SPARCS

Get to know other Positively members here.

Moderators: emmabeth, BoardHost

bendog
Posts: 2188
Joined: Fri Jan 07, 2011 7:42 am

Re: SPARCS

Post by bendog »

Nettle wrote:I have seen none of this, but might I be heading in the right direction by assuming he is using herding dogs in his demonstrations?

Because this sure as heck won't work with a large number of other breeds.
Right on the money as usual Nettle!

I'm still mulling all this over so will post my views when I've made them!
WufWuf
Posts: 1371
Joined: Thu May 12, 2011 7:53 am

Re: SPARCS

Post by WufWuf »

Bang on Nettle, I should have mentioned that he uses Border Collies (of course :P ). He also said that they have to "force" these dogs to take a break as they do not want to stop working, (It is possible that it's just because they are BC's or is it because they are addicted to the work, he did not say what his own thoughts were on that but I got the impression he might not be that experienced with other dogs outside of the research environment)
mansbestfriend wrote:I remember he talked about some of the processes of training for snake and turtle scent ID and discrimination. He touched on the "Clever Hans" effect, and how dogs/animals can pick up on unintended human cues.
MBF - This was from his talk "Applied Canine Olfactory Processing" the effect of clicker training on the brain was from his talk "It's not what you like but what you want that counts".
I am skeptical even not having heard his talk that you can create a true addict to training in the same sense you would refer to someone who is addicted to cocaine or alcohol.
Cocaine and alcohol are consumable substances that effect the brain and cause the user to become addicted to them however the brain can also become addicted to it's own chemistry (gambling addiction, shopping addiction, sex addiction etc) it is this mechanisim of addiction that SG was discussing.

It's not a new idea and it is already regularly applied in standard dog training by using some of the schedules of reinforcement (variable/intermitent). We often use the analogy that humans don't get addicted to vending machines but they do get addicted to slot machines in explaining how to have your dog more engaged in training and SG mentioned that slot machine programmers are experts at applying schedules of reinforcement to maximise the addictiveness of the slot machines.

I have maintained a high rate of reinforcement with Honey as it feels "right" to me for her. This does not mean that she will not work without a reward, I often ask her to do things without a reward as this is part of life and during a trick training session I can and do ask for a few (maybe 3/4) behaviours in a row before I reward. I don't however do this in every training session as I feel that the more "yay" moments she has the better considering how damaged she was and how low her confidence was. She really enjoys training and will initiate sessions herself but I usually keep them short and sweet and I do it more as a fun activity, confidence building experience and bonding time rather than because I have any great desire to perfect her tricks or increase her "obedience".

I do strive to be the best dog lady I can be though and I spend time pretty much every day reading as much as I can about behaviour and training research, ideas, techniques etc in order to grow, however as the lovely Nettle advised me some time ago I also strive to "question everything" :wink: . This is all part of my questioning the research and the application of it in my own "brand" of dog training.

This SPARCS conference made me far more aware of the effect training can have on a dogs brain chemistry and I found myself a little uncomfortable about the idea of creating an addict (which I personally do believe is possible, probably not with all dogs but I've met some training addicted Jackers* as well as the more bidable breeds). I found myself wondering how much this would increase a dogs stress levels and ability to "switch off". If I felt that applying this mechanisim of addiction would benefit Honey I would be remiss in not using it in my own training but as I said my greatest desire is to reduce her stress not add to it.

* SG mentioned that some breeds have a greater propensity for addiction, the breeds he listed were Border Collies, Belgien Malinois and ... Jack Russell Terriers :shock: :lol:. These breeds were all described as "neurotic" with a greater "need for excitement". I had this lovely vision of SG in a lab surrounded by a sea of crazy JRTS "saying nope it's clear this will never work go get the BC's" :mrgreen:
Operant conditioning rocks but classical conditioning rules
jacksdad
Posts: 4887
Joined: Thu Dec 10, 2009 10:48 pm

Re: SPARCS

Post by jacksdad »

mansbestfriend wrote:the "click" (secondary reinforcer)
which raises the questions.....

is the click actually reinforcing

and

is there really such a thing as a "secondary" reinforcer

there are some pretty smart people out there that aren't so sure about either of those.
jacksdad
Posts: 4887
Joined: Thu Dec 10, 2009 10:48 pm

Re: SPARCS

Post by jacksdad »

WufWuf wrote:Cocaine and alcohol are consumable substances that effect the brain and cause the user to become addicted to them however the brain can also become addicted to it's own chemistry (gambling addiction, shopping addiction, sex addiction etc) it is this mechanisim of addiction that SG was discussing.

