Donate
Kayla sits down with Sarah Stremming to discuss how she has been handling Barley’s tendency to generalize to other species while working on her wolf scat detection project in Alaska. This is a nerdy one, hang on!
Transcript (AI-Generated)
Sarah Stremming 00:09
Hey everybody. This is a crossover episode between my podcast, CogDog Radio and Kayla Fratt’s podcast, the K9Conservationists podcast. I’m Sarah, she/her, certified dog behavior consultant in case you don’t know me, and I’m joined by:
Kayla Fratt 00:26
Kayla Fratt, also she/her and also certified dog behavior consultant. I run K9Conservationists along with two lovely co-founders, and I’m a PhD student at Oregon State University.
Sarah Stremming 00:37
And Kayla and I were talking the other day about a problem that she was having with her conservation detection dog barley, and also what the solution was. So, Kayla, let’s start with you give me the background and setting people absolutely understand what it is you and barley are doing and what the problem was.
Kayla Fratt 00:58
Yeah, definitely. So Barley and I have the coolest job in the world, which is that I get to train him as a detection dog. But instead of finding drugs or missing people, or you know any number of other things that working dogs will do, Barley finds biological samples for conservation projects.
Kayla Fratt 01:18
So for my PhD, he and I, and my younger dog Niffler should be joining us next year as well, our finding wolf scats up in southeast Alaska, where we are looking at kind of the diet and movement of these wolves across a huge island archipelago in southeast Alaska. So we spent a couple of months prior to coming up to Alaska training barley to find wolf scat samples, and then also did a little bit of training to ensure that he was not going to find dog scat samples. And we got up here and everything was going really well. But Barley, as my listeners will know tends to be what we call a generalizer, a gambler, a guesser. He tends to err on the side of once he’s trained on a cue on an olfactory cue to kind of be a little liberal with what that may mean, he will source and alert to a variety of things that might not be exactly what I initially taught him. In a lot of cases, this is really helpful. Because in a lot of cases, we might be training the dogs on like three or four examples, because that’s what we’re able to get when we’re working with threatened or endangered species. And those might not represent the whole breadth of you know, hormone status diet, decomposition of those scouts. So it’s nice that he tends to, you know, guess a little bit is helpful. And sometimes he guesses a little bit too much, he can be a little bit too liberal.
Kayla Fratt 02:39
So when we got up here to Alaska, he was not only finding all of the wolf’s got samples, which was great, but he was also finding martin scats, and martins are kind of these little, mostly arboreal mustelids, they’re super duper cute, I think they eat squirrels and kind of look a little bit like squirrels, but in my opinion, like a squirrel, cat, sort of dealio, and there are a lot more of them on the landscape than there are wolves. So it can be kind of exhausting for him. And, and for me, if on any given day, we’re probably going for three or more hours and doing these big searches, if he’s finding all of these other things that we don’t really need him too, that can be kind of exhausting and frustrating for us. So that’s where we were in the first couple weeks.
Sarah Stremming 03:23
And to clarify and have everybody up to speed. When you when he alerts on a scat, you then need to approach and you can actually visually identify if he’s right or wrong. Once you approach what in this particular example is not true.
Kayla Fratt 03:43
Yes, yeah. And like, I would say, 90% of cases, it’s pretty straightforward to tell the difference between a mustelid scat and like a martin scat and a wolf scat. The wolf scats, they should be the same diameter as dog, or even bigger. So versus the mustelid scats are probably about the same diameter as your pinky. And they’re really twisty. So there are some cases when they’re really degraded, that it’s hard to visually tell the difference. But most of the time, it’s really, okay.
Sarah Stremming 04:11
And so I think that’s important just for us, when we get into the training conversation, for everybody to understand that you his alert does not look different. So you don’t know if he’s right or wrong until you get up there and visually look at it, but you can tell once you visually look at it.
Kayla Fratt 04:26
Correct, which is a huge difference. I think as well particularly like for our detection dog folks who are listening, you know, we’re not looking for something that’s clandestine, or trying to identify mold or fungus or something where you’re just not going to be able to see it. And in that case, we would have probably done barleys training really differently to prepare him for this. And quite honestly, I’m not sure probably wouldn’t be the right dog for those sorts of projects. I would use nuclear for that because niffler tends to err on the side of only taking his training very literally. That’s and that’s something we’ve talked about on the podcast before.
Sarah Stremming 05:00
I’ve heard you talk about that on the markets. And I think it’s fascinating. And so that might be something that I have you come back to CogDog Radio, to kind of dig deeper into at a later date, because I think it’s so interesting. And I think it’s something that shows up in big ways in the detection world because of the fact that they know more about this than us. And so we observe these really interesting patterns versus these kind of very contrived things we teach them to do for sports, for instance, are, you might not actually identify as many of those differences as clearly. And that’s something you and I are gonna talk about that later. We’re not going to rabbit hole right now, because we could rabbit hole right now. But let’s talk about that at a later date.
Sarah Stremming 05:37
So, circling back, here’s the text I received from you the other day, you said, “Have I told you about Barley’s, huge milk, not water, win?” And so I need to explain what that means. Yes, that’s about something that I say, which is basically providing a dog with a contingency that is not scary, not painful, but wrong, not the contingency they were seeking. So not scary, not painful, but not the thing they wanted. And the reason it’s milk, not water is because the example that I use is that because everybody’s experienced this, is that when you are in a public restroom, and you wave your hand under the faucet, and the water just doesn’t come out. That’s your desired contingency, not happening, nothing is occurring.
Sarah Stremming 06:23
And that’s akin to the way that a lot of trainers respond to unwanted behaviors, they just give nothing, right, you’re trying to just ignore the response. And just spoiler alert, that doesn’t work a lot of the time. Like just we believe that it works. And it’s fake. Like it just I don’t know about you, Kayla, but like it’s never worked for me. And it could just be that I’m too impatient. And so I don’t let it work. But well, we’ll just put that aside, if I’m trying to punish the behavior of or extinguish the behavior of you waving your hand under the faucet, if I’m trying to get rid of that behavior to stop using the the jargon, if I’m trying to get rid of that behavior. I have options here. I can spew I can make hot oil come out. Now you’re burned. Now that was very painful, startling, bad, you’re not going to do that again. And also, I’m going to get behavioral fallout, if you know that I did it, we’re going to have a problem, you’re going to maybe be scared of faucets, like it’s, you’re not gonna wash your hands again, basically, which is that’s not the behavior I was trying to get rid of. Right. So that’s an option. Or like, if I just made milk come out, if milk came out of the faucet, you would go, that’s weird. That’s not what I wanted. And there wouldn’t be behavioral fallout, necessarily, and the behavior would be gone.
Sarah Stremming 07:47
And so I refer to this as kind of being the milk so when your dog is seeking something, and you don’t like the behavior they’re using to seek it, rather than ignoring it, deliver on a different contingency, this is the human is broken protocol. Basically, the human does not work, the human doesn’t understand what I wanted. So it’s kind of like the example I use when I recorded this in a podcast years ago. And the example I used was my dog is yelling at me to make her dinner faster. And me stopping and petting her and telling her she’s when she does that, which is not scary to her, and not painful to her, and also very much not what she wants in that moment, and is something she wants other times, so it’s not flat, it’s not universally aversive. So you texted me and said that barley had a milk not water, when, which means that you kind of use this idea. So now, tell me what, tell everybody what you did.
