Decontextualised Science

[Flipping plankton]

Submitter comment: An AI-generated and completely nonsensical infographic titled AUTISM SPECTRUM DISORDER. Elements are labelled “MISSING VALUE &runcitional features”, “Historical Medical frymblia”, “Factor Fexcectorn”, and “RELU DROP-OUT Totalbottl REMICH N”. Also a woman is sitting “on” some sort of table but her legs are passing through the material and visible beneath it. Honestly, if you’re relying on the ALT text to make sense of this image, today is a day to be grateful you can’t see.

[All the world’s a phage]

[As strongly requested by the reviewers, here we cite some references [[35], [36], [37], [38], [39], [40], [41], [42], [43], [44], [45], [46], [47]] although they are completely irrelevant to the present work.]

Introducing IntraPaint: a free and open source program for combining digital art techniques with AI editing. I've been working on this project for months with the goal of making something that can actually be useful to artists, and I think I've finally succeeded. It's out now, and 100% free and open-source. You don't need any subscriptions or cloud services, it all runs on your computer, under your control.

See below for more information, but here's a quick preview of its most interesting capability: human-guided AI inpainting.

Source:

This is an embarassingly bad paper based on the use of two aversive based dog trainers who are clearly incompetant in the use of +R - yet the e collar training community is taking a victory lap.

The fact that this got past an animal ethics review is honestly appalling and the university should be ashamed.

Let's go over the main huge flaws with this study:

The positive reinforcement protocol they are using to compare to e collars is laughably incompetant - with no management or use of distance, desensitisation. Just use a leash? Maybe?

Use of aversives to stop chasing is, unsurprisingly, effective. Because, yes, painful consequences tend to work to suppress behaviour. That isn't the problem that people have with e collars.

The author seems to think that use of e collars is justified because it "gets results" in a short amount of time to stop dogs chasing - again, apparently we haven't heard of a leash.

"Aside from presumably pain-induced yelps in the dogs with e-collars when they received shocks, none of the dogs in any groups showed any signs of stress or distress." So we're saying a pain vocalisation isn't a sign of distress? That's your welfare parameter? Jfc.

Love how they didn't even try to get fecal cortisol baselines and just collected samples from a few dogs but not all of them. You tried, I guess?

The fact that the trainers for the study is "Ivan and Ivan's Student" is ridiculously biased. Acting as if both these trainers know how to apply +R effectively as opposed to their heavy aversive tool use is laughable.

"We designed our study’s methods around current trainer practices as well as the recommendations of our senior trainer while attempting to keep continencies as simple as possible." - yeeeah that's a conflict of interest. Especially when your "senior trainer" has an axe to grind with R+ people using scientific papers to support their training methods.

You're chucking treats in a bowl after a few sessions and you think that's going to stop a dog chasing a lure? Oh yeah, you got us. We just throw kibble in a bowl to stop dogs running into traffic. I am genuinely floored that that's how they think a +R protocol for chasing works. They have no business comparing the two when they can't even do one properly.

"We also did not test whether the food rewards we deployed were highly valued." Okay so the majority of my client dogs will go nuts for my meaty high value treats because they've only had store bought dry treats and the client is shocked at how food motivated their dog suddenly is..... So this study was relying on whatever the guardian of the dog brought with them lmao rookie mistake.

"Dogs that chase livestock may be shot or otherwise euthanized, and dogs that chase cars may be killed in traffic." Put. Them. On. A. Leash.

"The practical applicability of either the LIMA or LIFE models is likely limited, at least in the United States where dog trainers are not required to be licensed or certified. In a sample of highly rated dog trainers, more than half did not document any relevant education." Does that include your "senior trainer"? It does. Because he doesn't have any formal education in canine behaviour, welfare and science. You know this. You chose him to be your trainer for your study.

"The speed and effectiveness with which the e-collar inhibited the dogs in this study from chasing may justify the limited number of painful stimulations the dogs experienced if the object the dogs were chasing had been something that could directly or indirectly cause them serious injury or death." Or..... OR.... you could not use electric shocks on your dog and potential cause long term damage, fear responses ect.... And use a leash.

Just baffled by this whole study. My high prey drive whippet will go feral for a lure but only when she knows the context for when she's being allowed to chase. She can be at a lure coursing trial for the whole day not worrying about the lure because 1) I keep her at a distance where the stimulus isn't saliant enough and 2) she gets reinforcement for calmly watching from a distance.

“Only 9 of the 19 dogs provided a fecal sample on each study day.” So you didn’t collect the samples you needed for a decent sample size and you didn’t measure baseline cortisol levels either.

“Four from Group A, two of which were removed from analysis because of the exceeded shock levels." I’m sorry but why are you shocking a dog over 20 times and then excluding them from the analysis? Does that not demonstrate that in some cases e collars are ineffective even when you use them excessively? Because prey drive can overcome pain. Ah but of course you don’t want to ruin your biased paper by adding nuance do you?

There is so much amazing about this. It's an archeological museum in 530 BCE or so. Also, the exhibits are labeled in three languages. Also they apparently had replicas on display for some things, much like modern museums do.