The Surprising Science Of Motivation

[Aidan]

http://www.ted.com/talks/dan_pink_on_motivation.html

Not about dogs, but could be.

[corvus]

Thanks Aidan, that was really interesting. I listened to most of it twice.

I think it has applications in reward choice for the task we wish our dogs to perform. You don't always want your dog to fly in with maximum effort. They sometimes need to be able to realise that the rules have changed and work out what the new rules are, and it's hard to do that when you're focused obsessively on a massive reward. The rewards I use for clicker training are pretty ordinary and so small they are basically just a taste of food rather than food itself. I know I'm doing well when my dog stops caring about the odd dropped treat and cares more about getting another click. I use big rewards when I'm trying to build anticipation and enthusiasm for something my dog already knows.

[Aidan]

They sometimes need to be able to realise that the rules have changed and work out what the new rules are, and it's hard to do that when you're focused obsessively on a massive reward.

That's an interesting point. So do you think there is a cognitive process going on when that happens?

When I remember back to when my dogs were pups I can recall things like digging in the garden, and Django spontaneously coming up and digging beside me. I would point to where I wanted him to loosen the soil and he would dig there. Or when a salesman came around to measure up the house. I was helping him hold the tape measure in place. The next wall was short so the salesman didn't need help but Sabella jumped up and held the tape in place, standing against the wall.

I have dozens of stories like that as I'm sure others do also, but generally they become fewer and further between as the dog ages. I would say that generally my dogs will "offer" operants but they take far fewer initiatives now even though I did always encourage them as pups when they did things like hold tape measures or pull weeds.

I was speaking to a client who has spent most of her life, until recently, on sheep farms. She told me about some incredible things that her dogs have done on their own initiative, like fetching the fly-blown sheep when she went to get the stuff they put on fly-blown sheep, no command, no training, just watching and thinking.

And don't even get me started on cats...

[Bubitty]

They sometimes need to be able to realise that the rules have changed and work out what the new rules are, and it's hard to do that when you're focused obsessively on a massive reward.

That's an interesting point. So do you think there is a cognitive process going on when that happens?

When I remember back to when my dogs were pups I can recall things like digging in the garden, and Django spontaneously coming up and digging beside me. I would point to where I wanted him to loosen the soil and he would dig there.

Off topic but can you please tell me how you taught him that? I would love for my dog to be able to do that!

[Aidan]

When I remember back to when my dogs were pups I can recall things like digging in the garden, and Django spontaneously coming up and digging beside me. I would point to where I wanted him to loosen the soil and he would dig there.

Off topic but can you please tell me how you taught him that? I would love for my dog to be able to do that!

Well that's the very interesting thing about it - I didn't really teach him! I was digging (with my hands, for re-planting), he saw me doing it, he joined in, I got him started somewhere else and then all I had to do was point.

It falls under the categories of "learning by observation" and "mimicry", which is normally a phenomenon you would associate with animals of the same species but there are countless examples of dogs, cats, marine mammals and birds learning this way across species. I believe it is strong evidence of cognitive thought process.

I once had a cat observing me shape a dog to do something using clicker training. The dog did NOT figure out the thing I was trying to teach him, but the cat, after observing for a little while and seeing a chance to upstage the dog (presumably) just walked up and performed the completed behaviour! I should disclose that the cat had been previously clicker trained to perform a similar task, but still...

But back to your question - if you want to experiment I suggest you take him down to the beach (being a Goldie the presence of water will most likely remove any inhibition he might have), then get down on your hands and knees right in front of your dog (right in close), then start digging like a dog would dig. I've seen a few dogs taught to dig this way, it is one of the most easily replicated experiments on mimicry you can perform with a dog.

[corvus]

They sometimes need to be able to realise that the rules have changed and work out what the new rules are, and it's hard to do that when you're focused obsessively on a massive reward.

That's an interesting point. So do you think there is a cognitive process going on when that happens?