Kayla Fratt 08:44
Yeah, and I think it’s gonna be really fun to dive into what you could call this because I was thinking milk about water when I came up with this idea to test. But I think there’s a lot of other things you could call this as well. So what we started with is okay, you know, I’m a PhD student, one of my job’s is to mentor undergrads. So I could just collect these marten scouts. And then I can have a my own baby undergrad. And they can process those martin scats that Barley finds that maybe we figure something interesting out about the distribution or diet or something of these markets. So that’s cool. That was kind of the first step. I was like, Okay, great. We’ll just keep rewarding him for martin actually.
Kayla Fratt 09:26
Well, my very first step was to not reward him at all was to kind of go with extinction. I tried that for a couple days. And it wasn’t really working very quickly. I have done an extinction protocol with detection dogs who are making incorrect alerts, unproductive alerts, false alerts, whatever you want to call it. But that was like a six week training protocol where we had a ton of known samples. We stopped everything else and just went to the drawing board and did an extinction protocol. I didn’t want to have to do that with Barley here.
Kayla Fratt 09:54
So we first did a little bit of extinction. Then I was like, Well, okay, I guess we just could collect the Smart scouts that’s not the end of the world. But that was starting to slow us down quite a bit. And one of the other things that we run into is physically, we reward the dogs generally with ballplayer in the field, toy play of some sort. Sometimes it’s tug, sometimes it’s a Frisbee, whatever it is. That means that if the dogs are finding like 20, 40, 60, 80 scats in a day, even if you’re being pretty conservative with your toy play, that’s adding a lot physically on to their day when again, Barley is averaging like 14 ish kilometers a day when we’re in the field. Anyway. Yeah, they put their bodies through a lot. And we tried to make sure that they’re ready for it.
Kayla Fratt 10:41
But anyway, so it’s like, okay, that’s, that’s, that’s not quite working either. And again, then he was like, great, I went from kind of guessing to now I’m actively finding scats for you a lot more. So we kind of had an increase in response to that, which is fine. And luckily, Barley and I have worked together enough for long enough that this whole experience and as far as I could tell, was not that frustrating for him. So then what we started was okay, I am going to start giving him a couple pieces of kibble for every martin scat, he finds. And then he continues getting his ball like normal for wolf scat, and started doing that. And within a couple of days, his response to martin scat decreased pretty dramatically. It has not completely extinguished, which I actually think is really interesting. And I actually like, which we can talk about that in a moment. But um, that’s basically the procedure. That’s kind of it, you know, we just swapped out, if I identify that he’s found a martin scat, he gets a couple pieces of kibble, if it is a wolf scat, he gets his ball. That’s it.
11:40
And so again, this is the reason I asked you this, you walk all the way up, you look at it, and then you either give kibble or you give the ball? Yes. Is he ever totally wrong? Does he ever identify something else that you’re just like, I don’t know what to do?
Kayla Fratt 11:54
Yeah, yeah. Every once in a while, he he does, he tends to bring us to weird dead things, which, you know, he’s trying to find a burden back carcasses. So not totally wrong. But like, he’s brought us to a couple of dead toads, which it’s like, it was probably don’t smell all that similar. You know, I don’t know if he’s ever brought us to something like completely random, that I couldn’t place he did. We dropped a Sharpie a couple of weeks ago. And when we were going back and backtracking to try to find our Sharpie, he didn’t stop and an alert to that. But I that had our odor all over it. And he’s never done that with any other human origin item.
12:35
Okay. And the reason I asked you that is because when we’re doing something like this, we need to account for kind of all the possible responses. It sounds as though it is alerting to Martin Scott alerting to wolf scat. And then the other responses are so infrequent that they don’t they so and that’s not worried about the others. That’s important for everybody to note, like from a behavioral sense, is that anytime you’re having a response that is basically negligible, because it just doesn’t happen often enough to matter. Your response to it also doesn’t matter. And in sport dogs, which is one of that’s my kind of hobbyist part of dog training that I come in, on, if it’s a negligible error, but it costs you something like a trophy, people tend to really fixate on it, and they cause problems for themselves. So that’s just a hit for everybody to think about. Okay, so you had two responses, and then you started to give to different contingencies. So you started to say, if you what you have identified as Martin, you will get kibble. And if what you have identified as Wolf, you will get the ball. And you basically right away saw a huge decrease in alerting to Martin.
Kayla Fratt 13:52
Yeah, yeah. And I wish I was keeping better dog training records at this point. And not just you know, we have all of our fields data sheets. And I wasn’t keeping track of this, because I think it would be an interesting case study had I actually been really tracking all of his alerts. But I but I wasn’t. But yeah, really, really quickly. And I think why is because it is just clear. You know, it is just, you know, he is getting a response from me that he understands I am acknowledging what he has done and like interacting with him, and then we can restart the search versus kind of trying to pull him up out of an alert and convince him to start searching again, without anything is a little bit more of a little bit more of a discussion between the two of us.
14:39
For sure. And so let’s dig it. I want to dig into what’s going on with this. Yeah. Because it will be easy for us to just say, well, then clearly the kibble in this scenario is punishing. And it’s easy for us to say that. I think by definition We could say that because there’s been a decrease in that response due to the contingency that you have delivered. But I think it’s more interesting than that. Like, and here’s this is the thing about the quadrants. When we say punishment, we’re talking about a quadrant of operant conditioning. And I tend not to talk about the quadrants. And the reason I don’t is because they don’t work in isolation. There’s never only one. And when we talk quadrants, we’re almost always talking about one for some reason. And so when that’s just not how life works.
Sarah Stremming 15:31
So, yeah, we could say that the addition of kibble was a positive punisher essentially, we could also say, and I don’t think this is true, but I think people would say, the removal of the ball could might be operating as negative Punisher. I don’t think I don’t think that’s happening, because you’re not actually removing it. It’s whether you deliver it or not, but I think people would think that way. So it’s important for us to talk about, by definition, the behavior did decrease with this added contingency. And so yes, there’s much more going on here. Here’s what’s more interesting than that, I think, is that if you think of the entire chain of what’s going on barleys doing a whole lot of stuff, in order to land on the final link in the chain, which is playing ball. Yeah, that’s the final link in the chain. And when we talk behavior chains, each link is reinforced by the next link, right? So the ball, being the final link in the chain is reinforcing all the way back to the first link in the chain, which is whatever he’s doing when he first starts searching. Gosh, we could speculate all day about what the searching part is unlike does that mean that then he’s actually changed that first link in the chain? I don’t think he has, I think there’s a place in there, where the chain like changes into one contingency or the other. And so I think that I’m gonna put a pin in that and just kind of say, if we think about this as behavior change, now we’re thinking of something that’s more interesting than just punishment.
Sarah Stremming 17:14
But what I want to ask you is, when you try to just withhold reinforcement for the martin alerts, that is failing, that’s going to fail miserably. And what might be interesting there to explore, is that I think the reason that would fail, because technically it shouldn’t, if it’s consistent. Technically it shouldn’t, if it’s consistent, and technically it shouldn’t if opportunity for wolf identification is plentiful. So that’s kind of one thing. That’s that’s a variable you don’t have control over. Yes. But the reason it’s not is because of the all the reinforcement that happened in kind of the second half of that chain, basically, yeah, whether or not you delivered reinforcement at the end.
Kayla Fratt 18:01
Yes, yeah. And I want to come back to kind of the place in the chain that this is affecting now, because I have a couple different guesses based on some observations, but we’ll we’ll put a pin in that. So I agree. And I think like maybe can I tell a story that I think might illustrate why withholding reinforcement doesn’t seem to work super well, in this particular case.
