100% Recall

[Staranais]

So I don't think it's quite as simple as conditioning. Mind you, it's the best thing we've got.

I suspect, from what I have observed, the dog turning towards you as soon as it hears the recall signal is a classically conditioned response, an automatic reflex that the dog just can't help, as long as you've taught it correctly. But after that initial orienting response, there's plenty of time for the dog to weigh up whether or not he wishes to return - and whether he comes back depends on what he thinks you might do to him to reward or punish him, compared to the value of the distraction & his chances of getting it.

[Aidan]

I hear this described as a classically conditioned response fairly often, I'm not sure where that began and there is some truth behind it (there are some respondents conditioned in a recall) but turning your head is an operant. That said, operant conditioning can be an entirely subconscious process even for humans, the most cognitive creature on the planet.

If a dog is still weighing up the consequences of chasing the bunny as opposed to the consequences of returning, this is not a fluent behaviour under these stimulus conditions.

[Staranais]

You're more up on the terminology than I am, Aidan, but my understanding of was that the conditioned response was involuntary response, as opposed to operant where it's a voluntary response. In a good recall, I think the initial orienting response to the owner is involuntary - just like orienting to a loud or squeaky or (conditioned) clicker noise is involuntary. Actually turning around and returning to the owner is operant.

I could be using the terms incorrectly. However, I do think those stages (orienting and actually deciding to return) are different, and once you have the automatic orienting down, your recall is more than halfway to being reliable.

[corvus]

If a dog is still weighing up the consequences of chasing the bunny as opposed to the consequences of returning, this is not a fluent behaviour under these stimulus conditions.

Ah, that's what I was getting at when I said I think it's easier to condition some dogs than others. I can't remember now if I said that in this thread. Erik doesn't weigh up consequences, he just recalls. He's had a lot less training and practise than Kivi and he generalised a lot easier. Kivi is a matter of practising the recall in every single freaking situation imaginable. It gets hard with high levels of arousal because he doesn't often get that aroused. The situation with the possum carcass was probably influenced by the fact that he was at the bottom of a very steep embankment and so his initial momentum didn't get him far. I actually think he may have turned around to go another way with a less steep hill and gone "Hello again possum!" on the way and that was it. He wasn't coming after that and I had to go and get him. Haven't had to do that since he was a wee puppy.

I find it hard to accept that his recall isn't conditioned because 95% of the time he doesn't think about it. My rewards are not higher value than what he is currently doing, and he recalls like he's meant to. But the 5% of the time he doesn't recall is more or less random. He's as likely to blow you off in an everyday situation with low distraction from which he has recalled scores of times without a hitch as he is when he's chasing birds or something. All I can say is it doesn't make sense to me. There are probably details I'm missing and always will because he's a good 30m away when I call.

[Aidan]

Ah, OK. Yes, once you have the automatic orienting down you've got possibly the most important bit. Orienting can be a reflex (unconditioned response via the autonomic nervous system), and there are respondents involved in the example under discussion, but we are describing an operant (via the central nervous system) which has been conditioned using the 3-term contingency of operant conditioning: Antecedent > Behaviour > Consequence.

The terms "voluntary" and "involuntary" could be a bit misleading here, we are really talking about ANS vs CNS, whether the hippocampus is involved or not. Things like blood pressure, heart rate, salivation, neurotransmitters etc are respondents and things like head turning, running, barking etc are operants. The early textbooks will tell you that respondents cannot be conditioned using consequences (operant conditioning) but bio-feedback disproved that (to some extent) and there is a lot of cross-over and multiple things going on at once within a complex organism such as a dog.

Does a dog weigh up whether he will recall or not? Possibly, but it's not that simple. We do know that, done correctly, we can condition a dog to come back for a schedule of reinforcement using very ordinary reinforcers. Does this mean that we have made ourselves "more interesting than a kangaroo" - I highly doubt that. If the dog isn't conditioned to recall then the best we can hope for is to bribe him back, in which case he WILL weigh it up and we will need to be more interesting (valuable?) than the kangaroo. If we've been through a process of graded exposure and generalised the response we can get a very high probability of compliance even for trials where no reinforcer is available.

