[mer.petkova]

Thanks so much for the recs, Tracy!

Our regular trainer is Kristen (and sometimes Kate) at High Fidelity! And Obi’s fitness instructor is at Canine New England, so you most definitely got us right 😀 😀 I’ll let you know where we end up matching probably by MaxPup3 🙂

Cheers,
Mariela

[mer.petkova]

Hi Tracy,

Just stopping by here to say thanks for all the fun and support in MaxPup2 and that I’m very much looking forward to reconnecting again for MaxPup3! Obi is getting a few weeks break while I’m in Europe, but we should be able to wrap up a few loose ends we have here. I hope you laugh at this, but I’m using my downtime here to make some agility flashcards for us 😀 (and I was thinking the flashcard life is well behind me!).

Also thank you for offering to suggest someone who is a good fit in the area!

Best,
Mariela

[mer.petkova]

Hi Tracy,

You were super right about getting Obi all riled up while tugging! I thought I had him excited in general by doing some body play and food play before, but some chasing and tugging got us even higher. He flew through the tunnel, then when I tried jump and tunnel he was blind to the jump 😅 More tugging it is!

One thing about our class is that it’s in a small training facility. He will never be able to jump full height there for one. And we rarely sees distances of more than 6ft between obstacles. Plenty of distractions working around other dogs though, so that’s where I see a lot of value (video from tonight)

(btw he saw a pinwheel for the first time tonight in class!)

While I figure out where I can train with him at a bigger venue (and when it makes sense to start that, and who the right people to do that with are). I’ll work on increasing distances at home. We have a decently big yard albeit sloped 😅

Any thoughts on that are appreciated!

Also, he did fill in these past few weeks! I like to think fitness is also contributing to the changes.

Lastly, Duke’s team are clearly missing out. But hey if you are passing by MA on the agility circuit, Erin’s team seem wonderfully responsive and their calendar is book able until Sept or so (that’s when the working dog study ends)!

[mer.petkova]

Hi Tracy,

It’s been a bit of a week here 🙂 Too much love exuberance from my Berners put me off balance and I took a chair arm rest to the ribs (11/10 would not recommend, makes laughing very hard and our househould is all about laughter). But since I managed to wrangle a snarling, mounting dog off Obi on our morning hike I figured I’m healed enough to wrap up what we’ve missed from MaxPup2 (Obi’s fine!).

We’ve missed the “find the jump game”, which surprisingly came up in class a week ago and he ran past the jump when I was ahead of him 🙂 I tried very hard to get ahead of him in the game here, but he is surprisingly fast! I liked his commitment and focus ahead a lot and we got decent lateral distance (though not obvious at 0.5x)

And speaking of lateral lead outs, apparently I recorded our first session last week. I remember being surprised at how easy it was for him to get it, and then coming up with some handling with the second jump on the spot. Not sure what I was thinking beyond that 😀

Thanks for nerding out with me on the neuroscience/dog behavior intersection. Its constantly on my mind and I’m compiling relevant research, I’d love to connect about it when something clear and cohesive comes out of it (these things take their time)! Incidentally, Erin Hecht here is running the Canine Brains Project and right now they seem to be focusing on working dogs — I’ve signed up Obi as a participant 🙂 https://sites.harvard.edu/caninebrainsproject/publications/

That’s the Sawtell paper I mentioned, but again — very dense read: https://pubmed.ncbi.nlm.nih.gov/31735497/

Cheers,
Mariela

[mer.petkova]

Hi there! I have to admit I harbor the hope that brain and behavior, as complex as it looks to be, is fundamentally understandable and there are simpler principles we’re getting at. Perhaps it is my initial training as physicists (we love solving problems for spherical cows in vacuum), perhaps it is working with a simpler” model organism in neurosicence, and perhaps it is seeing all the incredible practical successes in dog behavior and training 🙂 I’m listening to the BrainCamp lectures by Kathy right now on performance, attention and the emotional state of a dog and I can’t help but see the parallels with our lab’s work, which focuses the neurological basis of zebrafish behavior — zebrafish are neat creatures, they start from a single cell and 7 days later it’s an animal which has 100,000 neurons and can swim, hunt, run from predators and we can study those behaviors in the lab and even look at all the neurons in the brain as it processes information:

I’m bringing this up, because the lab has been around for just about 20 years and studying the behavior of the animal, but just in the past five years or so we started paying attention to the *internal state* of the animal and find the same things: zebrafish have *attention* which sets their performance and they can switch between attention states depending on their internal state. The brain’s wiring diagram doesn’t change dramatically at any moment, and at any moment the brain is capable of *all behaviors* because they are all in the wiring diagram, but what gets recruited depends both on the sensory stimulus *and* the internal state (the soup of neuromodulators such as dopamine, serotonin etc prioritizes what behaviors are being recruited at the moment). It’s very cool the very practical approach in dog training mirrored in the basic science of a (at first glance) very boring and “simple” organism like the zebrbafish.

