This weekend I was repeatedly asked the question whether it is a good idea to train with so many treats. The argument was that "if he [Noch] always gets a treat, won't he always be expecting a treat?". The worry here was that treats would forever be a training "crutch".
I understand this worry as it is one of the hardest things to grasp when one starts to clicker train. When I first started out trying to clicker train I treated too little and with too long intervals (a very common problem apparently). I also trained for too long periods of time. This resulted in slow learning and frustration and extinction of behaviour (extinction = when a behaviour stops entirely). I believe this reluctance to reward a lot is rooted in our way of perceiving learning and how to go about it (think: school).
The thing is we do is to fade out treats once the behaviour is learnt and only occasionally treat. We put the behaviour on an so called "intermittent schedual" (unpredictable and random) as it is called in more scientific terms. An intermittenet schedual is the best way to encourage a behaviour to be strong and long-lasting (think: casinos, they reinforce on an intermittant schedual).
Things (off the top of my head) that aren't regularly rewarded anymore: calm behaviour indoors, drying paws, putting on his lead (this has become self-rewarding), pooing outdoors, peeing outdoors, ignoring "normal" calm people walking by on our walks, following me after the command "come along now" (after he has been allowed to sniff an interesting patch of grass), sitting before allowed to play offlead (self-rewarding nowadays).
Back to the "won't he expect it all the time" part of treating often. I actually think this is an excellent and important part of clicker training. You are effectively shaping attention and becoming very exciting for your dog. You are pairing commands/praise from you with food. This eventually results in your dog excitedly waiting for opportunities to impress you so that you will praise-and-treat. Which is excellent! Why would I want things to be different? Every time I say "Yes!" or "good boy!" I am rewarded with a snap-turn of Noch's head as he focuses on me and the treat that is on its way.
This is excellent because:
a)I know he knows that he did something right
b) I can gauge how "difficult" and distracing he finds the environment. I do the "Noch" check. His normal behaviour when I call his name is to give eye contact and then come to me. So to do the little Noch "test" I call out his name and if he doesn't hear me, he is over-threshold, and I know then to avoid other triggers too and not to train anything at that point (he won't hear me anyway).
In the end your job is to find what motivates your dog. You find it, and you use it. Some people are lucky that their dogs will happily work for pats or toys. But for many dogs (the majority) pats and toys aren't really worth much, especially in the beginning when you are still building your relationship with your dog.
Play with toys sometimes has to be taught to be enjoyed (as is the case with my dog Noch). Noch and I have only recently come to the stage where I can reward ball fetching with a toss of a new ball. It used to be that this would result in the break down and finally the extinction of the fetching behaviour. For him, and many other canines, food is number one on the motivation scale.
There are of course some things that can top food: chasing cats, chasing squirrels etc. but these things are hard to systematically reward with. If I could, I would, for it is even better than food when it comes to motivation!
Another aspect of the food-worry is that the dog gets spoilt. Which is indeed a unnecessary worry as the food, even though given very frequently, is never given "for free". It is always given with the thought "reward good behaviour". (Anyway, define "spoilt" if you have a well behaved dog it surely isn't spoilt in a bad sense! Read Pat Miller's article that I posted just before this post, she does talk about how clicker dogs behave differently to coercion trained dogs and how this can be seen as misbehaviour to a coercion trainer).
In my case I wish I didn't have to use it as much as a lure. But in some cases here in the city I have to as it is the only thing that sometimes can break him out of his "trance". But even this is something that I do less and less as he gets better and better.
And that brings us to the last aspect of this. Which is that our dogs are the ones to decide when they are ready to be "weaned" off food rewards. Personally I prefer to always be challenging Noch mentally so I will always train using many treats or anything else he find motivating. The biggest difference is that I will be using less treats on our walks (happily, this is already happening). Also, as Leslie McDevitt points out in her book "Control Unleashed" there are plenty of life-rewards and other rewards that are not food that can be used once the behaviour is learnt.
Inga kommentarer:
Skicka en kommentar