에피소드 47: Dr. Douglas Elliffe

6:37

And that they tend to be a bit less sensitive to the allocation of reinforcers between alternatives than strip matching would predict. So quite a bit of our stuff since has been, you know, about understanding whether, why, why that might be true. And, and, you know, just to kind of cut to the chase, we think the problem probably relates to how well they're able to, to discriminate, which alternative the reinforcers are coming from. And there's various things we can do to make them better or worse at doing that. So I guess we, we would probably say now that what they're doing is matching to what they perceive to be the relative reinforcers on the two alternatives that might not be actually the true, you know, or the arranged ratio of reinforcers on the two alternatives. One of the things I liked best that I did was about maybe 15 years ago, we did something when there were four alternatives rather than three. And also that every 10 reinforcers, that is about every six or seven minutes, the distribution of reinforcers between those four alternatives changed in a random way. And the interesting thing was how well pigeons were able to track that kind of rapidly changing four way distribution of reinforcers. So I guess where we come from, where we come to with this is that animals are trying to predict what's, what's likely to happen in their future and that they will, you know, behave in order to, you know, to, to try to maximize, well, possibly to maximize, but to follow the way that they think things are likely to change. And of course, coming to what I think you're likely to want to talk about mainly, that will apply to positive contingencies, you know, to things being presented, but it will also apply to negative contingencies like things being removed. So that would work equally well with negative reinforcement. So that's a little bit on that. Yeah. And, but it's basically there, there was something that like the pigeons or whatever the animals are, that they don't always match the behavior to the reward rate. They don't match it perfectly. Right. Yeah. But, but I'm, I'm, I'm saying that our current theory is that they match it to what they perceive the reward rate. Right. Right. Right. Which is interesting. But they are wrong about that in various ways. Yeah. So what, what could be like a takeaway from this for a, for a dog trainer, you would think? A takeaway for that, for the, for a dog trainer, I think I would, the most important takeaway is that I think Herndstein's right, that all behavior is ultimately choice behavior, that there's always something different that we could be doing.

9:31

And there will be reinforcers attached to that. And those might be reinforcers about avoiding bad stuff happening or getting good stuff to happen. And, you know, I'm, you know, I'm going to say that it doesn't necessarily matter which of those we're talking about, according to the principle. And so that, yeah, that's probably jumping ahead to something that I wanted to, you know, to talk about later. Let's leave it then. Let's leave it. Let's go with the, let's, let's go with the, give me a version training program. Yeah, well, could I, could I, I just make a couple of general points about punishment and aversive contentment. I love it. I have it on my list anyway, so go on. Yeah. Yeah. So it, it, it seems to me that most animals behavior is sensitive to, to, well, is, is controlled by avoiding bad things happening just as it is, just as much as it is by trying to get good things to happen. And I was, I was, I was thinking just before talking to you, I was, I was standing out on the side of the street, having my cigarette because I was nervous before coming to talk to you and thinking, why is it that I don't just blunder off into the traffic? Yeah. And the reason why I don't do that is of course, because I know something bad would happen. So nobody's been saying, good, you're staying on the footpath. You're safe there. Yeah. So a lot of my behavior, and I think, you know, a moment's self-reflection would lead us to think that most of everybody's behavior is actually controlled by avoidance contingencies rather than by, you know, by leading to something good happening. Right. Because if we're not alive, who cares if we can eat, right? Yeah. Yeah. Yeah. Yeah. So there's, there's among some behavioral psychologists, and as I see now, you know, among quite a lot of dog trainers too, there seems to be an almost utopian vision of the world where there is no punishment and, and everything must be, you know, must be controlled by positive contingencies, by presenting something nice. And you can kind of trace that back almost intellectually to B.F. Skinner, I guess, who, who, who, again, had, he, you know, Skinner was an unabashed hedonist and, and, and wrote this utopian book called, called Walden, which was, which was imagining a world, you know, where there were only positive contingencies and everything was controlled like that. And I say that that might be a very nice world, but I also think that it's probably not the world that we live in. And that, and that when we're trying to, when we're trying to modify the behavior of any

12:13

other organism, whether they're animals or whether they're other, other, other humans, we would be foolish not to take advantage of the, of our evolutionary history, which will be about learning to avoid bad consequences, as well as trying to produce good consequences, learning to get good consequences. It's just, it's so deeply programmed in us, right? I mean, it's, it's. Yeah, I think so. Yeah. Yeah. I think we're all like that. And, you know, you can, you know, you can also think back to people like, like Dr. Spock, you know, with, with raising children and the idea that, you know, you should never say no. And I think that that, you know, again, I think that this is, this is a very nice thing to think. Um, but I also think that it doesn't sort of really recognize realities and what it is that controls behavior. That's not to say that when we can, you know, if we've got an equal choice between equal efficacies of two different treatments. No question. Interventions. Then it would be nice to use the positive one. Um, it's certainly true that humans tend to feel that, that when they're choosing control by positive contingency, sorry, I should, yeah, by appetitive contingencies, by nice things happening, because there's a, there's a different technical difference between positive and negative contingency. Um, but, but about, let's, let's call them good contingencies. Yeah. Then they feel as if they're doing it of their own free will, whereas if they're doing it to avoid punishment, then they tend to think that they're being coerced. So insofar as that goes, it would be nice to go for the, you know, for the, for the good contingencies if we can do that. But we will not always be able to do that. And, and, you know, we'll talk later, I think about some of the, some of the pitfalls with the notion that you can decrease an undesired behavior only by reinforcing an alternative behavior to that. Yes. Maybe that won't be enough. I would love to talk about this as well. Yes. Uh, since we are, since we are on this a little bit, when we're talking about avoidance, we, we are, we cannot avoid to say something about fear and fear right now, at least in the dog industry. It's like, like there is this notion, this narrative that fear is a very bad thing and nobody should ever, no dog should experience any, any sorts of fear. But just as you gave the example of you hesitating to cross and making aware should you or not, there is some element of caution, which is based on, on some fear or am I wrong? Well, I don't think, I don't think you're wrong.

15:03

I think the first thing I would say is that, you know, coming from a kind of fairly strictly behaviorist condition, tradition, I would say that adding the term fear into it is almost a mental way station, you know, in, in the sort of term Skinner talked about it. So I would be wanting to say, here's the contingency that produces this change in behavior and that we can, we can reliably show that that's the case and putting this idea of fear. And so, okay. And a more mentalistic explanation would be to say the animal or the organism is, is, is fearful of the aversive contingency and is therefore trying to reduce the fear. And I'd say, I'm not sure that that adds anything to the explanation. Um, but I would say also that there, there are things, uh, of course, there are things in the world that's very reasonable for us to be fearful of and, and, and, you know, trying to, trying to deny that that's the case doesn't seem a good thing to do. And if there are actions that we can take that reduce our fear, then that's not irrational either. And doesn't seem to me to be particularly inhuman, inhumane either. Right. I listened to, uh, to your podcast, uh, with Michael Peroni, um, uh, who I also know. And, and, and he was, he was, he was talking about, no, the calmness of rats that he sees working on avoidant schedules and, and, and how this, this, this, in, in, in, in, in fact, doesn't seem to produce a terribly nervous and, and, and, and, and terrified rat that's difficult to handle, but they're at least as easy to handle as rats that are working solely on positive reinforcement schedules working for food. And that would be my experience of that too. Um, so, you know, I've, I've, I've seen rats working on, on, on Sidman avoidance, uh, schedules, or, or, or, or free operant avoidance schedules, which I've never done any, any, any of that myself. But that the, the difference between those and normal avoidance is, is that normal avoidance works as, as a series of trials. Whereas Sidman avoidance schedules, and it's kind of odd, but Murray Sidman became, you know, one of the, one of the, one of the strongest opponents of punishment. Yes. Uh, and that he did a lot of research along these, uh, these, these lines where Sidman avoidance, uh, schedules, uh, defined by a shock, shock interval, where if the animal doesn't do anything, this will be the interval, the interval between shocks and also by a response shock interval. So every response postpones the next shock for, for a certain amount of time. I remember that diagram. Yeah, yeah, yeah.

