In Because We Let Them: Part 1, on top of filling some space until I talk about my own training next week, I managed to tie a lesson I learned during one of the doggie training classes at the Austin Humane Shelter into a brief discussion of behaviorism and the 4 primary approaches taken to alter behavior. I also realized that this was going to be too long and I’d need more than my original 2 parts. Which screws up my scheduling but ah well, this is going to be a 4 update week.
In any case, the 4 major approaches to behavior modification I presented were (with some definition): positive reinforcement (providing reward), negative reinforcement (removing constant punishment), positive punishment (providing punishment) and negative punishment (removing a reward).
For each, I gave some dog and human related examples of each and I’d point folks to that article for details. Simply, each can be appropriate under certain conditions and sometimes a mixture may work better. I also looked at a few nuances of them that will have a bit more relevance today especially the whole concept of how random rewards can be more positively reinforcing than anything else.
Finally I talked about a fifth powerful approach to changing someone/something else’s behavior and that was to simply ignore it. This works best when someone is simply looking for A reaction but where the specific reaction is irrelevant. In this case simply responding at all (i.e. to a troll on a forum) is all the reinforcement that they need; ignoring them completely works better than anything else.
Today I’m going to briefly cover some other principles that are relevant and important, again, it’s clear that the topic of behaviorism can and does fill entire books. Then I’ll give a detailed example of how we apply all of this to stop leash pulling to put it into something approximating a real world situation. I’ll wrap up on Friday by trying to make things a bit more generally applicable and talk about humans, clients, etc. and how to go about using this stuff to deal with behaviors you want to change (in others).
There are a few more very important concepts that we regularly use at the shelter that I think are worth talking about, at least in brief and that’s what I want to do next. I’m going to try to avoid too many endless examples, today will be too long as it is. I’ll show some application on Friday.
What Is Your Action Actually Doing?
One important concept, that I brought up very briefly in Because We Let Them: Part 1 is that of what the behavior change strategy you’re employing is actually doing. Because sometimes, and this is more the case with humans than with dogs (humans being marginally more complicated), you get a situation where what you think you’re doing isn’t what you’re doing.
The example I gave before was an athlete who equates ‘more/hard work with success’; for that athlete a coach’s strategy of eliminating wind sprints (negative reinforcement, removing an aversive stimuli) could be taken by such an athlete as negative punishment (removing something that they enjoy). What is meant to be a reinforcer, for that athlete, ends up being a punishment.
Here’s a less obvious example, aimed more at the personal trainers. Let’s say you have a policy (as most do and more should) that if a client is 15 minutes late (but shows up) they get sent home or get an abbreviated workout. This is meant to be positive punishment, right? You’re actively punishing them for being late in hopes that it will prevent them being late.
But let’s say that that particular client is one who doesn’t want to work out in the first place. They could very easily parse what you see as ‘positive punishment’ as positive reinforcement. In their mind, they got rewarded for being late; either they got sent home and didn’t have to work out at all or only had to work out for a shorter period. (I actually had a client like this once, she saw only having a 30 minute workout as a ‘bonus’ for being late since she didn’t want to be there in the first place).
Your ‘punishment’ is actually a reward to them; what do you think will happen to their lateness? What do you think would happen if, in contrast, you subjected them to the worst workout they ever experienced with the understanding that every time they are late they are going to suffer. Given an actual consequence, one of two things is going to happen: either they will stop being late (to avoid punishment) at which point you might find a way to positively reinforce being on time…or they will find another trainer. Either way it’s not your problem anymore.
But my point is that humans are crafty and psychology can be tricky with a lot of individual difference that make things more complicated than they often seem. Sometimes the message you think you’re sending with a given approach isn’t the one you’re actually sending.
What Lesson Are You Actually Teaching?
A related example comes straight out of dog training but applies to humans as well: often the lesson you think you are teaching is not the lesson being taught. For example, when people are trying to teach their dog to sit they will stand there and yell ‘sit sit sit sit sit’ at the dog while doing a variety of things ranging from pushing the dog’s butt down while pulling up on a leash to using a food lure to get the dog to sit back and down. What most fail to realize is that dogs know how to sit, often you can just wait for them to do it spontaneously and then treat it. Soon the dog will start sitting to try and earn a treat. Then you add the command later.