Kayla Fratt 18:23
So our my podcast listeners will be somewhat familiar with the drink the great chico sapote, incident of 2022, I think it would have been, which was barley, was working with me in Guatemala, and our job there was to find the scat of a bunch of different carnivores there. And this project was a great example of where it was nice to have a dog who generalizes easily because our training samples were only from three or four species, but we actually wanted him to find, I think seven or eight total. So we were hoping that if we trained him on Ocelot scats, they’re close enough related to margay that he would also jump over and find mark a scouts and he did.
Kayla Fratt 19:03
However, then we had a day where our guides out in the forest picked a forest group for us called the chico sapote and shared it with us it kind of looks like an apricot, tasted like an apricot. So we ate a little bit and I give a little slice of it to Barley, because he’s my buddy we’re in the field and we’re all having a snack. And then the next day, Barley starts finding these things everywhere. And it was just one of those like oh my god we have like six or eight different field staff with us that are all following barley around we’re all like just glued to this dogs every move and every change of pace to find what we’re looking for. And he just kept bringing us to these chico sapotes. And throughout the day, he was getting more and more specific as we were extinguishing the behavior.
Kayla Fratt 19:49
So I was walking up to him confirming I was at chico sapotes to because I often in the undergrowth there couldn’t see what he was alerting to until I was quite close to him even though obviously these fruits don’t look much like scat, and a lot of times cats will bury their scouts under leaf litter. So very hard to see until you’re quite close, I was walking up to him, confirming what it was saying no search on, which is probably some sort of no reward marker something reviewing the behavior, you can, we can, we can parse that out if we want to. And he was over the day getting more and more specific. So he went from finding all of the chico sapotes to only finding open chico sapotes, to only finding ones that had been recently broken a little bit, it’s starting to look like he was finding the ones that spider monkeys had taken a bite out of and dropped. But it wasn’t quite going away.
Kayla Fratt 20:36
And the thing that got it to fully go away was that I started asking my tech, Toni, and our partner Ellen, to walk up to Barley first check whether or not he had a scat, and then I was approaching. So I think what was happening is that my approach with the ball on the pouch on my hip was enough of like an anticipatory dopamine dump or whatever. I can’t say for sure that that’s what it was, because I wasn’t checking his hormones out minute by minute basis, that it was maintaining the behavior. And as soon as someone else was doing the checking that extinguish it really quickly, we only had one day of this problem. And the next day we were back to back in business. Yeah. Does that answer your question? I hope what I think it’s illustrative.
21:20
No, it’s a perfect story. Because what I want to get at is exactly that is like, what’s interesting here is I’m even picturing like a chain, that is one color and it like changes color at a point. And like when the chain changes the different color like now, you’re it’s almost like there’s an upward stretch of chain. And then when the dog feels like reward is imminent, it’s a downward stretch of chain. And if you’re on that, you are reinforcing whatever just happened, regardless of whether you wanted to or not. And we talk sometimes with these about these dogs being, you know, having that hit of dopamine, or whatever it is, occur when that downward slope of the chain starts to occur. You know, those dogs like the dogs that we like for work, they’re in a place a lot of importance on that downward slope of the chain. Yeah, in such a way that, you know, dogs that we wouldn’t necessarily select for work. It’s almost like their downward slope of the chain is that it’s shorter links, like three links of chain versus like, yeah, these these work-y individuals. For those who don’t know, Barley is a Border Collie, of course. So the those super work-y individuals have like almost, they just have a super long stretch of chain. And so yeah, to be even more careful about understanding when your reward event has been done. Because it is this like entire link of the chain.
Sarah Stremming 22:55
So it’s I’m I’m even thinking about Felix and his competitive obedience stuff. And his dumbbell retrieve which the dumbbell retrieve is just a hunk of plastic, the dog has to fetch, sometimes on the ground, and sometimes over a jump, and he thinks it’s basically the best thing that’s ever happened. And as soon as he believes the dumbbell is in play, his behaviors begin to change. So me, me even picking a dumbbell up basically begins that next stretch of chain no matter what I do with it after that. And so being aware of that was really necessary for me to solve some of his problems that he was having in the ring in regards to his dumbbell feelings. So really, interestingly, you had a different problem. But not Yeah, it was a different problem. Because it Yeah, wasn’t a generalization problem. It was just, I’m finding this too. It was just like, what if I also found these it’s almost like the Sharpie find, it’s almost like just like, this is a really educated guess about the fact that you might like this like that’s yes. And again, that’s the type of dog we’re talking about too. Right. And so being able to not initiate that downward slope of chain is what let you solve that.
Kayla Fratt 24:18
Yeah, yeah. And I believe I had Kim Brophy in my ear when I was like, Okay, we’re just gonna send Ellen and Toni. I think it was Kim Brophy that I had most recently heard or maybe she was the first one who brought up the idea of this anticipatory dopamine dump. I know that she is not the, you know, the the only person who has ever said this, and certainly not the first but she was the one that I I was thinking of, and yeah, we actually we joke a little bit that our dog union likes to try particularly Barley is the head of the dog union, obviously, and that they are constantly trying to add duties for which they should be getting paid, as well. And I think that’s, you know, that’s what happens when once every six months he finds a dead toad. And then maybe he finds three dead toads in one day. And he’s just like the union is requesting this to be added to our duties and we get fair pay for fair work.
25:13
Well, and that attaches to, again, him being, you label us a couple of different labels. Yeah, I know what you mean, because I do listen to your podcast, but it’s like, you call it a gambler. We’ve called him very liberal. Like, it’s basically a dog who’s like, he’s gonna take a shot. If there’s a shot, he’s taking it, right. Yes. And I bet a lot of the time you do reinforce it, whether you pay for it or not, because you initiate the downward slope of the chain to approach to see whatever your thing is.
Kayla Fratt 25:40
Yeah, because I have to. And yeah, that’s why this is an ongoing thing. And again, goes back to this idea of, if we were approached to work on a target that was invisible, I don’t think he would be the dog that I would say, of the five dogs we have available within our program. Barley shouldn’t be on this project. He’s got a lot of other strengths, but not hyper specificity is not one,
26:04
Well, kind of like he, it was a good choice for you to use one kind of cat scat to help him learn how to find another kind, like, I mean, it’s like, he’s great for that. He’s not great for certain other things. And that’s, that’s interesting. And that’s just that’s kind of about selection for task. But this is why though just not giving him the ball, when you get there and identify it will fail. Yes, yeah, the entire downward length of chain still happened except for that last part. And I think that inexperienced trainers who have maybe not had their hands in training, like maybe people who are still in an education kind of role, or phase of learning, there’s no book you’re gonna read this in, you have to go out there and get your hands dirty trains and dogs to start to understand these things. And there are dogs, for whom that that downward length of chain is so short, that withholding reward for those certain alerts would probably get rid of those alerts. But I would argue, Kayla, that for those dogs, they would stop searching altogether.
Kayla Fratt 27:11
Yeah. And that is, I mean, you asked me what is often done in the detection world when dealing with this sort of problem. And I think one of the things that is a little surprising maybe if you’re not in the working dog world, because I think a lot of people are aware of the fact that the working dog world can culturally be a lot more, you know, balanced, heavy handed, whatever you want to call it. I don’t know if I’ve ever heard anyone suggest punishing a dog for a false alert, like, like I must have, there’s got to be someone out there who says it. But I don’t know if I’ve ever heard it as advice. And that is because the potential for Fallout where you’re teaching the dog not to source odor or not to alert rather than the dog learning not to alert to this given thing is so high that it’s just it’s not a it’s not a smart place for punishment, even if punishment is on the table for you as a trainer.