Well, I'm inclined to think so. Say I've been teaching Erik at home to roll over. When we go to the park, I might ask him to roll over. He's in a whole new place and I can practically see him sorting through his repertoire looking for the right behaviour to match the cue. He hesitates and sometimes he even goes to do something else and then corrects himself before he's finished. If I waved a tug toy in front of him and asked him to roll over, he would just down. That's a good way to get to attack a tug toy. He wouldn't even try to think about what I had asked him. But if I ask him to down during tug, he certainly knows what it means and will do it with gusto. I reckon it's because he doesn't need to think about that one.

Leslie McDevitt often talks about default behaviours as a way for dogs to "ask" for something they want. I've found with Erik that a default behaviour loses some of its usefulness after a while. Once it becomes the thing to do for any situation, his brain switches off and he does it around the highest of distractions without even really knowing it sometimes. I guess that was the point of the default behaviour, but to begin with I was using it sort of as a "slow down and think" cue. Once he no longer had to think about how to get what he wanted, he stopped slowing down and engaging his brain when I asked for a down. Now I've switched to sit as the "stop and think, please!" cue. He knows it well enough to do it when he's aroused, but not so well that he can do it without thinking about it. It's almost like when you talk to someone for whom English is a recent second language. It takes them a little longer to respond sometimes because they have to do some mental translation.

I don't know if that's the same thing as being creative or thinking outside the square, but I do think at the moment that there are times when my dogs think consciously through things and times when they just react. Erik is great fun to shape. He gets that a click for a lean one way probably means he'll get a click for a step in that direction, and another click for several steps in that direction. Kivi seems to me to frequently be unaware of what most of his body is doing when he gets a click. He gets tunnel vision and so it's a real challenge to teach him things sometimes. If he doesn't know why he's being clicked, he tends to throw himself on the ground and whine. I've seen the moment he realises how to win the game. His eyes light up, his ears come forward, his tail comes up and he starts going great guns.

I could teach Erik something new with tug as a reward if it were easy. Like, dive under my leg to get to the toy. I used to argue with people over food rewards with my last dog. I rarely used them because she just couldn't THINK around food.

When I remember back to when my dogs were pups I can recall things like digging in the garden, and Django spontaneously coming up and digging beside me. I would point to where I wanted him to loosen the soil and he would dig there. Or when a salesman came around to measure up the house. I was helping him hold the tape measure in place. The next wall was short so the salesman didn't need help but Sabella jumped up and held the tape in place, standing against the wall.

I have dozens of stories like that as I'm sure others do also, but generally they become fewer and further between as the dog ages. I would say that generally my dogs will "offer" operants but they take far fewer initiatives now even though I did always encourage them as pups when they did things like hold tape measures or pull weeds.

That's funny, as I've noticed the same thing. Erik was in the garden with me on the weekend literally pulling weeds up with his mouth while I was pulling weeds up beside him with my hands. Kivi used to do that when he was younger, but he's not into it anymore. Similarly, Kivi has become increasinly stressed about shaping as he has got older. We have switched entirely to targeting because it was just not that fun for him, despite the fact that he was quite decent at it as a pup and seemed to enjoy it.

Interesting...

[sas]

Just watched the video and read through your responses but I still don't get how this is relevant to dogs especially the ROWE theory?

[Aidan]

Just watched the video and read through your responses but I still don't get how this is relevant to dogs especially the ROWE theory?

Not everything on the video is relevant to dogs. Dogs don't work for Google, for e.g, but they do herd sheep, use the ATM, and retrieve pheasant in thick cover.

I suppose the fundamental relevance is that not everything that dogs do is what Dan Pink refers to as "mechanical", and that behavioural science has shown that "if/then" rewards are sometimes poor motivators for cognitive tasks that require you (or the dog) to think outside the box.