[Aidan]

Ah, that's what I was getting at when I said I think it's easier to condition some dogs than others.

There is no doubt at all. Your average Golden is not your average Borzoi. Skinner suggested that genetics will provide variance in responses to different reinforcers, which is the simplest, most reasonable explanation that I have seen.

He's as likely to blow you off in an everyday situation with low distraction from which he has recalled scores of times without a hitch as he is when he's chasing birds or something. All I can say is it doesn't make sense to me. There are probably details I'm missing and always will because he's a good 30m away when I call.

I'd guess that he was more sensitive to discrimination, whereas Erik readily generalises.

[Staranais]

The terms "voluntary" and "involuntary" could be a bit misleading here, we are really talking about ANS vs CNS, whether the hippocampus is involved or not. Things like blood pressure, heart rate, salivation, neurotransmitters etc are respondents and things like head turning, running, barking etc are operants. The early textbooks will tell you that respondents cannot be conditioned using consequences (operant conditioning) but bio-feedback disproved that (to some extent) and there is a lot of cross-over and multiple things going on at once within a complex organism such as a dog.

From my understanding, the CNS would by definition be involved in any learning experience, classical or operant or whatever. Whereas the ANS is merely peripheral like the rest of the PNS. Not really sure what you're getting at with the neuroscience?

[Aidan3]

In that case I've not explained it very well! The distinction is that the stuff the ANS takes care of are Respondents. If you use classical conditioning, that's the sort of thing you are working with. The ANS does not take care of head-turning, it might do something that causes the dog to turn it's head but the ANS does not do the head turning.

I'm not convinced I'm doing a very good job of explaining this (which may be a problem for me with exams coming up!) so I've turned to Google and uncovered this:

http://www.psywww.com/intropsych/ch05_cond...nditioning.html

[Staranais]

In that case I've not explained it very well! The distinction is that the stuff the ANS takes care of are Respondents. If you use classical conditioning, that's the sort of thing you are working with. The ANS does not take care of head-turning, it might do something that causes the dog to turn it's head but the ANS does not do the head turning.

I'm not convinced I'm doing a very good job of explaining this (which may be a problem for me with exams coming up!) so I've turned to Google and uncovered this:

http://www.psywww.com/intropsych/ch05_cond...nditioning.html

Hmmm, read the link, but still not convinced that the distinction is real sorry Aidan. And that link is sort of wrong, no offence, but the ANS isn't the part of the nervous system responsible for emotions. The sympathetic branch controls the release of adrenaline & noradrenaline, yes, & the CNS may interpret the body's reactions to these hormones as fear or stress, but that's as close as the ANS comes to getting involved in emotion. I believe the limbic system and frontal cortex (both CNS) process emotion.

An increase in sympathetic tone (ANS) is probably involved with the orienting response, I buy that. But the two divisions of the ANS are working at all times in any mammal. And I doubt any change in ANS tone is strictly necessary for classical conditioning to occur - adrenaline release and classical conditioning need not go hand and hand. And the CNS will also be involved as an integral part of classical conditioning - it must be, or else all you would get would be a reflex arc, no learning.

The behavioural response in operant conditioning will obviously involve the CNS, but also must involve the somatic peripheral division (voluntary motor control), as well as generally have ANS peripheral involvement (if the dog runs back to you in a recall, the symp ANS is involved. If it anticipates the food reward, the parasymp will ramp up. etc etc).

Sorry, but just makes no sense to me. I'm not always 100% with behaviourist lingo, but I do know a little mammalian neuroscience. The only difference I can see between the two in terms of the peripheral nervous system is that an operant response must generally (always?) involve somatic peripheral system, whereas a classically conditioned response need not.