17:52

And, and what you'll get is animals responding in quite an, an efficient way, responding within the, the response shock interval, therefore postponing the next shock and actually never receiving any shocks. And, uh, you know, like, I'm kind of really perfectly happy about this. And, and there's lots of things humans do that are kind of, you know, analogous to that. Yes. That we pay the phone bill on time so that we don't, you know, so that we're not punished by having the phone removed. And, and, you know, we do this without feeling, feeling particular fear of the phone company or anything like that. Or, or even an electrician would go up on those high level voltages and they know what they're doing so they're not like afraid to go to work in the morning. No, they're not. Um, but, but, but they are, you know, sensibly weary and, and, and, and, you know, because They should, right? Yeah, they should, they should. So, so how, how they behave has been very well controlled by these avoidance, avoidance contingencies. So the world arranges of aversive contingencies, bad consequences that we can avoid just all the time. And I think it's, it's an unrealistic utopian vision to pretend that those aren't there. Yeah. It's a, it's a very difficult one because, uh, on top of everything, I mean, it's like a very serious war in the dog training community. Like, like it's insane and, and they, you know, the, the one side wants to just be like, no, there is a scientific consensus that this is, we, we know better, we have better ways right now and we don't need to use any form of aversive. Um, we have differential reinforcement, which, uh, Yeah, well, you know, we'll, we'll, we'll probably come back to some, some, some possible problems. Yes, I would love it. Yes. Problems, problems from a theoretical perspective. But I would, I would say just to, you know, to get to a little bit, a little bit of, of a conclusion about some of this, that there is of course a moral and ethical question about whether we're justified in arranging aversive contingencies with other humans, with other animals, things like that. But what I'm, what I'm kind of reacting to, and why I was quite happy to come onto onto your podcast is the idea that there is a scientific consensus about these things being co-opted to serve that moral argument.

20:24

And so far as I can tell, um, that, that based on largely cherry picking the evidence, you know, so that, that, you know, I don't want my science to be co-opted to serve, you know, to serve this, this, this aim when I don't think that the, that the research is nearly as clear as that, as they might be adding. Yeah. And you know, like I, I also like when I do training, I, I do a lot of play based training. So I'm, I'm like very, like for my whole time that I've been training dogs, it's been probably 40 years now. And play has been a, a really big deal for me. And the reason I'm bringing play is because like, when you watch two dogs play, there is, it's like a play, playful aggression, if you want to call it that. Right. Yeah. There is challenges. There is, it's definitely aversive based, if we can say it this way. And there is infinite number of benefits of how they develop and how they learn their social skills because of those experiences. Yeah. And, and, and even with humans, when, when we think of most interesting games, you know, that there is, they're challenging games and there is some element of competition and added some level of discomfort. We're not talking abuse by no means, but that's kind of what can make a game very interesting. Yeah, look, absolutely. Sorry to cut him. So, so I take it that you're talking about things like situations where a big dog's playing with a little dog and the little dog goes a bit too far and nips, nips the big, big dog. He gets annoyed, you know, and, and, and therefore, you know, bites back or something. And the little dog learns, don't do that again. And the consequence of that because they don't do it again is then they can have the interaction playing with the big dog and everything's good, right? Yes. Yes. And then, and then they make some, learn how to cooperate, how to make up again. And then, yeah, it's very, very, yeah. So, so clearly dogs are, you know, a very kind of social animals just like us. And, and, you know, we, we, we learned that there are things to avoid under different, you know, under different contexts because they lead to not very satisfactory consequences. Don't, don't swear in front of your grandmother, you know, do that with your friends. So different things are being reinforced and punished under different circumstances and, and, and that's, that's how we learn to interact in our social world just as it is how we learn to interact in our physical world, I think.

23:16

What about the, like, if we, if we stare a little bit with, I guess it's in my head right now about the differential reinforcement and, and, and, and basically the, the effectiveness when, you know, like, like the, again, it's always, it's not always, but very often is taught as the, that's the alternative to aversive method and, and it can do everything and we don't need to use any form of aversive ever. Which that, that word ever is where it gets very tricky because, as you said, if we have the choice to choose not to use and be successful, why would we, right? But, but, but like from, from your perspective, why, why do some, I mean, do you think some behaviors that are especially like problematic, dangerous behaviors, you know, they, they are often resistant to strictly differential reinforcement alone? Yeah. Yeah. Yeah. So, so there's, there's, there's, there's a number of reasons, you know, a number, a number of things that I think would bear on that, on that question. But I'll, I'll start off by saying that I think Einstein's right that all behavior is choice. So that in principle, you can reduce an undesired behavior by reinforcing an incompatible alternative behavior. That's fine. And I guess that's what you're talking about when you say differential reinforcement. That's what they mean. But if you were to, you know, say that, say that you're trying to, trying to train your child not to put their fingers in the electric light sockets, right? Then, then you, well, first you've got to get that right in one go, right? So that, that, that really, you know, this, this is a behavior that I need to eliminate now. It's not a game. It's not a game. Yeah. And, and, and so one of the, one of the things we, that we know about aversive learning is that when you get one trial learning, which is something that's often talked about, you know, where things are just learned immediately. One consequence and behavior changes. Now, most, most of this comes from Pavlovian conditioning rather than operant, operant conditioning, but nevermind the same sort of thing applies. These are almost invariably about aversive contingencies as opposed to positive contingencies. So it will take, you know, if you were trying to train your child not to do that, you know, by, by, okay, put your finger somewhere else and I'll give you a reinforcer for doing that. Then it's going to be sort of terribly inefficient to do that because there's such a range of behaviors that are other than putting your finger in the light socket.

26:08

Do you want to, in the, you know, the power socket, do you want to try to reinforce all of those individually or do you want to say, look, here is a behavior that you must not do anymore. We're going to try and cut that one out. Okay. So try to do that. And, and, and one of the, one of the, I think, dangers. So in, in, in human, human stuff, we often talk about, well, a good idea when you're getting a bad behavior going on is distraction. Right. And you could think of distraction as, as being, providing an alternative behavior that they can engage in, which will give positive reinforcement. That'll make sense. But the trouble is that, that you've got this danger of producing a kind of behavioral chain where, where what you get is, is, you don't actually do any of the distraction until the unwanted behaviors beginning. Right. Until the thing that you want to get rid of starts. So person, dog does something you don't want it to do. And you provide, and you immediately step in and say, don't do, you know, here's this thing instead, go and do this. Right. And that's an opportunity that they didn't have until they did the undesired behavior. So what you've built up is a kind of chain where the undesired behavior is actually being reinforced by the second link of the chain, which is the opportunity to do the other, you know, the new behavior. So. Is this where you're going with the, like the, the behavioral momentum? Tom, Tom, Tom, Neil. Yeah. That's, that's, that's, that's really a separate question. I was going to count to that later. So this idea about, about chaining. Gotcha. I just think a problem. Okay. Because, because we know that you can, you can, you can train long behavioral chains where the only primary reinforcer, like, you know, the only access to food or whatever it is, is at the very end of the chain. And the thing that, that, that reinforces the earlier links, the things they do first, is getting in position to earn the primary reinforcer. Yeah. So there's a whole lot of condition reinforcement going on. So what, you know, so what you can have is say, um, uh, child pulls the cat's tail. Okay. The, uh, bad behavior. You want it not to do that. And so what you do is to say, no, don't do that. Come and have a biscuit or something like that. Yeah. And so the child knows, okay, if I pull the cat's tail, then what that means is mommy will offer me the chance to go and get a biscuit. Yeah. So actually, you know, it's kind of going to work in theory. If you provide the alternative behavior all the time before the unwanted behavior ever begins to happen, then that could work in principle.