And you brought up a super important point which is social cues and learning, we humans are so good at learning because we have language. IMO, we don’t have language for the pups, but *because* learning happens very quickly when it is in a natural setting for the dog, a short *clear* session with high rate of success substitutes for the langugage we humans have and sets up the clear message. Which is why, again, I love how you break down the exercises in max pup. The communication is beautifuly clear 🙂 Also, idk if you know that saying “Love is like a fart, if you have to force it, it’s probably sh*t” and I think it applies to dog trainig too 😀

There is cool articles on the genetic basis of breed-specific behavior https://www.cell.com/cell/pdf/S0092-8674(22)01379-4.pdf relating to your comments about the differences between the whippets and papillons. But then again, as you say there is also a general difference between the working lines vs the show lines within the same breed, so it is even more refined. I only have to look at my two Berners to be reminded how much genetics can make a difference 😀

You brought up Hebbian learning, and there is actually some super cool research in electric fish I’m learning from a collaboration. Here’s the deal: electric fish live in murky waters and send out an electric signal that they use to map the space around them in order to catch prey. To do this they measure the electric field of the environment, subtract their own electric signal and what’s left is the perturbation caused by the prey. So what the collaborators at Columbia U (Sawtell lab), showed was that Hebbian learning (it’s technically anti-Hebbian, but whatever) will let the animal learn it’s own signal over time on its own, *but* the details of how all neurons are wired up in the system (i.e. which neuron type talks to which neuron type) makes this learning happen 100 times faster!!! So it isn’t just the local rules of synaptic plasticity, but the genetically wired architecture that makes learning happen so quickly. Happy to send you the papers, but I have to say they are too dense and technical and not the most effortless read. But all that is to say that your dog maybe learned 100 faster that dropping the ball in front of the jump is an option 😛

BTW, given that you do read science papers I want to make sure you know about scihub (just put it in google search) which is a great website that uses a bank of academics-submitted credenetials that open up the paywalls to the journals. So you can read articles that are behind paywall (if you type in the url or the doi number). I hate the journal paywall system, it’s a horrible pyramid scheme.

————————————————————
Videos!

We tried the backside proofing:

and I tried to add countermotion:

I have a sense that it gets reeeealy hard for him when he sees more than just the wing of the jump, but also when he sees the cross bar and if he gets ahead of me and needs to make a choice front vs back.

I’ll flip the wing to the other side to practice it tomorrow, but this was our first try.

Cheers,
Mariela

• This reply was modified 1 year, 2 months ago by mer.petkova.

[mer.petkova]

Thanks for the threadle feedback Tracy! Not recorded, but you were totally right — my open arm prior to the release was causing some stay confusion 🙂 we’ll clean it up!

We did the grid with 8 inch jumps yesterday and since it was set up already I did the moving jump version today.

fixed grid:

moving jump:

He just keeps bouncing!? Should we modify the distances or do we leave him bouncing?

Neuroscience talk:
Yes, your description is a very familiar way of thinking about learning and motivation. We’re playing with a different point of view, where RPE plays the role we know it to have but it isn’t the primary driver of learning, rather it is fine tuning.

This all started with an experiment where a graduate student raised zebrafish and blocked all brain activity for the first 6 days of the animal’s life, then removed the chemical blocker and 1. the animal performed all the behaviors a normal zebrafish does, and 2. you could actually see the brain come online when the blocker goes away (one of the supplementary movies in the paper: https://www.ncbi.nlm.nih.gov/pmc/articles/PMC10774424/). So what this tells us is that the animal behavior is very much genetically built in already and there is very little neural rearrangement going on, this did cause some stir in the community of course because we have been attributing a lot of neural rearrangement to environmental factors.

What this then boils down to is the idea that animals learn very quickly behaviors that they are wired to do genetically, basically it’s effectively one-shot learning. Executing the behavior correctly is a matter of fune tuning the system. For example we quickly get the idea of “use the racket to hit the ball over the net” in tennis, but it’s many repetitions necessary to build the muscle memory and coordination (fine tuning). On the other hand a lot of the research on learning is done in mice and yet, a mouse learns a maze ridiculously quickly, or food avoidance when it’s toxic — to the point that these are so quick that no one in a lab studies them because they are “too good at the task already” and instead focuses on lever pressing which is such an unnatural thing to ask of a mouse that it requires thousands of repetitions (taking fine tuning to an extreme and calling it learning).

