Do You Always Reward?

[aussielover]

Just want to get others thoughts on this.

I always give a reward if my dog obeys a command. It might not neccessarily be a high value reward such as food or a toy, but I do always give verbal praise at least. With recalls I always give a high value reward.

I have been told that it is optimal to only reward intermittently. does this mean you can't use verbal praise somtimes?

[Aidan3]

Intermittent rewards make a behaviour more resistant to extinction than other schedules of reinforcement and they are also fairly practical once a behaviour is fluent (already very reliable).

One of the problems with rewarding every time is that you tend to reward sloppy responses, so it's good to have a picture in your head of what is a sloppy response and avoid rewarding those. If you are competitive or building a behaviour that requires a high level of performance (e.g recall) then it's good to have some clear definitions on what you require and what your dog is currently capable of with that behaviour.

[aussielover]

Ok so for something already well established like a sit- if the dog is slow to respond, no reward, not even praise?

I kind of feel bad not even giving verbal praise :D especially when she sits there looking up at me with those big puppy dog eyes and waggy tail... LOL

I always use a food or toy reward for recall, but will use some dry food for a slower recall, but yummy chicken or cheese for a lightning fast recall. If she is in the mood, i will use a squeaky toy as a reward as well.

[W Sibs]

Once they know it (something like sit or drop), then nope. i don't reward them for doing a half arse job either... lol

[Agility Dogs]

Intermittent rewards make a behaviour more resistant to extinction than other schedules of reinforcement and they are also fairly practical once a behaviour is fluent (already very reliable).

One of the problems with rewarding every time is that you tend to reward sloppy responses, so it's good to have a picture in your head of what is a sloppy response and avoid rewarding those. If you are competitive or building a behaviour that requires a high level of performance (e.g recall) then it's good to have some clear definitions on what you require and what your dog is currently capable of with that behaviour.

x2

[Staranais]

Yes. :D Although I count the release command as a reward for some purposes.

[JulesP]

It would be very rare that my dogs didn't get a reward and would be not on purpose. But the reward might be going through the door. Lots of verbal praise happens too. I don't reward with food at all around the home.

[Aidan3]

Ok so for something already well established like a sit- if the dog is slow to respond, no reward, not even praise?

Not unless you want more slow sits :D You get what you reinforce!

[Sue & Waldo]

I wouldn't have used up many rewards today. :D

[bedazzledx2]

agree and also permission to do the next behaviour in the chain becomes its own reward...ie a nice present on a great recall is permission to do a finish for which I may give a hand touch which is the reward for the great finish.....

Yes. Although I count the release command as a reward for some purposes.

[Mrs Rusty Bucket]

I do intermittant rewards on established commands, but with new ones, I give rewards every time she makes the standard, though I do up the standard required from time to time. And I reward in a variety of ways depending on how new the command is - food for new commands, play for agility and some food, food and play in obedience but ideally only when a sequence of commands have been completed eg "exercise finished".

If I don't think the standard is good enough I say "oops" or "nope" in a neutral tone so she knows she's getting nothing for that effort.

If you want to string sequences of commands together without rewarding eg competitions, trick displays, herding - you can't reward for every successfully performed command.

[Paul Bright]

If your dog has already mastered certain commands do interval praises

[corvus]

I reward every time, usually with a primary reinforcer, but not always.

Ken Ramirez had a lot to say about variable reward schedules. He told a horrifying story about an inexperienced trainer that used secondary reinforcers with an orca during a show for about ten minutes straight. An experienced trainer had done the exact same thing earlier that day, and the inexperienced trainer was so inspired he decided to give it a go. At the end of the sequence, he reached in to scratch the orca's tongue, which is a good secondary reinforcer, and the orca bit both his hands off. The chief thing that went wrong in Ramirez's view was that the orca didn't trust that the inexperienced trainer would deliver the primary reinforcer, and after ten minutes of being asked to work without one, it was getting pretty peeved.

Obviously this isn't going to happen with a pet dog (I hope!), but Ramirez has a systematic method of introducing a variable reward schedule, which is much the same system as introducing a secondary reinforcer, and he uses that with any animal, including his own pet dog. I don't have my notes from the conference with me right now, but I can post an outline of the method later if anyone is interested.

Personally, I don't bother. I always have treats with me, so I always reward. For sloppy responses I usually use an LRS - pause for a couple of seconds, then cue something easy, reward, try again with the sloppy behaviour. I don't often do that, though. My partner is determined to corrupt all his cues and rewards sloppyness hither and thither. He accuses me of being a training nazi and insists he wants his cues to mean multiple things. Given I worked so hard to get him rewarding in the first place, I live with it. For the moment. There are some cues I just don't tell him about.

