Public Dog (31, 32, 33) & Nosework

Gimme continues to do well in this class.  We have one more Public Dog class and then are off for the rest of the winter.  I'm noticing that Gimme's attention is wandering and she's a bit more emotional and needy lately.  Its too soon to chalk it up to a false pregnancy; only four weeks from her hot date.  So I'm wondering if she is simply reacting to my stress getting ready for the big trip.  Then again everyone has an off day, even brilliant Gimme.

She wasn't totally off, just took her longer than usual to get focused.  It probably didn't help that Tor had a run-amuck moment and went blazing past us (about ten feet away) a couple times.  She likes him and has walked with him... so it shouldn't have been too stressful for her.  There's also a new dog in class that has a growly way of "talking" and that may be uncomfortable to her.  His owner uses a fair amount of corrections (though a lot less than when I first met him and encouraged him to call Ursula).  He's not that reliable yet, so we made sure to set up away from him.

We started working in the tennis court with our dropped leashes and just did parallel (lines 15 feet apart) walking 20 feet, doing a behavior and then returning to our station for treats.  Gimme did okay at this, she just lacked really solid focus or her usual precision.  After we went out on the grass and I was further away from the others (in particular the instructors), when she did a really good bit of LLW, I rewarded her with PB.  After that I had a lot more focus and she did some really lovely things.  And, I didn't use the PB all the time, just for the best efforts.

That leads to something I've been planning to talk about.  Ursula thinks that sometimes peanut butter raises Gimme's arousal level.  We've had this talk before and basically Elizabeth's answer was to never use it.  Obviously that's not going to happen.  At the time I sewed some loops of elastic in my treat pouch so I could have a peanut butter filled film canister, to dip my finger in and give Gimme stealth PB.  I did that for several weeks and never had any problems with Gimme becoming too aroused.

Recently this came up again and I was very frustrated, because it was sounding like Gimme could never have peanut butter and this was coming from Ursula, who's opinion I have a lot more respect for.  I had a long talk with Ursula after class and we realized we weren't communicating well.  Ursula doesn't think I should let her lick it off the Go Toob.  She thinks that Gimme gets a more aroused that way because she doesn't know when its going to end.  Which explains why there was no issue when we were doing stealth PB, because it was a blob off my finger and that's it, plus no licking.

Ursula also thinks I shouldn't give it to her when she is already aroused or overly excited.  I mentioned teaching Gimme to behave better about getting treats (no jumping up, no putting her feet up on things to get closer to the treats, etc).  Ursula doesn't think that will help, that arousal is the issue, not behavior.  After our talk, I thought it through and decided to continue with that plan anyway.  My thinking is that once she is more clear (requiring consistency from me) about how to behave to get the treats she's earned... if she can't keep four on the floor, then she's too excited/aroused for PB and gets her regular treats.

So I've been doing two things.

1. I've been working on stimulus control with her target stick.  This is an ongoing project that I work on for awhile and then get bored with it and slack off.  Now what I'm doing is using that to teach Gimme how to win PB.  I stand next to the table where there is both the Go Toob and a dish of regular treats.  When Gimme does her best work I reach for the PB.  But, if her feet come off the ground - the PB stays put and she gets other treats.  This has proven very challenging for her.  She's working at it, but its still really hard and she's still sorting out the rules for how to make the PB come to Gimme's lips.
2. I'm also using peanut butter more in training, not less.  I'm using it as much as I can for differential style reinforcement.  When she puts forth a superior effort, I want to give her a super reward.  I notice once I get a really good effort that I reward with PB, the quality of all the efforts that follows goes up quite a bit.

Time will tell how these two approaches work.

Being Random

One bugaboo I have in training is an inability to be random and since she's a genius, Gimme picks up patterns very quickly.  That means she starts guessing what I'm going to say based on the pattern she intuits, which interferes with her development of listening skills and true stimulus control.

So I went online and located a Random Number Generator.  I'm going to try using it when I want to train a set of skills that are similar and require Gimme to focus and differentiate between one cue and another. 

On this RNG, you can specify the number of integers you want, such as between 1 and 3 and how much output you want - quantity 25.  Push a button and within moments you will get a string that is random. 

Now for my purposes I would want a roughly equal representation.  So I cut and paste the string of numbers into Word and replace each number with a cue, such as 1=high, 2=five, and 3=pretty.  Then as I use the replace, it will tell you how many replacements were made and from that you can gauge if its distributed in a roughly equal way. 

I did it twice.  The first time wasn't even close, but the second time was 9-9-7... which was just fine. 

K9 Genius Strikes Again

Gimme was insisting that if I wasn't going to play nosework with her, then she needed more training... she does love it.  So we dabbled with some behaviors.  Then I decided to see what it would take to change the cue for her left and right high-fives.

I tried the traditional method - new cue, old cue, behavior, then reward...  After a dozen attempts, there was no sign that Gimme was getting it.  I tried another half dozen times with the treat in the new cue hand, to draw her attention to it - still nothing.

So I decided to just reteach the behavior, using the new cue hand.  She offered the behavior after just seeing the new cue, on the third try!!!  We did it a dozen times and by the end she was offering it 80% of the time on just the cue. 

Too damn smart!

Follow up -  The next time we worked on this... it seemed really confusing to her, so I've decided to leave the behaviors' cues the way they were.  Just means I'll have to work harder on getting them to a reliable verbal -- which is a good project to work on, especially given our difficulties with stimulus control.

Class Update

I'm a bit behind reporting on classes.  They continue to go well and Gimme is learning a lot and doing very well on her special challenges.  Last Thursday was an especially great class.

We were working in our cubicle and she could see other dogs through the opening, though basically ignoring them.  A couple arrived late and when they went by, they let their dog stick its head in our cubicle (not supposed to happen).  Instead of behaving badly, Gimme's eyes got big and then she turned to me and demanded treats.  I was so proud.  She was challenged, made a good decision, knew it was a good decision and then demanded recognition - gotta love that. 