Edited May 5, 2010 by Staranais

[Aidan3]

But the two divisions of the ANS are working at all times in any mammal. And I doubt any change in ANS tone is strictly necessary for classical conditioning to occur - adrenaline release and classical conditioning need not go hand and hand. And the CNS will also be involved as an integral part of classical conditioning - it must be, or else all you would get would be a reflex arc, no learning.

My replies and the article spoke in broad terms and I think you've taken a more black and white view than was intended. I wasn't trying to give the impression that any part of the nervous system stopped functioning at any time, and I certainly didn't imply that all classical conditioning involved adrenaline release (but adrenaline release is respondent behaviour).

The behavioural response in operant conditioning will obviously involve the CNS, but also must involve the somatic peripheral division (voluntary motor control), as well as generally have ANS peripheral involvement (if the dog runs back to you in a recall, the symp ANS is involved. If it anticipates the food reward, the parasymp will ramp up. etc etc).

Yes! That is it, and I thought I had made a reasonable attempt to convey that, you'll get both operant conditioning and respondent conditioning and the two are interdependent but "turning the head" is an operant. If you've conditioned head turning as part of a 3-term contingency, it's not a classically conditioned response.

The only difference I can see between the two in terms of the peripheral nervous system is that an operant response must generally (always?) involve somatic peripheral system, whereas a classically conditioned response need not.

OK, that'll do I guess Burch and Bailey describe respondent conditioning as affecting "the reflexive actions of the glands and smooth muscles" which I think is probably adequate.

[Staranais]

My replies and the article spoke in broad terms and I think you've taken a more black and white view than was intended. I wasn't trying to give the impression that any part of the nervous system stopped functioning at any time, and I certainly didn't imply that all classical conditioning involved adrenaline release (but adrenaline release is respondent behaviour).

Well, by definition, anything involving the ANS either involves the sympathetic or parasympathetic nervous systems (these are the two component arms of the CNS), and these necessarily involve the release of catecholamines (adrenaline and noradrenaline) and/or acetylcholine. If classical conditioning always involves the ANS, then it must always involve either catecholamine or acetylcholine release, and their action at certain receptor types on certain nerves in the periphery. (Incidentally, to lighten the mood, I should mention that I can do a dance describing all of these receptor subtypes - oh the joys of vet school!)

So perhaps I'm taking a black and white view, but it's also true that those neurological terms refer to fairly specific, well defined, black and white phenomena.

Yes! That is it, and I thought I had made a reasonable attempt to convey that, you'll get both operant conditioning and respondent conditioning and the two are interdependent but "turning the head" is an operant. If you've conditioned head turning as part of a 3-term contingency, it's not a classically conditioned response.

No, I agree it's not, and sorry if I lead you to believe I thought it was. The only part I think is classically conditioned is the orienting.

OK, that'll do I guess :D Burch and Bailey describe respondent conditioning as affecting "the reflexive actions of the glands and smooth muscles" which I think is probably adequate.

Yes, that's basically saying the same thing in different terms. Smooth muscle and glands are innervated by the ANS. Although respondent/classical conditioning must necessarily also affect the CNS as well as the ANS.

Edited May 5, 2010 by Staranais

[Aidan3]

My replies and the article spoke in broad terms and I think you've taken a more black and white view than was intended. I wasn't trying to give the impression that any part of the nervous system stopped functioning at any time, and I certainly didn't imply that all classical conditioning involved adrenaline release (but adrenaline release is respondent behaviour).

Well, by definition, anything involving the ANS either involves the sympathetic or parasympathetic nervous systems (these are the two component arms of the CNS), and these necessarily involve the release of catecholamines (adrenaline and noradrenaline) and/or acetylcholine. If classical conditioning always involves the ANS, then it must always involve either catecholamine or acetylcholine release, and their action at certain receptor types on certain nerves in the periphery. (Incidentally, to lighten the mood, I should mention that I can do a dance describing all of these receptor subtypes - oh the joys of vet school!)