Sarah Stremming 28:05
I want to dig into that a little bit more. Yeah. I wasn’t surprised to hear you say that. Even in circles where heavy handed training is quite the norm. I’m thinking police, I’m thinking military. It’s not the norm here for these certain responses. And I wasn’t surprised to hear you say that. And I think that almost is its own episode. So like, let’s not super rabbit hole into why punishment works better some places and not others or not even just punishment, like putting that aside, but corrections like being harsh with them. So much of it is about, we don’t understand this, the way that they can understand this and we never can. And so when you come in with some kind of harsh response to them, when they actually did almost everything right, all you do is diminish, I think their willingness to try next time in a way that so withholding from a dog who’s downward length of chain is long, isn’t going to do much for you, it’s probably not going to get rid of that behavior. And it’s probably also not going to damage your searching. But if you walked up there and you like gave him a collar pop or something. Now what you’ve done is you’ve attached potential other contingencies to the downward length of chain causing stress causing anticipation. If you’ve ever been in a situation where you’re not sure how it’s gonna go and it could go great and it could go terrible.
Kayla Fratt 28:47
You look when you get a text that says Can we talk?
29:40
It’s exactly right when you get that text you go bleh, your initial risk. I think it’s so normal so healthy for everyone to go out. I don’t know Right? Which is why just you guys send like a smiley emoji send like a be like, hey, it’s not bad because we don’t If
Kayla Fratt 30:02
I do this sometimes when I call friends and then hang up, because I realized that I’m in the wrong time zone, and I’m calling them at 11pm. So then I immediately send them a text saying, not an emergency. I’m fine. I just wanted to talk. And then I realized that it was 8pm for me and 11pm for you have a good night.
30:18
Exactly. We do this with each other socially all the time, because we really are aware of that yucky feeling of having your Downward length of chain potentially have a fork in it. When you introduce that fork, you create behavioral fallout. Yeah, in a way that you don’t usually, if you just leave off the last chain, which is what you’re talking about when you withhold.
Kayla Fratt 30:41
Yeah, although it’s interesting. Gosh, this is just where it gets so dog dependent, too, because what I see with Barley when I just withhold is a lot more frustration sort of behaviors. So he often if I tell him, tell him to just continue searching, and that is my only response, he will often go back and knows the target. Again, he might alert again, you might circle back, not a ton of times, but like he does kind of like I would I would call like he continues to ask the question. He’s like, Are you sure. And then Niffler, my younger dog is a little bit more likely to even if it’s just a hey, but not right, let’s let’s search on or go search or, you know, whatever it is that I say he is a little bit more likely to you see a decrease in energy, you see a decrease in enthusiasm when he goes on, he’s a little bit more likely to just get into quote, take that withholding, and and you see a decrease in the remainder of the chain.
31:36
And that’s yes, I like this. This is the individual and I just had like three more 300 more thoughts. But like, you’re still not creating, you’re not creating the kind of fallout that makes you that makes people stop doing this altogether, the way that it made totally people probably tried the correction route. They still I’m sure because the fallout was too big to often.
Kayla Fratt 32:01
I’m sure rookie cops still do it, you know, every year? Because it is the reasonable thing to try. If you know very little about animal behavior, like, yeah, I probably would have tried it if I tried to train a detection dog when I was 19.
32:16
If it’s in your repertoire of skills, it’s a reasonable thing to try. So talking about what qualifies as fallout, here’s the problem with inducing frustration, how often do they work harder when they’re frustrated? Again, if they’re the kind of dog that is reinforced by that downward length, they are the kind of dog also that works harder when you withhold that final piece of the chain. People use it to their advantage all the time. And they break dogs like Niffler doing it. And then they blame dogs like Niffler because he doesn’t have the work ethic, right? I’m thinking about my Icelandic sheepdog, Rayya, and the fact that if I withheld reinforcement for her getting almost the entire chain, right? She would be like, Well, I have better things to do. I knew I had better ideas than you. And she that’s why, you know, that’s why she’s not signing up to work for you, Kayla, cuz she’s got the stability, but she just would not be practical, right?
Kayla Fratt 33:18
She has her other hobbies.
Sarah Stremming 33:20
She’s just she has a lot of her own good ideas. Okay, she can she’s an entrepreneur on her own. She doesn’t need to work for anybody else, supporting anybody else’s dream because she has her own dreams. So. But I said, introducing a potential fork in the chain creates very real fallout. But what have you done other than you have produced a fork in the chain by giving the kibble in the ball?
Kayla Fratt 33:41
Yeah, yeah. And it’s somewhere in you know, going way back to the quadrants, you know, it’s like, okay, we’re decreasing the marten finding behavior, but we are still we’re still providing an appetitive stimuli, you know, we’re still giving him food for it. And the way that I would anthropomorphize and try to get into Barley’s head about it is that he, it seems like he is now deciding whether or not he wants a snack and whether or not that behavior is worth it to him. And he is kind of toggling the Martins on and off throughout the search, because what we have seen over because it’s now been about a month that we’ve been living in this contingency for him is that generally he finds a couple of marten scats early in the day. And then generally, if the wolf scats are really thin, so if we’re going more than 15 or 30 minutes in between finds, he’s then starting to find marten scats for us again.
Kayla Fratt 34:34
So I think there’s some amount of like a cost-benefit analysis going on in his head about whether it’s worth it and if the wolf scats are plentiful, why would he slow down for marten scat, because he’s got a jackpot waiting for him at the wolf scat. And I would imagine, I can’t say this one for sure. But the wolf scats are also probably a little easier to find. They’re bigger. I would imagine the odor cone is larger and more salient to him. So also, why would you work harder for less pay, but at least that contingency is really clear to him so he can make that cost benefit analysis. And the last thing I’ll say on that before, I know you’ve got 34 things running through your head now is, I love this for us, because I can continue my undergrad projects.
Kayla Fratt 35:18
Now our protocol is we, we gather the first five, intact marten scats that we find on a given day. So I can still do that if we want. And I know he’s still searching. So if we go a half hour without finding a wolf scat, or goodness forbid, like three hours without a wolf scat, but he’s continuing to find a marten scat, I have evidence that he’s continuing to work, he is still on task. There’s just nothing there. And I love having that data. For us. It’s so helpful. So I’m, I’m pretty sold on how this has worked for Barley, we’ll see, you know, going forward, how I might deal with other dogs or other problems. But I, there’s a lot of upsides to how it’s worked out, and I wasn’t sure it would.
36:01
And this is really valuable. And this is why I am not sure it can qualify as a punisher. Because the behavior is still intact. Yeah, it is simply that you’ve attached the behavior to a contingency that he knows like you sense clarity. There is one chain, which is so let it which is search and find marten scat, alert to marten sca,t receive kibble. Okay, so that’s one chain, very simplified. The other one is the entire same thing except it’s Wolf. And then it’s receivable. So he’s really aware that there are two pathways to reinforcement here. And it’s your motivating operations I play that are going to as well as some of the antecedents because I think probably if both Marten and Wolf are available, he’s going to choose Wolf, even if he’s tired. Am I wrong about that?
Kayla Fratt 37:00
No, I bet. Yeah, that would be my understanding of it.
Sarah Stremming 37:03
Because he’s going to choose ball, even if he’s tired, because that’s who he is as a person. Yeah. Yeah. Same with Felix going to choose ball over food, right, no matter what.
Kayla Fratt 37:14
Yeah, and that is a good point for our listeners. Because if you’re listening and you got a dog who’s a little bit more even on reinforcer choice, you probably are not, your mileage may vary. This working so well, because that that discrepancy between what he’s getting is huge and do not get me wrong barley loves food. Barley is biggest behavior problem as a person is dumpster diving and trash riding, like he is a hog, and the ball is 10,000 times better, no matter how many times he’s had it that day. And that’s not the case for all dogs.
