No flight cancellations this time: hooray, Jak is home! I survived just over 153 hours this round, 50% longer; I’m not as much of an emotional mess, but I’m more physically exhausted.
However, I think I made another mini-breakthrough on Tashi’s biting behavior while Jak was gone. Giving him time outs when he becomes unmanageable is not a new idea for us, but I was thinking about his obviously very strong pack instinct, and so the last few days I’ve been actively working on teaching the words ‘alone’ and ‘together’ and labeling consequences in advance. Biting gets him “alone time” — fifteen or so minutes in our currently-empty spare bedroom (it’s awaiting a renovation opportunity). Now, when he starts objecting with teeth, I can say “No bite. Tashi bites, Tashi goes alone time”, or something in that vein, and so far about nine times out of ten it shuts the biting off instantly — a better success rate than anything I’ve tried to date.
Here’s hoping it lasts. (My previous best approach, “Karawynn mad”, has lost some of its efficacy and now works more like seven times out of ten. Commanding a “down” also works, but is a lot slower and takes three to five repetitions.)
Can’t really use “alone time” when I’m outside with him, though. Often after I’ve played with him and I start to go inside, he doesn’t want me to be done, so blocks me by jumping and biting hard. If he can get his teeth in my dress he will try to run off with it like it’s a toy — he’s ripped holes in three dresses that way so far, most recently just two weeks ago. (His preferred style of play is all teeth. His slightly larger dog friend Chewie rolls with it, but for humans it’s a hard pass.) I can’t enforce “alone time” when he’s outside without a leash; he can just run away from me. But indoors, I might have the problem licked, so to speak. Fingers crossed.
So anyway, Tashi has been less of a pill this time around, but I’m still clearly not capable of caring for him and myself without help over the long term. Saturday I hardly ate anything because I ran out of leftovers and had no energy to cook. (People in US cities will probably find this baffling, and wonder why I don’t just order takeout. But options for prepared food are a couple orders of magnitude more limited here. The end result is that if I want to eat, I have to cook. From scratch.) I haven’t gone out to the garden since Thursday, which means food is definitely going to waste and who knows what new pests have shown up, leaving me with lingering guilt and dread, but: triage. Yesterday (Sunday) I worked on pets and chores for about four or five hours before I just flat ran out of spoons and collapsed — but I did at least get a giant plate of nachos in first. (It’s guacamole season: the avocado tree next to our back patio is dropping an average of one fruit a day right now. The challenge is always to get to them before Tashi does.)
The one upside, from my perspective, of Jak’s emergency trip to the States this month is that he was able to bring back more talking pet supplies that otherwise I would not have until probably the beginning of February.
It’s true that these last couple of weeks I was on the verge of running out of buttons, but I’d already been feeling the limitations of insufficient hextiles for a long time. Once I knew Tashi was going to progress quickly and switched to the Connect system, I should have overbought on tiles. It would have kept things more flexible and reduced the number of ‘do I put [word] down now and move it in a few months when I get more tiles, or do I wait to give access to [word] for a few months, until I get more tiles?’ dilemmas. Of course, it’s much easier for people who live in the US to order more equipment as needed, which would make that less of an issue.
Anyway, Jak returned with thirteen new hextiles: doubling my existing board space to twenty-four tiles, plus one to go on the counter for the cats. Also, 36 new buttons, bringing me up to a total of 84. (Much more than that and I’ll have to buy a second speaker, as each speaker only has memory space for 100 words.)
And: two cameras! It will probably take me and Jak the rest of this week to get them set up, but that’s so much better than waiting until February. I cannot overstress how important those cameras are going to be.
Once I switched to the Connect system in late October, I tried to keep up with accurately logging all the activity, but after three days I basically threw up my hands in defeat. Jak, who is far less dedicated/maniacal about this than I am, didn’t even make it half a day before getting really annoyed.
The Connect system logs each button word, and groups presses within a certain number of seconds (you can adjust the number) into a single ‘event’. Everything else has to be done manually: marking who (of the five household members) was responsible, writing any notes about context and documenting any spoken words that were part of the interaction.
It would be different if button-pressing were an occasional phenomenon. But Tashi by himself deliberately presses many tens, sometimes over a hundred buttons a day. Not counting the times he sometimes tromps across the board space while playing and accidentally trips a word or three. Jak and I seldom leave the house at all and even more rarely together, so (unlike with pets who are left alone most of the day) button usage is spread out and can interrupt me at any time.
Without cameras, if no one is in the room watching at the exact right moment, I have no way to know if any given press was intentional, accidental, or even which pet was responsible. Rikki, who has kinda-sorta figured out how to press buttons, but not that different buttons have different sounds and mean different things, is now randomly patting at, lying on, and (so thrilled about this development) biting buttons at various points throughout the day. Even Gracie, who can chase a feather teaser right across the whole board without tripping a single word, has been known to lie on the buttons a time or two. If I’m napping or running an errand, I don’t even know which presses are modelling from Jak.