Video or it didn't happen :-)

You've gone well beyond anything I need to know, perhaps all respondent behaviour does involve catecholamines or acetylcholine? I do actually have to study this but not for a little while. I'll have to learn your dance :D Although it's probably different for non-mammals and very simple organisms. <- ETA: I meant what happens in the brain and nervous system, not the dance! I may be a very simple organism, but I am definitely a mammal :D

Yes! That is it, and I thought I had made a reasonable attempt to convey that, you'll get both operant conditioning and respondent conditioning and the two are interdependent but "turning the head" is an operant. If you've conditioned head turning as part of a 3-term contingency, it's not a classically conditioned response.

No, I agree it's not, and sorry if I lead you to believe I thought it was. The only part I think is classically conditioned is the orienting.

The perception, what that does in the brain and nervous impulses, yes, those are respondents.

Edited May 5, 2010 by Aidan2

[jackie_a1]

I don't think there is such thing as 100% recall. You cannot say that. Assuming your dog will ALWAYS return to you when called? So you are dismissing environmental factors, for example an interesting scent, another dog, another animal, a frisbie, Al Pacino could walk out and call your dog you never know. You cannot predict your dog will always return to you it's an impossible statement to adhere to.

Animals are unpredictable, as are humans. I always say I could never harm or kill a person HOWEVER, what happened if that person attacked me or threatened me etc I cannot with one hundred percent certainty say I wouldn't harm or kill them because it hasn't happened.

If what you're asking is has my dog demonstrated the ability to come back so far 100 percent of the time then maybe some people will answer yes. But if you are asking can a dog have 100 % recall period. No the answer is no.

No dog can have 100 percent recall.

[Staranais]

Video or it didn't happen :-)

You've gone well beyond anything I need to know, perhaps all respondent behaviour does involve catecholamines or acetylcholine? I do actually have to study this but not for a little while. I'll have to learn your dance Although it's probably different for non-mammals and very simple organisms. <- ETA: I meant what happens in the brain and nervous system, not the dance! I may be a very simple organism, but I am definitely a mammal :p

Well, I can't imagine any situation you'd need to know the dance (or the component receptors) to train a dog. Prescribing medicines - well, perhaps I'll just do the dance in my head instead of actually in front of my clients when I'm deciding what to give the dog. They may lose faith in their veterinarian... :p

Seriously though, I'm not sure how useful it is to analyse dog training at the level of neurology. Seems kind of like analysing baking at the level of chemistry, or car mechanics at the level of atomic physics. You can do it, perhaps it's sometimes useful, but you know, going into that much detail often just muddies the discussion (or ends up with the science being oversimplified to the point where it verges on untruth).

[dandybrush]

100% recall with my coolie, always has been. Even the other day when he was off the lead on a walking track and horses rode nearby which got his interest he came immediately when called.

as for the goldens well they are a different story consequently they stay on the lead on the walking track.

can i ask how you got 100% with you're coolie, with mine he comes in everyday situations but soon as its a bit different and more exciting to be elsewhere he wont come.

Does anyone have any exercises i could do to work on his recall and i do not intend to buy a whistle, i want him to come even if he doesnt want to come if you know what i mean

[corvus]

Seriously though, I'm not sure how useful it is to analyse dog training at the level of neurology. Seems kind of like analysing baking at the level of chemistry, or car mechanics at the level of atomic physics. You can do it, perhaps it's sometimes useful, but you know, going into that much detail often just muddies the discussion (or ends up with the science being oversimplified to the point where it verges on untruth).

That is so true. If there's one over-arching thing I have learnt since starting my PhD it's that neurology is extremely complicated and anyone trying to apply it to behaviour and how to affect it or understand it needs to be very careful.

Having said that, I do think it can help. There's a lot of info out there about rewards and the brain that I think aids understanding. But I find it a little bit nerve-wracking when people start talking about neurology in relation to dog training. It's like watching someone swallow a sword.

