Thursday, January 28, 2016

Dog training tip o' the day: No-reward markers


Today I want to talk about no-reward makers (aka NRMs).

In some circles, mentioning NRMs can elicit boos and hisses. This is because a NRM is technically a positive punisher. That is, it is actively applied in order to reduce the likelihood of a behaviour happening again. Many people are leery of punishment in dog training, and with good reason. So the use of NRMs is often discouraged.

But, I kind of like NRMs in certain situations.

A no-reward marker is intended to be a way to communicate to the dog that the thing they are doing at the moment is not going to earn them a reward. It's sort of the opposite of a clicker. Using NRMs comes naturally to people -- most of us say no, nope, or try again if our dogs make a mistake during a training session. It is a way to offer a bit more information to a dog if they begin to go down an undesired path during a training session.

In a perfect world, NRMs shouldn't be necessary. A dog should be set up to offer the desired behaviour during training from the very beginning through management and a conscientious approach to the session. In a perfect world, dogs should be crystal clear in what it is you are asking of them. However, I am not a perfect trainer.

I occasionally rush progression, or stall too long on a step, or just get lazy and let my criteria slide. My dog might begin to offer undesired behaviours that run the risk of being reinforced by subsequent steps within a behaviour chain before I've had a chance to address them. When that happens, mistakes can become entrenched within a chain and can be difficult to remove once there. If I remain silent and withhold a click, my dog may grow frustrated at not being offered a clear picture of what it is I'm looking for. She might grumble, stress up or grow frenzied in her responses if I remain silent. If I let her know that she is offering something that I am not looking for and that will not be rewarded, she can gain some clarity and try something else.

As the trainer, it's me who shoulders the responsibility for this lack of clarity. If I found myself having to rely a great deal on NRMs, it would be in my best interest to step back and assess where I'm going wrong in my approach. However, as the occasional stopgap measure, I find them useful.

In brief, NRMs:
... Should not intimidate or demotivate a dog. Some dogs will wilt if they are used. For these dogs, they are the wrong tool for the job.
... Should not be used to stop unwanted and/or nuisance behaviours.
... Should be cheerful & motivational or neutral. They should be free from disapproval.
... Should prompt the dog to stop what it is they are doing and check in with you.
... Should be used sparingly at best, and avoided if possible. Relying on them in every training session is unnecessary.
... Should be avoided when you are teaching a new behaviour. Errorless learning is the better choice. You want your dog to understand what is right well before you focus on what is wrong.
... Are best suited to polishing behaviour chains wherein a mistake earlier in the chain runs the risk of being reinforced by subsequent steps in the chain.

So, that's my spiel on NRMs. What do you think? When do you use them? Have you changed in your approach to NRMs over time? Have you ever put much thought into how you use them? Let me know in the comments.


Happy clicking!

Thursday, January 14, 2016

Dog training tip o' the day: Stimulus Control


Dog training tip o' the day: Do you have a dog that starts throwing every trick he knows at you when training time starts? Are you trying to isolate a single desired behaviour out of an assortment of offered ones? Time to work on stimulus control!
Stimulus control: 
(from http://www.clickertraining.com/glossary/17#term21269)
"A conditioned stimulus becomes a discriminative stimulus (or cue) when it is followed by a specific learned behavior or reaction. The response is said to be 'under stimulus control' when presentation of the particular stimulus fulfills these four conditions: the behavior is always offered when that cue is presented; the behavior is not offered in the absence of that cue; the behavior is not offered in response to some other cue; and no other behavior occurs in response to that cue."
When we teach a dog something new, the first thing we look to do is get the behaviour. Once we have the behaviour, we add the cue. Next, we polish the behaviour and fine tune it -- you can also add a new cue to this improved behaviour at this point, if desired. Then, you need to get this new behaviour under stimulus control.
Since stimulus control comes later in the learning process, it's often something that novice trainers forget about or choose to bypass because they're pleased with the behaviour as it exists already. But if it's ignored or only completed part way, you can end up with one of those dogs that throws everything they know at you, ad infinitum, as you reach for the cookie jar. Sound familiar? Some people find enjoyment in their dogs doing this and will reinforce this "throw everything at the wall and see what sticks" approach. If you do, that's fine! However, you may never end up with clean behaviours, or with a dog who eagerly waits to hear what you have to say.
To get a behaviour on stimulus control requires that you go back to the teaching process. Pick one or more simple behaviours and methodically establish the criteria for each cue. This means:
- you want your pup to sit when you say sit
- you don't want your pup to sit when you say something else
- you don't want your pup to sit when you say nothing
- you don't want your pup to do something else when you say sit
Only reward if your dog is performing the desired behaviour when you use your desired cue. If they do something else, whoops, no reward this time, nice try buddy! Try again! If your pup is making mistakes multiples times in a row, or more than 10-20% of the time, the exercise is likely to hard. Try to make it a bit easier on them. Keep these proofing sessions very short and spaced a few hours apart, at least.
You can have fun during this process. You can also take it to new heights by alternating what you're doing when you offer your verbal cue. Spin or hide out of sight or whisper or sit on a couch or work in a new environment to change up the picture for your dog and to further cement the "sit means sit" association in your dog's head. Once your dog has a handful of actions under good stimulus control, he & you should have a significantly easier times going forward in training because he understands that specific behaviours are linked to specific cues and only cued behaviours are rewarded.
Problems with stimulus control are often seen as problems with over-arousal during training sessions. Stay tuned for my next Dog Training Tip O' The Day for some more ideas on how to address over-arousal.
-----
This was something that was asked as part of my recent call for submissions for dog training tips. Thank you to everyone who asked about it. If you have a question of your own, feel free to submit it here, on this page or via private message. Happy clicking!

