Do dogs generalise badly?

Re: Do dogs generalise badly?

[quote author=Boogie link=topic=7888.msg111482#msg111482 date=1411237926]
It's like having ADHD

Us ADDers don't have problems focussing - we have problems focussing on too many things. Our senses work overtime - so that pencil on the floor has to be picked up before we can even begin to listen properly!
[/quote]

What a great way to think about it Mags. Joseph's (has HF Autism) ideal learning environment would be a class room with white walls, no posters or work hanging up and no windows, visual overload is a big problem. Anything where it shouldn't be causes total lack of focus/stress and has to be sorted out before he can move on and settle down. Very stressful for anyone with these conditions x

 

Re: Do dogs generalise badly?

[quote author=Merla link=topic=7888.msg111223#msg111223 date=1411160884]
Interesting. I think you're right that the problem is not actually 'dogs don't generalise' but rather, that dogs notice everything, visual, aural and, er, smelly, and don't prioritise the cues in the same way we humans do (we're obsessed with verbal communication!). So, in effect, the 'generalising' is about improving the 'signal/noise ratio' for the dog, so the poor love has some hope of working out what cue we floundering humans are trying to associate with what action!! Basically, it's a process of elimination for the dog : 'OK, that changed, but she still seems to want the same thing. Mmm, what is she doing that's staying the same?' Except, obviously without the internal monologue :

Most of the time, when I think my dog is doing stuff wrong or not 'getting it', she's actually way ahead of me.
[/quote]

Merla nailed it.

We can reason and visualise scenarios. We can use logic and the benefit of others' experience (what we're told or read) to rule things in or out. Dogs can't - they have to rely on their own observation. They have to run their own experiments, each and every dog, from scratch.

We have to make their experimentation easy. When training a new thing, get it really good and reliable before changing the context. Make that context change as small as possible. Teach from scratch again (will be faster than first time). Repeat.

Dogs differ in how readily they generalise. My first dog could make leaps. He tolerated sloppiness. He generalised. My current dog, Obi, is the opposite. He notices everything and demands precision. I have to be a much better trainer with him.

 

Re: Do dogs generalise badly?

So a different version of this conversation is happening on facebook - actually two different versions on two different groups.

One person is arguing that when we train we should make many small variations all the time. Change the way we stand, what we do with our hands, whether we have both feet on the floor, which way we are facing and so on. To reduce the chances the dog incorporates something we don't mean to be there in the cue. If we train like this, then changing location (or context) is less dramatic because there is a greater chance that the cue has already been isolated.

 

Re: Do dogs generalise badly?

Well there is probably no research on this (PhD anyone?) but I think it's essential to teach a basic response in a stable environment, before varying anything. Keep everything as constant as possible. Once you've got a good response (don't ask me to define that) then start varying contextual factors in a systematic way - ie one at a time.

In training we generally talk about the '3 D's':
- Distance
- Distraction
- Duration

Vary the exercise on one dimension only at a time.

It you think about the dog's job, it is really hard. All they know that is that occasionally food appears and that's nice so they'd like it to happen again. They ask the question "what made that food happen?". We have to make the answer as easy as possible.

 

Re: Do dogs generalise badly?

The discussion is centred around the 3Ds (or 4) being a bit of a "Red Herring" - and no variation is why there is then the need to generalise as extensively as we do. I think that's what is being argued anyway. It doesn't feel right cut and paste, but if people are on facebook, the discussion is in "clicker revolution".

 

Re: Do dogs generalise badly?

On the Lab Site FB page? I had trouble finding it (ie. I failed).

 

Re: Do dogs generalise badly?

No, it's group "clicker revolution" - I'd love you to look if at all possible. Here is a link, though it might not work:

https://www.facebook.com/groups/ClickerRevolution/

You might have to join the group though...

 

Re: Do dogs generalise badly?

The argument is also in the book I think. I might be able to summarise it, though would probably get it wrong. I'll have a go if you can't see the group.

 

Re: Do dogs generalise badly?

I have to join the group. Don't worry about summarising it (thank you for offering). I'm on my way to bed here (only 9:30, how pathetic) but will ask to join the group

 

Re: Do dogs generalise badly?

After you mentioning the group earlier, I applied to join out of interest and was accepted after a few hours. No credentials needed phew!

 

Re: Do dogs generalise badly?

So I joined the group. To be honest, I don't really understand what is being advocated there and how it is different. I will have a go at saying what I think they are saying... There are a few statements against the DDD concept (had to chuckle at the fact that I had emphasised the very thing that the FB thread would like to see gone from our vocabulary) and against the concept of proofing (which is what DDD is about). But to me they are still advocating proofing, just without saying 'proofing'. The difference is that they advocate beginning to proof, varying everything except the cue, right from the start of training a new behaviour. So it's kind of 'early onset proofing by varying every little thing'. The contributors seem to be stressing the importance of keeping the cue absolutely consistent (while varying all else) and I totally agree with the need for our cues to be 100% consistent, but that is a no-brainer and I cannot think of anyone who would disagree with that.

One person expressed the view that the use of the term 'distraction' is arrogant as it implies that everything else apart from us is an irrelevance - the idea that we should be the centre of the universe when training is arrogant. Actually, I DO think that the trainer needs to be the centre of the universe when training However I also recognise that we are not training for the dog's benefit (mostly) but for ours and we should be grateful that they are willing to give us their attention at all. We need to work for their attention, and make it worth their while to give it.