29:01

But the thing is that I don't think anyone will ever do that because you'd have to spend all your time kind of dishing out positive reinforcers for other behaviors all of the rest of the time. And, and so very often you're going to be providing more opportunities for that when the undesired behavior has already started to happen. And therefore I think that you've actually reinforced the undesired behavior. I'm not saying you can't do it, but you would have to be just enormously dedicated. And actually reading it in, in, in some of the dog training literature, it talks about, oh, it's really hard and it'll take a long time to train these things. And we'll, yeah. So, well, yeah, damn right. Of course it's going to do that. If you, if you only trying to do it by reinforcing an alternative. And especially if, if we're talking some instinctual, strong, innate. Yeah. Like, right. And then, then we're going even in a different level of stubbornness of persistence of a behavior. Correct? So, yeah, well, I think so. Yeah. So, so in a way that might be a nice segue into the Kiwi aversion training stuff because it was one of the, one of the issues that I think came up there. Do you like me to talk about that now? Well, just to say a little more, because I think we end up something interesting here. Like in, like in your opinion, like in, in cases where differential reinforcement kind of fails to, to do its job. What, what factors are at play would you say? Okay. So I've talked about one of the, well, really two of those factors already. I think one, one of the most profound will be this inadvertently training up a chain of behavior. Okay. It starts with an undesired behavior. There'll be that, there'll be the, the real kind of non-specific thing that, okay, if all behavior is choice, then in principle, any behavior other than the one you want to get rid of should be reinforced. Yes. And, but there's an awful lot of those. And so that, you know, the reinforcement density for each of those is, is likely to be really low and therefore that's not likely to be particularly strong. And you haven't actually taken the reinforcers for doing the unwanted behavior away. Right. So, so the, so the, the idea there is almost that you're just trying to distract them to get them to do something else instead. And really what you want is, is to get the undesired behavior, your animal, your organism, child, whatever it might be, to learn that that undesired behavior doesn't have positive consequences anymore.

31:37

It seems to me that might be most efficiently done by having it learn that it has negative consequences, not just neutral, because you often don't have control over what the positive consequences that we're maintaining that behavior. And that is a problem. That's a big problem. Yeah. Right. So, and, and, and there's, there's a third phenomenon and this, this is where I want to talk about behavioral momentum, which is, I think, you know, not an area that's all that kind of widely recognized. And so far as I can see, it hasn't really come into the animal training literature at all. So this is a line of research that began by Tony Nevin in the 1970s. And the idea there was that we've tended to say, okay, reinforces, reinforces, reinforces strength and behavior. Yep. So reinforces increased response strength. And what Tony speculated was that, well, yes, but the way that we traditionally measure response strength is in terms of response rate. How much of the behavior do we get? Yeah. And his, his, his, his speculation then was, well, maybe that's just one aspect of how strong the behavior is. And the other aspect, the other dimension of response strength, if you like, could be how resistant is it to change? So if we put it on extinction, say, then how long does the behavior persist before it goes away? And that might be correlated with response rate. And of course, we've, we, we tended to think that will be the case, that these are just different ways of measuring response strength, that the more of a behavior you get, also the more persistent it will be. But whether that's actually true or not is an empirical question. So Tony suggested that maybe we'd be better to think of behavior as having momentum and that you could consider momentum to be strength. And in, in, in classical physics, momentum is the product of mass and velocity. Yes. So he's talked about, yeah, the example with the Newton's law, right? Yeah. Yeah. Yeah. Yeah. So, so maybe Tony thought behavior has both of a velocity like property. That is how fast is it going response rate, but it also has a mass like property and that might be resistance to change. And so in a series of, of studies, mainly with pigeons, but with other animals too, and also with, with, with humans, he was able to demonstrate that that's so, that these are actually independent properties of behavior. That things that you do, that, so often things that you do, interventions that you do will increase both of those two dimensions.

34:13

But there are, there, there are circumstances where it doesn't, where you can increase the rate and at the same time reduce the persistence or the other way around. Okay. And so one of the, doesn't make it easy. No, it doesn't make it easy, but it does make it easy when he started to break down, what was the, what was the difference between those two things? And what he identified to cut to the, cut to the chase. And it wasn't just Tony, lots of other people work in this area too, quite a lot in New Zealand actually, was that the, the rate like, the rate properties, velocity like properties were, were controlled by the contingency between responding and reinforcement. Okay. So the more you reinforce the behavior, the higher the rate it will get. Fine. But that the resistance to change ones were controlled by an almost Pavlovian contingency, a classical conditioning contingency, about the relationship between the stimulus context and reinforcers. So what that kind of means is that it doesn't matter which behavior you're reinforcing in a particular stimulus context, you will increase the, the, the resistance to change properties of all behaviors in that stimulus context. Including the one that you're trying to replace. Including the one that you didn't want. Yeah. So if you reinforce an alternative behavior, I'm not saying this is not a good thing to do, but there are pitfalls, things to worry about. You will indeed reduce the rate of your, let's, let's just call them behavior A and B. So A is your target behavior. You want to get less of A, so I'll reinforce behavior B instead. And you'll reduce the rate of A, right? Because Einstein's right, all behaviors choice and they're matching to the, but at the same time, what you've done is to, is to increase the overall reinforcement in that stimulus context. And what that can mean is that you lead, is that your behavior A becomes of lower rate, but of greater resistance to change. It becomes more persistent. So you end up thinking, I've reduced the behavior, this is good, but I've actually made it harder to get rid of it completely. So Tony, with humans and with various colleagues, particularly Bud Mace, did some of this sort of thing in institutional settings. So trying to eliminate unwanted behaviors like food stealing, say, and doing that by reinforcing an incompatible behavior and found that actually when you took the reinforcers away, the food stealing persisted much longer for that group, even though it was of a lower rate.