Then later in the same class, one of the students was working on the stairs.  Our cubicle was at the bottom of one side of the stairs and I was working Gimme on something that was hard, when the lady comes down the first flight, onto the landing and before realizing we were there, started down the flight that would have brought her right into our cubicle!  She was so focused on training her dog that she got turned around and started down the wrong side on the stairs.  Again Gimme got big eyes, totally forgot what we were working on, but made a good decision, turned to me and demanded the payoff.

And even better - after these two episodes, she remained calm and went right back to working.  There was a time not long ago when she would have spent the rest of class on high alert.  This is evidence of really good progress since both dogs were within 5 feet and intruding on our space.

The thing we were working on was related to listening skills, again.  Ursula had me pick six cues that Gimme knows well, that she could do in class and in a small space.  I picked: kennel, chill, sit, down, touch, paw and melon.  Touch is a nose touch to my hand and paw is shake hands.  Chill is to down and put your head down, wherever she was.  Melon is to push her cube with her nose.

Then Ursula strung them together in different logical sequences.  The idea was to cue the first one and the instant Gimme did it cue the next, and the instant she did that cue the next, and so on to the end.  The last behavior got clicked and treated.  If she made a mistake, we broke it off and then tried it again, starting at the beginning.  We did each sequence about five times (correctly) before moving on to the next.

The sequences we did were:

• chill-touch-down-melon-sit-paw
• kennel-chill-touch-down-melon-sit-paw
• sit-kennel-chill-melon-down-touch-paw
• paw-down-melon-kennel-chill-touch-sit
• down-touch-melon-sit-kennel-chill-paw

Even though it was hard mental work to focus like that and especially really listening part, Gimme seemed to enjoy it.  I times I fully expected to see steam come out of her ears.  This girl really loves a challenge.

Learning Styles

This week our class schedule is all turned around, because Ursula is leaving town (for a herding trial). So instead of nosework class tonight, we'll have it tomorrow morning. At first I was really worried about what affect the change to our routine might have on our ORT Friday evening. Now that I've thought about it, I think it will work out well. After nosework class tomorrow morning, we'll go to the fort for a walk. So Gimme will then have a nice relaxed evening after nosework and a walk. And since what we are working on in Ursula's class is proving to be a bit challenging (and frustrating), I think its better that we did that tonight, rather than tomorrow night.

As I've talked about so many times, Gimme has a real challenge with stimulus control. I recently learned of a model for learning styles that I think applies (been discussing it with a canine behavior researcher in Australia). One thing I really like about this model is that it talks about both the pros and cons for each strategy. "Briefly" explained (blue-green are my comments) it goes thus: 

-------------------------------------------------------

There are two different learning strategies. The model describes them as either/or; I think its more likely that they are on a continuum. I also think dogs may change strategy or slide up and down the continuum due to various factors. This model surely applies to humans and other critters, it just came from research on dogs. Also, though its described as a "learning strategy" in general, it seems to have the most impact on the stimulus control aspect of learning - which is also what piques my interest. The two strategies are:

• the stimulus-response strategy (doers)
• the cognitive strategy (thinkers)

In SR strategy (doers) the dog responds quickly and habitually to a stimulus. These dogs are more focused on consequences and less likely to think through what to do before taking action. They hear a cue and immediately respond without necessarily processing what they heard, often doing the last thing that got favorable consequences. They are behaving by rote in response to a stimulus, any stimulus, and then check to see what result they got. Do keep in mind this describes a learning strategy, so we are talking about the pattern of how the dog learns - it doesn't mean the dog is always that way. IE this isn't etched in stone and yes, it can be modified with training.  Also, just because the doers are not called thinkers, has no reflection on how smart they are - this only applies to how they learn a certain aspect of training. 

In cognitive strategy (thinkers), knowledge of the correct response is linked to a general rule rather than a particular behavior. These dogs might be slower to respond, but are generally more accurate early on. They think about the stimulus and pair it to a known behavior to get a known result.  While the thinkers start out with an advantage... as you will see below, it doesn't seem to last. The doers catch up and in some ways may surpass the thinkers. 

Both groups of dogs end up with an understanding of what different cues mean and the same level of accuracy. The doers eventually pick up the discrimination through association and then become much more thoughtful about their responses. The thinkers are thoughtful right from the start. Thinkers are a little bit faster to learn the stimulus control aspect of the task, but doers are usually super motivated. Thinkers with lower motivation sometimes get bored or distracted and take longer as a result. A focused doer that is not too aroused may pick it up in about the same time, but the tendency is to get over-aroused and then easily frustrated and make more mistakes. And I think vice versa - make more mistakes and get frustrated.

My researcher friend does say that of the dogs in her research project, all but one of the doers tested as overall optimistic. That one doer exception was just neutral, not pessimistic and did turn out to be moderately dysplastic. Every pessimistic dog has been a thinker. She tried to explain to me how the evaluation for optimistic vs. pessimistic is done, but I confess I couldn't get it.

I had an idea that this might be one time when traditional training might provide a "benefit"... When traditional training was in vogue, I think the doers as a result of punishment, had an overall suppression to their behavior. Then being risk averse would cause them to slow down and become more thoughtful - acting more like thinkers. Please do not construe this as any inclination on my part to start using traditional training methods - that stuff is no longer in my training bag, nor do I intend make room for it. Its just a thought I had and my friend thinks it is plausible.

-------------------------------------------------------

So where this is most interesting to me is that Gimme has really strong doer tendencies. As this model describes, she tends to hear a cue (stimulus) and throw out a behavior, without necessarily processing what she heard.  We've put a lot of effort since last year into what Ursula calls "listening skills".  Ursula noticed the improvement right away when we started classes this time around.  Gimme is  also very highly motivated and attends more to the consequence than the stimulus itself. Because of that, she makes more mistakes, her rate of reinforcement falls off and then she gets frustrated.