It's not a new idea and it is already regularly applied in standard dog training by using some of the schedules of reinforcement (variable/intermitent). We often use the analogy that humans don't get addicted to vending machines but they do get addicted to slot machines in explaining how to have your dog more engaged in training and SG mentioned that slot machine programmers are experts at applying schedules of reinforcement to maximise the addictiveness of the slot machines.
I have not heard of variable/intermittent schedules of reinforcement described/explained in terms of addiction. I am going to have to listen to his talk and think about this. but my gut feeling is addiction is not the right word to be using here. that may sound like I have already prejudiced my self to what he has to say, but I am not. I am far more interested in finding useful and solid information than being personally right. So, if he can make his case I am open to his argument/observations.
WufWuf
Posts: 1371
Joined: Thu May 12, 2011 7:53 am

Re: SPARCS

Post by WufWuf »

jacksdad wrote:I have not heard of variable/intermittent schedules of reinforcement described/explained in terms of addiction
JD a google search of "reinforcement addiction" will bring you to plenty of articles explaining how addictions are effected by schedules of reinforcement.
Operant conditioning rocks but classical conditioning rules
jacksdad
Posts: 4887
Joined: Thu Dec 10, 2009 10:48 pm

Re: SPARCS

Post by jacksdad »

WufWuf wrote:
jacksdad wrote:I have not heard of variable/intermittent schedules of reinforcement described/explained in terms of addiction
JD a google search of "reinforcement addiction" will bring you to plenty of articles explaining how addictions are effected by schedules of reinforcement.
But that isn't the same as them creating addiction. something just doesn't feel right about this. again, going to have to watch the talk and spend sometime thinking about.
emmabeth
Posts: 8894
Joined: Tue Oct 17, 2006 9:24 pm
Location: West Midlands
Contact:

Re: SPARCS

Post by emmabeth »

I am interested to know more, but currently busy/knackered so someone needs to prod me with any info they find!

I think I have witnessed with some dogs, the click becoming more important than the reward, certainly with Rocky who has been clicker trained since 5 months old and is now 14... BUT... really huge massive BUT ...

Whilst I do think he finds the click MORE important than the reward... he still clearly wants the reward. The click being more significant to him means he will do clicker training for ANY reward, we joke you could reward him with dust but the truth is not far off that really, he'd work as hard for something he adores and would sell his Grandma for as he would for something that is merely 'food', the only qualifiers are it must be edible and not something he actively dislikes.

Try without the reward nad whilst he will perform a bit, he rapidly gets visibly frustrated and annoyed and then won't play (not that we've done that a lot obviously!) When we tried a variable schedule of reward, but clicking every time, his responses to behaviours he knows pretty well, became quite sloppy adn slapdash - basically if he wasn't assured of a reward, then his concentration and accuracy went out the window and he'd throw in half hearted attempts or 'nearly' stuff because well, why try hard if you might not get anything!
West Midlands based 1-2-1 Training & Behaviour Canine Consultant
jacksdad
Posts: 4887
Joined: Thu Dec 10, 2009 10:48 pm

Re: SPARCS

Post by jacksdad »

emma, I think that means the reward is still more important than the click to rocky. if the click was more important, than when you clicked and didn't provide a reward you wouldn't have gotten the frustration.