36:58

So there is a problem that if you reinforce an alternative behavior, then what you are partly doing is increasing the whole reinforcement density in that stimulus context, in that environment, say, and therefore you run the risk of increasing the persistence, the resistance to change of your undesired behavior. Now, there's a way around it. So, yeah, I think you can do it. The most salient of those will be that if you only introduce the competing behavior when the unwanted behavior has already started, then you've really set up trouble for yourself for both of those last two reasons, the chaining reason and the momentum reason. Because now as you're talking, I can see how they're paired. Yeah. Yeah. Right? Yeah. Yeah. Yeah. So I think that the mechanism for the two is different, but that they both lead to the same outcome, which is not an outcome you want. That's super, super interesting. And if a behavior has that strong momentum, does that make it also probably more resistant to extinction? Yeah, absolutely. I mean, hard to extinguish. Totally get rid. Yeah. So there are various ways to measure resistance to change. Resistance to extinction is the most obvious, but there are others. Like, like if you decrease the deprivation of the scheduled reinforcer. So if you're using food as your reinforcer and just start with an animal that isn't very hungry, then that's another way to measure it. allow them to satiate on whatever the reinforcer is that you're using. come out the same way. Extinction is actually probably the one that's least often used. And the problem is that arranging extinction, you've got to consider how discriminable is the change from the reinforced condition to the extinction condition. And that really takes us back to Skinner's original, not very good study that led him to reject punishment, right? Whereas that was that was confounded. Yes, they doubled down on that so much after that, too. Look, they really did. So just maybe for your listeners, do you want me to go over that one? Absolutely. Absolutely. What I think really happened? Yeah. So what Skinner did, this was with rats, and he had two different groups. And both groups were trained to press a lever using variable interwar enforcement. So intermittent reinforcement, unpredictable reinforcement, by no means every response being reinforced is probably about one in 10, given the schedules that he used. So he did that, and then both groups went into extinction. But the difference between the two groups is that one group just got reinforcement first, extinction second. And the second group, experimental group,

40:58

got reinforcement first, and then extinction plus punishment. And I think that the punishment was kind of poor slaps from the lever. So the lever, you know, sort of flicked and hit the rat's paws, and it did a lot. I don't think it was electric shock in this case. And that happened for every lever press. And so that group stopped responding much more quickly under extinction. But then what he did was to take away the punishment conditions, so take away the poor slaps, and found that all that group recovered and started responding again. And his claim was that actually you got about the same number of responses in extinction for the two groups. And therefore all punishment was doing was just temporarily suppressing responding, and therefore it's not worth doing. You don't need to do it. But the problem was this, and the problem was that actually adding the punisher when every response was being punished increased the discriminability of the transition. So if you think of what the first condition was like, variable interval schedule reinforcement, so most responses are not reinforced. And then you start extinction, and that's going to be relatively hard to detect. If you're only getting a reinforcer for one in every 10 responses on the average. You're stubborn now. Yeah, well, yeah, but it's going to take you quite a long time to notice. Hey, I haven't got a reinforcer for ages. They must have gone. Yeah, so that'll take a while. But adding a signal to say things have changed now, so adding the punisher, adding the poor slaps, making that transition easier to detect, and therefore you got a faster change. But also, when you took those away again, then what that did was apparently to reinstate the original conditions. The poor slaps have gone now. The punisher's gone now. So what does the rat think? Oh, the reinforcers are probably back. Yes? And so therefore they started responding again. Okay? So to test that, you know, the punisher's gone now. So what that interpretation says is Skinner was confusing the the direct sort of as a consequence of behavior effects of the punishers with the discriminative effect of the punishers of the fact that they you know, they act as a signal for what to expect. And so to test that out, somebody, and I think it might have been Bill Estes, but I'm not sure of this, did exactly the same thing. But rather than use poor, rather than use a punishing stimulus, he used a neutral stimulus. It was something like flashes of green light, and got exactly the same results. Okay? So what the rats learned was the green light signals that there are no reinforcers

43:48

available now. Take the green light away and behavior comes back. So Skinner's interpretation of that just wasn't right. And of course, naturally, you know, being Skinner, a great deal of respect for Skinner. My whole field wouldn't exist without Skinner. But because he was, you know, kind of morally committed to the idea that everything should be able to be done by positive reinforcement, he's delighted to find punishment doesn't work, and therefore I can reject that. But I think he was just, you know, misinterpreting what the data said. And this is common in so many branches of science, but perhaps particularly psychology, where people remember the first dramatic finding, and they don't remember all the refutations of it that came up later, all the things that show, well, actually, you know, we couldn't replicate. That's probably not true. This is not really right. And of course, if it serves you, if it serves you, I could call it a moral agenda, or perhaps a political agenda, then you're likely to pick the piece of research that you think fits with that. This is the difference, I think, between being an advocate and a scientist. No, that's really important thing to share with us, for sure. Yeah, I imagine you've probably come across that before, but anyway. Yeah, no, no, that's, I mean, that's really where it starts. But it's just how we're talking about persistent behaviours, and this narrative, that notion is as strong as ever, even 2024. Yeah. Like, in some minds, this is just how it is. Well, if you, for moral reasons, I don't discount those. I think these are, you know, people have things that they think are just wrong, and they say, I don't want to do that, no matter what the evidence says, I don't want to do that. But it's only human if you can find a piece of evidence that seems to support that, that that's the one that you will pick. But of course, you know, science is always, it's always changing. We may come to different conclusions about the effectiveness of punishment in the future. No, who knows whether that will be true. But at the moment, all the evidence would seem to be that it changes behaviour in a way that's exactly symmetrical to reinforcement. Reinforcement increases behaviour and punishment decreases behaviour, just as Thorndike thought in the first place. Right. Right. What about, what can you tell us about, like, how well, how well differential reinforcement generalises across different environments and what, because that's a, that matters, right? Yeah. It's very easy to get a result in a certain specific environment.

46:53

Yep. But that's not your objective. You want to transfer that. And I... Yeah, you really do. So this is always a problem with training any new behaviour. So what we're, fundamentally, what we're talking about here is a phenomenon called stimulus control, which is that what the law of effect really says is that reinforcement makes behaviour more likely to occur again in the same context where it was already, where it was previously reinforced. And so that would, you know, apply just equally to punishment, it would apply to, you know, to differential reinforcement. Procedures, anything you might do. So what we know is that, I mean, it isn't kind of doomed. It may well be that things will, that behaviours will generalise to new situations. And according to stimulus generalisation kind of theory, they're likely to do that more or less according to the extent that the two environments are similar to each other. So the more similar it is, the more it should generalise. Standard stimulus control stuff. But you can train an animal to generalise simply by training it in more than one context. So if I reinforce a pigeon's key picks to a yellow light, and I change the light to green, then I would expect to see a decrease in responding, no doubt about that. But if I've trained it in, you know, to pick both yellow keys and green keys, then I can probably try blue keys and lots of other things and it will still pick those. Because effectively what I've said is, don't bother attending to the colour of the key, it's not important. It's like the concept of... Yeah, yeah. So the key to getting behaviour to generalise to new contexts is train it in more than one context. Yeah. And then there is the other thing that we need to make sure that we can maintain that behaviour, whatever the desired behaviour over, you know, to last longer than just this initial training that we are doing. Yeah. Yeah. I find one of the, like for me, one of the issues I personally have sometimes with differential reinforcement is the the go around a problem or like there is times where I find that I would really like the person or the dog or the kid to know that I don't like him to do that instead of just choosing to do something else. Like the whole errorless learning if we want to go that far. And like, okay, there might be a cable or something that, or the outlet, the electric car, as we said. And I think it's important to, in some cases, to know that if you're playing with that, that's not good for you. Yeah, I think so. Right? But that's very much avoided. That's like nobody, at least in the dog training world,