Because this is our biggest training challenge, we've been focusing on it, using the learning of "touch" and "paw" to emphasize it.  One of the elements of stimulus control is that the dog waits to hear the cue before acting and that is by itself hard for Gimme.  In the case of targeting, the prop is there and Gimme tends to see-prop-do-behavior.  Since this way of targeting that I'm learning can use the same prop for different behaviors, its a handy way to make it clearer to Gimme to wait to hear the cue.  Of course with her doer tendency, when she hears the cue - she just leaps into doing without actually paying attention to what the cue meant.  So overall, this is very hard stuff for her.  When she gets frustrated, it definitely impacts her abilities.

Please understand this isn't a case of her not understanding the cues "touch" and "paw"... she's got those down cold.  Remember what I described in my blog the other night?  Gimme was switching back and forth between them with great accuracy, as well as "feet" and "bring" with a bit less accuracy (because they are newer behaviors).  However, at that time we weren't doing any waiting for the cue, so she wasn't dealing with frustration.  And just for the record, there have been times when I've noticed this same issue with "sit" and "down"; I am 500% sure she knows the difference between those two.

In class we have been doing short bursts of training with down time between them, but its still been hard for my beautiful girl.  Tonight I flashed back to how worried Michael would get at our early agility trials and how I'd play the "get it" game to keep him busy and having fun (and thus, not worrying about what was going on around us).  The get it game is really just tossing small bits of treats one at a time in any direction, with the cue "get it".  Real easy for the dog to win and burns off a lot of energy (and frustration), while keeping them busy and entertained.

This worked really well for Gimme tonight, so by the time we did the next training burst, she was really refreshed and did much better.  The get it game worked so well, that I decided when I clicked/treated during the session to toss the treat and let her pounce on it.  That also worked and kept the difficulty of the task from building up frustration.  So our last little session was the best of the evening.

Gimme LOVES training and learning and wants to do it all the time.  She thinks training is loads of fun and quite possibly the best thing she ever taught me how to do.   She is not convinced that we couldn't do this any time, all the time, whenever she is awake.  Some tasks are harder for her - specifically some concepts - and, of course, that is frustrating.  Still she is making good progress.  I believe when she finally grasps the concepts of stimulus control, training is going to take another quantum leap in the fun department.

I love that no matter how hard it is and no matter how frustrating it sometimes is... Gimme never gives up and always keeps trying.  She is certainly the most determined dog I've ever worked with.  I really love her attitude.   And, as I've said at least quarter-million times, she is definitely the smartest dog I've ever worked with.  Not to mention (except I will) that she is also the most beautiful, most loving and most wonderful dog I've ever known...

Listening Skills

One of the issues I've had with Gimme is that she has poor listening skills.  She tends to make assumptions about what I am asking for, without actually listening to the cue I used.  She loves offering stuff and so would rather run through her extensive repertoire of skills until I pay for one.  Naturally this isn't particularly useful.

And yes, I am certain that she actually knows the cues I'm talking about.  She is learning 53 cues with more to follow, and just about half of them are strong enough that they should be on cue.  We've worked the drill to put them on cue and she has demonstrated before that she understands how cue attachment and cues work. 

From what I'm seeing, I think she just prefers to do things her way.  There's a reason I bought her a little train conductor's hat and its not because she ever intends to let me wear it...

So tonight I decided to try something new.  Leslie McDevitt had given me the idea to use the Gimme A Break game as a reset for Gimme's frustration bark-a-thons and that has continued to be successful.  There are times she will quit barking, and default sit in front of me "talking" to me in the funniest way (sounds like a cross between a pigeon cooing and someone moaning).  I use judgment about her attitude when she talks - is it just talking (I "get" being verbal, being a yakety-yak type myself) or is it frustration?  If its frustration leaking out, then we wait for actual silence, otherwise I ignore it.

Anyway, the thing I did different tonight was to cue a behavior and if she gave me anything other than the behavior I cued, then I go to sit down and wait for a reset (Gimme comes and offers me a default sit to ask to work again).  My theory being that if she isn't listening, then she must need a reset.  It took a bit for it to sink in, but then she was really trying. 

There were a few times I'd give her a cue and she would start to do something else, then catch herself and give me the right behavior.  I rewarded that copiously... since she was definitely trying to do what I wanted and clearly showing that she understood.

By the time I ended the session she was doing very well.  I was even able to alternate between "finish" and "behind".  Both behaviors start in front position and are very similar - only differing in direction.  For finish she passes on my right, ending sitting in heel position; behind, passes on my left, ending sitting in side position (my right side).  In the past, no matter which I said she would just repeat what she'd done last.  So being able to alternate and/or mix it up between two such similar behaviors is a huge step forward in understanding. 

If we can get Gimme's listening skills to match her thinking skills, there will be no stopping this kid.  Naturally I'm thrilled with this step in her understanding.  She doesn't know it yet, but ceding a tiny bit of control to me will make training a whole lot more fun. 

My Ahah Moment

Tonight I suddenly figured out something that had escaped me before.  I've been wondering why it is that Gimme is so good about stimulus control on some things and so awful on others.  There are things that I've put a cue on that she never questions and sticks with me and then there are the others...

It occurred to me tonight that the behaviors which Gimme has difficulty ceding control are all behaviors we created through free-shaping.  Every last one of them.  I suspect that in her mind, part of the picture of that behavior is the offering of it.  I'm definitely going to talk to Ursula about that.  I've been diligently working on the things she and I discussed in that private, but we are only making a smidgen of progress. 

Any behavior we created through luring, targeting or guides - those she generally is good with.  Though if she gets frustrated during a training session, all listening goes out the window.  Boy I'd sure hate to have to give up free-shaping.  Gimme is so good at it and so creative.  She's also mighty creative in coming up with clever embellishments on any behavior I'm teaching through other means as well.