To bring this to Obi and the training sessions, interacting with me in the context of trainig is like being in a game of tennins and knowing that the ball should be hit with a racket, then doing all the agility sequences is all about fine tuning the movements and context. There really isn’t *that much* of a surprise for him. The short sessions make sense in this context because 1. learning actually happens very quickly, 2. animals switch between exploring and exploiting states all the time, e.g. a behavior can be very successful on many repetitions (exploitation state, looks like “paying attention”) but after a few they will switch to exploration state (try other things) which looks like distraction, or fatigue, etc. I’ve come to appreciate the MaxPup setup precisely because it seems to set us up for clear quick learning followed by an exploitation state to reinforce the sequence and if the human doesn’t get greedy with it we can walk away early before we switch to a new exploration state (and the clarity of what we’ve just done gets muddled).

Anyhow, everything following the actual experiments with the fish is mostly active brainstorming for rethinking learning and animal behavior in light of evolution and systems neuroscience 🙂 thanks for letting me thought vomit, hope it’s not entirely un-parseable.

BTW today I learned that Karen Pryor was helping one of the faculty here to design training assays for his rat experiments!!! 🙂 Small world ..

[mer.petkova]

Hi Tracy!

Work travel got us quiet, but we are back 🙂 We played with the threadle/serpentine/tunnel proofing these past three days.

Day1:

Day2:

Day3:

It’s Obi’s bday today! Hard to believe he is a year old already <3 I dragged him to his first fitness appointment and no surprises there – he had fun learning tucked sits and kickback stands (on platforms :))! Now we are committed to 100 mile drives on Mondays for awhile 🙂

I’m curious about Brain Camp (stalking it here — https://braincampflix.com/), and the dopamine comment you made (for Obi and Skye, ofc … not because I think it might be useful training strategy for tackling emails 😀 ). What was the rationale for timing the length of the sessions with dopamine release? I’m vaguely aware that the patterns of dopamine expression are different for classical and operant conditioning (where motivation plays a role), but I’m curious to hear more.

Thanks!
Mariela and Obi

[mer.petkova]

Hi Tracy,

We had more luck with the wraps vs go with your suggestions! The distance, in particular helped a lot 🙂

We used the moving target to help him point his head down, too. I loved his stays even when the toy was on the ground in front of him!

We did better with his out game as well:

I’m back to keeping things super short and fun for him (5 min timer for the whole session, or 4/5 reps of something). It keeps both of us wanting more next time 🙂

I’m noticing he doesn’t like to shift his weight onto his hind legs and I’m policing his sloppy sits. We have a fitness appointment for him in a week to start working more on these, and a sports vet (in June! — wow they book out). So hard to tell with these youngsters if this is a puppy thing, or something else (and ofc we always worry about the “something else”).

Hope you had a good weekend and thanks for your help!
Mariela and Obi

• This reply was modified 1 year, 3 months ago by mer.petkova.

[mer.petkova]

Right — the speed up (+) idea makes total sense. Thanks for the tip!

We’re celebrating a turn-away on the minny pinny!! 🙂

For the decelleration to wing wrap Obi was turning wide and going more into the center of the jump. I think he expected a Go!, he has been primed to go staight on jumps. Maybe things will settle in after another rep later.

We also tried the Out game and he picked up on leaving me for the jump right away. Staying with me and ignoring the jump was a different story. He had two mistakes in a row and I ended up rewarding the third anyway. Then cut the session. I was pointing at the jump with my chest in both cases and I think it was fair of him to read it as go take the jump? Curious what you think!

Thanks!

[mer.petkova]

>> So true! LOL!!! But will eventually figure it out and I am sure it will replace me in the future LOL!!

I spend a lot of time with machine learning models and AI and I hoghly doubt it! But Gemini is going to get a boost soon in terms of how much data it can process, so we could give it a try with a couple of videos with feedback from you as input and ask it to comment on a new video 🙂

I actually do like using Perplexity a lot as great way to get context on new information/ask questions, it searches the web and also gives you references/links to where the info came from. Works really well for science stuff.

Thanks for the feedback on the diamond!

p.s. I had a complete revelation as someone new to using recording for troubleshooting – I played the video at .5x speed and I could see my cues/Obi’s decisions etc SO MUCH BETTER. What a *duh* moment ….