50:13

when we're talking about differential reinforcement, we always say, well, we will convince the dog to do B and C, but when the dog doesn't know that A is a not good thing to do. Right, right. So, and I have a problem with that. Yeah. Sometimes. Yeah. Yeah. Yeah. If you think of, you know, things like academic knowledge, as that might go, you do need to know that some answers are wrong, as well as that other answers are right. Yes. Right. Right. Right. And so we're really talking about the same kind of thing. And just to go back to the point I was trying to make at the start, not very well, the world arranges these aversive contingencies all the time. Right. So, so we need to, why would we not take advantage of that, that, you know, that evolutionary history? And, and yes, you do need to know if you want to stop behavior A, and there are 99 other behaviors that can go, that could, that could be done instead, you don't want to reinforce all of those. No, you really want to try and make it efficient by saying anything else will be okay, but don't do this one. And so it, and when you're talking about maintenance, so I talked a bit about, about Sidman avoidance schedules before, you know, where the, where the rats pressing a lever to postpone a shock. And if it keeps on doing that every so often, it'll postpone all the shocks indefinitely. And what that means is it never receives any shocks. And the sort of interesting consequence of that would be, well, you could go away, you could change the rules, you know, you might never have been going to give it any shocks, dog doesn't know, you know, the animal doesn't know that. So because each of those, each of those behaviors then is perhaps being maintained by negative reinforcement, because it's the avoidance, or at least in, you know, in the perception of the animal, the avoidance of something that might be going to happen. Yeah. So you can make an argument that long-term maintenance is going to be easier when you've, you know, when you've trained it in the form of avoidance, because you don't always have to be there giving the periodic reinforcers. You will have to do that if you've trained a behavior by positive reinforcement. So, of course, you know, what we hope with human behaviors is the kind of the naturally occurring reinforcers that the wanted behavior will produce will eventually take over, so that we don't have to be there. You know, it's often called the middle class mousetrap, right? You know, that you do these sort of things because they lead, you know, because they lead to good consequences. If for some reason

53:09

in your history you've never learned to do those, then we have to reinforce those explicitly so that you do and then you come into contact with the usual consequences. But I think they might be a bit different for, you know, for dogs, so. Yeah, I mean, like, like it's so different with a human, you can, you can still instruct that. Yeah. And, you know, a dog has to experience it. It just has to. That's right. Yeah. So you cannot say to the dog, if you do that, I won't help you. That won't work. So you can, and actually, if you try to do that, if you try to, you know, and I noticed that some of the more extreme of these dog trainers say, you know, pretty much never say no to your dog, right? Yes. Just as Dr. Spock used to do with never say no to your children. You can only use language with the dogs insofar as you've trained it to respond to a particular sound that means something. It doesn't, it doesn't have syntax. It doesn't understand the meaning of the word. It just knows. And you, you could argue this is true of humans too, though that takes us to somewhere different. Um, but, um, but what, yeah, where I was, I was going with that is if you try to do it by sort of telling your dogs things, then you're going to be doing it in a kind of a firm cross voice and wagging your finger and things. And so you could argue that what you're actually doing is, is preventing aversive contingencies based on something that the dog hasn't done yet. Yes. So that, so that you, you'll be punishing whatever the behavior was, which might have been sitting and attending to you. So that's obviously not what you want. So I think with, you know, with all of these things, it just, just need better understandings of what contingencies are actually operating here. You know, I mean, one of the, one of the, the examples I use with my stage one students is, what contingency is it when we send somebody to prison? You know, so you could argue, and, and, and of course, what we think we're doing, we think we're punishing the undesired behavior. But if it's the case that, that, that removes them from a more difficult real world social situation, puts them, you know, puts them into one where at least they get fed and things, we might be reinforcing it. Right. And so we really, we really need to, you know, we can't just our priority to start, this would be an effective contingency, because it would be effective on me. Right. Everybody's going to be different and every dog's going to be different. Very much as a timeout with a kid. Sometimes the kid is like, thank you. I got out of this.

55:48

Yeah. It's so fantastic. I would love that. I can go and sit down and read my book. You know, I can do this. Yeah. Great. Yeah. Lovely. So all, all that has come into it. I also think that sometimes because it's a, to, to, to do right, to do it correctly, that there is so many factors, so many things that need to happen and it takes time to happen. Yeah. And, and meanwhile, in some way, at times, not always, but at times we may very well end up just delaying to address something that's kind of urgent, that actually, uh, uh, it can be dangerous to self or others or, or, um, and we're kind of putting it in, it's lingering in there, you know? Yeah. Yeah. Yeah. So if they never experience any consequences, if they're, if they're just prevented from doing the behavior in other ways, then, then if, if there is some kind of, you know, I'm wary of talking this way, but if there's some kind of motivation to do that behavior and instinct, say, um, then that would have gone away by just distracting it, by getting it to do something else instead. Yeah. I will give you one example of, I don't know when we were talking, if I mentioned this or no, but, uh, back in the nineties, I used to be a, uh, guide dog trainer. Yeah. And, um, back at the time we had a very specific training to make the dogs aware that cars are not something you mess with. So as long as there was a car moving in front, even if the, the blind person tells the dog to go, the dog knows better not to go. Yeah. And that's, as we are talking, it cannot happen by explanation and instruction. That's some direct experience. Yep. So today, um, I, I still have that very interesting about, uh, uh, how, how things are done, especially because it has shifted very much to, to that extreme of everything is positive and there is a differential reinforcement. So what happens is we, they're there, the way they're teaching the dogs is, well, you stop at the curb and then you're not crossing, but you never know that the car actually can be a problem for you. Yeah. And that, like, like me being a guide dog instructor at one point, this is a very scary thing, especially that now we, because they, what happens today, they are relying very much on the, uh, on the reorientation skill of the, the blind person. And that means that they are probably way more selective today who can have a dog than back in the nineties. Um, and, and when you're a blind person, you, you know, like you're not just blind that there's like a bunch of diabetes. I mean, you have a package of, of, of problems. Um, and now the dogs only know how to go from like,

59:16

let's say from home to the bank and from home to the grocery store. We used to train the way that we used to train and it was trial and error. Dogs were like, I, you can drop me anywhere with a dog that I have trained and I can navigate everything I need to navigate. And it's a like today, especially with having the more and more electric cars that have no sound whatsoever that creates a very different problem for a blind person to where they can tell a dog go because they do not hear traffic. And that dog's going to go because they have never experienced anything, uh, uh, you know, association with the car to be careful of wait. That's interesting. I was, I was just thinking about just the same thing and even, you know, maybe even more, uh, I don't know whether you have this so much in the States, but, but, but we have a lot of, of, of, of e scooters here, you know, being electric, electric scooters being ridden on the footpath and they're silent and they're, you know, they're actually terrifying, even if you can see, you know, cause, and so, so the dog really, you know, you have to train the dog. These are a danger. You have to avoid those. Right. Yeah. I don't, you know, maybe, maybe there are people cleverer than me who can see how you could do that entirely by differential reinforcement, but I can't see how you could. That goes also like, I don't know if you do anything. I don't know if you have this, like, like in the States, in some States, not everywhere, but in some States we have like a lot of bad snakes, like rattlesnakes and so on. And there is this snake. No snakes. Really? No, none. Wow. You're blessed. Very, very lucky. It's one of the reasons I still live here rather than Australia. Yes. So we, we do, you know, that there is actual training that goes on with like snake aversion training. Snake avoidance. And, and of course there is a, uh, quite a few trainers that do it in force-free way to where they are just deflecting the dog to do something else as we're talking, but the dog never really understands that the snake is a problem, can be a problem. I think you're right. And, and, and so, yeah. I mean, the dog- It's successful. It's, it, it, the dog, most of the time, most likely will go and do whatever it's reinforced. Yeah. Yeah. But they still really don't understand that the snake is not your friend. Yeah. So the only way they can learn it is if they're treating the, the snake as a discriminative stimulus, but reinforcement for doing this other behavior. And that could, that could,