[corvus]

And dogs don't value work that is meaningful in the bigger picture. :D

But the experiments he spoke about at the beginning with the candle problem were, I think, relevant to training dogs. If they know they are working for a reward that is worth a lot to them, they might be more motivated, but not necessarily faster at learning. In fact, it may be that they learn slower than if there are no expectations of rewards.

I think this has implications with the way drive training may be used, as well. Drive training seeks to increase motivation and anticipation for high value rewards. It's not always a good state for a dog to be in to learn something new. There is a bit of stress involved in being in that state, and like Pink says, tunnel vision. My mum has a dog that has very little interest in food rewards and way too much interest in toy rewards. It took a long time to even set her up so she could be trained, because around food she was not interested and around balls she was so interested she'd get the tunnel vision and just could not think past "BALL. WANT." It was a real challenge to find the right motivator for her that was not too boring and not too arousing.

ETA I realise that the drive bit generally doesn't come in until the behaviour is learnt. That's the point I was making, and what I said earlier about teaching something new with teensy, barely there rewards and bringing in big rewards once its learnt to encourage enthusiasm. I have heard people say that it's very easy to train a dog that's nuts about balls, for example, but it's not. As above, you have to prepare them before they can think beyond "ball". As opposed to teaching things with teensy bites of food where with many dogs you are lucky enough to hit on just the right motivation level straight off the mark.

Edited February 27, 2010 by corvus

[Aidan]

And dogs don't value work that is meaningful in the bigger picture. :D

Hi Corvus, I'm not sure what you mean by this? Could you explain it for my benefit? Taking it at face value I can think of a few examples that would contradict this but I'm not sure that I understand what you are getting at.

Drive training seeks to increase motivation and anticipation for high value rewards. It's not always a good state for a dog to be in to learn something new. There is a bit of stress involved in being in that state, and like Pink says, tunnel vision.

Most of what we train dogs to do is mechanical stuff - sit here, jump over that, bring this back here, stay over there. For those sorts of tasks you really should be "taking the thumbtacks out of the box" to use the candle problem analogy. Make it easy, then the bigger the reward the faster you get there (as a rule).

But where I'd like to go is where the dog is being innovative or taking initiative. Beyond that boundary where operant conditioning fails. It's nothing new, we just don't know that much about it. Where we have working dogs who are bred to perform intelligent tasks that they perform better than a human could perform them we are reaching into that realm. If you look back into the origin of any dog you will find this.

In some fields, and it might surprise some people to hear me say this, operant conditioning CAN hinder performance. But what do we do instead?

[Kelpie-i]

Interesting thread Aidan. I haven't looked at the llink yet (too tired from dog training ) but I will. But going by the input in the thread it's seems like an interesting phenomenon (perhaps not the correct word to use). Dogs have allelomimetic tendencies so it's only natural that they would/could mimic certain behaviours without being trained 'mechanically'. I like the word INITIATIVE because it shows that the dog is thinking on his own and making a decision. How did the dog know to hold the fly-blown sheep without being taught or told to? Instinct, cognition...a combination of both??

I believe that many of the usual everyday routines, which may or may not be 'usual' for all households are products of some sort of cognitive function which may have come about because the initial offering of the behaviour was pleasing or useful to us, therefore we reinforced it.

Many years ago when I my poodle was alive (wow some 20 years ago now). I was getting ready to go out and was busily drying my hair with the hairdryer in the bathroom. The telephone in my room was ringing but I couldn't hear it. My poodle came running into the bathroom barking with a sort of urgency and simultaneously running to the door and back. I switched off the hairdryer to hear the phone ringing. I had never taught him this, so how did he know to come and let me know that the phone was ringing? And what would have been his motivation to do so? Some dogs bark AT the phone or just ignore it all together, so this was intriguing for me.

I will read the link.