[poodlefan]

Yes I always reward.

However, I build duration between rewards. I may ask for more behaviours, a longer focus or performance etc.

There is one exception. I reward each and eveyr time I recall my dogs outside of formal obedience recalls (which I've not done lately anyway)

Corvus:

Personally, I don't bother. I always have treats with me, so I always reward.

If you were planning on agility trialling you would need to rethink that. However, the obstacles tend to be rewards in themselves and you can always praise.

Edited November 10, 2010 by poodlefan

[corvus]

Corvus:

Personally, I don't bother. I always have treats with me, so I always reward.

If you were planning on agility trialling you would need to rethink that. However, the obstacles tend to be rewards in themselves and you can always praise.

I don't always reward with treats. I was lumping secondary and primary reinforcers. At the moment with Erik I'm working on switching from games to food and back again precisely so we can get back to agility. We're on a hiatus due to over-dependence on treats in early training sessions. E is a routine-driven creature and it basically unravelled everything I'd built in him. Lesson for Corvus: don't build on shaky foundations! Ramirez had an awesome talk for introducing different rewards and teaching an animal to accept their varied sprinkling through a training session. He says he tries to have at least 6 secondary reinforcers and he treats them like training a behaviour until they are extremely rewarding in themselves. Grey Stafford says a secondary reinforcer or a bridge are only as good as the last time you paired them with a primary reinforcer.

[poodlefan]

We're on a hiatus due to over-dependence on treats in early training sessions.

Howard has pretty much zero interest in the obstacles. I'm training him to see most as big toy ejectors. That's worked well with tunnels - he now drives to them. Now to build drive for jumps. Agility training a Whippet is no cake walk - every now and again the "this isn't fun anymore" flag runs up and you have to rethink making it a game.

Thank God he works for both toys and treats. A stationary toy is not much of a motivator - it has to be tossed.

I haven't been training lately either. Need to pull my finger out.

[Luke W]

This thread reminds me that I reward too much.

Intermittent reinforcement with slowly increasing criteria really is a skill I need to work on. Sometimes it's hard to recognise the 'best' reponses.

[Tassie]

I'm training him to see most as big toy ejectors.

:D pf I found the Susan Salo jumping method helps to build a lot of value for jumps - and/or SG

[corvus]

We're on a hiatus due to over-dependence on treats in early training sessions.

Howard has pretty much zero interest in the obstacles. I'm training him to see most as big toy ejectors.

Haha, that's about where I decided it would be best if we pulled out until our reward system was back on track again. Erik decided agility was all about food and when I tried to bring toys back in for jumps he told me he gets food at agility, thanks, and could I put that stupid thing away and bring out the treats already. This quickly attacked the rest of my training like a cancer and suddenly training full-stop is all about food. I totally moved too slowly on it because I saw it happening several weeks before I tried to do much about it. Another lesson for Corvus! We are back to the flirt pole and shaping fetch of tug toys because somehow my tug monster is now a clicker addict and I never introduced much toy work into my clicker training. Fascinating how easy it can be to botch the whole thing! :D Never mind. All the pieces are there, just gotta rebuild it.

[poodlefan]

We're on a hiatus due to over-dependence on treats in early training sessions.

Howard has pretty much zero interest in the obstacles. I'm training him to see most as big toy ejectors.

Haha, that's about where I decided it would be best if we pulled out until our reward system was back on track again. Erik decided agility was all about food and when I tried to bring toys back in for jumps he told me he gets food at agility, thanks, and could I put that stupid thing away and bring out the treats already. This quickly attacked the rest of my training like a cancer and suddenly training full-stop is all about food. I totally moved too slowly on it because I saw it happening several weeks before I tried to do much about it. Another lesson for Corvus! We are back to the flirt pole and shaping fetch of tug toys because somehow my tug monster is now a clicker addict and I never introduced much toy work into my clicker training. Fascinating how easy it can be to botch the whole thing! :D Never mind. All the pieces are there, just gotta rebuild it.

Have you considered throwing the food? Clean Run and some other stores sell really great food tubes and pouches that can be thrown or used as tugs. A friend of mine with a really food driven dog uses a throw pouch with a lot of success. If Eric can learn that all food doesn't necessarily come from your hand (initially) it might help redirect his focus.