Reducing Reliance On Food

[Luke W]

Barkly is super focused and happy if I've got food in my hand, on the bench, etc. As long as he knows it there, he's willing to work.

But we've got some problems.

1. As I try to move to an intermittent reinforcement schedule, he gets frustrated and 'demand barks', winges and carries on.

For example, he'll do a sit, then a down, on the third or so command without a treat, he starts getting grumpy. Sometimes he'll bark, sometimes he'll start whining, sometimes he'll just decide he's had enough and walk off.

When he does his little tantrum, I try to break the session for 15 seconds (a negative reinforcer I guess) or so and up the rate of reinforcement when we resume.

2. I'm trying to get the food off me (out of my pockets) and getting him to run to a treat bag. In addition to the problems in '1' above - his focus when heeling is almost non-existant when I don't have have food in my left hand or left pocket.

Any suggested videos, books, or any advice at all in reducing his reliance on food - especially his need for immediate gratification?

Does Bob Bailey or Ted Turner cover this stuff?

BTW - I'm not very good at explaining what the problem is.

I'm not sure if I'm just trying to move too quickly. Should I just concentrate on 2 commands per treat for a while (a month say), then move to sometimes using 2 commands/treat sometimes 3 commands/treat...and just really gradually increase like that over a period of months?

How do I keep focus in heeling without food in my hand/pocket?

How many steps and turns should he be heeling between treats?

BTW - He's 10mo old.

----

Added

Oh - I also want to swap food for tug - but he still seems more food focused than tug focused - although he does like tug - just not as much as food!

Edited January 1, 2009 by Luke W

[kelpiechick]

Hi Luke,

(how cute is your avatar!)

Without really knowing exactly how you train, it sounds to me as if you have been using food as a lure rather than just as a reinforcement, which can easily happen without you meaning it.

Although I prefer never to lure when teaching a new behaviour, it is OK to do so provided the lure is faded very, very quickly - if not then you get the kind of problems you are describing- where the value is totally with the food, rather than transferring to the work/activity/interaction with you - sorry, I'm not very good at explaining what I mean.

But you can't go from a lure to an intermittent reinforcement schedule - you need to get rid of the lure first - to rewarding with 'hidden' rewards, then to variable schedule. It's OK for him to know that the food is coming, but he shouldn't know where it is all the time.

When he does his little tantrum, I try to break the session for 15 seconds (a negative reinforcer I guess) or so and up the rate of reinforcement when we resume.

This particular part of your post worries me - not the fact that you are having a 'time out' but more the fact that you are then upping the rewards - if I were Barkly I would be thinking that 'tantrums' do indeed pay !

Can I ask a couple of questions ?

Are you clicker training at all?

If so, I would be looking at concentrating on teaching Barkly to 'learn' in a different way with a lot more shaping. Maybe give the obedience stuff a 'rest' for a bit and teach some tricks, etc. - so you can work on building up the reinforcement schedule you want with an activity that is a lot of fun, but not as 'meaningful' as something that you may want to compete in one day.

Clicker training would also give you the opportunity to 'hide' rewards in various places. Your dog can offer the response without the presence of the food, but then receive a reward - which can be sometimes off your person as the 'click' will mark the correct behaviour rather than the food - mix up where you keep the reward when training inside or in your yard.

The other thing about clicker training that I like is that you can introduce 'jackpots' where multiple rewards are sometimes issued very quickly for an extra good performance - gives the dog something else to think about!

Also, where is the reinforcement zone you have created for Barkly? Think about where he is usually when he gets the food - right at your side? in front?, etc.

It might be helpful to find a way to reward him away from your person sometimes - bigger pieces of food you can toss (ravioli is good for this!) or a food tube to throw, etc. That way he learns that not all rewards come from your person so it isn't so important if (he believes) you have no food on you - or at least food that he can't see.

Sorry I can't help with any of the heeling questions as I don't do formal obedience with mine, but I have seen the same problems you describe with agility dogs where a lure has been used for too long.

Going from tugging to food to tugging is a whole other topic in itself !

I'm sure there are a lot more experienced people than me on here to help you but I hope some of this helps.

(and yes, Bob Bailey would be a good reference for shaping behaviour)

Edited January 1, 2009 by kelpiechick

[Luke W]

For heeling, I have indeed used food as a lure to keep him at my side - and I probably didn't fade it quickly enough. And I'm now paying the penalty.