[Vickie]

ETA I realise that the drive bit generally doesn't come in until the behaviour is learnt. That's the point I was making, and what I said earlier about teaching something new with teensy, barely there rewards and bringing in big rewards once its learnt to encourage enthusiasm. I have heard people say that it's very easy to train a dog that's nuts about balls, for example, but it's not. As above, you have to prepare them before they can think beyond "ball". As opposed to teaching things with teensy bites of food where with many dogs you are lucky enough to hit on just the right motivation level straight off the mark.

I have to disagree with this. Agility people used to think as you have written above, they used to teach a behaviour & then reward to make it fast. Now we just start fast. Dogs can & do learn in a high drive state. I have had so many people come with dogs who have moderate motivation for food. When they see a ball their eyes nearly pop out of their head. It always goes the same.

Me: why didn't you tell me your dog loves toys?

Them: oh, you can't train him with toys, he can't think.

Me: give me 5 minutes

Them: OMG, I'm going to train him with toys all the time.

I find the higher their drive for the toy, the easier they are to train & the quicker they learn the behaviour.

Same with sheepdogs. A keen working dog is on just about the highest drive there is. There IS no opportunity to say, hang on, tone it down & learn & then you can get back in this state. They learn in drive, they have to.

[Vickie]

It is so interesting to speculate on their thought process.

A couple of years ago, Trim & I put all the sheep away after training. Walking back from the paddock, about half way back, I realised she wasn't with me. When I looked back to the paddock the sheep were running around. I was mad as it was out of character for her. Before I got back to the paddock, she ran to me, then back & forth, very distressed, I think she even barked a few times, also out of character. LOL, she reminded me of Lassie, trying to tell me something.

When I got closer to the paddock I saw a terrier in amongst the sheep. By the time I turned to her to tell her to bring the sheep to me she was behind them & brought the sheep (along with terrier who was exhausted by this stage). We grabbed the terrier, found it's owner & went home.

I will never forget that day as long as I live. I can only assume she knew it was wrong & that we had to fix it.

Edited February 27, 2010 by Vickie

[corvus]

Aidan, Pink was saying that the conventional reward of, say, money for humans was not that great a motivator compared to feeling like they were working towards something bigger than themselves. I can't remember the exact words. I don't know any dogs that work for the greater good, but I'd love to hear about it if you do.

I've been a bit vague, coming at this from a number of angles simultaneously, although I will happily bow to Vickie and Aidan's superior knowledge if they still disagree with me.

Agility people used to think as you have written above, they used to teach a behaviour & then reward to make it fast. Now we just start fast.

That's a very refreshing comment! I have to say, since I got Erik I've realised that you certainly can teach a dog that is very aroused and build speed and enthusiasm into it from scratch. I generally train Erik in fast, high-powered training sessions where he's nice and excited and putting in heaps of effort. He doesn't just roll over when told. He dives into a roll and is back on his feet in an instant (as opposed to Kivi, who lolls around on the ground waving his feet in the air and then finally flops over and lays there on his side until he's sure you're not going to ask him any other down-related things while he's there). I was thinking that with Erik (and my mum's dog), you don't want them as aroused as they can be, at least until you've taught them to be able to think through that. It's not like you can walk out with an extremely high value reward, get them revved, and then expect them to even perform something they know all right in other contexts. That goes without saying, right? Erik's ability to think through his arousal has improved steadily as he's practised, and my mum's dog Jill was the same. But Jill especially took some preparing before she could think past the ball. Like encouraging things she was already doing and rewarding with a throw just to get her in the right state of mind to think about doing something for the ball.

I was also thinking that there have been times that I've deliberately tried to bring Erik down to a lower arousal state because it just doesn't suit what I'm trying to teach. Like lying still to get his nails done. As soon as you bring food rewards into the equation he's slapping his paw into my hand and wagging his tail madly when I want him to be calm for once. And I was also thinking of when I've taught Erik things that are a bit weird to him. Like walking backwards. It seems like a normal mechanical skill, but for him at least it wasn't obvious what he should do and it was hard to kickstart it. At least in the very early stages of something like that, I think it's better for him to be a bit calmer than usual. I find that he progresses in leaps and bounds if I start slow with clusters of clicks and then a break, and then ramp it up once he is onto the right sort of movements. Things I can kickstart easily that aren't very weird still get taught as high energy sorts of behaviours right from the start, but some things Erik just seems to need to problem solve.