1:02:09

in principle that could work, but learning the discrimination is going to be relatively slow. And it's also one where, you know, you really can't afford any mistakes. Right. It has to be. So I would think that you boost that by making, you know, making the snake actually aversive, not just a discriminative stimulus. Yeah. And, and, and, and, and, and in fact, you know, you could argue now, I don't know where this, this would happen, but if, but if the snakes become a discriminative stimulus where it signals to the dog, you can get reinforcers for doing a different behavior, then the dog could come, and it's a positive discriminative stimulus in that case. So that can itself become a conditioned reinforcer. So you could end up with a dog that's seeking out snakes. Right. The chain thing. So it can then avoid them, you know, and do that. And you say, it's not actually what you want. Yeah. And to go even further with where you're going, I just thinking now, if for some reason, the owner is not paying attention, so they don't get reinforced. Yeah. And now we have this little moment of extinction burst where the dog is like, hey, no, no, there is a snake here. Yeah. I need to get paid and actually tapping on top of the head of the snake. That's right. Right. Yeah. So that is, as we were saying before, that's, that's one of the kind of theoretical advantages of avoidance in that the reinforcer is always there if you don't get the, you know, if you don't get the aversive consequence, even if the owner's not concentrating or whatever, the reinforcer will still be there. Super cool. Let's, let's talk a little bit. I, I, I'm dying to hear the, like everything you have to tell me about the, the Kiwi studies because we did a very different, like what we did, we just published a study. I don't know. I sent you the link probably. I read it. Yeah, it was good. Yeah. We, we focused mostly on a, on a predation sequence on the, the chase instinct, not, not so much on the, the, the aversion. And one of the, the really good things that came out of the study that a lot of people just cannot accept. And that was the fact that whatever discomfort was, it was very short lived. And there was no, like the bag of fallouts was just not there. But let, let go, go with your, your, because it's like a so big and so like, through like, and I, and I know there is a caveat to it, but I will just let you. There is a caveat to it, but it, it's interesting because it, you know, you know, cut to the chase on that. It doesn't say that you shouldn't use aversive methods here.

1:04:59

It does say that you need to do it right. Okay. So, so doc here, the Department of Conservation, I'll call doc, have, have this program that you can put your dog through and get it licensed so that your dog, you're allowed to take your dog into, into Kiwi habitat. Okay. So the Kiwi is our kind of national bird. We're very sentimental about them and things. And yeah, we, we treat them as treasured, as protected. They are threatened. All our numbers are coming back and dogs indeed will kill Kiwi. There's no doubt about that, but it's also true that, that one of the advantages of having dogs that are able to go into Kiwi territory is that pigs are a big danger to Kiwi wild pigs. And so there's a lot of people here who do pig hunting with dogs as a hobby. And so if they can have their dogs license, then they can take them in there, they can kill the pigs and they'll reduce the pig numbers. And that will be better for the Kiwi. So it isn't just, you know, be nice for the dogs to go to places. Yes. This, this would be a benefit to the Kiwi. So the, the, the kind of training protocol that they've used, they, the first thing to know is they don't train with live Kiwi. So they train with what they call Kiwi props. And I would call Kiwi related stimuli. So things like Kiwi feathers and Kiwi feces and nesting material on a, a silhouette of a Kiwi, a carrot Kiwi. I think they used a, a stuffed Kiwi under some circumstances. So lots of Kiwi related things. And the idea is you put an, an e-collar on your dog and it gets walked past these. And if it shows any interest in those, which of course they always do because of the smell, um, then the dog gets shocked, you know, and, and learns to avoid it. And, and it's usually only done once single shock. And they say, uh, one trial learning, this will work. And so what we showed, um, this was, um, this was one of my PhD students, um, PhD project. The first thing we did was to show that, yeah, that works, that they will, they will avoid the props, the Kiwi related stimuli, right? No trouble. And then we did it, you know, we did a big kind of analysis of a whole, because that was, that was, that was being done where my student conducted the research herself. And then, then we looked at a whole lot of training records that they had and found that all that worked and that there were, there were no differences between breeds or anything or ages that we could detect. Nothing like that. There didn't seem to be any welfare implications. The owners were happy. Everything

1:07:29

seemed good. But then of course we had a concern because the idea of this is of course not to get the dogs to avoid the Kiwi, Kiwi feathers and things like that. It's to get them not to attack live Kiwi. So we thought, well, we better just check this out. So will that training generalize to avoiding a live Kiwi or not? Because it's likely that there'll be, you know, there'll be quite a strong kind of instinctual drive to go and do that. They're hunting animals, you know, of course. So naturally, you know, because of the protected status of Kiwi, we couldn't do the training with a live Kiwi. So what we decided to do was a kind of analog where the target bird was a chicken, an ordinary domestic chicken, and did all just, all just the same training, you know, so, so chicken nesting material and a stuffed chicken and all that sort of stuff. Train that. Okay, the dog avoids those stimuli perfectly well. But then the critical test is, does that mean it will avoid a live chicken? And the answer to that was no, at least not reliably. So a much higher proportion still attack the live chicken. Then, so nearly all of them avoided the props, but that, but, but quite a, quite a number still attack the live chicken. So that would suggest, you know, that therefore aversion, you know, avoiding the props does not necessarily generalise to avoiding the live Kiwi. So, at that stage, you know, you kind of have to say, well, this is, this is not, this is not an effective basis for licensing these dogs to go into Kiwi territory, that there is still a risk that they're going to do the predation. And that, and that, that risk is, you know, relatively large, where we want it to be pretty close to zero to, you know, to do that. Right. Now, that's not to say that the method's wrong, uh, because there are various ways that the, the, the procedure might be tweaked. With the most obvious one of those being, you've got to train with a live Kiwi. Right. Because we, we then trained the dogs. So this, this is our last condition. We trained the dogs in just the same way, but this time the target was a live chicken and they all avoided the live chicken. Virtually all of them did. So that worked. So Doc's assumption, you know, the original training protocols assumption that training with the stimuli, things related to Kiwi would be just as good. The answer to that was no, it's not just as good, but it does work if you train with the live Kiwi. Exactly. Doc, you know, for understandable reasons, you know, there's, there's all sorts of, you know, protected status of Kiwi for various different kinds of reasons.

