Thursday, 4 April 2013

Science Based Dog Training (With Feeling)...but without the science?

I was shocked the other day...

To find that a trainer I respect and admire in many ways, is doing something wrong.

I asked him about it and I was really disappointed that not only did he seem a little irritated (and that may just be my interpretation of the written word) but his explanation was really, off - it didn't actually cover the question I asked and he contradicted himself.

The title of this blog actually comes from a blog written by Dr. Ian Dunbar, a man I really admire (and he is a super lovely guy too!).. and I fully agree with this concept.

This is what he says, about training getting too technical:

"Dog training is in danger of losing its soul. Far too many trainers have adopted impersonal, quantum consequences (clicks, treats, jerks and shocks) in lieu of verbal feedback. Trainers have become technicians, which although beneficial for refining timing or learning how to set criteria, lacks feeling when teaching people to develop relationships with their dogs."

From: Dog Star Daily

I ought to point out, Ian isn't the trainer in question - its actually Zak George I am referring to, and I hope he won't mind me writing this blog and asking these questions, hes a very pro-internet minded guy and he is more than welcome to reply, by emailing me, (and I'll quote  that in full on here) or by video blog, however, I'm easy.

This is the video in question: (actually it isn't its another where he says the same thing, the one I wanted was the one from his training class but I didn't seem to find it using the blog search tool!)...

It's all great until you notice near the end he says you should reward around 80% of the times you click......

No. No no no no no no.

Now, I fully appreciate that what Zak wants is people to engage with their dogs, consider the relationship with their dogs, enjoy time with their dogs.

Much as Ian Dunbar said, dog owners are not technicians, they are not programming a computer, this should be fun and easy going..

I totally agree.

However if you are going to use a tool, use it correctly.

The click means 'yes!', if you teach that the click means 'yeah, maybe, a bit right, I kinda liked it but no cookie' sometimes, then that click no longer means 'yes'....

I am not a clicker purist - I really am not, I talk to my dogs when clicker training, I will encourage them, I will lure them sometimes, I make the task fun and interesting and we have a ball. I am most certainly not some robotic clicker-die-hard who is cold and uninteresting.

But I do use the clicker properly - click = treat. Always.

Zak IS correct that a variable schedule of reward is more effective than just giving rewards all the time, sure it is, I am not going to fly in the face of science and say it isn't.

But in applying that concept, WITH clicker training, it is the CLICK that you withhold and thus as well, the reward,  not JUST the reward.

That way you are saying to your dog 'yeah, that wasn't quite good enough; and if you WANT to, you can teach your dog a word that means 'yeah, but try a bit harder buddy'.. I use 'keep going' in a bright, interesting tone of voice which my dog understands to mean 'you are on the right track but I want more'...

I asked Zak, he generally responds on his new videos, pretty quickly, and here is that question and response:

Why have you chosen, in teaching clients to use a variable schedule of R.. to withhold the reward following the marker... instead of withholding the marker?
I find that this confuses people (and some dogs) and prefer to teach people that the click is ALWAYS followed by the treat, and phase out the click for behaviours that are reasonably well learned and where we are now working on improving the quality/intensity of the behaviour.
Well, I happen to think that it is important to teach "yes" and "no" to our dogs in the same way we would our children. While clicker training is part of my class, it is not the end all be all approach for me. I find it less confusing (as do the dogs, I believe) to be quite clear in our communication and to let them know where they went "wrong" as well as right. Good question.
· in reply to Emmabeth1980 "
So... sorry Zak but I don't think that really answers the question - I put it to you here, as I have on your video comments...

You say that teaching yes and no is important - I agree (though I avoid saying 'no' wherever possible because if I am saying 'no' a lot then clearly, I haven't put in enough work and the dog doesn't understand me as much as I thought or hoped)...

I think its important that we are clear in our communication, so I agree with you there too..

So why do you disagree with me, that clicking, (which we have conditioned our dogs to understand means 'yes' because it is followed by a reward), and then NOT rewarding, is not confusing or indeed 'lying' to our dogs?

I am writing this blog because I believe Zak to be a trainer open to ideas, continuously learning and up for discussion/exchange of views - I do hope I am not about to be proven wrong though!

I am fully in favour of science based dog training with feeling - but I think its important that we don't chuck out the 'science'!

So... Zak, from one trainer to another, are you open to discussion?

Edited to add:

In the interests of clarity and not leaving a thing 'unfinished', I paste the responses to the shared link on Facebook, here, as I suspect that link will be hard to find in a few days time.
Thanks to everyone who took the time to respond and discuss!

  • Emma Judson Do share this around, I'd like some constructive and interesting discussion to come of this
  • Zak George · Friends with Risë VanFleet and 2 others
    I feel where you're coming from Emma. Furthermore this is not the first variation I've introduced to status quo clicker training (or dog training in general). You are not the first to point this out. I fully acknowledge this is at odds with traditional clicker training. I advocate clicker training, not because there is anything special about clicker training in and of itself, but because it teaches the human to learn about impeccable timing when they have that button to press in their hand. This is a great tool, especially for beginners. The paradigm shift here is that I view clicker training as more valuable for teaching people about timing than I do teaching our dogs obedience (though it facilitates both of course).