Kivi doesn't think things through, which is why he gets target trained as much as possible these days. He has tunnel vision and I find it very difficult at times to kickstart new behaviours. If I could bring him down a bit, I would sometimes! But he's the low drive dog, so I guess I should be happy he's so focused and use it to my advantage with the targets.

[Aidan]

I like the word INITIATIVE because it shows that the dog is thinking on his own and making a decision. How did the dog know to hold the fly-blown sheep without being taught or told to? Instinct, cognition...a combination of both??

That's a good question and it's made me think about it a bit more. The owner told me she had asked the dog to get the fly-blown sheep a couple of times before, and after that whenever he saw that she had the stuff he would just go get the fly-blown sheep without being asked.

OK, there are antecedents there (the treatment, the flyblown sheep in the mob) but to put all that together so quickly and get it right, you have to wonder about the level of cognition? You also have to wonder, had the dog not found the work reinforcing, would he have put this together? So there definitely needs to be some instinct. I'm pretty sure I could train a gundog to fetch a sheep but it would be hard going and the dog might die of old age before we got to this level, why is that? Working gundogs display similar levels of intelligence in their field of work, they have similar levels of drive, they even do a similar thing (bring another animal back to the handler).

My poodle came running into the bathroom barking with a sort of urgency and simultaneously running to the door and back. I switched off the hairdryer to hear the phone ringing. I had never taught him this, so how did he know to come and let me know that the phone was ringing? And what would have been his motivation to do so?

Very intriguing, I'll add some thoughts when I reply to Vickie's post next.

[Aidan]

A couple of years ago, Trim & I put all the sheep away after training. Walking back from the paddock, about half way back, I realised she wasn't with me. When I looked back to the paddock the sheep were running around. I was mad as it was out of character for her. Before I got back to the paddock, she ran to me, then back & forth, very distressed, I think she even barked a few times, also out of character. LOL, she reminded me of Lassie, trying to tell me something.

I guess you could call that "intelligent disobedience"? Similar to what guide dogs are expected to do (although a bit of behaviour analysis does tend to remove the mystery of "intelligent disobedience" in most cases).

My retriever, Django, has shown several examples of this over his life. One stand-out was when my other dog, Sabella, was lost in the bush as a pup. It had been a couple of hours and it was only a small reserve on a hill so I decided to walk to the top of the hill to call her, thinking it would make me easier to find. We got a fair way up the hill and Django, completely out of character, just stopped and dropped. I tried to coax him along but he wouldn't budge. So I pulled. Then I tugged. Then I yanked. Then I dragged him. Then I yelled at him. Then I got down and grabbed his scruff to pull him to his feet.

Then I stopped. Django was not the sort of dog to stop, and certainly not the sort of dog to stubbornly dig his heels in. So I started walking back DOWN the hill, and he followed me. I turned and started walking back UP the hill, and he dropped to the ground. I got the message. After about 5 minutes of walking back down the hill Sabella came barreling out of the scrub looking very pleased to see us.

Certainly in that case there was significant pressure, "IF you don't come with me, THEN I will yank/yell/scruff you until you do" but he stuck to his guns despite the clear and present threat.

ETA: and yes, I did feel like an a__hole after that and I learned to trust him, in 9.5 years he has not let me down since.

Edited February 27, 2010 by Aidan

[Aidan]

Aidan, Pink was saying that the conventional reward of, say, money for humans was not that great a motivator compared to feeling like they were working towards something bigger than themselves. I can't remember the exact words. I don't know any dogs that work for the greater good, but I'd love to hear about it if you do.

Ahh, I get you now. Does my example above (intelligent disobedience example) seem like something done for the "greater good"? I think it does, hard to critically analyse something that is still (after 6 or 7 years) so close to my heart.

