This is going to be controversial, I’m okay with that. Punishment in the world of training is defined as anything that reduces the intensity or frequency of a behaviour. Putting this in context, if your dog is barking at a dog walking past the window, and you happened to sneeze at the same time. The barking stops and never occurs again, the sneeze punished the barking at the dog. Obviously, punishment can be nasty, and can be used as an excuse to do nasty things to dogs. I’m not talking about the extreme examples of punishment here, the things that will obviously cause harm, pain or scare your dog. The tag line for my business is ‘stress free learning with compassion and understanding’ and I maintain that throughout this post.
Ethically, I subscribe to a least intrusive minimally aversive (LIMA) training methodology, with Susan Friedman’s humane hierarchy in mind with this. I always start with the least intrusive, easiest to apply that the dogs will enjoy and that handlers will find easy to implement. I will give this my absolute best before I escalate my approach to be any less fun for the dog or more work for the client. That said, sometimes there are scenarios when protocols have to be escalated, even to the point of using punishment. I illustrate that with the following case study.
Pumba starts a night off in my bedroom, then when a family member goes to bed after I’ve gone to sleep, they open my door and he has free run of upstairs and some of downstairs. That routine cannot change. When I go to bed, Timon is out downstairs and we have no way of securely preventing any potential trigger points between the two dogs unless one dog is shut in a room upstairs or shut in a room downstairs. The evenings are Timon’s time to be free downstairs, which means that Pumba must be shut in a room upstairs. Unbeknownst to me, when the family member opened my door when they went to bed, they had been giving him a small treat. This led to Pumba anticipating the reward and whining, scratching and yipping at the door in order to have the door opened and receive his midnight snack. To start with, this began when he heard Timon going to bed downstairs and the family member coming up. Gradually, it got pushed back 15 minutes before the family member went to bed at their earliest. Now, his behaviour obviously woke me up and I wasn’t appreciative of being woken up between midnight and 2 in the morning to be a glorified door stop. I needed the behaviour to stop, and quickly. As soon as I found out about the bedtime snack, I put a stop to it.
I had the following options:
- Wait for the behaviour to go through extinction and for it to stop naturally when the reinforcement stops (this was an issue because I couldn’t guarantee that the treat was the only reinforcement he was receiving when the door opened. He could easily have gone down to the kitchen to clean up after our dinner, which would keep the behaviour maintained. If the other family members greeted him and gave him a belly rub, that was more reinforcement and the behaviour would still be maintained. These potential alternative reinforcements were entirely out of my control so I couldn’t do this)
- Stay up or wake up before Pumba starting trying to leave the room and train Pumba to settle in his bed quietly and then go to sleep after the session (I like sleep, I can’t honestly say that I would have enjoyed this option or that I could have kept it up whilst training effectively. Furthermore, it could easily have shot me in the foot by functioning as a reinforcer for Pumba and the last thing I wanted from the situation was for it to turn into him demanding a midnight training session!)
So, I was left with punishment. I wanted the behaviour to stop and any attempt to prevent the behaviour from occurring in the first place, or to teach an alternative behaviour weren’t feasible in the situation. I chose to use a timeout. Now timeouts in this scenario can take one of two forms; I left the room or Pumba was removed from the room. Leaving the room myself would have had no impact. He didn’t care I was there, his reinforcer didn’t hinge on anything from me. Pumba being removed from the room changed the context sufficiently, removing the link to the reinforcer, and changed the antecedents to set up alternative behaviours really nicely.
So, when Pumba got up, I would get up at the first sound of any whining (a slightly whiny exhale) to predict the noise. I opened the door, stepped outside of the room and asked Pumba for a hand touch to move him. I moved him into the bathroom directly opposite my bedroom and he got closed in there for 30 seconds. I used the hand touch because I didn’t want to physically force him into the bathroom, our hand touch has a huge reinforcement history and although I suspected it may be damaged from using it for this context I was confident that I could fix any damage it caused. It took 3 repetitions to stop the whining. Not 3 nights. 3 repetitions.
If you’re a trainer that only uses reinforcement and is derisive of those who acknowledge the uses of punishment, consider how much of the possible training you’re shutting off. Efficacy in training is important, the best clicker training programme in the world is redundant if it’s not effective at changing a situation. There are some scenarios where we can take our time, things like trick training can take years and no one cares. However, behaviours that are affecting the lives of the humans or the welfare of the dog need some sign of change within the first session. You won’t get human buy in if you can’t demonstrate efficacy quickly for behaviours that are that serious.
Punishment can be bad; I’m not denying that. It can cross ethical lines I’m not okay with it. Where that line is for each of us differs and that’s okay. But should you write it off entirely? I think you’d be doing dogs and humans a disservice by doing so. Do you think punishment should be entirely avoided in dog training? Or you do you think it’s a necessary part of it on occasion?