For other behaviors, I've mostly lured (sits, downs, spins, move to heel, etc) but faded the lure very quickly. The stand I finally mostly shaped (after so many failures luring, shaping, lifting him up etc).

I think I suck at shaping, but I'm getting better. I think my reinforcement rate when shaping isn't high enough and I don't have a good feel for when to pay and when to up the criteria. Barkly tends to get frustrated and barks when I try to shape him (as soon as I try to up the criteria).

I have succeeded in shaping him to put his front paws on a small tuppaware container and follow me round in circles with his back legs - but it took a while.

Re the time out thing - I try to end sessions on a high and to prevent the demand barking, I have to increase the payout rate...catch 22 hey.

Where does he get food? - where ever the excercise finished usually Eg..for a recall and front, he's in front. For a return to heel, he's at my side. Sometimes I throw food for him to catch - eg. For 'go to mat' - I'll often throw the food to him.

Thanks for your thoughts!

Added

There's a Delta trainer nearby that hold 'tricks training classes' (a prelude to doggy dancing). She's a pretty good shaper. Maybe I should get some help with my shaping skills.

Edited January 1, 2009 by Luke W

[ness]

Not sure I can help much but I fell into this trap with my first dog and it was hard work to undo that reliance. Having said that she got her CD (under the old rules) and CDX in straight numbers of trials.

My 9 month old BC baby (aside from being a different dog to train) I have made certain from the time she was 9 weeks to not fall into that trap. I will ask for behaviors wherever I am and not necessarily when I know I have food or a toy to reward. Sometimes her reward is a good game of chasey. If she is learning something new I might lure it a few times but after that I won't. She is more toy motivated then food however will certainly work for both. I have worked hard with her to be able to reinforce with a large variety of items so she never knows what she might get. I think that in part helps at reducing the reliance on food. If you can build value for a game of chasey for instance you always have a "reward" available.

I don't know if its your problem or not but I found with my first that part of the whole reliance on food issue was handler driven. I was afraid to ask for stuff unless I had food on me/food available to reward her immediately. I had also inadvertently trained the food as part of the cue to work.

With my older girl I have worked with lots of jackpots and containers of food placed around where I train. Initially within sight and once she did as I asked (not matter how slow to start with) we would run and she would get rewarded. You can eventually place the containers out of sight so the dog never knows where it might get a reward from.

Not sure if this helps or not. I am a bit slack with my old girl now (although she is competing semi-successfully in the UD ring) - hoping 2009 is our year for a couple of passes but I know its a hard trap to get out of if you have been in it for a while. She is now 8 years old and I think was around 4-5 when I realized how food dependant she had become in order to perform.

[Luke W]

Not sure I can help much but I fell into this trap with my first dog and it was hard work to undo that reliance. Having said that she got her CD (under the old rules) and CDX in straight numbers of trials.

My 9 month old BC baby (aside from being a different dog to train) I have made certain from the time she was 9 weeks to not fall into that trap. I will ask for behaviors wherever I am and not necessarily when I know I have food or a toy to reward. Sometimes her reward is a good game of chasey. If she is learning something new I might lure it a few times but after that I won't. She is more toy motivated then food however will certainly work for both. I have worked hard with her to be able to reinforce with a large variety of items so she never knows what she might get. I think that in part helps at reducing the reliance on food. If you can build value for a game of chasey for instance you always have a "reward" available.

I don't know if its your problem or not but I found with my first that part of the whole reliance on food issue was handler driven. I was afraid to ask for stuff unless I had food on me/food available to reward her immediately. I had also inadvertently trained the food as part of the cue to work.

With my older girl I have worked with lots of jackpots and containers of food placed around where I train. Initially within sight and once she did as I asked (not matter how slow to start with) we would run and she would get rewarded. You can eventually place the containers out of sight so the dog never knows where it might get a reward from.

Not sure if this helps or not. I am a bit slack with my old girl now (although she is competing semi-successfully in the UD ring) - hoping 2009 is our year for a couple of passes but I know its a hard trap to get out of if you have been in it for a while. She is now 8 years old and I think was around 4-5 when I realized how food dependant she had become in order to perform.

Fortunately, I've noticed the signs early!