I think the click only appears more important because the click signals something good is about too happen. food, a chance at a toy etc. it's not the click that is rewarding/important, it's what the click represents. take away the really super cool thing that follow the click...click isn't so fun anymore.
emmabeth
Posts: 8894
Joined: Tue Oct 17, 2006 9:24 pm
Location: West Midlands
Contact:

Re: SPARCS

Post by emmabeth »

Yeah... but if we work without the clicker, hes less enthusiastic, and bummed out if the rewards are low quality. It really does seem as if its the clicker that really makes the deal for him - if I try to reward with say, crusts of bread (tiny bits, crumbs really!) he's not interested. Pick up the clicker and he's right there, hes totally into it! Weird old boy!
West Midlands based 1-2-1 Training & Behaviour Canine Consultant
mansbestfriend
Posts: 301
Joined: Mon May 20, 2013 7:35 am
Location: Queensland, Australia

Re: SPARCS

Post by mansbestfriend »

http://www.damninteresting.com/clever-h ... ath-horse/ (???)

I'm still trying to decide whether a reaction like "see the possum equals chase the possum" is self-rewarding or simply like a dog brain on auto-pilot. Is it like a human involuntary reflex response, or different?

More questions than answers. :)
The journey of a thousand miles begins with a single Sit.
WufWuf
Posts: 1371
Joined: Thu May 12, 2011 7:53 am

Re: SPARCS

Post by WufWuf »

jacksdad wrote: WufWuf wrote:

jacksdad wrote:I have not heard of variable/intermittent schedules of reinforcement described/explained in terms of addiction



JD a google search of "reinforcement addiction" will bring you to plenty of articles explaining how addictions are effected by schedules of reinforcement.



But that isn't the same as them creating addiction. something just doesn't feel right about this. again, going to have to watch the talk and spend sometime thinking about.
Sorry took me a few days to get around to finding something that might help clarify the process of addiction.

http://www.helpguide.org/harvard/addict ... _brain.htm
Operant conditioning rocks but classical conditioning rules
jacksdad
Posts: 4887
Joined: Thu Dec 10, 2009 10:48 pm

Re: SPARCS

Post by jacksdad »

emmabeth wrote:Yeah... but if we work without the clicker, hes less enthusiastic, and bummed out if the rewards are low quality. It really does seem as if its the clicker that really makes the deal for him - if I try to reward with say, crusts of bread (tiny bits, crumbs really!) he's not interested. Pick up the clicker and he's right there, hes totally into it! Weird old boy!
bear with me, I know much of what I will say next you know, but I am going to say it anyway to help you and others follow my thought process.

The clicker -
Until we give it meaning, generally speaking the clicker has no meaning to the dog. when a dog first hears a clicker the dog may try and orientate to it, look for the source of the sound, maybe even startle a little at first. if all you ever dog is click, click, click at whatever interval you chose to, assuming the dog isn't sound sensitive (at issue out side this discussion) the dog will habituate to the click and eventually ignore it.

But if paired with food...click, food, click, food, click, food etc, the dog learns that the click predicts something. in this case food. this, in case anyone doesn't know is simple classical conditioning at work. think Pavlov's bell causing dogs to salivate.

Sequence of events -
I recently listened to a very fascinating 90 minute webinar on K9 compulsion. It has been observed that for some dogs a part of helping the dog can be as "simple" as eliminating 1 part of the sequence by preventing it from every happening again. for example of X, Y, and Z are the sequence of events that then trigger the dog to go into a compulsive behavior, simply removing X or Y breaks the sequence and opens a window to replace the unwanted behavior with the wanted. but what if X Y Z in other context isn't bad or harmful. what if a sequence of events gives the dog information about things about to happen?

Putting my thoughts together.

Rockey has a VERY strong reinforcement history of the click being followed by something high value. And you have at some point phased in the odd lower value and/or random rewards so Rockey never quite knows what to expect, except that Click = something to eat. So similar to the compulsive dogs you have an X Y Z sequence, but rather than Rockey going into an unhealthy compulsive behavior, he is conditioned to expect something to eat because the clicker is making noise. due to reinforcement history you can vary what he gets steak this time, apple the next, bread crumbs, and back to steak. But if you take out the click, then there is nothing to tell him "something cool" is about too happen or might happen. rather he is just suddenly getting old boring bread crumbs in an entirely different sequence of events. there is nothing in the click less history that tells him that ya bread crumbs this time, but steak might happen next time. On the other hand when he hears the click there IS a history that tells him that while this time bread crumbs happened, next time might be a chunk of steak. Which keeps his interest and willingness to work active.