37:47
So, oh, my gosh, we can talk for the next 10 hours, because the discrepancy between the two and contingencies also is a big deal. If there isn’t a big gap between those you will not see differentiation in in the behavior you just want.
Kayla Fratt 38:04
I believe that if I was giving him salmon skin instead of kibble, we would be getting a much higher rate of response to marten still.
38:12
Right, because that’s closer to the ball or like a less preferred toy versus the ball, right. So if your target instead of fetch, yeah, if they’re too close together, you’re not going to get a difference. So the action that think of it again as a chain, and think of the final behavior in that chain needing to be different enough, from the other behavior, for it to make a difference, pack a punch, had them care about it. If you were training only with food, you would need to have, you know, the eating behavior is the same. So you need to have what they’re eating be ready different for this to pack a punch.
Sarah Stremming 38:52
So when I initially thought this was a protocol, I called it never wrong, sometimes more right. Basically, meaning nothing that you do is going to be wrong, everything you do is gonna get payment. But what I actually want you to do is going to get better payment, and it does not succeed in punishing those behaviors that are getting the lesser payment unless the lesser payment is actually aversive to the dog.
Sarah Stremming 39:19
So if we just kind of keep that in your head as I kind of explained what it is, I might have somebody do this in like a shaping protocol. One of the really good ones this works for is trained it off to back up to a foot target. Most dogs will at some point want to turn back around and walk back to the foot target rather than rather than keep walking backwards. And if you introduce this sagevue kibble when you turn around and hit the target, but I give you steak when you back up to the target. It works pretty nicely. Still, motivating operations come into play and the dog is fatigued because hey, you’re trying to teach back up and you’re having a hard time, recognize how hard it is physically for the dog to do, keep your sessions very short, they start to fatigue, they start to turn around more anyway, because walking backwards now hard for them again, that cost benefit analysis that the dog is kind of doing on the fly.
Sarah Stremming 40:13
In this exercise in a workshop, friend of mine who’s probably listening and knows who she is, was training a malinois to walk back. And she had decided she was going to use these hunks of freeze dried raw for the right, you got it right treat. And she had the very clever idea of using banana chips as the, you know, you’re not wrong. But here’s your here’s your consolation prize. And when she gave that malinois a banana chip, I mean, that’s odd looked at her like, you better sleep with one eye open like it was such a stark moments of the dog being angered actually, by that response. So you also want to make sure that your other response is not actually punishing and actually aversive. I think that I mean, that dog did not turn around and hit the target. Again, she ate a banana chip. And she gave the look of death to her person. And then she deliberately walked backwards and hit the target. Like she was like, How dare you? So that’s not smart, either. And they’re so that’s hard to do. There has to be discrepancy. Yeah. And they both still need to be repetitive. Yeah.
Kayla Fratt 41:23
Yeah. And depending on who your dog is, and how they interact with heat, or fatigue, or whatever, you know, how many different reinforcers are you going to have with you at any given moment, and particularly for us when we have to carry our reinforcers for miles? That could be quite limiting? One of the questions I have for you is do you think that part of why this works is so you know, we’re talking about this, the forking in the behavior chain? I wonder if part of that what’s happening here is we’re just filling in the blank of, hey, so yeah, if you if you want to test this, if you want to do anything, right, but you’re ending on a marten scat, instead of ending at a wolf scat, this is just clearly what’s going to happen here. And now they understand that, versus again, when we’re kind of giving nothing dogs like Barley who tend to guess and dogs who care so much about their reinforcers, where it makes sense to them to want to continue guessing, like until there’s a clear answer of what happens when you do that. It’s really, I wonder if it’s harder for them to really make the decision about whether or not it’s worth it? Does that make sense as a question?
42:33
I think so. Let me answer it. And then if I didn’t answer it, then it didn’t make sense. So I’ll answer it with a thought I have, which is that if you had if you had from the beginning, given Barley kibble for marten alerts, I think your marten alerts would be even less and potentially gone. I think we cannot discount history of reinforcement. When we’re having these conversations. It isn’t only about any given scenario, or instance. He’s now got hundreds of alerts on. I mean, I’m just guessing, like he’s got so many alerts under his belt on just this project alone. Yeah. And I don’t know how many marten alerts got a ball in the beginning. But that is still in his history. If we, you know, if the matching law is coming into play here, which of course it would be, then it’s going to take an equal amount of payment with kibble to kind of start to rewrite that history for him.
Sarah Stremming 43:46
And so if you had been very clear in your head, what your contingencies would be from the very beginning, I think a few things would have happened. One is I think you wouldn’t be getting as much marten alerting as you are now. And I don’t think you would have had a period of it being a problem at all. Yeah. And so I that’s something to think about for later. But the other thing is that you may have seen some other behavioral fallout you didn’t like, probably not in Barley because he will search for a very long time to get a ball. But you could have seen other behavioral fallout. So when I taught my “never wrong sometimes more right,” I was very clear about this is best for situations where the learner does not trust that reinforcement is imminent. So I was kind of designed for the dogs. I had a hard time hanging out through the fumbling of the novice trainer that was trying to teach them versus like that isn’t the situation here. And so it’s more just like no this. There are two chains available to you here. And probably you’re going to shoot for the chain that ends in the ball because I know you and I know you as a learner and that’s the thing that you asked Should you want? And if you had showed him that from the beginning, I don’t think would have happened. Did I answer your question?
Kayla Fratt 45:05
Yes. Yeah, I think so. And I think that also leads to I mean, a, now that I know how well this worked for marten, I think this is going to be my go to with him for for any, because this was this will come up again. I know him there, this will not be the last time that he makes a guess. And it won’t be the last time also where he ends up in a situation in which some guesses that he makes get rewarded. And I’m like, Oh my gosh, you brilliant, perfect boy, thank you so much for knowing that when I tried to on ocelot, I also meant margay.
Sarah Stremming 45:38
That’s the thing! You don’t actually want this gone from his repertoire. So you know, still you do want it and I think giving him kibble from the beginning just would have given him clarity, I don’t think it would have gotten rid of his generalization as a searcher.
Kayla Fratt 45:55
And I think, in an ideal world, what I would have done is maybe thought a little bit hard about okay, Kayla, you’ve got a dog who generalizes very easily, maybe rather than just thinking about dog as a discriminatory stimuli that you’re introducing in early training, you need to include more carnivores. And maybe if I had been thinking that way, I would have introduced marten much earlier in training back when we were doing lineups. And back when he was getting the opportunity to compare between the two and get rewarded for one and extinguished for the I mean, it’s not even really extinguishing because he’s just, he’s never been rewarded for it. Right. But I wasn’t I didn’t think of that, at the point that I was starting the training for this program in the way that maybe now i will i honestly I didn’t think that marten was going to be this this much of a problem for him with well, they, you know, they’re they’re quite different.
46:51
It’s interesting that it’s a huge front. I mean, that that is just an interesting thing about just –
Kayla Fratt 46:56
Well, and I think me fumbling around me fumbling around and trying to decide what to do about it. You know, when I had my day of being like, Oh, this isn’t a problem, I’ll just find all the marten says, well, that dug that hole, I think if I had stuck to my guns with extinction, it probably would have gotten to a similar place to where it is now, it probably would have just taken a little bit longer. And I think it would have been more frustrating for both of us. The compromise of giving, giving me something clear to do made the experience of him making these false alerts or unproductive awards, or whatever, less frustrating for me, too, because I could just be like, Oh, here’s a cable, great, let’s go find more, you know, and I knew what to do. And I didn’t have to be frustrated or like trying to coax him into leaving the marten and moving on.
