Okay, so this is what the games actually look like that Rico played. So, Rico's been sent in to retrieve, Rico's been sent in to retrieve one of the toys that he already knows and was given a word that he was already familiar with. You'll notice no one's in the room and of course, that's the control for any input behavioral input that the humans might have. Since Rico has to go into a room where there's nobody there, it can't be that Rico is using any kind of cues because the people aren't watching. So Rico brings back one of the known toys using one of the known labels, and now Rico's gonna be asked to bring back another toy that he all ready knows the label for. Okay. So, he brings that one back no problem. And now is a crucial test. He's never seen that white rabbit in the middle of the room before. And they've given it a new name and now the question is does Rico bring back some other toy or this weird thing he's never seen. See, he plays with it. He's not quite sure what is that thing. He hears the word again. I don't really know what this is. He looks at all the different toys. Wait, it can't be that, it can't be that. Are you sure? Are you sure? Wait, what? Maybe I didn't hear you right? >> [INAUDIBLE] >> No, no, it's something weird. Okay, all right. So it must be this thing I've never seen before. He brings it back. Incredible. So again, the only species where this type of behavior or cognition has been observed and the inferential reasoning that's required, is in dogs, of all species. Okay, so of all the things that were tested, there was positive evidence for all of them and I think this really just shocked a lot of people including myself, that dogs might be possible of this type of inferential reasoning. I think one of the interesting things going forward will be to test more dogs, a variety of dogs, to see is this something that's really special about border collies. Because all of the dogs, including Chaser. Chaser is a dog that was trained by John Pilley and John, who's at Walford University, did an incredible job taking Julianne Kaminsky's work and extending it even further. Chaser was able to use the principle of exclusion and inferential reasoning to learn over a thousand words, and showed very similar abilities. But the interesting thing was, when John Pilley went and got Chaser as a puppy for the purpose of this research. He didn't somehow choose her as some special dog or try 10 or 15 dogs before he could finally identify a dog who could do this. He literally just went and got a puppy and the first puppy that he did this with, was incredible and turned into this dog that knows a thousand different words. So that suggests to us that this isn't something special about Rico and Chaser. This is probably something that lots of dogs have the ability to do. Just like when Oreo was using gestures, a lot of people said well do you think it's something special about your dog? I said no, I think lots of dogs can do this, and I imagine that's the exact same thing. Rico and Chaser have given us a window into the minds of lots of dogs, it's not just about them. The neat thing is that Adam McClosey has taken this further with his colleagues. And he's developed a game and published on it where it's not, it's actually a cup game. And basically, you show dogs where food is not, and they have to infer that if it's not in this location it must be in the other location. And that is one of the games that you can play through Dognition as a laboratory game with your dog to see if your dog is capable of the principle of exclusion. Now of course that's not required, but it's fun if you wanna see what your dog can do. So, given that we discovered this amazing thing that dogs are capable of potentially fast mapping, people like Julian Kaminsky continue to look for other evidence of inferential reasoning in dog. And this is another gorgeous example where dogs show the ability to potentially understand what somebody else could perceive visually. And this is related of course to theory of mind. We've talked about dogs understanding communicative intentions or what it is you want and you're trying to communicate. This is dogs understanding what you can and cannot see. And in this case what happens is the dog, it wants to play fetch. You ask it to retrieve a ball, but the trick is one of the balls is behind an opaque occluder, which you, as the person cannot see through. And then the other ball is behind a transparent barrier, that you can see through. So, the dog can see both balls, but you can only see one. And when you say fetch, the question is, which ball does the dog bring back? Does the dog bring back the ball that you both can see, or only the ball that it can see? And what they found was, what Julianne found was that dogs tended to bring back the ball that you both could see. Interestingly, when they ran the control, when you were on the same side of the dog, and you say fetch, dogs would tend to chose randomly because now both of you could see both balls. So this is another example where dogs are inferring when he's saying fetch, he must mean the ball that he can see. So, this a really interesting study, and it's gonna be fun to see if this replicates, at what level dogs are able to do this, and where this type of ability might take us, in terms of our learning about how dogs make inferences. The next example is Phillip. Phillip was a dog trained by Joseph Topel, who worked with Adam McLosey in Hungary. This is an incredible story of a dog that is a service dog. And they wanted to see if they could train Philip to be really flexible in solving problems for the people he was helping. And instead of using sort of your typical repetition and training, and sort of get a fixed behavior that when you make a command, the dog really reliably does it. They wanted to see if Philip could actually imitate novel actions that you wanted him to perform. So they taught him to do as I do. And what they found was that when you said, after several training steps, when you said do as I do and then you showed him a novel action that he'd never seen before. So let's say you spin around and you take a bottle and you move it from one location to another and you say do as I do, Phillip would spin around, take the bottle, move it from one location to the other even though he'd never seen you do this before. That is incredible. And subsequently the Hungarian team has trained dozens of dogs to do this and I think it shows the potential power of a cognitive approach and helping us think about how we can potentially teach dogs, which we'll focus on in our lecture to come. Okay, so those are some of the big discoveries that have been made about dogs understanding their social world which I think have really knocked the socks of lots of people including me. And to summarize, the discovery of remarkable social skills in dogs has really opened the door to study a whole host of questions about dog cognition. Rico and Chaser have been shown to learn the labels of objects through inferential reasoning, using the principle of exclusion. This type of learning is highly similar to fast mapping observed in human children, but not in other apes. It's gonna be really interesting to see as people continue to probe those abilities in dogs and see if it's just a certain breed that does this. But my guess is that lots of breeds are capable of this. Dogs have also shown skill at making inferences about others can and can not see. And of course, Philip that we ended on has shown the ability to do as I do and learn things very quickly, with very little practice in incredibly flexible ways. So, I think taken together, really we've learned a lot in the last ten years about how dogs understand their social world. And it's more than just they read our gestures, they're doing all sorts of interesting things in solving problems together with us.
