Clicking Versus Luring and What happened to Rewards?!

Hi all,

What I've noticed at training recently is that we are often confusing getting information to the dog about what it is doing right (or wrong) and motivating it do the task with exuberance and consistency- the speed and accuracy part of agility. These two are separate goals though historically we've used praise and exuberance from the handler or food to try to achieve both goals. I guess I am trying to emphasise that we need to reduce luring behaviours and cheerleading when the dog is learning a new behaviour or behaviour chain, as it is providing the motivation but isn't giving clear information to the dog about what is correct.

Luring has its place as an essential part of our toolkit to motivate the dog to do a behaviour that may not come naturally (weaving through 3 poles or sitting with its bum on a plank and its feet on the ground). It is essentially used at the start of shaping a new behaviour to encourage the dog to offer what we are looking for. However we need to cease it as soon as possible, a good rule of thumb is only lure three times as waving the treat in front of them for extended periods of time does not help get essential and timely information to the dog about what it is doing right. And it can sometimes result in shaping incorrect behaviours- the dog thinks the behaviour it is being rewarded for is following food- so that is the only behaviour it offers. And often the cue to start that behaviour is us reaching into our bumbags or pockets. So the dog pays no attention until that cue is used.

This often looks like the dog is training us- but we initially gave the incorrect information and the dog is now responding consistently to that. So in all luring doesn't permit us to give timely information about what the dog is doing right- it only provides a motivation. Same with cheerleading; same with squeeking a toy etc. If the dog is having problems with the exercise without luring it usually indicates that we are asking for too big a step and that we need to break the task down into a smaller more do-able task for the dog- if it is not understanding weaving three poles without the lure after a couple of times then we need to break it down to going between two poles without the lure- click and reward that and once that is understood without the lure then add the next step in the task (sometimes we break the behaviour right back to moving toward the poles and reward that).

The same principals apply to more general behaviour, such as if the dog is not coming when called without a cue such as squeeking a toy or reaching into our pocket/bumbag for food. Then we need to go back to recall training so that the dog learns that coming to you is more rewarding than the things in its environment (that can be very hard if the dog is chasing a rabbit or some similar activity that is inheritently very rewarding to the dog), however if we have established the basis by giving timely information about the correct behaviour and then reward that behaviour to motivate it to happen again, then we are starting off well. For coming when called there are several starting points- marking (clicking) the moment the dog turns to you when its name is called is a good start. Building up a come or here command from close in until it is very consistent. Then the same exercise with increased distance and then with distractions is the usual progression. Often we haven't really trained the latter two with marking the correct behaviour and rewards, we simply expected the in-close come to also translate into a distance exercises with lots of distractions. When we find it doesn't then train for it- we don't want to have to train for every occurance or distraction but we need to give the dog as much information about what a cue means as is possible. And coming when called is one of those behaviours we are often too quick to put on random reinforcement.

Random reinforcement is an excellent tool, and necessary in dog training, however it shouldn't occur until the dog first knows the desired behaviour very well. And indeed the same principals as above should apply for jumping an entire course of 15-20 obstacles. If we lose the dog's attention partway through it can often be because we haven't shaped the behaviour of keeping the dog's attention and drive for 20 obstacles before rewarding. Essentially in agility we want to build up several behaviours into a string before rewarding. And even more difficult is that the order of behaviours we ask for is constantly changing. This is why I will often suggest that people reward after a few obstacles on a course and why in our classes we start with singular obstacles and teach correct performance and then build up to varying sequences of two and three and four more gradually. This is so that dogs understand that these various behaviours they are being taught are then expected to be performed off our body/verbal cues need to occur without immediate reward, but with the knowledge that a reward will be forthcoming. (I'd have to say that I think we do have a weakness in our current training between 6-10 obstacle seqences done in the intermediate classes and the 20 obstacles we expect dogs to do on course. The only time we train this many obstacles is in our DOY rounds and then we don't place enough emphasis on reinforcing at different points in the course for the starter dogs. But in reality we simply don't reward enough!) But the key to building a behaviour chain and/or variable reinforcement schedules is that the dog must "know" each individual behaviour well, that is it should be able to repeat the behaviour at least 20 times on cue without error before you attempt to put it into a behaviour chain or reinforce it randomly.

Anyway now I am getting off the current topic- which is that the clicker is a tool that provides timely information to your dog in training behaviours and behaviour chains. The article I'm directing you to below simply talks about how the clicker can provide timely information (tells your dog he/she got it right!). It only motivates because we have paired it constantly and consistently with a reward- most often food but also with toys and praise. We've conditioned the dog to know the click means a reward is forthcoming. But the clicker's essential job in shaping behaviours is as providing timely information of what the dog was doing right. It marks correct information quickly and consistently.

The article primarily talks about what makes a good marker of correct behaviour (mostly clicker in preference to voice commands at the start of teaching a new behaviour)- but I think it also demonstrates that luring/bribing uses an action such as reaching into the bumbag as a correct behaviour marker (the dog has associated the reaching into bumbag/pocket with a reward forthcoming). This has several problems; first it is a lengthy marker and not always consistently applied; secondly, and more importantly, the dog is often doing something wrong at the time we reach in; and if we then proceed to lure the dog for some time or a number of obstacles the dog essentially has no idea what it was doing right.

I hope I haven't confused people further with this long garbled essay- I guess I got carried away! But I'm seeing this as the basis of a number of problems. I'm also seeing it is occurring because we are pushing forward too quickly without ensuring the dog understands and can repeat the initial behaviour, whether it is weaving, jumping, coming or whatever before we attempt to randomly reward or place the behaviour into a chain of behaviours. We haven't reinforced the behaviours enough at the start. And unfortunately we don't have time in our sessions once a week to really develop and proof a lot of these behaviours as much as we'd like so you will find that we (the instructors) will provide information on how to try to achieve these behaviours outside of club We also encourage the use of the email list to share our experiences in shaping these behaviours at home so that we can all learn from each other and troubleshoot. I also strongly encourage people to shape some tricks with their dogs with the clicker as this helps build your relationship with your dog, generally improves attention, it encourages your dog to offer behaviours (which I believe means that behaviours like recalls are more likely to translate into situations you haven't explicitly trained for), and it allows you to practice shaping behaviours that are fun and don't matter so much. So I hope this lengthy monologue that started out just talking about the role of luring and clicker and turned into goodness knows what helps some.

Cheers Tracey

See the article Information vs. Motivation - The Essence of the Clicker by Gary Wilkes