I've been a bit vague, coming at this from a number of angles simultaneously, although I will happily bow to Vickie and Aidan's superior knowledge if they still disagree with me.

I like it. I like it a lot. Keep it coming!

Agility people used to think as you have written above, they used to teach a behaviour & then reward to make it fast. Now we just start fast.

That's a very refreshing comment! I have to say, since I got Erik I've realised that you certainly can teach a dog that is very aroused and build speed and enthusiasm into it from scratch. I generally train Erik in fast, high-powered training sessions where he's nice and excited and putting in heaps of effort.

I find there are a lot of things that I just can't teach while the dog is highly aroused because the pathway to the end result is not clear in my mind, but the things I am good at I often prefer the dog to be aroused. A lot of it comes down to personal skill, how you work with a particular dog (e.g Kivi vs Erik), what the dog is good at etc I suspect Vickie has become very adept at working with dogs in this state, particularly with the sorts of things she teaches, through her experience. I am comfortable around highly aroused reactive dogs, teaching them to walk nicely on leash, come when called and that sort of thing. If I had to put a working BC through a set of weave poles I would struggle, it's just not what I am experienced with.

Kivi doesn't think things through, which is why he gets target trained as much as possible these days. He has tunnel vision and I find it very difficult at times to kickstart new behaviours. If I could bring him down a bit, I would sometimes! But he's the low drive dog, so I guess I should be happy he's so focused and use it to my advantage with the targets.

It would be interesting to see Kivi working a fit-ball or soccer ball. A task that is a good match for his genetic heritage, that he might find intrinsically rewarding. I've seen some remarkable problem solving ability watching dogs that are allowed free-play with a large ball.

[persephone]

With our sheepdogs- their reward/motivation is the chase- if they do not see it this way- they are of no use to us as workers. If they work well, and are responsive- they get rewarded by being allowed to do their work

I have trained a sheepdog who didn't enjoy the work- but would obey my every command. She was hard to actually work with, as I had to think for her and give her the direction.

i have never tried that again

[Are You Serious Jo]

I am fascinated by examples of intelligent disobedience and how the thought processes can work with domestic dogs to do this. As a cognitive researcher it is a challenge to try to fit these anecdotes into what we know about the way dogs think. Herding is a huge draw to me and one of the things that developed my love of research into how animals think. So keep up the stories as they help guide experimental design for the future and are great to read.

But in terms of learning in drive, some of the techniques individual herding dogs learn they have learned while in a high drive state, but I suspect that the range might be limited to behaviours that are linked to the activity they were selected for, e.g. working with stock, which is obvious really. I can see how operant conditioning would hinder a dog of lower intelligence when working, but a dog that somehow does recognise when it is essential to disobey would still do so, as seen by all the examples over time. I guess my take on it when training would be to teach the basics and allow the dog to explore the boundaries a bit more if you feel your dog does have the ability to think for itself.

[m-j]

I can't remember the exact words. I don't know any dogs that work for the greater good, but I'd love to hear about it if you do.

Australian Working Dog stories has quite a few anecdotes where dogs have done things where there is no direct reward in it for them but has been beneficial to something/one else.

In regards to motivators inducing tunnel vision in dogs, My thoughts are what you are teaching the dog to do and the "naturalness" and / or complexity of the behaviour being taught could have an impact on this along with the dogs enthusiasm for the motivator. An example is I was teaching a food obsessed Lab I had, to do a foot target through pulsing, touching the foot naming it and then after many reps holding my hand above her foot waiting for her to touch my hand. After many unsuccessful trials I had to remove the food and teach it with pats and praise, I don't think she even realised that I was touching her foot with the food present, everything else was incidental the only thing she saw was the food. Maybe she thought that I was just conditioning her to another secondary reinforcer

cheers

M-J

edited for spelling

Edited February 28, 2010 by m-j