47:43
Right? So you’re actually spending less time to on this less important chain, because you’re just giving it food and moving on because it’s See he understands like milk going into the fossa is a contingency that makes you go oh, I’m going to another fossa like just okay, we’re moving on. Also, like milk is kind of, you know, yeah, technically punishing, because very much not what the person was seeking. But like, I don’t know, what if it was like sparkling water that came out? Like, technically, you could wash your hands? Yeah. But it’s not what you wanted.
Kayla Fratt 48:14
Water that’s just a little bit too cold, like you’re too cold, like, I don’t like that.
48:21
Yeah. But if it’s the only faucet, you’re gonna still wash your hands there. And that feels more like the kibble is like, yeah, there’s only been marten for a minute. So I’m gonna go ahead and alert on that. And I don’t think they’re thinking that complex. But just reinforcement history is such a powerful thing. Yeah. And you and this might be for our eventual later rabbit hole that we go down as far as him being a gambler. And that, that being that type of dog, if you are, it’s kind of like, I would advise you, I’m not advising you. But if I were asked to, I would say I would look at each project as like, is this one in which I want him to generalize? And then here’s my protocol for that, which is basically trust that he will and pay him for doing it. Yeah, versus Is this a project in which I need him to discriminate pretty finely. And so I’m going to spend more time in the discrimination phase with more options, because I know he’s a gambler. And so I’m going to use more options to teach him to be more deliberate in his discrimination on this project.
Kayla Fratt 49:29
Yeah, yes, absolutely. And I think that like when we did we did a project in California about a year ago now where it was a multi carnivore project and we ended up being really lucky for Barley where they were somewhat interested in the dogs finding coyote scat but didn’t want a ton. So we actually had two different dogs on the project and because of who Barley is, I said, Great, we’re gonna let Barley find black bear, puma, bobcat, and coyote, and because of who Scotty is, our newer dog, we’re going to teach Scotty Yes, bobcat, no Coyote, because those two are almost impossible to identify in the field and tell the difference.
Kayla Fratt 50:10
So if I had wanted Barley to just find one of the two, I would have needed to spend a lot of time training him to feel confident. And I would have with known as samples in like a setup, to feel confident that he was actually staying true to what we needed, versus Scotty doesn’t seem to have the same kind of extreme level of gambling, guessing, liberation, whatever we want to call it. And it was so cool to be able to take Scotty out on these trails, where you could see just bombs of scat everywhere, and Scotty was going from one to the next to the next, and then just finding the bobcat and alerting to it and it would have taken us even though the scats were visible, it would have taken us so long to try to decide which ones we wanted without him doing it. And again, now I keep trying to bring in the different, the dog differences. And I know that’s not the point here, but it’s just fascinating and knowing how that fits with what the project goals are. And therefore how that influences your training and what’s available, you know, if you don’t have all of your discriminatory samples available, um, because they’re a protected species, because they’re invasive or you know, permits, whatever it is, then it’s nice to be able to pick a dog whose tendencies align with what you need the training to do. Or if you’re in a perfect world, then you can do all the training you want. And I have confidence that if we needed to barley could get there, but it would be a lot harder and would require a lot more training samples. Does that mean, I think that makes sense?
51:39
I think it makes perfect sense. And I think that it’s okay to keep coming back to like there are differences in the individual because of course there are and that really matters here. I think there are people who would say that there would be dogs that would be that would have zero fallout from even having a correction. Like they would just say, No, it would be fine. And I think that we kind of, you know, in that particular example, in this particular situation, the data would kind of point to them being wrong. However, individual really matters because it you’re basically saying what is the search task? And then who are you? And then how am I building a plan for you?
Kayla Fratt 52:21
Yes. And how feasible are our plans, based on what we can acquire, which is something that we run into, I think in the conservation dog world more than most other detection trainers, if you’re a human remains detection dog trainer, usually your problem is getting your hands on the human remains. That is hard, but getting your hands on pig, or something else like that doesn’t tend to be as far as I understand it nearly as challenging. And you know, if you’re employed within a police department or the military, you have these things available and being provided to you in a way that are not always like I have more than once have needed to take to Twitter and dedicate don’t do it like calling all bear biologists does anyone have a freezer full of black bear? Because I need some to train on for an upcoming project. You know, we’ve we’ve had to call zoos and try to get our hands on stuff. So sometimes for us getting those discriminatory stimuli into our possession to actually do the training is a lot harder than we would like, like the dog training is not always the hard part for us.
53:25
Yeah, and that’s also a factor. Like, that also matters. And I think that people you know, people in the sport world often kind of use that almost as an excuse. They’re like, well, you know, I for instance, this is this might get me into trouble. But here we are. I have a lot of friends who compete in Mondioring, a protection sport. And I have conversations with people in that world regularly about the heavy tool usage that is happening on the protection train dogs, and why it seems to be so heavy and why it’s hard to get away from and that sort of thing. And the biggest. One of the biggest things that I hear come back is about decoy scarcity, the decoys, the guy were in the suit of the dogs, and basically saying if I only have you know, access to this person, this often, then I’m needing to basically cut corners to show clarity.
Sarah Stremming 54:31
And I think where I might get into trouble here is people saying it’s not cutting corners, it’s whatever. But if you are, you know, I am a trainer who understands the use of punishment protocols understands the use of negative reinforcement, honestly very, very well and could write you a protocol for the bytes court thing that did not involve your tools. And it would probably it would require you to have more access to the thing that you are working on, that’s probably true. And so then it’s kind of like, well, now we use this because we only see this person so often like that’s, that’s one of the big conversations I have. And then to relate it to like a world that I am more familiar with. So that, you know, the people were telling me to stay in my lane right now can just be satisfied, let me step back into my lane, which is dog agility, competition, and people kind of lamenting that they don’t have the equipment, maybe they don’t have a dog walk. And so they can’t train a running dog walk because they don’t have a dog. Or they can’t fix their stop dog walk because they don’t have a dog walk and me kind of saying, Well then use your ingenuity and your cleverness.
Sarah Stremming 55:38
And that’s where that’s what you’re talking about right now, is you’re saying, Okay, if we need to train discrimination, then we need to get other scat samples. So who are we going to call to get the samples? And if you can’t get the samples, then what’s the next route you’re going to take? Yeah, right. And just being clever kind of in the production of your plan. And then also, being willing to pivot within the plan is so important to what you do is so important to sport training, but also what my actual job is, which is behavior modification, really important, right? Like not having access, not having access to something can be a problem, or it can push you to other solutions. And it wouldn’t excuse you continuing to try to use a punishment protocol here.
Kayla Fratt 56:27
No. It wouldn’t, it wouldn’t make it a better option. It wouldn’t make it more expedient or clearer. It, you know, it would be okay, maybe my list of options is shorter, but it wouldn’t actually make it a better option.
Sarah Stremming 56:41
Yeah, and so for instance, I was involved in a conversation online, not long ago about search and rescue was a mantrailing, like, live find kind of training situation, talking about the use of an E collar to keep the dog on track, basically. And I, the that, almost that excuse, and I don’t actually want to say excuse, I would like to say the reason that was cited was essentially I can tell when the dollar is starting to go after bird, a squirrel a dog scent something else I can tell when the dog is no longer on scent. I have questions about that. But saying I can tell when the dog has stopped doing the actual work. And this helps me tell the dog to get back to the actual work without redirecting the dog’s attention to me, which is what a positive reinforcement protocol would do. And when I looked at it and dug it apart, it all always comes back to the foundation is the problem with foundation training is the problem or the conditions or the problem.
