Gambling Effect Question

Discussion in 'Labrador Training' started by athao04, May 6, 2013.

  1. athao04

    athao04 Registered Users

    Joined:
    Feb 10, 2013
    Messages:
    34
    Hello,

    I have a question about putting the gambling effect into practice. I use a clicker to train my puppy. Say I am trying to reward only 8 out of 10 times, when it comes time to not give the treat, do I still click and treat or do I just have my puppy try again. For example:

    "Titan, Sit." He sits, I click as soon as hit butt touches the ground and I treat.
    "Titan, Sit." He sits, I click and treat.
    "Titan, Sit." He sits, I don't treat.

    At the time where I dont treat, do I just acknowledge that he did a good job with a "good boy" and then start over? Or do I do nothing and just start over? I don't want to click and not treat since that would over time diminish the sound of the click being something good. I also don't want to not acknowledge him since he would think he didn't do it right.

    I want to make sure I am doing it correctly. He sits very well and recalls very well, but he does it even better with food. I usually let him out in the yard without a leash (we have 13 acres of land) and so it really gives me a lot of opportunities to work the recall.

    Thank you!
     
  2. Oberon

    Oberon Supporting Member Forum Supporter

    Joined:
    Mar 6, 2013
    Messages:
    14,194
    Location:
    Canberra, Australia
    Re: Gambling Effect Question

    Whenever you click you have to treat. So be selective about when you click :) The decision about what to click/treat and what not to depends on what you are aiming for. It's always good to have a specific goal in mind. Do you want the recall to be faster? The sit closer? The recall to be longer? Choose one dimension (longer recall, or faster recall, or faster sit, or longer sit, or closer sit) and reward better and better examples of that for that session. Work on one dimension at a time (i.e. not faster plus longer plus neater). Having a goal will mean you can gradually shape a better response. You reward only behaviours that are closer to your goal. The dog has to keep thinking and working for the treat. You don't ever want to push a session to the point of boredom though - so stop while your dog is still focused and enjoying it and do something else or another, different exercise.

    With the things that your dog has learned well, then, yes, you want to avoid treating every single one. Reward some with food and just praise the others ('good boy'). Try to reward the best examples. If all your dog's sits etc are excellent, then you can reward randomly (maybe a third or half), but otherwise be selective and choose the good ones (based on your goal which will define what 'good' is for you - faster, or longer etc).

    With your sits, for example, stop rewarding as soon as his bottom hits the ground. Wait a third of a second, then click/treat. Then wait half a second, then one second. You are gradually shaping a longer sit (so the dog learns that he doesn't just sit and then immediately run off). Don't always make it longer and longer and longer before you click/treat - do half a second, then one second, then half, then two, then two, then just under two. Overall you are shaping a longer sit, but without making the dog think 'Gee, this only ever gets harder and harder and harder, what's the point, I'll give up'.

    Another thing to do is to make it harder for the dog to work out if you have food or not. The dog should not know if there is food or not till after he has done the behaviour. Once he has learned things, stop having food in your hand for those exercises. Put it in a treat pouch, or your pocket, or on a close by wall (and after you click run together to get the treat). Only reach for the treat after the behaviour. The treat doesn't need to come the millisecond after the click - the click is a promise of a treat and you have a couple of seconds to deliver. Timing the click precisely is the key thing.

    Sounds like you are really going well with Titan and are working as a good team :)
     
  3. lois

    lois Registered Users

    Joined:
    Mar 24, 2013
    Messages:
    147
    Re: Gambling Effect Question

    Quick question Poppy is nearly 8months old is it too late to try clicker training with her?
     
  4. athao04

    athao04 Registered Users

    Joined:
    Feb 10, 2013
    Messages:
    34
    Re: Gambling Effect Question

    Thank you for your response Oberon! I really appreciate it! :) I never thought about waiting just a bit longer so he knows to hold the sit. I guess that would only make sense since he's probably understood sit means butt to the ground, but maybe not so much don't get up right away. It'll be weekly goals from here on out. Thank you!

    Lois - I dont think it's too late at all. Poppy is still young and I'm sure will do wonderful. With my last dog, we never did formal obedience things with her. I finally started clicker training her when she was 11 years old and she picked up very quickly. :)
     

Share This Page