• This reply was modified 1 year, 3 months ago by mer.petkova.

[mer.petkova]

Thanks Tracy!

You are right, he wasn’t actually upset with me or I’d have been nose bumped in the face 😉 We will give the suggestions a try!

Today was a beautiful day and we justh ad a lot of fun with the diamond game.

I have one question – when exiting the tunnel and he heads for the wing where I intend to either blind cross or front cross is the correct verbal a wrap or a soft turn. I was watching the video just now and realized I was using wraps, but it looks more like left/right turns for him. Edit: I rewatched your runs with Elektra in and cleared it up that it should have been soft tur cues. Oopsie.

Seems like I’m not yet there to goof around on the spot with verbals, and I just need to stick to a pre-determined concrete plan so I don’t muddy those for him. (like going out and practicing blinds for the figure 8 before getting him out to play).

Thanks!
Mariela

p.s. AI never tires, but sure it gets confused!

• This reply was modified 1 year, 3 months ago by mer.petkova.
• This reply was modified 1 year, 3 months ago by mer.petkova.

[mer.petkova]

Thanks for the feedback Tracy! We’ll go back to these later this week.

In the meantime I tried rear crossing and I think I got Obi upset with me! The evidence:

I was actually really happy that he was facing forward on the “go” and that he went ahead of me. The strategy there was to run along and stare in his eyeballs until he looked at the jump and then throw the toy.

The first attemt on right turn on the rear cross also started well – he got ahead of me and I could start turning away, but once he saw that he ended up with spinnies. I also don’t think he was reading this as a right turn (any right/left turn-away is very very hard). He has an easier time reading cue + body when he is on my left side and I cue a right turn (which I tried after two failed attempts at the rear cross).

I shouldn’t have tried the left rear cross in retrospect, but I wanted to see if he might do better there. I succeeded in making him sticky to me, even on the “go’s”. Impatient human! 🙂 (that’s really my only actual frustration with the session — that I went too long into scientist-debugging-mode)

All and all I think I need to proof the turn-away right/lefts before attempting the rear cross. I’ve done a few sessions with the minny-pinny for that — keep at it or try something else? 🙂

[mer.petkova]

Hi Tracy,

We did parallel jump AND the backside! I committed horrible verbal crimes, but I was too busy having my brain explode seeing Obi catch onto the back slice. (note to self: back = back slice, push = backside wrap). I don’t have this with Skye.

At 0:45 I forgot completely about my arm, because I was just staring at him take the backside line 🤯. FUN!

I’ll do it the other direction next (with left send to the wing).

We did the grid too, I’m totally loving his stay. Proudest moment was when on rep3 I called out the ‘catch’ cue instead of the release and he knew there was a cookie about to fly his way so he didn’t move 🧡
The jumps are 6 feet apart. Is the distance OK?

BTW — The pool noodles are split lengthwise and wrapping around a jump bar, hair ties are pretty great way to keep them from popping off the bar!

Thanks a lot!
Mariela and Obi

[mer.petkova]

Hi Tracy!

Hopefully Contraband is enjoying some RnR after that tunnel crash and getting put back together! It all happened so fast! Silly agility pups trying to carry on – ‘I’ve got three legs, mom! That’s one more than you do. We got this!’

We followed up on your serpentine suggestions — Obi had a much easier time. Watching the videos, I saw I should be opening up my shoulder better (even if he read the setup right. I’ll progressively make the angle zero and try the front cross too. Any other suggestions are welcome!

p.s. I finished the sesh with some food play. I’m looking over Denise Fenzi’s course on building relationship through play, so we are experimenting with that. Both dogs seem to love it 🙂

[mer.petkova]

Hi Tracy!

Obi and I have been watching lots of UKI this weekend, and got a bit worried about Contraband — is he ok? We are squarely in his fanclub — after all, he is our demo dog!

We are home and worked the tunnel exit handling game at the baby level:

I saw that 1. I baby the tunnel send a lot,2. I’m repeating right/left too much after he already has turned and I need to be giving him info about the next thing, and 3. that he still peeks towards me after the tunnel on the “Go”. I ended up doing a toy race game in the end, and will do wind in your hair tomorrow before coming back to this tunnel exit handling game.

We also got our pool noodles so I made bumps and tried the serpentines. He was more successful with the angled jump, and his right sends were also better/easier than sending to the left. On my end I saw that I wasn’t walking a parallel line to the jump, but was veering off to draw him in. Oh, and any added motion/speed on my end propelled Obi to blast right through the serpentime 😉

Thanks for thinking about these,
Mariela and Obi

[Author]

Posts