.

For one child, the punishment involved, when he engaged in the behavior, holding his hand down on the table for 3 seconds. When we started, he did this about 4 times per minute. I compared two  possible treatments (one involving only a reinforcer, and the other which included the same reinforcer and the hand hold). I assessed and determined the function prior to prescribing the hand hold treatment. I compared the two for two weeks. At the end of two weeks, when the reinforcement alone was in effect, the behavior was at a rate of 2 per minute. The treatment with the punishment had it at .25 per minute. We elected to continue the treatment with the punishment procedure, and in another 2-3 weeks, the behavior was at a near zero rate (in the environment where it had been applied). For this kid, the problem behavior inhibited his ability to communicate appropriately, among other things, because he does not speak, and he needs to use his hands to  do so. When he wasn't engaged in the problem behavior, his hands were free to learn to do appropriate behaviors.

For the other child, when engaging in the behavior, he would get very messy, and he required a lot of clean up (you can take your guesses as to what this behavior might have been, and if you really want to know, you are free to ask me privately). Previously, they had been giving him a bath to clean him up. He loved baths (he's one of those kids who basically treats the tub like a swimming pool). Showers, he didn't like so much. So, we switched from giving him a bath to giving him a shower. I believe he had 3-4 showers before he started choosing the appropriate behavior exclusively (and reinforcement was available for that). We tried the reinforcer alone first, and the item we selected was his most highly preferred activity. We only added the shower after we had tried that alone. Given that one of the potential options on the line for this problem was a residential treatment facility, and we were able to keep him in his home, I do feel justified in having done so. Compare to the case of a dog whose life or home is on the line, I might agree it is justified in some cases to use a mild aversive.

I used -R to start Pan's "hold" for the formal retrieve.  How do I know it worked?  Because I only had to use it for one session and he figured it out, something we'd been trying for almost a year using clicker and food and he never did correctly.  Previously he would take the dowel and just barely hold it, lolling it with his tongue or letting it fall out of his mouth.  After about three reps using -R (paired with +R...so pressure, then the hold turns off the pressure, then mark, release, reward) he was taking and holding the dowel firmly (I could tap on either side and pull on a string attached).  After the two minutes these three reps took, we moved on and have not had to repeat the -R exercise again.  After two weeks of sporadic training he has a full formal retrieve with an excellent pick up and a firm, calm hold behavior.  Right now our problem (as with everything he does) is that he's enjoying it *too* much and vocalizes, "leaks" drive.  When he sees me grab the dumbbell he gets super excited.  We actually have to make the retrieve exercises less fun for him now.

What should be taken into account is that this is a breed, and these lines and pedigree specifically, that finds security in biting and holding.  Biting full and holding firm and calm is a way of unloading drive and/or relieving stress.  This is true for putting a bit of pressure into the retrieve work, pressure in obedience (and then releasing and "rewarding" with tug games), and in protection work.  If "biting" has a bad connotation for you then use the word "gripping" instead.

 For me, for my dogs, I wouldn't do that to teach a hold. For me, there is nothing about that behavior that makes it that important to justify the use of an aversive. It took me weeks to get Callie to do even a very simple retrieve, just picking up the ball off the ground from directly in front of her and giving it to me. Perhaps because it just isn't that important to me, the decision I considered was whether I would continue with her on trying to teach it,  and whether I was going to continue her in flyball. She got it, finally, and just Friday started to do a dead ball retrieve of about 10 ft. in class. Saturday night at home I finally got retrieving over the jump. If it hadn't happened, I probably would have given it another session of classes since we did have progress,  but after that, she would have been done.

Then again, I have entirely different dogs. Luke is, for all intents and purposes, bred for therapy dog work (pretty much, his mom did or does therapy work, his breeders run a program for children with disabilities, they almost kept him for it). In his case, the side effects of aversives would risk making him unsuitable for his job. Callie on the other hand was bred for nothing specifically, but she also displays a pretty low resistance to extinction (meaning, how long, when a behavior is put on  extinction, she will persist in doing that behavior). Two or three times that she doesn't get the reinforcer, she'll walk away. I'd hazard a guess that with aversives, she might very well walk off even quicker.

Definitely depends on the dog.  My dogs see way more pressure (both mental and physical) in their first 30 seconds of any given round of protection work than Pan did when he learned the hold.  Learning how to come through pressure, how to control the situation in spite of it...that is what makes them solid, stable nerved working dogs that work with power and confidence.  That is why they are bred for the balance of drives they possess and the appropriate thresholds.  I don't want to be breeding dogs that need to be constantly coddled for fear of damaging their nerves forever or take weeks trying to shape something that they can quickly pick up on in about ten seconds if I just show them what I want.  If the dog can't handle any pressure, it's not a German Shepherd dog.  The point of our training is not to have a nice end result and say "look what my dog can do!" but the training itself exposes the strengths and weaknesses in the dog and that cannot be achieved if I were to say I am training every single exercise one way and that's it. 

Also, almost everything we're asking the dogs to do is based on genetics and instinctual behaviors.  Take the retrieve for example.  Regardless of what methods are used, training a working line GSD to retrieve fast and hold firm and calm should not be hard, it should be a fairly quick and smooth process.  IMO it should not take more than a few weeks start to finish (from the dog having never seen a dumbbell to a dog doing formal retrieves with a dumbbell on the flat and over various obstacles).  The dog should possess high prey drive, clear-headedness, and a natural desire to do obedience with the handler.  Now, if someone asked me to train my dog a behavior that doesn't really fit the dog's breeding and genetics, like say how to go to ground or how to point in the field, I would probably not consider -R or +P for training because in this situation I don't think it's really fair, the dog doesn't really have a concept of those exercises and in some cases they may go against his nature.