Sarah Stremming 57:47
So you know, and then they were saying, Well, this is necessary, because we sometimes have to search for hours and find nobody, most of the time we find nobody. Sometimes we have to search in the heat, blah, blah, blah. And so again, it’s conditions and its cleverness, it’s finding ways to utilize either positive reinforcement contingency or another contingency that that does not produce behavioral follow. And so that’s just why I appreciate having these conversations with you is the ability to kind of go, Okay, I need to solve this problem. This isn’t barleys problem to solve. This is my problem to solve. Right? It’s not his problem to solve. It’s yours. And so you then you then putting, yeah, putting your brain into it is really is really important and really good. And I think that if you were going to use some kind of correction, like an E collar, whatever it would be vital for you to see when he decided it was marten and not wolf, and you can’t.
Kayla Fratt 58:44
No. Yeah, there’s there really, I don’t think it’s a way for me to even if I were – Maybe, maybe maybe maybe if we went back to like an indoor training scenario. And I had, you know, a scent wheel setup. Maybe that, but not out in not out in the woods. Yeah. And then actually, so that was something I think I’ve noticed and I, I wrote down, this is now going back probably 40 minutes. But you said something about, you know, we don’t know exactly where this chain is forking on us.
Kayla Fratt 59:18
But one of the things that I think I am seeing, and again, very hard to say for sure. I have seen a bar Lee several times since starting this protocol to do what we would call a nose hook. So he goes from kind of like casual searching behavior. He’s trotting out an even pace, he’s usually kind of got his mouth open. He’s not necessarily actively sending because he’s out there for three hours, his tails up to a nose hook. So he’s kind of sharply turning his mouth so close goes into active sniffing mode. I’ve seen him do that several times now and then do it for like two seconds and then go back to normal searching. So he is his attention is being grabbed by something. And then he’s dismissing it and going back to work. And then the other thing I have also seen him do with marten and I know This one is happening with marten, I can’t say for sure what’s going on with the nose hook, because I don’t know what he smells. But sometimes he will also fully source to marten, and then stand there instead of performing his alert. And not really quite finished the chain, he’ll just kind of like stand there and like, look at me, sometimes wag his tail a little bit, and if I say alright, but if you want to keep searching, he’ll just trot off and keep going. And I’ve done it enough times where then I’ve gone back over and seen what it is. And it’s usually marten.
Kayla Fratt 1:00:29
So there is like this point, potentially at the nose hook, where he’s able to figure out, oh, that caught my attention, but it’s not what I’m looking for. I’m giving up on it. Or he gets there and he realizes, oh, shoot, I did it again. It’s now what I was looking for, really, I don’t really want cable, but also, I made it all the way here. I don’t want to do the workout of lying down. So I’ll just stand here and see whether or not I’m gonna get paid for it. And obviously, I’m putting a lot of words in my dog’s mouth right now. But I spent a lot of time thinking about what words my dog would say. And I think they’re relatively accurate as far as maybe not like he’s not necessarily having these thoughts in his prefrontal cortex the way that we are. But I think like the, the cost benefit contingencies, whatever are leading him in these ways.
1:01:10
I would say so. And I think it’s important to think like that, actually. And I think that the answer is, and especially when it comes to anything sent related, is that we can’t perceive their world of sent with no hope of doing it. We can only guess and theorize about it. And so Theorizing the best that we can then leads us to the plans that we have. And you know, that’s the best we can do. So then that’s where I think we have no business actually trying to utilize any kind of harshness on them. Because, like, yeah, how dare you you don’t even know what’s going on. versus, you know, we could I could make, it’s easier to make an argument for certain other cases where I do feel like I understand what’s going on. In these cases, I think there’s a big reason that our attempts at correction or harshness have failed, whereas they’re not failing everywhere. And that’s not me endorsing, right. They’re not failing everywhere. And they do fail here.
Sarah Stremming 1:02:20
Yeah. So we’ve been talking a while. I think we should wrap it up soon. But I do have What the f is in NRM in the notes. Because that kind of came up in our text conversation. And so I just want to give kind of my take on what that is, because I think a lot of people would say, Well, I’m not going to be harsh, but I am gonna, I am going to indicate to the dog that no reward is happening. When dog trainers say no reward marker, which by the way, so NRM stands for no reward marker, it’s kind of a dog trainer, a jargony term, not a science term. So that’s your first red flag, but it’s supposed to be a signal that tells the dog, they’ve lost their chance for reinforcement. Basically, my perception of it is it is either a conditioned Punisher, which is what that is, by the way, so if it tells the dog reinforcements been removed, that’s a condition negative puncture. Or it’s actually aversive in and of itself, which it usually is, if it’s an or like any noise like that, because they don’t universally kind of don’t like that. Neither do we. That’s why that scary game Operation like buzzes when you turn to the wrong set, right? I mean, just dating myself, traumatized by the game Operation.
Kayla Fratt 1:03:46
Winincing for everyone who can’t see me.
Sarah Stremming 1:03:49
And so not surgeons, either of us, by the way. I’ll point out.
Kayla Fratt 1:03:53
That’s my sister.
1:03:56
See, that’s fine. So anyway, long winded way of saying it’s either an aversive stimulus itself, therefore, it is the punisher. Or it can act as a punisher. Or it’s a conditioned punisher or it’s noise. It’s just noise. Yeah. Well, yeah.
Kayla Fratt 1:04:15
Or, and this probably me like, because I wasn’t necessarily trying to do an overlord marker, but I think generally what I say to my dogs is no, let’s go find more. So it’s like it’s a it’s a preamble. It’s noise. Before I give a real cue and my let’s go find more. It’s a ridiculous cue. I should just say search. But I choose to interact with my dog sometimes in a way as if they speak English and they do not. But yeah, I think it can I could see a world in which it could actually be cute transferred into no becomes the cue for the dog to continue searching and to go find something else. apps. But that would require me being careful about not saying it in a way that causes it to become a Have a punisher or become aversive. And it would probably require almost intentionally performing a cute transfer.
1:05:07
I deliberately teach my dogs that the word no means try something different. I don’t I don’t use it when I’m acquiring new behaviors, because I think I need to be a better trainer than that. But I will use it if I believe that if a dog tries again, they will get it right. So you can have literal cues, that mean try something else. And still, I’d like to call it what it is, which is a cue to try something else or a cue to keep searching. I think if you just sent search, you could leave out the rest of it. I’m also not going to criminalize you saying the whole preamble because I would really too, because that’s how I am. I’m always like, Oh, no, you don’t need don’t do that. Okay, come over here. All right. And then I like given the real cue, right? Like I Yeah, exactly like that. So I think it’s fine.
Kayla Fratt 1:06:04
And actually, now that I’m thinking about it, I’m sorry, I cut you off. Um, sometimes when I say search, it’s because I have not visually located what he has found. And I think he’s alerted to maybe like, sometimes he’s alerting to like three pieces of hair that are held together by a pinhead of fecal matter. And he is correct, but I’m not going to collect it. And if I can’t find it, I still don’t reward him. So he’s still in some amount of variable reinforcement for scats. Because sometimes I can’t be sure. But sometimes what happens is I say No, buddy, let’s go find more. Let’s go search. And he goes back and he sources harder.
Kayla Fratt 1:06:45
And I’ve got a video I can share, send you to, to release with us if you want that’s hilarious of a scat that is under like two inches of moss, like moss has grown up and around fifth scat and completely obscured it. So I think if I were clear about this, and I’m not sure I am, there could be a difference between I see what you have found, I would like to do to find something else. And I don’t see anything. So if you’re really, really sure you can try to convince me. But if you’re not sure, then like let’s keep going. And he does seem to insist, when there is something there. And if I say no search and there isn’t anything there, he believed. And I will then reinforce him for not going on and searching for not leaving it when he was right. And I was wrong. And it just gets really complicated. I have no idea I would have to sit down with a pen and make a diagram.
