Okay, so as I've recently learned, most of my training turned out to be bribing with food. I have suspended training for the last few days well I read up. Question: Can you De-Bribe a command or are my commands soiled and sullied permanently? As an example, I say sit, and hand gesture and *sometimes* he will in *certain* environment, he was bribed in, even with no food he will obay. But usually No food NO response. If it is possible, Help would be most appreciated. I must admit discovering my error (Thanks to you kind folks) was a massive blow to my mental state. Raising a puppy is HARD. I Keep reminding myself, dogs have been with humans for centuries. Hopefully all is not lost. (I'm sure it is not) (4.5 month old puppy Sora) Thanks again!
Yes it is! I made this mistake with Tatze. She’s still a bit ‘show me the money’ for some things. But new things are easy to teach the right way - just be sure no food is visible and the food comes from different places so they can’t wait for your hand movement before they do what was asked. Easier said than done lol. If he has no response just ignore it and try again later. Don’t keep trying as this will water down the cue. People who are much better at this will come along soon with much better explanations .
Gosh! I fell right into this scenario with Benson, he worked it out quickly. For example, I would be stood there trying to get him out of the car (he LOVES the car )Yep I would give in and offer a treat, with a wag of the tail he would hop out. It took me a quite a while to get my head around the difference between luring and bribing to get the desired outcome. Now I understand the concept I am far more relaxed about it. So don't worry, take a step back and just enjoy your training
Ok, so hypothetical scenario... I say 'sit'. Sora then complies. I THEN go off to get a treat from a counter top or other place? If so do I reward if the sit if it is held, or what if he moves? I know the reward must be FAST after the action. Also do I later faze out the treats completely or alternate. Just trying to break this down. (lets pretend I'm real dumb ^_^) If you could for example give me your hypothetical scenario? Thanks for your patience!
Yes, mark the sit with a ‘good’ the moment he does it, then go and get the treat. Don’t phase out the treats completely but make them more unpredictable- but this is a few weeks down the line, treat every time at first The mark (or click) must be fast, the treat can be a bit slower. This article will help - https://www.thelabradorsite.com/5-steps-to-charging-a-clicker/ .
And a golden rule is that whenever you mark or click you must follow it up with a treat. A marker or click is a promise that a treat is coming. If you don’t keep your promise you’ll diminish the power of your marker/click. If in future you don’t want to give a treat every time that’s fine - but don’t give a marker or click either.
When I’m not treating I praise instead, but try my best to keep it random. This is after the action has been thoroughly learned. I get different things wrong with each pup. With this one it’s the ‘down’ that I’ve somehow taught to be a ‘jump and thump down’ .
Please someone else correct me if I am confused, but my understanding is that the click (or mark, e.g. saying ‘yes’ or ‘good’) for most trainers ends the behaviour, i.e. releases the dog. So it marks the exact moment that the behaviour is good and correct (eg when bum touches floor when training a sit), and then it’s ok if the dog gets up after the click to receive their treat. As long as they have learnt that the mark/click reliably predicts a treat, then the purpose of the click is to say ‘what you are doing RIGHT NOW is perfect and will now be rewarded’. The idea is that the treat reinforces the behaviour that the dog was doing when it hears the click, rather than the behaviour it is doing when it eats the treat. I’ve read debates about whether this is desirable in long calm behaviours like a stay, where some people prefer to reward in position then release, so they don’t use their normal mark/click for that. I’d be interested in hearing other people’s opinions about that, as I think there are different opinions as to whether a mark should end the behaviour. It makes sense to me, as if e.g. you are training a simple sit, the dog should be rewarded for sitting, not for ‘staying-sitting-whilst-I-walk-over-here-to-your-treat-bowl-to-come-back-and-give-you-a-treat’ as that is a completely different (and more difficult) thing. So I wouldn’t worry about the dog getting up after the sit and before the treat, as long as you have marked the sit with a good/yes/click before it gets up. Although others might disagree!
It’s correct that the click ends the behaviour. It’s ok if the dog gets up after the click but before the treat (eg if you’ve clicked for a sit, down etc). But if you can time your delivery so a treat is given while the behaviour is actively being performed (or the dog is maintaining the position you want...like in a stay) then you don’t need to click - you can go straight to a treat (no click). We use clickers and markers because most of the time we can’t accurately deliver a treat at the exact right moment (but we can accurately make a sound). If we could be 100% accurate with our treat timing with all behaviours then we wouldn’t need markers. But we can’t, so we do need markers.