We've been working on backing up using the guides.  She figured it out so quickly.  Now though she's in embellishment mode.  Who knew there were so many ways to move backward.  I swear she just sprinkles a little pixie dust around and that's why she is so bewitching.

Another project we have going is teaching her to close the door - after all, she is the one that keeps leaving it open.  We are on day two and she already has a strong idea of what to do.  What a character...

Nosework (5/3)

Class tonight was very interesting. I realize now that Joyce doesn't quite "get" Gimme. She talks about her reactions as if Gimme is a sensitive little wilting flower, instead of a tough little tomboy who is also the Empress of the Cosmos. She is sensitive, but to frustration, not other stuff. She doesn't get her feelings hurt, she gets annoyed (read that royally p.o.'d). We talked a lot as a class about the ORT.

Joyce was much more focused on handling and teaching our dogs to commit to odor and stay there until we get there and pay up. She hasn't pushed that in the past.  I wish we had worked on it sooner - especially before the ORT. She did say that her students that failed all had fast dogs and/or were testing two odors. So tonight we did a lot more moving around the dogs and were not quite as fast to pay up, letting them wait a second first.

So while we were working things, I kind of did what made sense to me based somewhat on Joyce's comments mixed with my own understanding of Gimme.

For instance, Joyce commented that when dogs hit the end of the line it feels like a correction, so we should be going with them. Of course, even with my longer than usual line (25ft), Gimme can hit the end in a nanosecond. Joyce is always saying I have too much line; yet, I find it very useful because I can play it out to let Gimme check out something briefly and then reel it in as she turns back to me. I looked at the video again and each time when she snapped at the line was just after she'd hit the end. Joyce thinks the unintended "correction" hurts her feelings and undermines confidence... my take is that it annoys Gimme and is frustrating to her. Snapping at the line is her way of telling me to get with the program.

The other issue at the ORT is that Gimme wanted to check out the room and I stubbornly stayed near the boxes and wouldn't let her go. Joyce still thinks I should keep her near the containers and not let her snoop, and kept telling me to move her back tonight. Given how easily frustrated she gets, I've decided to use that long line to give her more freedom, and then gently keep reeling her back to the boxes.

So tonight I really focused on my line handling. I let her have more freedom to snoop around the perimeter even though we were set up for a container search and when she'd turn back in my direction, I gently reeled her back. I also made it a point to keep tension on the line as much as possible, so if she did try to squirt off in some direction, she was getting the sensation of a steady pull as the line slid through my hand, not a jerk when she hit the limit. My impression was that Gimme seemed more content with this way of doing things.

When we had the threshold hide, I let her go a bit further past it and then gently used the line to encourage her back toward it. She did a really nice job.

I also talked to Joyce about how much freedom I have at an ORT or a nosework test to do management things once we enter the search area and before the search starts. Turns out I have more latitude than I'm used to from other sports. As long as I don't hold up things, I can use the time walking to the start line to play "whazzat" with her and I can be rewarding her as I go. So that would help to prevent a buildup of frustration and perhaps lessen her need to snoop once the search starts.

We can hold them behind the start line for up to ten seconds. Before I've just been holding her for a good long pause and then letting her go. Tonight I counted to myself "one-banana, two-banana, three-banana..." I couldn't stand it and sent her to search after six-banana. Ten seconds is really a very long time.

After class I explained to Joyce what I was doing and why I might have seemed to not being paying attention to some of her direction during class. She understood and said it was worth a try. In an ORT I could certainly let Gimme snoop, realizing that I'm on the clock and eating up time. Which, fortunately isn't an issue with Miss Gimme, the Nose That Knows...

A funny from class - Gimme's nosework search cue is "wherezit". So tonight after she found one of her hides and Joyce had set another, I gave her the cue and one of the girls in class started telling me where it is. We've been in class together all along, but this time I happened to be pretty close to her - I guess she'd never noticed our cue before. We all got a chuckle.

We came home and did a nice session on some skills. I am still seeing that Gimme just doesn't want to give up control of behaviors. Once they are on cue, she still wants to get paid for offering them and finds it very frustrating that I'm so obviously failing to live up to her understanding of how the world works.

It must be hard to be a genius surrounded by such gross ineptitude.

Busy Girl

I've come to the realization that I'm just going to have to keep a separate calendar for Gimme's schedule.  I keep running into conflicts between what I have planned and what I want to do.  I feel like a soccer mom who is desperately trying to keep up.  All Gimme wants to know is whether her new purple leash is in the car and do I have enough cheese.

We've been keeping busy lately with a little bit of plinking here and there on different training things.  A lot of focus is going into nosework... which is how I discovered the calendar double booking.  The trial I planned to go to in April is on the same weekend as the Sdao/Pouliot seminar we are signed up for.  So now am looking at a couple others - March or June, both in Oregon.

We are playing with the Project behaviors again - which we haven't done since October.  I have been working off and on with the target stick "wait for the cue" and its coming along.   I still haven't fixed the "apple" box, so that will have to wait. 

Grape - dead bug (on back, feet in the air)

I wanted to do this since I now have interlocking floor matting to cushion Gimme from the hard floor (and provide better traction for other things). Since we haven’t trained this for three months and since it wasn’t on cue when we left off, it took awhile to get Gimme back to offering it. Even though she has been doing the Grape in my lap when she wants attention (which started 4 days after we let Michael go). She offered me a lot of "pretty" instead (we’ve been free shaping this), got frustrated and barked some. I kept clicking a "down" and when she was solid on that, I started clicking for movement while she was down. Finally in frustration she did a movement that was almost a grape and she got two treats. Next thing I knew she was throwing a LOT of  grape behaviors at me. I started putting it on cue and will continue in the next session. She sure is cute.