    I encourage the random dispensing of treats after 2 to 3 weeks of solid 100% reinforcement. I suppose the thing that I've evaluated over the years is that by clicking and rewarding intermittently, the attention begins to shift from the clicker to the person, which as you know, is what I'm all about, relationship based teaching. If one ALWAYS rewards after the click, training will ALWAYS have some element of superficiality (my opinion of course). I would say that to a degree you are right, that indeed the clicker begins to mean less to the dog when randomly following up the click with a treat. But what I seem to have not communicated to you accurately is that by "weakening" the click, we are exchanging for strengthening and better fusing communication with the human and not the tool.

    For example, when our dog says, "Hey you clicked! Where's my treat?" We are there to say, "Keep paying attention to ME and (not the clicker) and you'll find yourself getting rewarded frequently". This is how I begin to transfer the emphasis of the artificial sound of a clicker to a training approach rooted in love, and genuine, UNFILTERED communication.

    I appreciate, as always, your participation in the discussion!

    -Zak George
  • Jo Law I agree with most of what you say there, Zak. The only thing I'd disagree with is those times where you don't reward (with food, at least) is where I don't click. I mostly use a clicker when I'm teaching a new behaviour. Timing wise there's no better method of communicating. Once a dog understands what to do, and when to do it, that's when I do away with the clicker. I'll always reward verbally, and sometimes I will add food to that, sometimes a toy, sometimes 10 seconds of telling him what a special dog he is in my squeakiest voice (rarely that one in public though) etc.

    It sounds like we do pretty much the same thing, only you still click and I don't.

    My dog was 8 when I got him, zero interest in training, very little trust of people, no experience of positive cooperation with people, and he viewed people as mostly the things that stopped him doing what he wanted.

    The clicker is so powerful because I now have a dog who says 'yay! Training!' when I reach for my clicker, who loves just being around me and doing things with and for me. It's great, and a large part of it was having that neutral, emotionless communication tool to help him get past bad experiences.

    I wouldn't ever want to click and not treat, because that click is SO powerful, and I don't want to affect that.
  • Deborah Campbell Burrows I think this muddies the water. I think the anti clicker brigade and newcomers already have a hard time understanding how it's supposed to work (simply click takes a photo of what you like = reward every time you click) that they are going to feel confused by this. It's the same as some camps saying click twice but feed once or press the clicker down to mean one thing and up for another/the next. It's really complicating what is a beautifully simple way of teaching an animal. We can and should mix in fun giggles, roughty toughty, happy voices as well - for sure.
  • Sally Bradbury At the recent WOOF!! conference Susan Friedman showed a video to illustrate why you never break your promise to your dog that click = reinforcement. It was of a gorilla in a zoo being clicker trained for husbandry behaviours. Some way in to the session the novice trainer made the mistake of clicking and not feeding and he started to become agitated and eventually lost his temper and threw himself at the bars and stropped off.
  • Emma Judson Now I stamped my foot (sorry it wasn't meant to come over quite like that but reading back it seems it has) and demanded an answer I have to apologise in advance for not replying in massive depth..

    In short, I agree with Jo and Deborah and Sally - particularly Jo, I know she trains in a really similar way to me - I think I can safely say that for both of us, we have experienced the clicker being the first thing that a rescue dog REALLY clearly grasped as a positive communication with a super crystal clear meaning, and as a result we are perhaps more keen than most not to risk damaging that.

    So as Jo says, we do do pretty much the same thing, moving the dog on from a constant reward to variable reward, we just do it by withholding the click as well as the treat rather than just the treat.

    Anyway I am glad we can discuss stuff, I think it is super important that we all can, and keep on doing so, so thankyou for responding everyone!


Sally Bradbury said...

Completely agree Emma. I am not a clicker purist either but if I am teaching with a clicker then the click is ALWAYS followed by a reward, even if I click by mistake.

When moving to a variable schedule of reinforcement then I start to withhold the click and click only the better, improved performance.

Clicking and not rewarding is the equivalent to giving someone a cheque that bounces.

Deborah said...

I've been taught by clicker trainers and clicker resources that every click must be followed by a reward, and as above and as per the comments so far, we can introduce variable reinforcement if we really want to by withholding a click - never withholding the reward. I imagine Zak realises there's been a boo boo but it would be clearer to express that, perhaps in a follow up video. I heard that the message about having fun is a poke at a well known online clicker trainer. Clicker training is mechanical and can appear cold, there's no getting away from that. Perhaps there is a point to be made about re-introducing more 'rough and tumble', kisses and happy voices as rewards, if the dog values them or even to break up the mechanics more.

Anonymous said...

Open your mind guys. Don't get blocked in by science's best understandings of the past - REAL scientists are always looking forward for new and better ideas. Clearly Zak's style of training works wonderfully and all the dogs he works with seem to really enjoy the process.

@Deborah - Clicker training absolutely does not HAVE TO be mechanical! It's just done mechanically by the majority of trainers. All you have to do is be more fun and genuine, but still use the click as a marker. Dogs understand much faster if you genuinely communicate with them instead of coldly tossing a treat or word at them. I think you will be pleasantly surprised if you give it a try!