[corvus]

Ness made a good point in suggesting rewards other than food. We moved to life rewards pretty quickly. Things like sitting before crossing a road when on walks. Downing before having a good game of tug or throwing a toy. Kivi is not very play-driven, but he will do things for a game a few times before losing interest. Even things like getting cuddles he will be good for. In fact, we have more troubles with him over his dinner now than anything. He'll stand there and stare at you dumbly when you tell him to sit, just waiting to see if you'll give it to him anyway. He acts dumb, but he's not! I think any reward you use that isn't food helps tell Barkly that it's worth doing it just in case. We also use things Kivi likes doing a lot. If he wants a cuddle, we ask for his paw. He loves that one and will do it just for fun, really. Maybe I'm dreaming, but I think it helps them think beyond "behaviour=food".

[Staff'n'Toller]

I found once I started using food as the primary reward to get anywhere with fading food I had to re-instate my release word, it becomes fairly important. If I am teaching something new I will revert back to rewarding when necessary however once I start putting the food on a schedule all food comes after the release word which signals the finish of the exercise.

I think you are probably going too far too fast as is the general consensus here. If he is barking at 3 cues, release and reward at 2- but you worked that out for yourself. I move pretty quickly from intermittent to variable myself.

I would be concerned about the demand barking, if you are trying to free-shape then it is more probably frustration barking "I DON'T UNDERSTAND WHAT THE HECK YOU WANT!!!"

I also believe smart dogs don't appreciate being asked to do things over and over, Ptolomy will tell you that Tollers will do things once, maybe twice if you're lucky and after that you need to move on or they shut down/bugger off on you.

If you are sure that he is demanding treats then I actually don't time-out for long, as soon as the dog barks I say nothing and turn my back for 5 seconds, then spin back around and ask calmly for something, if they bark again I immediately turn my back; rinse; repeat. I can't stand demand barking so I nip it in the bud fairly quickly using that routine but it will take longer for Barkly because it's fairly established behaviour.

Cheers,

Mel.

[Luke W]

I found once I started using food as the primary reward to get anywhere with fading food I had to re-instate my release word, it becomes fairly important. If I am teaching something new I will revert back to rewarding when necessary however once I start putting the food on a schedule all food comes after the release word which signals the finish of the exercise.

I think you are probably going too far too fast as is the general consensus here. If he is barking at 3 cues, release and reward at 2- but you worked that out for yourself. I move pretty quickly from intermittent to variable myself.

I would be concerned about the demand barking, if you are trying to free-shape then it is more probably frustration barking "I DON'T UNDERSTAND WHAT THE HECK YOU WANT!!!"

I also believe smart dogs don't appreciate being asked to do things over and over, Ptolomy will tell you that Tollers will do things once, maybe twice if you're lucky and after that you need to move on or they shut down/bugger off on you.

If you are sure that he is demanding treats then I actually don't time-out for long, as soon as the dog barks I say nothing and turn my back for 5 seconds, then spin back around and ask calmly for something, if they bark again I immediately turn my back; rinse; repeat. I can't stand demand barking so I nip it in the bud fairly quickly using that routine but it will take longer for Barkly because it's fairly established behaviour.

Cheers,

Mel.

Nice post - thanks S'n'T.

I have trouble maintaining use of a release word. Must start doing that!

So it's:

Sit

Down

Free

Treat

Stand

Free

Treat

Heel

Sit

Free

Treat

???

I think he demand barks when I'm not treating enough and barks in "what the hell do you want" mode during free shaping!

Your point about not wanting to do things over and over again might be close to the mark too - I don't have a heap of behaviors yet, so he does a lot of sitting and downing! Too much probably.

The 5 second turn around...that's pretty much what my 15sec time out is like.

Thanks again.

I really need dog school to start up again!

[Luke W]

PS - I should add...my 'intermittent' is a bit 'variable'. Sometimes it's 2, sometimes 3, some times 4...

[JulesP]

Luke I reward prior to free most of the time. A treat after free is a treat not a reward. The only time I treat after free is if I am running to a jackpot box. After a free I might play tug or give a good patting session.

[Luke W]

Luke I reward prior to free most of the time. A treat after free is a treat not a reward. The only time I treat after free is if I am running to a jackpot box. After a free I might play tug or give a good patting session.