Friday, January 8, 2016

Dog training tip o' the day: On the use of tools to facilitate your training


Dog training tip o' the day: Tools can make your dog training better. However, without a proper plan to fade the use of the tool, it can quickly become a crutch.
Tools can give you a leg up when it comes to controlling your dog around distractions. They can be things like collars, harnesses, leashes, targets, platforms and any other things you use to guide your dog to perform a desired behaviour. They allow for greater control and communication and can be used while you work on the requisite training to fill in the gaps. Then, most importantly to me, you need to fade the use of the tool.
We humans are excellent tool users. Sometimes, however, we get a little lazy. Applying a tool can often provide an immediate change in behaviour (sometimes for a honeymoon period, sometimes in perpetuity) so it can be tempting to continue to rely on it on an ongoing basis. To me, this is, well, lazy at best and unfair to the dog at worst.
Ideally, the goal when using a tool on a dog is to get the tool off the dog as fast as you can. Commit to harnessing the extra power the tool affords you and work with your dog whenever you're using it. Don't fall on your laurels and forget to actually train your dog. As with everything, the additional work you do with your dog in the short term pays off in the long term.
Happy clicking!

Wednesday, January 6, 2016

Dog training tip o' the day: Extinction & extinction bursts




Dog training tip o' the day: Extinction & extinction bursts
Extinction refers to the process of no longer providing reinforcement for a behaviour and in the absence of reinforcement a behaviour will cease. For instance, if Pavlov stopped offering meat powder after sounding the tone for a period of time, the dogs would cease to salivate since the association between the tone and the food is no longer being reinforced. This is why ignored behaviours often stop -- the dog is no longer being reinforced for offering them.
However, some behaviours are self-reinforcing, and therefore very difficult to extinguish. For example, a dog often finds barking to be a pleasurable response to various stimuli (barking is FUN!) so even if you ignore a barking dog they're very unlikely to stop this behaviour since they're reinforcing it themselves. That's not to say that you can't train a barking dog to bark less, but it requires a different approach than to ignore it. You need to find and stop the source of the reinforcement while rewarding an alternative behaviour instead.
Then there are extinction bursts. If a dog no longer receives reinforcement for a behaviour, they will increase the intensity of the behaviour in an attempt to earn the reinforcement once again. If the behaviour is still not reinforced after this more fierce series of attempts, it will become extinct.
You can also use the increased intensity of behaviours during extinction bursts to improve them by riding the crest of the burst and reinforcing when the dog is still fervent in its response. We use this to improve hand touch or targeting behaviours and more.

Monday, January 4, 2016

Dog training word 'o the day: Premack.

This post is ever so slightly verbose, so here's a nerd photo of Cohen to compensate. 
Dog training word 'o the day: Premack.
From Wikipedia: "Premack's principle suggests that if a person wants to perform a given activity, the person will perform a less desirable activity to get at the more desirable activity; that is, activities may themselves be reinforcers. An individual will be more motivated to perform a particular activity if they know that they will be able to partake of a more desirable activity as a consequence. Stated objectively, if high-probability behaviors (more desirable behaviors) are made contingent upon lower-probability behaviors (less desirable behaviors), then the lower-probability behaviors are more likely to occur. More desirable behaviors are those that individuals spend more time doing if permitted; less desirable behaviors are those that individuals spend less time doing when free to act."
In dog training terms: It's no secret that performing certain actions can be rewarding to dogs. Chasing squirrels, jumping up, sniffing a tree can all be intrinsically rewarding for certain dogs. If you ask your dog to perform an "unfun" behaviour prior to being released to perform a fun behaviour, the unfun behaviour can be reinforced by what follows. Often the environment has a stronger ability to reward our dogs than we ever could hope to offer with the cookies or toys we may have in our pockets. So use it to your advantage! Ask for eye contact and a bit of self control before releasing your dog to (safely) run amok!
The beauty of Premack is that it also benefits from classical conditioning, conditioned emotional reactions and the transfer of value. For instance, eye contact can be reinforced by being released to chase squirrels. If the eye contact is routinely rewarded by extremely valuable reinforcers, offered eye contact becomes considerably stronger and, at times, innately reinforcing itself due to it now being associated with huge rewards and excitement. The "unfun" behaviour is now fun. Nice!
Like all classical conditioning, there is a degree of regular maintenance that is required to keep associations strong between action and reward. But once you learn to use it to your advantage, maintenance becomes second nature.
Sorry for the possibly obtuse, overly-sciencey sounding post today. Hopefully this nerdy photo of Cohen will make up for it. Premack has been on my mind a lot these days. I use it a great deal to build off-leash reliability and control, which to me is the pinnacle of dog ownership.