In any case, let’s say that on the 5th ‘sit’ the dog sits and you give it a treat. Most would think that the lesson they just taught was ‘sit’. But the actual lesson taught was that the dog can ignore you the first 4 times and still get rewarded for sitting on number 5. Assuming you want the dog to sit on command, that’s the wrong lesson to teach or reinforce. They can’t get the idea that they can ignore you, eventually listen and still get rewarded. Cuz then they start getting uppity and you don’t want an uppity dog.
Here we get into timing: dogs have a working memory of about 5 seconds. If you don’t treat or punish within that time frame, they have no conception of what they are being treated or punished for and it doesn’t do any good. This is why yelling at your dog when you come home and find poop on the floor does nothing; they have no clue why you are yelling at them and you should realize that it was YOUR fault for not getting them out on time, not THEIR fault for having to poop. It only works if you get them within about 5 seconds.
Now, if you catch them doing something wrong and punish and/or reinforce immediately, they can make the association. So if you catch them chewing your shoe (that they think is a toy), you can take the shoe away (negative punishment) and give them an appropriate toy (positive reinforcement) with lots of praise. This teaches them that chewing shoes is unacceptable but chewing their toys is the right thing to do.
It also means that if you’re trying to teach your dog to do something on the first command, you give the command ONCE. And if it doesn’t work, you need to wait 5-10 seconds for the dog’s brain to reboot and then try again. Then 5-10 seconds more when it fails the second time and try again. And then after 10 minutes of getting nowhere you give up and realize that you should have gotten a cat.
Mind you, here humans are vastly different because, in premise anyhow, they can link reward or punishment to an act that occurred long ago. Anybody with a girlfriend knows that they are still paying for that thing they did at that party….4 freaking years ago. And you can invoke punishment or reward for acts that didn’t occur immediately because you can explain that ‘I’m giving you consequences for something you did last week’ and, being big brained humans, they can link the two non-temporal events. In premise they can anyhow.
Ok, that was a big tangent. My point is that you need to make sure that the lesson you think you’re teaching is the actual lesson being learned. Sometimes what you think is going on and what someone else thinks is going on aren’t the same things and the lesson gets garbled.
In any case, this leads to the next bit of importance.
Anyone who has trained dogs knows that the absolute key to training is consistency. At the shelter, there are two upshots of this. The first is that all BRATT’s are trained to use the same cues in the same way. A given dog might be walked by upwards of 20 people in a given week. If we all used different training cues or words or approaches, all the dog would get is confused. Even if someone comes in with 3 decades of dog training, it doesn’t matter; they do it the AHS way or not at all. It’s for the good of the dogs.
Trainers or coaches working in a situation where athletes may be working with different people on different days should consider this; if your approach to coaching and cueing isn’t standardized in the facility with everybody teaching everything the same way, you will confuse people. Because if trainer A teaches one style of squat and trainer B teaches another, or even if they both teach the same style but they teach with different cues, people don’t get the consistent input that they need for things to make any sense.
So one trainer is telling them to look down, another to look up. Or one is cueing to ‘move the knees out of the way’ and the other is saying ‘push the floor away’ on a deadlift or clean. And the athlete/client just gets confused because the cues aren’t consistent (note that sometimes hearing the same thing differently can actually help with learning because people understand one cue better than another).
But far more important than this (and any parent who has argued with the other parent about punishment knows this) is that things like rewards and punishment must be consistent. Now, as I mentioned, there is some benefit to random rewards, it tends to be more reinforcing than constant rewards AFTER the behavior is established. So you reward the hell out of it until it happens consistently and then you go to a random schedule to reinforce it further.
In direct contrast, the application of positive punishment (if you use it) or negative punishment must be consistent and applied 100%, it can’t just be done some of the time. Simply, if you let the dog (or athlete, or child, or girlfriend) ‘get away’ with a behavior you’re trying to extinguish the lesson doesn’t take. Because the lesson you end up teaching is that they can get away with it some of the time. And depending on the person or dog, they may see how much they can push their boundaries. Worse yet, you may end up accidentally creating a random reinforcement schedule by allowing it to occur.
Here’s an example: bored with ‘sit’ and ‘free’ with ALFIE, I have taught him to play red light/green light. Red light means sit and stay until I give him the command to move with green light. I use this in a variety of situations, at doors (this is to teach him to never run through a door without permission) and at intersections (so in case he gets off leash I can stop him before he gets into traffic). I use it for other things.
So when I take him to potty or out to walk, I’ll get in front of him, open the door and give him a red light command. He has to sit and stay. Sometimes I’ll give him a ‘stare’ command (most use ‘watch’) which means he has to make eye contact with me. Then I’ll give him green light which means he can move/go through the door.