1:10:16

They don't want to do that. They don't want to train with a live Kiwi. Um, and also, partly because there aren't all that many Kiwi around the place. Um, so would it work? My guess is it would probably work if you gave them more shocks and with the original training, but I don't know that empirical question. We need to find out whether that would be true. I think we can be confident that it would work with a live Kiwi. But why I'm, I'm always wary of mentioning this is that that could be taken by the, you know, only positive training advocates to say, well, look, you know, I mean, they've shown it doesn't work. So you can't use it. They are saying that already. That's why we're having the conversation. Yeah. Yeah. Yeah. Yeah. Yeah. So they, um, they, um, I mean, they don't know about that result because we haven't published it, not because we don't want to, but, um, so the reason there really is there's a kind of, um, um, the reason for not having published it is because doc were enormously helpful and what they really wanted us to do was to kind of validate their program. And so my student, you know, given the kind of relationship she'd built up with them, she happy did indeed communicate to them privately and write reports and things to say, this day, uh, it's not working, uh, but she doesn't want to kind of run them down. And so I had an interesting conversation with another dog trainer, um, not as good as you guys, who was, um, again, recruited by a doc to do another evaluation of the program because they didn't like our evaluation, um, because it didn't come out with the answer they wanted. And essentially this guy didn't want to know. Um, so he, I was saying, well, look, here are the problems we found. These are the things you might have to do. So, and, and to paraphrase him a bit, what he was kind of saying was, well, yeah, but I really, you know, the, really, the most important thing is the kind of public relations aspect of this and it raises awareness and so forth. So that's all right. And it seems to me that if you're saying, you know, we've got evidence for thinking this doesn't work, then continuing with it anyway in a public relations sort of way is, is pretty questionable. It's a bit like, um, um, various studies have shown that, that the, the, the impact of reusing the same coffee cup all the time when it has to be washed and, and I bring, bring that back back that actually, if you take everything into account, the difference between that and the polystyrene throwaway cup is nothing like the size that they would see. But nevertheless,

1:12:57

people advocate for doing the other one because it, it, it kind of raises awareness and shows the right attitude. And so it's this difference between what's a, what's a signal of a behavior that we think would be good as opposed to what is actually effective if we look at the, look at it a bit more dispassionately than that. So, um, you know, the message from there should not be, it doesn't work. The message is doing it in this particular way didn't work. And that, and so there is a danger that with, with, with your study, um, what, what I think you've, you've shown quite nicely is that you're much more, much better able to prevent them from chasing lures. Yeah. But whether that would generalize to the live prey situation is kind of an open question. Yeah. And most likely I wouldn't even argue that it would. No, I don't think you would. Right. But, but, but, but we know what we can do. Yeah. So what, what, what all of those, all of those different studies and all their various ways have shown is that punishment does indeed affect behavior. Yes. But there's no doubt about that, but you'd need, you need to optimize the training. You know, you need to get everything right. And you need to con to conduct your, your study fairly. Um, so one of the things that I've objected to in that very widely cited, um, China et al paper, you know, you know, the one from, from Britain where they were. The Lincoln study. Uh, yes. Yeah. Yeah. Yeah. Yeah. The Lincoln study. So I wrote a, a, a critique of that for a lobbying firm in Great Britain. Um, and you know, again, they, they hadn't randomly assigned to groups. They, they, they, they, they'd assigned to groups according to what the probable, you know, the severity of the problem behavior that, that done all kinds of things that just meant that scientifically, there were a lot of, and, and, and uncontrolled variables where they just couldn't attribute the difference in the results they got unambiguously to the difference in the treatments. Um, and it was even very, like, like you, no matter how well versed you are in reading and understanding a study, you, it's not possible to, to understand clearly what was done with each group. It's really not possible. That's right. That's right. Yeah. And, and, and interestingly, as I seem to remember, the, the, the pilot experiment came out the way that I think both of us would predict that the e-collar was more effective. And so they had to try and find ways of doing it differently. So that didn't come out because it wasn't the result they wanted. And, uh,

1:15:38

I think this is, this is dishonest. Um, um, you know, it is very strange that we are, we are in this time because, um, I think it is, it's just such a fundamental laws on earth and we cannot just because we want to, we cannot get rid of it. Just like, I mean, we can say, well, well, let's, nobody needs oxygen anymore. We, we can do, we can live in a different way. It's like, okay, but, but until we can actually do, we do need oxygen and we better learn how to, how to be efficient and ethical because it's possible if we put the effort to, to educate ourselves and, and advance and do more research and, and so on, but to totally reject the idea, this is the, the mind boggling. Yeah. Yeah. Look, there's, there's, there's a kind of distressing tendency in, in quite a lot of areas these days where people are, are so morally sure that the, the, the, the conclusion that they want to come up with is right, that they, they feel that, that, you know, that, that end justifies almost any means. Right. You can, you can just, just pick the bit of bits of evidence you like and ignore everything else. So one of the, one of the, you know, one of the, one of the kind of principles of science, which is one of its strengths in the sense of discovering things about the world, but one of its weaknesses in terms of public relations is that it's always provisional. The results are always, you know, open to challenge when new findings come up, things might change. And, and, and taking that perspective makes you vulnerable to people who, for ideological or religious or whatever reasons, have a kind of a certainty of their own rectitude, you know, and, and, and so it becomes, it becomes difficult to, to, to fight these things. Do you think it's becoming even more problematic? Yeah. And, and it kind of almost doesn't seem if it's gonna, we will come back to some reason, do you think, or, or, or it's just progressively going into this direction that we... Look, look, look, I really hope to... I mean, eventually we have to reset it. There is just no way. Yeah. Yeah. Yeah. Yeah. So this, this, this probably takes us, you know, some way from where we're going, but to, to one of my, one of my other interests, which is kind of, kind of political and, and, and about concern with the rise of a postmodernist, you know, social constructivist ideology and things, which essentially says that there is no objective truth, that everything's just,

1:18:25

you know, that everything's socially constructed and in its most kind of radical form, that it's just a means of perpetuating the power of the powerful class and continuing to oppress. And this, this, you know, this, this is having a huge rise in, in, you know, academia, just, just sort of. Yeah. They're rewriting biology and evolution books. Yeah, absolutely. You know, I mean, I mean, you know, how many sexes are there, that, that sort of thing. So there, there's a whole, a lot of stuff where, you know, I think that, I think that as the public is getting more and more involved and more interested, and I think that, that universities are beginning to lose some of the kind of social. But isn't the public getting more and more confusing also? Confused? Yeah. Oh, it is. They're getting more confused, but they're also starting to think, you know, hang on a minute, what you're coming up with here is just, that's just rubbish, you know, none of us believe this. And so I thought, I think that some kind of reset towards, you know, you know, I want to say rationalism, but rationalism has a particular philosophical meaning that's different from that. But if, but if you, if you like, scientific realism, you know, that, that there is a real world, that there are things that are true and are not true. And it's not about, you know, how much we want them to be true and how much they might serve other interests. And that whenever we do something that is other than seeking what the truth is, we're, you know, we're letting ourselves down. We're letting humanity down. Yeah. And then you have the, like the, the whole social construct thing, like the, like people at home, they know, but, but, but you step out in public and now you have to put the different face and you have to go along. And it's a serious problem, unfortunately, but a serious problem. Yep. If we go back to, to the study with the Kiwis, how was, how, what did you guys observe as far as the, again, like any, any of the studies like the one, Dr. Daniel Mills and, and Van der Borg and like, like all this from early 2000s to 2015, there were a few studies that were done and, and they basically had just this insane list of huge fallouts that the dogs would be afraid of the owners. They'd be afraid to go to the training field. They'd be afraid to go on the street. Like there's just this crazy scare tactics. We didn't, we didn't see anything like that. Nothing like that at all. They were scared of the Kiwi stimuli, but that, you know. That was the objective. But that's, well, that's what we wanted. Yeah.