I use different event markers depending on what I want. I rarely use a clicker these days, although I believe they’re the most effective tool for precision, mostly I don’t find them necessary. That said, my “yes” is most like the click and I don’t mind if it ends the behaviour. I also use “good” as a continuation marker, meaning that I expect them to stay in place. The important thing is where you deliver the treat. Hannah Branigan says in her podcast how what happens between the marker and the treat is reinforced; the marker reinforces the precise thing, but whatever happens between that and receiving the reward is also reinforced. You can use this to your advantage if you mark an approximation of the behaviour then lure into position to feed. I need to think a bit more on the details of this, but my interpretation is pretty simple: You must pay after an event marker whatever the dog does when you mark. If you reward in place, your dog will learn to stay in place in anticipation of the reward. If you reward away from the place, the dog will learn to break position. So it’s up to you as to what you’re trying to achieve. Sometimes you’ll want to move your dog away, sometimes you’ll want her to stay put. You just need to adjust how you pay to achieve the result you’re after.
@Oberon you have cleared up a confusion I have had for a long time! But just to check, if you are able to reward in position so don’t need a click, does the treat still end the behaviour or in that situation should a separate release cue then be given?
I am not sure about this, my understanding would be to give a treat for a partially trained cue without a clear mark would be confusing? Regarding a release cue, this is something I have always trained, especially after a behaviour I have asked for such as a sit. This is because I need my dogs to know to stay in whatever position they are in until they are released. It doesn't mean it's the right way, just the best way to suit the work they do. So I would ask Bramble to sit in a heel beside me, and she will stay there for sometimes 5 - 10minutes. I do use secondary reinforcers, such as positive verbal language, and will even "click through" a behaviour. This is mainly when they is some momentum, ie running towards me. I find this useful in building confidence in chained patterns.
Secondary reinforcers (clicks, markers, ‘good boys’ etc) always and only have their power through their association with a primary reinforcer (something inherently desirable, like food). You most certainly can use your primary reinforcer on its own without using a secondary reinforcer - that’s not confusing (as long as it’s well timed!). A secondary reinforcer has no special power of its own (other than being easier to deliver in a precise way, and when at a distance, which is why they’re useful). Every time you use a secondary reinforcer without delivering the primary reinforcer you diminish the power of the secondary reinforcer a bit. It’s fine to use a ‘keep going signal’ to tell a dog they’re on the right track and this doesn’t need to be paired with food. A keep going signal is instead paired with and predicts the arrival of the secondary reinforcer. A keep going signal is therefore a tertiary reinforcer. It’s saying “you’re on track to get a click!”. I would not use a secondary reinforcer (click) as a keep going signal because you’ll diminish its power. Best to use something different that only promises (and always delivers) a click. I use clapping or “clever man!” as a keep going signal. Save your click for always promising and delivering food (or your primary reinforcer).
Would depend on how you’ve trained it. If you’ve been in the habit of giving treats for remaining in position your dog is likely to remain in position in expectation of another treat. If you haven’t been in that habit then he won’t. Just try it - your dog will tell you by his behaviour what you’ve actually been training for (and it might be different for different behaviours). Just as an aside.... Sometimes people think they’re training duration in a behaviour by reinforcing throughout the behaviour. So, I might want my dog to lie down in the same spot for 2 minutes and so I’ll give him a treat every 10 seconds until the 2 minutes is up. Have I trained a 2 minute down stay? No, I’ve only trained a 10 second down stay, and I’ve repeated it 12 times in a row.
Practical, yet makes perfect sense! Training for me at least has been difficult due to low attention span, but I'm always trying to be mindful of anything I'm doing that he may interpret wrong. I have no illusions of the road I face with Sora. A tough road ahead, but the persistence on my part will be well worth the gains. I've seen way too many adult dogs behave poorly or drag the owners on the lead. Not cool, and not what I'm working for. Thanks Oberon, and all of you for your replies! I'll keep on working with Sora!