Do you mark before treating?

So..

"Sit"

Click

Treat

Free

?

[kyliegirl]

when you first started training did you use the jackpot method? after one trick, treat, after 3 tricks treat, after a series of tricks a jackpot of awesome treats. After one trick another jackpot, after a series one treat, It keeps them guessing if they will get a big amount, nothing or something.

Vary the treats and their values, like kibble for sits etc, hotdogs for chain commands.. Make it very unpredictable when rewarding. It sounds like Barkly is too reliant on the reward being consistent and hasnt had a jackpot method. If you hold off treating now and then and reward with a jackpot, he will then get the idea that its unpredictable and will be worth it for the jackpot..

It sounds like you might have troubles weaning him off the overall reward system though which if you want to do obedience, you cant treat after every or any series of commands from what I know of.. he will need to know that the treats are a gamble so he keeps trying for the jackpot. toys are also good motivators, but if your dog isnt into playing with toys much then its not a reliable reward system..

I read Barkly loves water, you could use that to your advantage while at the beach you could use the premack method.. Come when called do some tricks and he gets to go play in the water or chase the gulls. Both are very rewarding to the dog and will work to benefit you on also training a reliable recall.

Edited January 2, 2009 by kyliegirl

[Levi]

Fantastic post, I am having the same problem with Bindi.

I have never clicker or food trained before I got Bindi and now that she is 15months old I find that I have to feed her constantly or she just walks away. I realise now that I have been grossly over rewarding her and she is now reliant on food. Even though she has been able to sit on command since she was a puppy, she still gets rewarded for this far too regularly. I guess learning something new is exactly that, learning , and with learning comes error.

Lately I have been making her wait for rewards longer after exercises she likes and knows well. Heeling is an issue though, she has never really liked it and she either drifts wide, lags behind or just walks away if I do not show her food (even just a food pouch is not enough if I don't feed her often).

At the moment while working on things she likes and knows, I am rewarding her almost exclusively with her favorite toy and although she still looks for the food, she is beginning to realise that there is no food, and that the toy is pretty fun too. Her interest is still a bit sporadic though and she can take it or leave it.

The heeling issue though, I am unsure how to rectify. She is just not interested in staying close to my leg. I am trying to break it down to taking one or two steps with her in heel position and then playing a game. However, even staying in position at all does not always happen unless there is food.

Any further suggestions?

Edited January 2, 2009 by Levi

[JulesP]

Yep always mark before treating. The mark in very basic terms means 'that is right something good is now coming'. The mark is never delayed but the food can be.

So you could go

sit

mark

drop

mark

stand

mark

treat

free

tug

A thing I have noticed with people using food is that they don't verbally praise as much as we did when we used non-food training. It has all got a bit quiet with a reliance on the food being enough.

Levi I would still be using food until doggie is staying there. Sometimes people try and fade the food too quickly or are not generous enough with the food.

[Luke W]

when you first started training did you use the jackpot method? after one trick, treat, after 3 tricks treat, after a series of tricks a jackpot of awesome treats. After one trick another jackpot, after a series one treat, It keeps them guessing if they will get a big amount, nothing or something.

Vary the treats and their values, like kibble for sits etc, hotdogs for chain commands.. Make it very unpredictable when rewarding. It sounds like Barkly is too reliant on the reward being consistent and hasnt had a jackpot method. If you hold off treating now and then and reward with a jackpot, he will then get the idea that its unpredictable and will be worth it for the jackpot..

It sounds like you might have troubles weaning him off the overall reward system though which if you want to do obedience, you cant treat after every or any series of commands from what I know of.. he will need to know that the treats are a gamble so he keeps trying for the jackpot. toys are also good motivators, but if your dog isnt into playing with toys much then its not a reliable reward system..

I read Barkly loves water, you could use that to your advantage while at the beach you could use the premack method.. Come when called do some tricks and he gets to go play in the water or chase the gulls. Both are very rewarding to the dog and will work to benefit you on also training a reliable recall.

Yep - use jackpotting.

Often I finish a session with a jackpot, or you a jackpot when he does a behavior particularly well.

I plan to go for obedience titles so I need to get him working well with the promise of rewards after however long it takes to get a pass.

[Luke W]

Yep always mark before treating. The mark in very basic terms means 'that is right something good is now coming'. The mark is never delayed but the food can be.