1:21:18

I mean, the idea was to get them to avoid those. So no sign whatsoever of them being scared of their owners. In fact, what they'd do would be to run to the owners. And it's, it's, it seems to me this is a different question really, but that e-callers, you know, as a means for delivering aversive contingencies have this huge advantage in that the dog doesn't know who's done it. You know, the, the, the, so you can have the remote control in your pocket. The dog doesn't see you do any of that. So they don't, you know, they don't. And, and, you know, I mean. And the intensity is very strictly controlled. And it's not like you're hitting with a stick or some damage, how they used to do experiments. Like there is no, like, like it's, it's really, it sends that impulse to the brain. We're in danger. Yeah. Make association. Yeah. Yeah. It does hurt. I mean, I've, I've, I've, I've, I've, I've, I mean, I've put one on myself. Right. And so, yeah, because I feel I should do that and know what that's like. And yeah, I'd rather that didn't happen again, but it doesn't do any, you know, long lasting damage. I mean, it has to be a visit. So we did, I did an earlier, an earlier study years ago, which was about, about trying to, trying to, trying to reduce unwanted barking in dogs, you know, when, when, when everyone's gone away by, by sound activated collars. And the idea was to assess this citronella spraying collar is, and to see whether that would, that would work because that would be more humane. And the answer was, of course, that it worked for some dogs, but not really very many. And we, and we gave them in the end, we gave them shot collars instead. And that worked perfectly. And that's what they wanted. Everybody was happy. And, but, but when we thought about it, this, this whole kind of notion that we want to produce something that's less aversive was never really very likely to work because the whole point is that it needs to be aversive. That's, that's, that's how it works. So, yeah. And I mean, one, one thing that, that, that, you know, opponents often bring up is the idea of learned helplessness, right? That, that, that, that, that if you apply punishers, then you end up with a, you know, an animal that just freezes. But that was, that was based on, you know, I, the original research used very intense punishers. Dogs were shitting themselves. Like, they're doing themselves horrible. Yeah. So nothing like that. And also the whole point. I mean, what, what produces learned helplessness is if the, if the aversive event is unavoidable, that there's nothing they can do to make this not happen.

1:24:13

That's what it means. And it's really non-contingent on anything either. Non-contingent on anything. You know, so you, you're just going to get that. And, and of, of course, and you can see this with humans too, when they were, they're in situations where terrible stuff happens and there is absolutely nothing they can do to stop it. Yeah. Then you will get problems. But that's not the case when you're arranging it as a contingency for a particular behavior. They can, that all they have to do is not do the behavior and that, that works. And even with the learned helplessness experience, that it's a very interesting moment, paragraph that talks about the dogs that have previously had experience of escaping, that they, of course, they were not responding because they know there is nothing they can do in the box. Yeah. But that did not carry over in their life because they know that they can. Yeah, that's right. That's right. And that's such an important piece that gets just shoved in the dirt, right? Yeah. Yeah. So, so, so far as I can see, all, all that research, although it keeps being cited this way, this is just not evidence. You know, it's not an argument against using contingent punishment. It's an argument, you know, and there's, there's certainly an argument against non, non-contingent punishment, which is effectively, that's just abuse, I would say. Correct. That's just like, I always give this example, like going into your house and it's booby trapped and you just don't want to walk into your house versus having a bad light switch and you just don't mess with the light switch. Yeah, yeah, yeah. Or fix the light switch. But most of the studies and most of the things, when, if you pay attention, they are always have that element of non-contingent punishment, which leads to, of course, it would lead to this kind of problems. Yeah. Well, yes. And so, so not, not only non, so non, non-contingent, yes, that will definitely happen. I think that'd be, that just says people are applying it badly. But of course, the other problem is lack of immediacy, you know, so that if you, and again, this seems to me that this is an advantage of shot callers in that if you're using other punishment methods, then all you're doing is punishing the dog once it's actually done the behavior you wanted it to do, which has come back to you. Mm-hmm, mm-hmm. Yeah. Now, none of this is to say that, and we might disagree here, so using shot callers and negative reinforcement as a method for recall, you know, getting dogs to come when they're called, kind

1:27:08

of wouldn't be my first choice, you know, but I would say using it as a method of reducing and unwanted behavior, the running away, you know, predation, obviously, things like that, I would say. Yeah, in those cases, I think it's actually the first choice. Yeah, no, for sure. Like, I, I tell you a funny story here and everybody, like I, I also make dog training videos, and one of the videos that I wanted to make was exactly this, like how, how would you teach a recall with an electric collar? Hmm. Hmm. But the dog that I had in mind to do it with never had electric collar on, so that was, you know, one of the, the factors. But I, I, it was, it was, it's one of my dogs, so I've done a lot of just positive reinforcement recalls, like just throughout his life. So now he's 10 months old and he likes to come to me. So we, we couldn't film the video. We couldn't make it because I could not make him not come to me. Yeah, right. Right. I mean, to the point that I would let another dog to play to kind of distract him, leave some food and toys, like did, did plenty of things to create a little problem. And eventually I'm like, okay, there's, you know, that, that's, that's what that's about. But having said that, just like how sometimes people, this is one of the, the, the, it's a serious problem in the dog community when, when we're comparing apples and oranges and we say, well, my dog is, he never chases squirrels, for example. But then there is that dog that never has killed a squirrel because never had the opportunity. But that instinct is so strong that if they have the opportunity, they will kill the squirrel. I don't know if they'll eat it, but they will kill it. And you cannot compare the two dogs and tell me that you will, the same glove gonna fit both. Absolutely. Absolutely. That, that, that, that, you know, somehow yours is a moral dog that wouldn't do these things. Yeah. So, so I guess that was, that was, that was one of the messages from the Kiwi stuff is, is, you know, the evidence from the chicken analog is that trained in the way they were doing. There is still this strong urge and also dogs, you know, when they're hunting, they get hurt and they keep going, you know, so they're, they're, they're, they're, they're bred to tolerate pain. Yes. And, and, and so under those circumstances, the, the urge overwrote it. But if we had actually, as, as, as we did in the final condition, paired the electric shock with the live chicken, that was enough to overcome it. So it can overcome it. But it's that, you know, you need to do the training right. You need to, yeah.

1:30:13

This was, this was a excellent conversation. Like, I'm very happy that we made it. And even with the time change and the confusion, but it was beautiful. That's okay. I mean, we, we always, we always get this wrong. We only have one time zone here, but yeah. I will keep in touch with you because like, I have some other ideas about studies and stuff and, and. Yeah, look, absolutely. I like, I like the way you, you've conducted your stuff. It's, it's very like. Thank you. Like there is like a very methodical way there. Yeah. So if you like, one of, one of my, you know, one, one of my ambitions and what, how I might hope to have an influence is to turn more dog trainers into what we might call scientist practitioners who are, you know, evaluating what they do more carefully. And, and, and it might be that, you know, scientists are trained or should be trained and thinking of the alternative explanations of the results, the confounding variables and designing things so as to try and rule those out. So that, you know, we, we love you guys to be doing research on what it is you do. Some collaboration. Absolutely. And maybe we can help, you know, send your master's students, send your PhD students to come and work with you. That sort of thing. That'd be really. That would be amazing. I think we are onto something. Thank you. Look, I really enjoyed that. Thank you very much. And I'm most flattered to be asked. I appreciate it very much, Doug. So, so like, thank you from all of us. Stay in touch. Have a good rest of the day. I know you have to go and teach. Yeah. I could have got a lecture in a while. Yeah. Let's go. Thank you so much. Thank you. Thank you.