Melon - push cube with nose

Again a project behavior we haven’t done in three months. Gimme knew what to do the moment she saw her cube. There was a short bit of skiing on the cube and batting it, but she settled down to using her nose very quickly.  I started putting it on cue and following our earlier Grape session, she seemed to understand what I meant. She was even pushing it with her paw or using her mouth to get it out of corners and away from the walls so she could then push it with her nose AFTER I’d cued "melon". Smart cookie.

My friend loves to alter pictures and such and add sayings and stuff... so I usually get things from her that have pictures of Gimme that she's taken before.  Just got this one...

Today's Training

Since our schedule has been off for the last couple of days, I decided to do a couple short sessions today, even though Friday is normally our day off.

I started with "around" using the guides.  As soon as I set them up Gimme jumped in (edited out).  I was sure she'd be eager to get in there when she figured out that was where the rewards happened.  I opened the guides and had her go out and re-enter, wanting to reinforce how she should enter.  This session was really short - only 7 click/treats.  Do note two things... first, she never left until I indicated the session was over (unlike prior sessions) and as soon as we were both in the circle she started moving around me.  I think the idea of the guides is becoming clear to her.

This next session is a bit of target stick training related to waiting for the cue.  The idea is to present the target stick and give her the cue.  Then toss the treat and wait before giving the cue.  If she comes back and touches right away, I do nothing and just wait her out.  The moment she pulls away, I give the cue and reward her for doing the behavior with a click/treat.  This is something I think I am slow to do and the video confirms my impression.  If I give the cue and she doesn't do the behavior in a reasonable time (for now I count five) then I am lift the stick out of range to indicate a loss of opportunity.  I also lift the stick out of range when she bites on it, which still seems to happen a lot, though it is diminishing.  At some point I can also start clicking her for not touching the stick when she begins hesitating more often.  I think her touch-touch-touch and then sitting is an indication that she is starting to figure out the game.  There are clearly some missed opportunities during this session; Gimme will just have to figure it out in spite of me.

Then we did a bit of noodling about stays and heeling around food on the floor.  I haven't done it for quite awhile and obviously we need to go back to it.  It wasn't training so much as testing. 

Because life isn't exciting enough around here... now that Gimme is coming out of season, Meaggi has decided to give it a go, for the first time in several years.

Copier Pictures

I've already sent off the 4 best pictures from the 9/11 photo shoot.  I had to tidy the office before taking pictures, so it looks better than any time in the last five years.  Of course now I'm sneezing my head off from stirring up all that dust. 

Gimme was very good about getting into position and sitting there waiting through all my picture taking nonsense.  She really does have an excellent sit-wait.  In the first picture, Gimme is patiently waiting after I pushed the button for the copier to take the picture of her cute little tushy...

In the second picture, she is clearly impatient to see the final results and is actually watching the paper come out of the copier.

Afterward I spread her pictures out on the floor and let her look at them.  She spent quite awhile examining them, before tearing a couple to shreds.  I should have gotten pictures of her doing that.

There's another picture I would have loved to get, but needed someone to take the picture... I didn't want to risk Gimme breaking her stay and jumping off to get to me since its a long way down.  I may get that picture some day, simply because Gimme is so photogenic and it would be a cool shot.  But obviously not until I have another set of hands to take the picture so I can make sure she stays safe.

I have a great Halloween costume idea for her, but was too busy with other things to get it done.  Maybe next year.

We've only done a little training in the last couple days since I was fighting a cold and needed to work some extra rest into my days.  The second session with the target stick went really well, so I attached the cue.  Next time I'll start working on the exercise that I learned from Ursula.  I'll try to videotape it.

We have a working slot in a freestyle seminar next weekend.  I have been so focused on the fruit project that I wasn't even thinking about it and now here it is almost upon us and I really haven't taught her any of the behaviors I wanted to have ready. 

So, I'm also using free shaping to teach Gimme to back away from me.  The hardest part seems to be getting her to understand that I want her to go straight away from me.  On the hardwood floor, I lined up with the strips and started by clicking any time she kept one foot between the strips, then two and finally three.  Still, I'm not sure she is getting the idea and it was kinda frustrating for her.  So I'm going to lay down two strips of duct tape to form a corridor for her.  Then as she gets the idea I can cut away at the edges making the tape narrower and narrower to fade it. 

Private Lesson & Target Stick

I wanted to do one of the exercises we did in our private lesson, but needed to attach the cue to the target stick behavior first.  Silly me, I thought Gimme would remember how to do her target stick behavior, even if we haven't done it for at least six months.  Whereas Gimme thought it was really nice of me to bring a stick in for her to chew on. 

So I quickly dropped the idea of attaching a cue to the biting-the-stick behavior.  I basically had to free-shape it from the start.  By the time we finished she was doing really nice nose touches on the end.  So if she does that again next time from the start, I will attach a cue to it.

PRIVATE LESSON...

We started with some discussions, based on some questions I had been thinking about. 

Q1a.  Are we complicating the wait-for-the-cue process by using her two default behaviors (sit and down) in the mixed sets?   When we did so much work on her relaxation protocol in classes, the down turned into a second default.

• Maybe we shouldn't use them in the mixed sets for wait for the cue training.  Its possible we'd get better stimulus control, but lose them as default behaviors.
• However, just because they are defaults, doesn't mean we shouldn't expect some level of stimulus control, we just may not get all four elements.
• If I cue sit or down, she only gets click/treat/reward if she does what I cued.
• If she offers either default I don't have to click/treat it, but could do so randomly.  I can also reward in other ways.
• If she offers a default in place of the behavior I cued, no reward.

Q1b.  (based on the last question) What are the four elements for stimulus control?

• If I cue X, you do X
• If I cue X, you don't do Z (or others)
• If I cue Z (or others), you don't do X
• If I don't cue X, you don't do X (doesn't apply to defaults or naturally occurring behaviors)

Q2.  Are there times when we should change the training order for the last three of the six stages of learning?