I do also agree that it is our job to help the dog understand what is the cue and what is the 'noise'. They have a hard task, working out what we want when all they know is that markers and food appear at certain times.

I don't think you can just train something at home and then expect it to work in all other environments. I also think we do need to start new exercises in environments that don't hold too many other enticements (I think the FB people agree with that bit).

Another view expressed on that thread is that we need to avoid simple catchphrases like DDD, and instead have people understand the principles behind learning and training. I agree that that is the ideal. But most people are hard pressed to grasp the most simple concepts in dog training.

Anyway, plenty of stuff to think about!

 

Re: Do dogs generalise badly?

Yes, this seems a good way of putting it:

[quote author=Oberon link=topic=7888.msg111870#msg111870 date=1411337424]
The difference is that they advocate beginning to proof, varying everything except the cue, right from the start of training a new behaviour. So it's kind of 'early onset proofing by varying every little thing'.
[/quote]

And yes, everyone agrees that dogs should be able to cope with different environments.

I'm just testing my understanding now - then I think what they are saying is that if, as you term it, early onset proofing is done, and the dog gets used to different environments (but without being asked to do anything, just gets used to them) then the cues work in the new environment. That is, we don't have to proof each cue in new environments? At the moment, when I read the standard material, it does suggest that a process of varying DDD applies to each and every cue that is trained.

 

Re: Do dogs generalise badly?

I have to say that most of this is now straight over my head. : I think my training is far more instinctive and less clinical than some of these theories are sounding.

 

Re: Do dogs generalise badly?

[quote author=heidrun link=topic=7888.msg111873#msg111873 date=1411339073]
I think my training is far more instinctive and less clinical than some of these theories are sounding.
[/quote]

I'd put money on that being the case! ;D ;D ;D And indeed every skilled trainer I've met.

Trouble is, for novices without any instincts at all (ie me - although hopefully they develop over time) that doesn't get me very far...

 

Re: Do dogs generalise badly?

Exactly what Rachael said.

I love behavioural science, but I think a lot of these kind of discussions are actually not very scientific. It is hard to explain, but sometimes I think we/people/us/me even, can get too wrapped up in how clever we can be. When really we should be looking at how we can best translate the essence of behavioural science for ordinary dog owners.

To me what is important is that dogs are trained as quickly as possible, as easily as possible, and as kindly as possible. By their owners. To this end, all dog owners need to understand the principles of learning and some basic behavioural terminology so that they can communicate with experienced trainers.

It has taken a decade to get people to understand they need to proof their training. And we are still not there yet. To my mind, this kind of arguing over semantics, just sets the whole process back a step and creates more confusion.

 

Re: Do dogs generalise badly?

Should probably add, in answer to the title of this thread - yes, dogs do generalise badly. We can dissect it into all the different causes and reasons, but the fact remains, dogs don't 'get it' if you change too many factors at once. Whatever the reason.

Maybe 'generalise' is not the best word, but it is a word that serves the purpose, and changing it now will create even more confusion

Dog owners need to know that dogs are poor generalisers, or they will get into difficulties.

 

Re: Do dogs generalise badly?

I think it's semantics too to a significant degree but I'm with Julie in that I need to think about all of this quite hard to make up for a lack of experience......

Mind you I would also say that time spent with experienced people discussing why they do or advise certain things is time extremely well spent....almost bridging the gap between experience/intuition and a set of instructions.

Where I work in IT we spend a lot of time bridging the gap between technology and business and I would liken that to bridging the gap between learning science and dog training as an art ;D

fun tho !

 

Re: Do dogs generalise badly?

See, I think training is all science. It's based on established principles. The principles behind it are quite straightforward. Yes, some skill is needed - mainly about good timing and observation, and some people are a bit better at that. But it's not an art. It's a science and it's done by systematically following known rules and steps.

 

Re: Do dogs generalise badly?

I also joined the facebook group to read this thread and I have to say I don't think I agree with their conclusions (but I won't say it on there as for a first post it will sound rather trollish!) There are a couple of things:
1. They seem to suggest varying the trainer's body-language and surroundings from the word go rather than starting in your garden and then working in other places. Surely this is making it harder for the dog? I was a primary teacher for many years and Vygotsky, a famous educationalist, talked about the 'zone of proximal development' from which other people developed the idea of 'scaffolding', in other words teaching something which is just a little harder than that already known and supporting the learner in the new learning until they can be independent. I would have thought a similar approach was the way to go with dogs.
I don't believe that the clicker trainers on the facebook group would really expect a puppy to do a sit/stay for the first time for 5 minutes in a busy park. I bet they do gradually lengthen the time they ask for this behaviour - the D for duration.
2. Sometimes dogs do understand the cue - they just choose not to obey it! And surely this is what proofing is all about, learning to obey the cue even if there is delicious rabbit poo or other dogs or whatever around.

 

Re: Do dogs generalise badly?

[quote author=Oberon link=topic=7888.msg111870#msg111870 date=1411337424]
I do also agree that it is our job to help the dog understand what is the cue and what is the 'noise'.
[/quote]

That's because we evolved from the apes, not the canines - so we 'chatter chatter chatter' where they notice and observe!

[quote author=bbrown link=topic=7888.msg111927#msg111927 date=1411385466]
fun tho !
[/quote]

And totally absorbing and fascinating!