I don't think clicks by them self are truly a reinforcer. not the way food is, or a toy. it is a classically conditioned marker. with a strong enough history the dog might work for a while without the food component but as we know with classical conditioning it has to be maintained to be effective. which is why dogs eventually show frustration and quite working on clicks alone. depending on the reinforcement history you might get to go a good long while with just clicks (not that I would actually suggest it). Steve White talked once about a chicken that due to it's reinforcement history had an extinction period of about 3000 pecks on the bar that used to deliver food, but now didn't before it quit trying.

My opinion ... for all it's worth and I am not saying I am right no if and or but.... is that IF a dog has a really strong reinforcement history of click = treat, they will appear to work for a while on the click alone and thus the click is mistaken as being a reinforcer. but It's not, rather your are simply with drawing on that big bank of reinforcement you built up, but sooner or later the dog figures out no reward is coming and goes on strike. When you eliminate the food from the click = food sequence, I currently believe you start an extinction process and you may due to history get away with treat less clicks for the short term, but eventually that banked history gets used up.

And while I do understand that you can create reinforcers other than food by using classical conditioning, those created reinforcers (which the click would technically be) HAVE to under go "maintenance" to maintain their power. food will always have a inherent power, toys and some activities may also have an inherent power, but "artificial" reinforcers loose their power IF you don't do some "maintenance" on them.

Also, because of the roll and purpose of the clicker, I am not sure why "you" (generic you, no one specific) would risk a valuable tool like the clicker by trying to use it as a "reinforcer". My opinion, leave it be what it supposed to be, a marker. The "that is what I want signal".

That is why I am not so sure that clicks are truly "reinforcers" nor do I think "you" (generic you, no one specific) should try and use them as such.

Again, just my 2 cents and thoughts.
jacksdad
Posts: 4887
Joined: Thu Dec 10, 2009 10:48 pm

Re: SPARCS

Post by jacksdad »

WufWuf wrote:Sorry took me a few days to get around to finding something that might help clarify the process of addiction.
http://www.helpguide.org/harvard/addict ... _brain.htm
Life is full of things we crave, enjoy, have to do again etc. but not all lead to addiction. I think the bar for something crossing over from "simply" being HIGHLY enjoyable and creating a desire to repeat and addiction is a bit higher than you maybe are worrying it might be.

I got my sparcs membership so soon as they release the 2014 talks I will try and set sometime aside to as soon as possible to listen to what the speaker had to say about addiction and training.
emmabeth
Posts: 8894
Joined: Tue Oct 17, 2006 9:24 pm
Location: West Midlands
Contact:

Re: SPARCS

Post by emmabeth »

Jacksdad - yep, that sorts it out in my head I agree with you! Not that I'd do it but I bet his responses would change dramatically if the rate of clicks that are followed by no food increased from the accidental rate it currently is (around um 0.001% of the time, very very rarely!) - I wonder if I'd see some intial frustration which in HIM would result in him trying harder, but in other dogs of mine would result in them being very upset and switching off - but then a big drop in performance.
West Midlands based 1-2-1 Training & Behaviour Canine Consultant
CarolineLovesDogs

Re: SPARCS

Post by CarolineLovesDogs »

I have seen a few dogs who the click seems to be extremely reinforcing and much more so that the primary reinforcement- it becomes a strong secondary reinforcement so strong that is almost more reinforcing than the reward! Working with dogs that are extensively free shaped (very adept at it, learning extremely quickly) when you click and offer the food reward, they will often immediately jump to the next step ignoring the food. There whole desire is earning the next click and finding out what the next behavior is- and not that food reward (or other reward)
Or at least that has been my take on those dogs' motivations. :)
Post Reply