Plus I don’t think there is any number of seconds, even ‘one’, that would accurately separate our modelling from Tashi’s (and now Rikki’s) interruptions, both deliberate presses and accidental (because he’s so excited that he’s dancing around between person and board trying to get attention). So many times I would have to go back to the app and split the ‘events’ apart and merge each one multiple times to group things properly — and worse, there’s an app bug currently where sometimes merging ignores the timestamps on each individual button and rearranges them into a different (wrong) order.
All of this meant that I would have to get onto the app (which, thank god I bought an iPad with a keyboard in August; there’s no desktop version, and if I had to thumb-type all this on my phone I’d just be like, fuck it) twenty or thirty times a day just to document stuff, at anywhere from two to twenty minutes at a time. Even as dedicated as I am, it was tedious, and it was taking over my life in a totally unsustainable way.
Oh, and also we found out that every time we logged onto the VPN — which we have to do occasionally, in order to access certain internet things from outside the United States — the Connect system stops logging presses. Luckily for me, Jak both figured out what was wrong (in our household division of labor, he takes point on technical wizardry) and also came up with a workaround: a way to connect the app to wifi that bypasses the VPN.
From my perspective, the hard data was merely nice-to-have; I cared about it more for purposes of actual science, for the researchers at UCSD’s Pet Cognition Project. But in talking to a few people already in the study, I gathered that CCL seems to be in a lull. They hadn’t yet responded to my original mid-September application (still haven’t), so it wasn’t like my data was actively being used.
So for sanity’s sake I bailed on messing with the app and simply kept a few notes of my own, which I am collecting here along with other records such as texts. But now I’m preparing to reset and take another shot at it, with the following improvements:
- Full-time cameras should mean I can review unobserved presses once a day, or even every other day, to assign agency in the app and make a guess as to deliberate or accidental. That way all the basic maintenance is corralled into specific times, and only for special or interesting interactions do I need to pause and record verbal words and context.
- As part of my ongoing crusade to teach Tashi manners, when he’s excited I’m going to start making him “sit” or “down” before I use the board, to cut down on the interruptions and the split/merge work. Learning turn-taking in conversation is a reasonable requirement for a talking dog, I think.
- Also, planning to try out a clever idea from someone on the Discord group and reserve one button for human-only use. I’ll put it in a human-only accessible place (TBD) and record a minimal sound, like a click or a low tone. In the app I’ll label it differently from the all-caps board words, like [modelling]. Then, each time Jak or I use the buttons, we’ll hit that one first, as a hint to future Karawynn that a human was responsible for the next event.
I’ve also reorganized the entire board, making a number of adjustments to the layout based on how I’ve observed Tashi using it. Along the way, I’ve eked out every last bit of space, giving me the ability to put up to 30 (or maaaaybe 31) hextiles down, for a maximum button count of 180 (186) (not counting any cat-specific tiles elsewhere). Though in reality not every hole on every tile will be used, so it will probably top out at closer to 170.
I wanted to find out how much of Tashi’s button choices were opportunistic, so as part of the redesign, I flipped the soundboard horizontally (from the diagram perspective) so those much-used sections (purple, red, orange) would be on the far side from his approach, and verbs, time, and questions (yellow and aqua) would be the first things he encounters.
I’ve been spacing out the chaos; I rearranged the in-use blue, purple, and red tiles three days ago, and by today Tashi seems fully comfortable with the locations again … just in time for me to wreak havoc on the yellow and aqua tiles today.
But even this preliminary board flip has taught me exactly how much location really matters. I originally thought Tashi was using certain buttons more because they were more desirable (food-related, ‘outside’) or more important to him (feelings). But I also noticed they were the first buttons he encounters from his usual approach direction (which is the right side of the above diagram).
The shift is dramatic. If it was, say, a 75-25 split before, I’d estimate it’s a 30-70 split now — almost completely inverted. And yet he still manages to make sense (most of the time)! For example, look at the difference between the morning presses on November 16 (pre-flip) and today, November 20 (three days post-flip).
I’ve put five of the new tiles down right away — that’s the nineteen that are bright on the plan above, minus the empty feelings tile and the two blue body-parts tiles at bottom. I’m holding at the current 43 buttons until I’m sure he knows where everything has moved to. Once I’m ready to start in on body parts, I’ll add the two blue tiles for that; he likely already knows ‘neck’ and ‘belly’ because those are his favorite SKRITCHES spots and I say them a lot.
The adjective islands in the center are on longer-term hold until he’s had more time to adjust, even though I’m already verbally working on some of those pairs (like ‘alone’ and ‘together’).
And that’s all for now; I’m finally caught up! Jak is napping — he had an exhausting week too — and I’m going to go lie down with him.