• Yes, there are times when we would.  Actually the last four stages are what makes up stimulus control. 
• Those last four are usually on a circular cycle where they are each being taught over and over, or in random order, as understanding and difficulty of behavior increase.  Its important to specify, that when I say random here... that doesn't mean within one training session.  It merely means that between different sessions, these don't have to be in a specific order.
• Different behaviors have a different logical order for teaching.
• Plus, some dogs do better with a different order.  For instance, a dog as smart and active as Gimme will be bored with most duration work, so it works better to teach her distraction and/or distance first.  Then the duration develops along the way without boring her.
• And, which "stage" you are working on may depend on the environment in which you are working at any given moment.  IE. What I train and expect in the living room will be different than what I can train and expect in a parking lot.  I would be more likely to train for distance or duration in a familiar environment, while distraction is present in a strange environment, so initially that is what you are training for.

Q3.  In classes we discovered that stopping after 5 good repetitions gave us the best results. Am I doing too many reps at a time and is that undermining the process?

• Probably yes – even though the sessions are still relatively short, 3-5 minutes, there may be too many reps in a set.  It's better to focus on repetitions rather than time.
• We did a plan for this private lesson, where for the most part, after 5 reps we'd break off that set and let Gimme relax and refresh.  Then after a minute come back for another set. In this way Gimme was able to stay with us for a session that lasted 30 minutes, even despite the distraction of a new location (never been in that room before), despite her knowing that nosework was happening elsewhere in the building, despite the brain intensive nature of some of the work and despite the fact of her being in season.  Woohoo Gimme…

Q4.  Does Gimme know the behaviors (peach and orange), does she simply believe she should do whatever prop she is looking at, or is it still a listening skills issue?

• We didn't really address this issue in depth, rather went into training and working on listening skills.  We initially found she couldn't do a number of well known behaviors, as well as the two fruit behaviors, in that distracting environment, so we had to show her that it still applied there too.  Then she did much better.
• When we set up a different chair (folding no less), after just a moment of attaching the cue to that chair, Gimme went right to work, if slowly, so she does know the cue.  Initially the quality was lower, nose under instead of head under.  Then as the lesson progressed she went back to putting her whole head under the chair.

In other blog entries I'll talk more about the exercises we did.  There is much to do and it has become quite a fascinating process.  Gimme always keeps me on my toes.

All Fruits Look The Same

I thought I might try something different today and got results that cause me to question whether Gimme actually knows the cues for peach (head under chair) and orange (sit on perch) as individual behaviors or whether she has a different understanding.

I set both of them up and tried randomly cueing between those two and sit and down.  The first issue was where I was standing and whichever I cued she went to the closest one.  Once I moved to where I was an equal distance from them she was giving me both of them. 

For a bit I thought she just wasn't in listening mode.  When she gave a wrong answer, I turned and walked a few steps away, as far as needed to get her to abandon the incorrect behavior.  But as the session went on - her responses to sit and down got more and more correct.  Still the peach and orange behaviors were really variable.  I thought for awhile that she was simply giving me the one of those two that she did last.  However, as I closely watched her, it became clear that she was giving me the one that she was looking at when I gave the cue.

This makes me wonder... is it still a listening skills issue or does she not know those two cues.  Is it possible she believes those two cues both mean interact with whichever prop is present, so that when two props are there she takes the cue to mean interact with the prop that is in her line of vision.  I just don't know the answer and Gimme isn't talking.

I had already scheduled a private lesson with Ursula for before our nosework class this coming Wednesday.  So am not going to train these two in the meantime.  Ursula says she has another listening skills exercise that will help Gimme to understand when she should and when she should not be offering behaviors.  I'll send her an email and let her know about what I saw today and we can perhaps add that to the session.

This has been an intriguing training process.  At times its very frustrating, but I know in the end it will be very valuable.  I know Gimme is a canine genius, but there is clearly a hump of understanding that we aren't getting past.  I wonder if her being so super smart isn't a bit of a liability in this situation.  I've always said no one ever needs help training a dumb dog and obviously the opposite is also true.  Clearly I need to raise my game - sure am thankful to have Ursula to train with.

Not Sure Who's Winning

Peach - head under a chair
We just did another session.

While Gimme's listening is improving (her accuracy in actually giving me what I asked for is up quite a bit from before our private lesson), she is still unwilling to give up offering.  I want to give her more cues to do things and earn rewards, but she won't stop offering long enough for me to get in a cue.  The idea of stop-do-nothing is just not in her plan of action.  With her its all about a plan of ACTION.  We got four shorter extinction bursts, but they were still there - two to five minutes each.

Upon Ursula's suggestion I tried clicking any time she was still for even the tiniest moment.  And for a couple of minutes I thought I was getting somewhere, until I realized what little Miss Smarty Pants was doing.  Every time after getting a treat, she was returning to exactly where she was when I last clicked.  She was trying to turn it into a free-shaping game.  I probably wouldn't have caught on, except one time she happened to be under the table when I clicked, so when she returned to that spot under the table, it was pretty obvious what she was up to.  

Makes me wonder if I should put free shaping on some kind of cue.  Maybe that would make it clearer to her when we are and when we aren't playing that game.  She loves free-shaping with a passion and I'm happy to do it with her, but we need to get over this hump in understanding as well.

While I'm fiddling with some projects tonight, I'm going to try something with her.  I'm going to put a bowl of treats where I can reach to it on the kitchen counter, so its not where I am and thus not an obvious cue for training time.  Then occasionally will give her a cue and if she does it, she'll get click/treat. 

I can also give her the occasional click/treat when she isn't doing anything.  That doesn't happen very often if I'm up and about, she's either trying to work me or laying on the couch waiting and watching for an opportunity to start working me.  (good Lord, I think I just described a Border Collie!)  If she thinks I'm working on a project, she may settle in a bit more.

She's a pistol...