But from time to time he’ll get rambunctious, he’ll see a squirrel or just not be paying attention and he’ll move without the green light command. And I will stop him, make him go back inside, red light him, get a stare out of him and then green light him. He doesn’t ever get to go chase, or potty or whatever if he broke red light without being given green light. Because then he learns that it’s ok to ignore me and not listen to commands.
Do keep in mind that, with dogs, we are trying to teach them to do things that not only aren’t necessarily automatic but that we can’t explain to them. I can’t sit the dog down and go “Look, I don’t want you to pull at your leash and if you don’t stop doing it, I’m going to take you back to the shelter.” It’s behavior change through rote repetition and that affects how you do things. You’re having to get them to do the behavior (or stop doing another one) over time.
As I’ll come back to in the final part of this mess, with humans, you can sit them down and put it in those terms “I dislike it when you do X, and if you do X again, I’m [pick one] firing you as a client, firing you as an athlete, breaking up with you, beating you with a wooden spoon, etc.” You don’t have to spend 3 months applying the various behavior strategies (necessarily) in certain situations because you can simply make it an issue of “Do this or else.” Assuming of course you’re willing to enforce the ‘else’ bit. But that’s tied in with this: any consequence you give in a setting like this has to be applied consistently. It can’t be done sometimes or the person just learns that they can get away with it sometimes.
And in some situations with humans you may be looking at something where there’s an issue of consistency and applying rewards and/or punishment in certain specific fashions. This is how schools work, and raising kids. There will be other situations that arise where you’re looking less at extreme consequences and more at trying to engender longer term beahavior change. And that’s where the consistency issue applies.
But again, with dogs it’s a bit more complicated. So while I may only reward desired behaviors randomly with ALFIE , I always make sure to give consequences for undesired behavior either through positive punishment (which I use exceedingly rarely and will talk about in Part 3) or through negative punishment (not letting him do something he enjoys).
He can’t learn that his actions only sometimes have consequences but that’s exactly what happens if you’re not consistent with how you apply this stuff, especially the punishment options. Because without consequences there is no change in behavior and without applying the consequences consistently, the message gets completely garbled. But it does bring me to the final general concept we use at the shelter which is NILIF.
NILIF stands for Nothing In Life Is Free and everyone reading this should burn it into their brain because it applies to so many things. In basic terms it means that a dog has to work for a treat; we don’t just treat dogs for the sake of treating. Yes, fine, we pet on them and love on them because that’s what keeps them healthy and teaches them that people bring good things for being good doggies.
But in terms of actual food treats or whatever, they don’t just get them to get them. Rather, they have to do something good to earn them. This is just a simple way to teach them that, generally, their actions have consequences good or bad. They have to do something to get something.
Because if you reward all the time, without any sort of context for it, the reward becomes basically meaningless. It would be like telling an athlete ‘good effort’ all the time even if it isn’t. The words lose their meaning because they are given without being earned. So in the kennels, a dog has to show us four on the floor and quiet to earn a treat. If they are up on the cage or making noise, they get nothing (negative punishment).
If we’re training them to do something, they have to show us the behavior, or at least part of it to get a treat. They don’t just get a treat for being cute (ok, not usually). Nothing in life is a free: it works with dogs, it works with athletes, it works with SO’s. To get a reward (or avoid punishment if that’s your approach), they have to earn it. If they just get it to get it (and here I’m assuming you’re actively trying to change a behavior, not just doing something nice for someone you care about or whatever) no lesson is learned.
Note that the same would hold true for punishment. If you just punish all the time regardless of action (think about a parent that spanks their kid for just about anything), it becomes meaningless noise (and look at how well behaved the kid who get spanked all the time are). The kid knows that it’s coming no matter what they do so there’s no point in doing anything differently. If, instead, punishment only comes in response to specific things, the lesson is heard far far louder. I got exactly one spanking in my life and I will NEVER forget it.
And wow did this get long fast. So I’ll cut it here. Tomorrow I’ll give a detailed example of how all of the information in the previous two parts fits together by showing you how to stop a dog from pulling. On Friday I’ll wrap it up and show some other application to training/coaching and life in general. Yes, I’m wordy. Probably because people keep reading my stuff and positively reinforcing my wordiness. See what I did there?
Again, comments are being turned off until the final part on Friday.
- Because We Let Them: Part 3
- Because We Let Them: Part 1
- Because We Let Them: Part 4
- A Dog Trainers Thoughts on Behavior Change
- The Causes of Diet Failure Part 1