So you could go

sit

mark

drop

mark

stand

mark

treat

free

tug

A thing I have noticed with people using food is that they don't verbally praise as much as we did when we used non-food training. It has all got a bit quiet with a reliance on the food being enough.

Levi I would still be using food until doggie is staying there. Sometimes people try and fade the food too quickly or are not generous enough with the food.

Ahh...but you are not treating after each mark. In my case the mark is a 'click' or a verbal maker ('boop') and the 'clicker should never lie' - a treat should always follow a click. Well, that's one theory any way!

[kyliegirl]

Yep always mark before treating. The mark in very basic terms means 'that is right something good is now coming'. The mark is never delayed but the food can be.

So you could go

sit

mark

drop

mark

stand

mark

treat

free

tug

A thing I have noticed with people using food is that they don't verbally praise as much as we did when we used non-food training. It has all got a bit quiet with a reliance on the food being enough.

Levi I would still be using food until doggie is staying there. Sometimes people try and fade the food too quickly or are not generous enough with the food.

Ahh...but you are not treating after each mark. In my case the mark is a 'click' or a verbal maker ('boop') and the 'clicker should never lie' - a treat should always follow a click. Well, that's one theory any way!

I would be against marking after each behaviour without a reward follow up.. Originally the click is supposed to be the "thats it youve done it!" indicator, therefore acting as a release and reward cue. If your clicking after every behaviour and not following up with a reward from the mark your going to confuse the dog on the true purpose of the clicker, which is to mark the correct behaviour and be rewarded for it.

your supposed to be fading the clicker not fading the reward from the clicker.

Edited January 2, 2009 by kyliegirl

[kelpiechick]

Originally the click is supposed to be the "thats it youve done it!" indicator, therefore acting as a release and reward cue.

True, each click should be followed by a reward, but the click should not be a cue for the behaviour to end, therefore it never acts as a release cue. (or shouldn't anyway)

Levi - I am not an 'obedience person' (although all my dogs have completed basic classes) so probably not the best person to give you advice about heeling but there was a really interesting article in 'Clean Run' magazine a couple of months back in the foundation training column where Mary Ellen Barry was teaching her puppy to heel by targeting a bulldog clip, which she then attached to her pants.

Not sure if this would be a good method to use for someone considering competition obedience, but I found it very interesting as I have never seen heeling taught like this before. Also took care of the food problem, as she built up value for the clip (paired nose touching the clip with reward) - however you would then have the problem of fading the clip.

But it certainly seemed to work to get the dog in desired position.

[Kavik]

Some people use the click = end of behaviour

There are several ways to teach heeling. With Diesel I've done: spitting food (with and without clicker), using a toy (at shoulder height first then hidden in my bra strap so he can't see it ) and at the moment working on getting the reward off me, so teaching him to run to my treat bag, then release him to treat bag when he is in heel position with attention on me. Since the best focus I get is at dinner time when he will look at me and I can get him to heel with great gusto before releasing him to his food I think this has the potential to work

[MrsD]

True, each click should be followed by a reward, but the click should not be a cue for the behaviour to end, therefore it never acts as a release cue. (or shouldn't anyway)

Absolutely agree, for my dogs the click *never* ends the behaviour - you get a behaviour with more drive & longer duration teaching your dog to work through the clicker. I also agreed with your post at the top of the thread too Kelpiechick , luring is fantastic but you HAVE to stop luring quickly & get the dog working without it otherwise you'll get nowhere fast.

Levi, next time you see Maree ask her to show you how to lure into position & then click & treat for the first step & then do it again without the food & then eventually getting her to work through the clicker. If Bindi understands the clicker (& you are consistant with it) she should pick it up fairly quickly although getting a long stint of heeling won't happen overnight of course. You've seen how Jarrah works in general, that's how I got him to where he is. Your biggest hurdle is your height & Bindi's size but Maree could help you with that as well.

If you get a chance, Steve Davies (from Sydney) is very good at teaching that sort of stuff as well, Im hoping he'll be at Camp Tailwaggers again this year as I loved his seminar last year & learnt alot .

BTW, I never release my dogs at the end of a behaviour, only when I have finished working with them. I expect them to be watching me & waiting for another command whilst Im working with them. The reward is my release.