E.Bursting Away

Peach - head under a chair
I started with a session of peach.  Gimme did well enough for the first couple of cues, but then went into doing her own thing, which got a big fat nothing from me.  So then we had another extinction burst.  The first time was 8 minutes, then 5, then 1.  That's a huge improvement over what she did Saturday evening where it was a solid 24 minutes.

At one point she was headed toward the chair when I said "sit" and just a moment before getting there, registered what I said and turned to sit instead.  I thought that showed she is developing some understanding of this new criteria.  Good girl Gimme...

She's lessened the insistent tone of her burst, while adding two new features:

• She apparently wondered if opening a crate door and going in to lay down might restore me to my prior functionality.  She pulls the door open with her paw and walks in - never saw her do that before.  She did it once each with the two different crates in the living room.
• She also tried more barking, but not at me.  She'd sit or stand in front of me, turn her face about 60-90 degrees away from me and bark several times really loudly.  Meanwhile she has her eyes cranked toward me, so she was watching for my response, but not facing me.  Another thing she's never done before.  Apparently she thought I might be offended by her barking at me.  She does get funny ideas.  

I also decided that when she does a peach that is better quality, with her head (not just the nose) under the chair, that she'll get a jackpot of two pieces of peanut butter toast.  Decided that close to the end of the session, so it's too soon to see if I got results.  Also noticed that if I cue peach and she only gives me nose-under, if I wait she'll offer head under.  Tells me that the concept is in her brain.  She's a peach, eh...

Orange - sit on a small perch
Gimme continues to believe that backing on or side-stepping on is a better option... she does that about 75% of the time.  This went well, we got three short bursts, a 4 minute and two at 2 minutes. 

Her clarity in responding to "sit" and "down" dropped off a little during this session, but overall it was still an improvement over a few days ago.  I may have confused her by also working on the limited hold of her down during this session. 

Her new offering for the burst this time was, naturally, very clever:

• She discovered that when you jump on and immediately off the perch while kicking it with your back feet, it'll tip over.  The first two times I righted the perch.  Then decided that she might be getting some reinforcement from me fixing the perch.  So then I didn't and she still jumped on it, actually from the other side and when she kicked off, that kick pushed the perch right side up.  Given how unstable the perch is when she does this, I'm amazed she'd do it more than once... no one ever said she lacked confidence.

Her eyes are bright and she clearly wants to train-train-train.  She'd do it much of the day if she could.  Even this mentally challenging stuff, she's eager to get to it.  She'd like to train a LOT more, but I am trying to stop before she gets confused or mentally fatigued. 

Later, just to keep her happy, I may do a short session on some simple behavior.  She is the Empress of the Cosmos, doncha know.  Its my duty as her human to accede to her wishes...

Peach Extinction Burst

As I said I would, I did a training session using peanut butter to increase the value of the peach behavior.  I made some toast, put PB on it and then cut it to crouton sizes, making 20 pieces.  I had cheese and steak bits for anything else I wanted to reward.

Gimme quickly decided that peach was THE behavior of the century.  I spent a little time refining it and asking for better peaches.  Sometimes she only puts her nose under and other times her whole head goes under, the whole head is what I want.  I didn't want to belabor the point though, since I think my click timing is part of the problem. 

When I had 6 toast treats left, I started asking for other behaviors.  That is when the extinction burst for peach happened.  Twenty-four minutes later... 

I defined the extinction burst as finished when she was no longer:

• offering any form of peach
• walking around the peach chair
• sitting as demand
• barking
• whining
• cooing
• talking
• screaming
• putting her paws on the table to see if there was any toast left
• nose bumping my hand
• stomping her paws around the room
• pouncing on the couch
• sitting at heel
• mouthing my hand
• licking my fingers
• trying to steal leftover toast
• whomping into a down
• stomping her feet while down
• anything designed to get my attention

Needless to say, that was one impressive extinction burst.  Too bad I didn't get that on tape.

Once it was done, she didn't offer me the peach behavior except when I cued it.  I did have to help her the first two times, but after that, she was right on it.  A surprising side benefit, suddenly the sit responses tightened up a lot as well.  I'm sure we'll go through this again, but right now I'm leaving her to percolate on what just happened.  I'm sure she thinks there is something seriously wrong with her person.

Private Lesson

Our lesson didn't turn out quite as expected, since Ursula got distracted by late arriving guests and forgot to bring her dogs.  So, she gave me a free lesson on the "wait for the cue" issue we've been struggling with.  Remember, this is the third stage of learning - the one referred to as true obedience.  Ursula sees it very much as a listening skill, one that very smart and very confident dogs often have difficulty with.

We first discussed what the consequences should be for a wrong response.  I knew I didnt' want to give a no-reward-marker because of the overall behavior suppressing affect of NRMs.  Still I wasn't clear on what to do that wouldn't be an NRM.  Ursula said to just calmly walk away until Gimme gives up the behavior and joins me, then turn back and give her a new chance to be right by repeating the cue.

Something Gimme had been doing and I'd been rewarding was - I'd say "sit", she'd give me a down (her new favorite behavior) and I'd wait, then she'd ponder and say, "hmmmm that's not getting a treat, now what did she say, oh yeah it was 'sit', I'll do that".  By rewarding that, she is still getting rewarded for a) not listening, b) not giving me what I asked for and c) poor latency.  So from now on I'll do the turn and walk away until she joins me - which will make it clearer and won't reward stuff I don't want.