As far as exercises to aid in recall go, hide-and-seek is the classic one. You can also play games where someone holds your dog and you run away from them and then the dog is released to chase after you. Leslie Nelson also recommends the name game, where you say your dog's name and reward them when they look at you over and over. I like check-ins. My dogs get a treat for deciding to come over to see what we're doing when they are off leash. Kivi gets this look on his face where he's suddenly like, "I want a treat." and he trots over and either sits or glues himself to someone's leg and waits expectantly for his treat. He doesn't usually get check-in treats anymore because he's such a pro. Instead he gets check-in training sessions. He comes over for his check-in treat and we ask him to do a couple of trick or practise a few things we're working on and then give him is treats. Erik isn't as big on treats at the park and gets tug games periodically instead. He likes doing tricks as well.

[Aidan3]

Video or it didn't happen :-)

You've gone well beyond anything I need to know, perhaps all respondent behaviour does involve catecholamines or acetylcholine? I do actually have to study this but not for a little while. I'll have to learn your dance Although it's probably different for non-mammals and very simple organisms. <- ETA: I meant what happens in the brain and nervous system, not the dance! I may be a very simple organism, but I am definitely a mammal :D

Well, I can't imagine any situation you'd need to know the dance (or the component receptors) to train a dog.

I'm studying behavioural science, including neuropsychology.

Seriously though, I'm not sure how useful it is to analyse dog training at the level of neurology.

It certainly wasn't my intention for us to go that far down that path, I was just pointing out that the head turning was not classically conditioned. I have seen people say they taught a good recall using classical conditioning, and certainly there would be some classical conditioning in a good recall, but the criteria people will reinforce in a good recall will normally always be operants, subject to operant conditioning. I think that is very relevant to the discussion. If you want the dog to respond quickly, selectively reinforce the most immediate responses.

[JulesP]

I've been fairly smug about my dog's recalls. They are border collies that I recall from moving prey (bunnies, sheep) and other dogs. I have however found their weakness - people sitting down having picnics. They do recall but it is not as easy as usual and if I don't notice the people in time, they do get a lick in. oppps.

[jesomil]

I think my dogs would have pretty much 100% recall excluding circumstances of them being badly hurt etc.

I cannot think of a single situation that they wouldnt recall from and they have been tested in many. My boy will even recall off a bitch in full season without hesitation. He is a very confident, active dog but knows it is worth his while to recall over anything.

Its not a matter of me having a reward.

I had a Afghan I trained in agility and trialled in obedience. I got her at 2 years of age and taught her a very strong recall and she was always off lead with me but I wouldnt say I could achieve 100% with her although under most circumstances it would be close.

I also had a young greyhound for a year who also had a solid, reliable recall but couldnt say it would be 100%.

I could let my sighthounds off anywhere I would let my other dogs off without problem but I dont believe I could achieve 100% like i could with my working dogs, more like 90%.

I didnt use methods requiring me to be more exciting than what they were after though as it simply wouldnt work. I use negative reinforcement always for recalls as that has provided me with happy, confident, safe dogs to take out.

Edited May 7, 2010 by jesomil

[Vickie]

I didnt use methods requiring me to be more exciting than what they were after though as it simply wouldnt work. I use negative reinforcement always for recalls as that has provided me with happy, confident, safe dogs to take out.

I have been hesitant to enter this discussion...but now I have an ally , I will.

I agree that I could never offer my dogs something more exciting than what they are after. I guess I use negative reinforcement as well (although have used positives too).

I start the first time they ever ignore my call. I just simply go & get them grab their collar & pull them a little toward me, then let go & let them choose to follow, if they don't I repeat the process, and again & again, until they believe they have no choice. They learn early that they never have the option of not coming as I will go get them. Shine probably has the best recall I have ever had in a dog (so far it is 100% but I could not possibly test it against everything). She was taught to recall off sheep before she ever worked & she has never not responded to my "that'll do" command. She has recalled off dogs, people, flying ducks, cattle & just last week 3 very speedy kangaroos. I whistle & generally my dogs will about-turn in full stride on hearing the whistle. I very rarely reward them for coming back, but may give verbal praise sometimes.

I have always felt that having a recall based purely on reward makes it optional for the dog. Flame me too .