1:07:41
You’re just still coming back to reinforcement history, he has a history of being reinforced if there’s a certain amount of it there. So if he knows, like he knew under that loss there was there was enough that he doubled down on the response. Yeah, try something different. could translate if the reinforcement history lines up to try harder.
Kayla Fratt 1:08:03
And right now he’s able to split it. Right now I can give the same cue and he can decipher it. And I want to figure out how to keep it that way. Because if he started learning that, yeah, go ahead.
Sarah Stremming 1:08:18
That’s okay. It’s because your cues are what drives the behavior, right? Your reinforcement drives the behavior. So saying, search when he is certain that there’s reinforcement here, because there’s an amount of scat that is attached to consistent reinforcement history, he will translate that into try harder, show me again, show me harder, versus if it is three hairs that he’s found. And you say search, he doesn’t have reinforcement history attached to that. So he goes, Yeah, okay, I’ll go back to searching like it is still your reinforcements driving your behavior, not your cues. Not. Yeah, even though we want it to be the cues.
Kayla Fratt 1:08:57
Yeah, and well, I think some scentwork people would argue, and I think they’re broadly right. I’ll stop bringing more things to the table after this; it’s because odor is our cue in most cases, not the words. So, if the volume or consistency of that odor is something that predicts the reinforcer consistently, he will double down on it. But if it’s a kind of fussy odor, then he won’t. Maybe. Maybe.
Sarah Stremming 1:09:27
Yeah, this is we’re just going to this is just I acknowledge that we could talk forever. This comes back to this is a conversation we had a long time ago about amount of odor versus like and because we’re talking about who we’re talking about handler scent discrimination for the utility exercise in for scent discrimination, utility and obedience. And I was asking you about it because I was troubleshooting Felix’s learning about exercise and I was having a hard time with it. And what it comes down to in Why I think that exercise is hard is because my scent is on everything. What we’re trying to teach him is to find the highest volume of the scent in the pile. And what is possible is that obviously, it’s possible to do, because yeah, there’s all these successful obedience dogs doing it. And also your dogs doing it. You just told me that you know how to do it, if you had, you know, if there was a pile of scat, and then there’s also hairs with you know, molecules like every, like he’s not alerting on those he’s learning on the biggest amount. And when he finds the biggest amount, he will double down until you know, I’m sure. Yeah, and it’s about it’s about reinforcement history. Odor, as cues is still about reinforcement history. Yeah, it’s about attaching reinforcement to odor. And then if your goal is to have them alert only on certain volumes, then that has to be part of your training process as well. discriminate between, you know, don’t tell me about that tube and the scent wheel that has a hair in it. Tell me about this tube and the scent wheel that has, you know, that’s full, right.
Sarah Stremming 1:11:12
So it’s like, yeah, deliberately teaching them that piece two is possible, and could theoretically be done. Yeah. And I think it has been done with explosives, dogs, they do a lot of this sort of stuff.
Kayla Fratt 1:11:24
And there was actually a really fascinating paper out of Nathan Hall’s lab where they found, I can’t remember who this was in collaboration with, but I’m pretty sure it was that of Nathan Hall’s lab, dogs were consistently being trained on, you know, say two kilograms, because that was what was issued from the military store, when you were going out and training with your dogs. And the dogs were not spontaneously generalizing between them. And maybe actually, the problem was, it was like to gram. And then they had a training protocol where they put like two kilograms out, and the dogs were not alerting to it. And they were like, oh, you know, that’s a problem. Because we’ve trained them to find this low concentration, and they weren’t not spontaneously generalizing to the much larger odor of concentration. And then again, these same scientists and I can find this and send it to you, it’s actually a pretty readable paper, then showed that I believe it was about a 10 fold volume discrepancy that the dogs did not seem to spontaneously generalize between. And I’m sure that varies a lot based on like odor dynamics, and volatility and surface area, and all sorts of other things. But there has been a little bit of research into this. And it’s really cool that like, to some degree, the discrimination happens with your training, if you’re consistent about your volumes. But if you wanted to be sure about it, you probably would want to do the discrimination as well.
1:12:39
And all of this to say, to kinda round it out. We owe it to them to come up with these good plans, and these clear contingencies that are not, you know, upsetting to them in any way, but are still very clear. Because they can do incredible things. And they are doing it like Barley generalizing wolf to marten is incredible. And we don’t actually know why he did that. And we can speculate why marten seems wolfy enough to him. Yeah, probably about diet, like, it’s, it’s really interesting. And, I mean, again, like, I’m not going to open any new rabbit holes, because we’ve gone down plenty of them. But like, it’s physically painful, though. Just, I know, just respecting them enough and respecting what they can do to put this in upfront and go, Okay, I respect that this is the response you’re giving me? Yeah, it’s not the response that I need you to give me. So how am I going to communicate that in a way that is fair, consistent, and, you know, nobody’s upset about it. I’m not yelling at you. Like it’s just it doesn’t need to look like that.
Sarah Stremming 1:13:53
And I think in the detection world, that’s not as much of the problem as it can be in the sport world for sure. And definitely in the behavior mod world, but it’s respecting that like their response didn’t come out of the blue one of my my training partner agility coach, Megan Foster, loves to say nobody gets it wrong on purpose. Just reminding myself that every single time I’m embarking on a plan or going out to train the dog, nobody gets it wrong on purpose is, you know, that’s it. He’s not getting it wrong on purpose. In fact, he’s Barley like, he’s king B. He’s getting it all right all the time. Like as far as he’s concerned. So.
Kayla Fratt 1:14:30
Yeah, you know him. Well. I’ve never met him, but you know him.
1:14:36
I saw him once in Denver, where he did come to the bat. Yeah. He was exactly as expected.
Kayla Fratt 1:14:45
Yeah, he screamed the whole time.
1:14:46
Yeah, exactly. He was perfect any whoever was holding the ball was his God. So he just Yeah,
Kayla Fratt 1:14:54
He totally ditched me.
1:14:57
So, all to round it out, have cool conversations like this with your friends who are training dogs to do cool stuff. That’s not the stuff you train, like, I don’t train anything like this.
Kayla Fratt 1:15:09
No, no. And I stopped myself 14 times from asking you if there was similarities between like, Oh, could this be like when you’re doing weave poles? Like, no, we’re not doing that.
Sarah Stremming 1:15:20
And I stopped myself 14 times from like the same thing from being like, Oh, and also this complex behavior chain over here that I’m teaching like, it is like weave poles, if they miss the entry, and you let them finish the entire set, you’re screwed. Except for that you have to consider the individual who’s gonna be to harmed by you not letting them finish the entire set. And then also, like if Rayya misses the entry but finishes the entire set. I have to pay her or she’s just not gonna weave next time. Like it’s yes, it’s a lot like that. It’s a lot like that.
Kayla Fratt 1:15:49
Yeah. And remembering which dog you’re training at any given moment is always exciting.
1:15:55
Well, I never forget, I’m training Rayya on because my watch is reminding me that I’m in a loud noise environment and that I need to get out of it. So I don’t forget. Thank you, Kayla, for bringing that to you. And I can have these conversations with I’m just going to drop because this is a joint release. I’ll say that you can find me over at sarahstremming.com.
Kayla Fratt 1:16:15
And you can find us at k9conservationists.org.
Sarah Stremming 1:16:18
Thanks, everybody.
Kayla Fratt 1:16:19
Thanks, bye.