Ursula suggested that I sometimes play a listening game with Gimme -- in particular when she is not in training mode.  I can give her a cue for a behavior she knows well without any other prompting (i.e. not saying her name to get her attention first) and then if she does it, say "good" and give her a treat, petting or some other reward.  I won't use a clicker since that gets her in training mode and I can't have the treats on or near me.  So, if I want to reward with a treat, I can repeat "good" on the way to where there are some treats.  BTW we are using "good" specifically because it is not our click-word (ours is "yes"), so if my timing is bad, I'm under no obligation to pay up.  For instance, I told her sit and when I said "good" she had started to lay down.  I didn't want to reward that so she got nothing and it wasn't a case of breaking the click=reward bargain.  (obviously you won't muddy the water with this approach for a dog that was a clicker novice)

Then the second thing we worked on was to get Gimme to give me a behavior only when I cue for it.  Ursula suggested that I might get better results if I tried the extinction approach, with a bit of the mixed sets.  I had worried about using extinction, since I didn't want to risk weakening the behavior.  In reality, using extinction won't weaken the behavior.  The moment she stops offering the behavior, then I'll cue it and she'll get rewarded for it.  At that time, its essentially on a variable schedule of reward, which makes the strongest behaviors of all.  Well duh...

We tried this with her orange behavior (get on a box and sit) and it worked well.  We were starting to get some good results.  We tried it with peach (head under a chair) and it didn't work as well.  Although Gimme clearly knows peach, she doesn't seem to be that fond of it.  I think that's because its not a particularly active behavior, which is Gimme's preference.  So before we work on that again, I'm going to train the behavior and raise its value by using Gimme's favorite reward.  You know the one - peanut butter. 

So that's some work cut out for us in the next few weeks.  Ursula agreed with me that Gimme has difficulty with this because its not on her agenda.  Gimme's agenda is "its all about Gimme-me-me-me".  We talked a bit about this, because Urs wanted to be clear that we aren't talking about dominance stuff - which she and I both agree has nothing to do with dogs. 

Rather, its about giving up control, which has to do with all species.  A dog that wasn't clicker trained or lacked confidence won't have as much of an issue because they've never really had any control, so there isn't much to give up.  THAT certainly doesn't describe Gimme.

The Project - Day 9

I have multiple lists of all the things I want to get done before we leave when dawn cracks tomorrow morning. I just don't want to have any detractors. I want it all to go smoothly.

The closer it gets to tomorrow, the more nervous I get. I've competed before in obedience, in conformation and a LOT in agility and I don't ever remember feeling this nervous. The only explanation I have is that I just want to do well so people will be impressed by Gimme and love and admire her as much as I do. I have an emotional investment in this debut, wanting people to see the adorable, loving, smart and funny darling that I've been blessed to live with. I want them so jealous.

I have to keep reminding myself that she's just a puppy and that almost anything could, and probably will, happen. Whatever - I know it will be fun - I always have fun with Gimme.

Apple - back into a box, back feet only
I did a session of this and again Gimme persisted in the front feet first approach, even though I very carefully did not click that, not even a freebie. Toward the end she was getting it right 9 out of 10 times. Then when I switched so that I was on the other side, relative to her approach, she went back to the front feet in first approach.

When she gets it right now, she is actually approaching it head on and then turning away with her front and essentially side-stepping in with the back feet. We got very few backing in today. Somewhere along the way I must have muddied the waters, because she was backing in a lot better and more often before.

We won't get to work on this for a couple of days because I don't have room in the car for the box w/platform with all the other stuff I'm taking for the weekend. When we do come back to it, I may rethink how I set it up so that we get better success. She's so eager I don't want to frustrate her. It occurs to me that I really need to get the mirror in place so I can see more clearly what is happening and click more accurately and timely. That may be all we need.

Orange - get onto a small perch with all four feet
Not trained today.

Grape - dead bug (on back, feet in the air)
Not trained today.

Peach - head under a chair
Another WOOHOO... its on cue! Despite a slow start (its pretty hot today and the training building doesn't have the best air flow), when Gimme really got going, she was very steadily putting her head under the chair. AND she was putting it in further than before. Before she put her nose under and I accepted that; today was her whole head. A couple times when I clicked a bit late it was clear that Gimme was ready to go under the chair and come out the other side, if need be. 

                            *** No behavior is ever good enough the way I
                            design it, Gimme always wants to embellish. ***

What I was most proud of was how quickly this went on cue. Just 15 reps between the three different stages and she had it. Then we did the test: sit, touch, touch, sit, peach. Again she was really slow the first time, giving me that sideways look. After that she responded to "Peach" quickly for two more mixed sets.

Then just as suddenly, her understanding of the cue was gone. Ursula said it wasn't gone, she was just mentally fatigued, probably exacerbated by the heat. I knew she was right when I thought about it, because the speed in response to "sit" dissipated too, though "touch" was still fast (but its a less demanding behavior). Sure enough when I tried it again later, she went right to it.

Melon - push cube with nose 
Not trained today.

I now have two behaviors that are ready to go to the third stage of learning: waiting for the cue before performing the behavior. Ursula calls that the "true obedience" stage, because in free shaping the dog is in control. When they learn to wait for the cue, the control passes back to the person. When the dog isn't getting rewarded for what they want, they may try to get those rewards by offering their default behavior. So its important when you are working on the wait for the cue, that you refrain from rewarding a dog's offering of their default behavior.
There are four different ways to teach the third stage:

1. Extinction - simply stop rewarding the behavior unless you have cued for it. Be aware that you will likely get an extinction burst and it can take awhile for the dog to give it up. You will see a lot of frustration. Once they stop offering, then cue it again and reward performance on cue.
2. Extending the cue - when you are naming a behavior, you start by saying the cue while the dog is performing and saying it earlier and earlier over multiple repetitions, until you are saying the cue before the behavior has begun. You can continue extending how early you say the cue, until you have created a wait.
3. Mixed sets - repeating the mixed sets, interspersing the newly named behavior in with well known behaviors, will teach the wait, especially as you add time between the cues. Of course, you must have other behaviors on cue that are well known to use this option.
4. Click for waiting - actually click/treating the momentary pauses between repetitions will help the dog to understand that "not doing" has value too.

Of all of these, I think I like the fourth option, click for waiting, the best. It seems to me that it would be the clearest to the dog. Also, by applying that while doing option three, mixed sets, it should go very quickly.