After a six week training course, this weekend Luka and I passed the American Kennel Club Canine Good Citizen (CGC) test. The course was really for me, not for Luka. He could have easily passed the test with a good handler. I was the one who needed the training.
One of the challenges I faced during the fall was managing Luka’s behavior during interactions with other dogs. When we came across another dog on one of our daily walks Luka tended to bark and growl even from a distance. I wouldn’t say he was aggressive but he definitely wasn’t friendly and it made those interactions very intimidating to me. We had worked on this in Austria but I was obviously doing something wrong.
I spoke to a local trainer and she explained that Luka had probably learned to assert himself around other dogs out of necessity. We don’t know anything about his life before being rescued but she theorized that he may have had to fend for himself and is therefore wary of any potential threat, especially strange dogs. She reminded me about the proper way to desensitize reactive behavior with classical conditioning then lots of repetition with operant conditioning. The problem was that as the weather turned colder and snow started to fall from the sky people seemed to stop walking their dogs. The opportunities for exposure to strange dogs rapidly dwindled and repetition was out of the question. She suggested we take a Canine Good Citizen class which, besides being good practice for us both in basic handling, would give us a chance to work regularly around lots of other dogs.
So every Saturday morning we met with other student teams and practiced for the test. I quickly learned how to capture Luka’s attention around other dogs and hold it in exchange for lots of high-value rewards. In very little time I gained a great deal of confidence in these situations which, as it turns out, was the piece I was missing. Now when we meet strange dogs we fall into a practiced routine which gets stronger and smoother every time.
Here is Luka and I taking the CGC test:
And here is some additional footage we recorded over the weekend of Luka demonstrating several “distance” commands that were not part of the CGC:
Forgive me, interweb, for I have sinned. It has been four months since my last blog post.
I could recite any number of excuses for not posting but the truth is Luka and I were dealing with some challenges that I didn’t want to write about until I had a chance to gain some perspective. I’m happy to say we have emerged from our cocoon of silence and are better for it. I hope to catch up in future posts but first I want to share some cold, hard facts.
In November I started keeping detailed statistics about Luka’s alerts. Every single alert (or missed-alert) was recorded and tallied by day in a spreadsheet. Each alert was categorized as follows:
- Expected: This means that at the time he alerted I wasn’t surprised. My internal clock that tracks when and what I last ate, level of exertion since, and general trend of the CGMS during that period were consistent with the potential for a low blood sugar. In each case I confirmed with a finger stick and rewarded Luka, ate some Skittles, and recorded the alert on an app on my phone.
- Unexpected: This means that at the time he alerted I was dubious about his sincerity. The CGMS typically showed me either steady in the 90’s or well above 100. I honestly believed these were false alerts. The meter confirmation of Luka’s diagnosis came as a complete surprise. I have another word for these events: LIFESAVERS. These are the potentially serious episodes that neither technology nor insight could have predicted and, before Luka, would have dropped my glucose to dangerous levels before I caught them.
- False for attention: I’ll be talking about these more in upcoming posts since this was THE big challenge I struggled with. Suffice it to say that Luka learned that the alerting behavior brought him attention whether I was low or not. This was, in fact, the reason I started keeping such detailed statistics – so that I could measure whether things were getting worse or better as I tried to retrain his behavior.
- False NOT for attention: These were really interesting and it took me a long time to understand what was going on. In these cases Luka was definitely not trying to get attention but still alerted confidently. More about this and false alerts in general in a future blog post.
- Missed/Prompted: These were times when I discovered I was low either from a random finger-prick or because the CGMS caught it. In each case I intentionally got close to Luka so he could smell me clearly and eventually he picked up the scent and alerted.
Here are the numbers from the past 80 days:
Total Alerts: 241
Total Expected Alerts: 129
Total Unexpected Alerts: 112
Total Missed/Prompted Alerts: 22
Hit ratio: 92%
These numbers are shocking to me. Luka correctly alerted me to an average of 3 alerts per day. He protected me from 112 potentially serious hypoglycemic excursions missing a low only 8% of the time. I couldn’t be more amazed or grateful for his presence in my life.