So what did he suggest that we do to figure out what Oreo was doing? Well, it's very, very simple. There, just like with the kids, there are two hiding locations and the dog, Oreo and Daisy, knew they might be able to find something in these hiding locations we had shown them that there can be food in these two places. And once they knew that, then what we did, is we occluded or we hid the two hiding places from their view. And then we took a piece of food. And, in the case of an ape, if we were doing this with an ape, we would use a banana. But of course in the case of a dog, we'd use some kind of treat. And then we, behind the occluder so that the dog can't see where you're hiding it, and you're taking the perspective of the dog we're talking about, we would hide the food in one of the two hiding locations. And then we would remove the occluder. So, now the dog knows there's some food hidden somewhere, but they don't know where it is. And the first thing we discovered is was if we didn't give them any information and we just let them search for the food, while they didn't choose using their nose or some other sort of non-social information, location of where the food was. So, if I just let Oreo or Daisy go, they would choose at random, 50 50 chance is 50 50, because there's only two, it's like flipping a coin. So, that's what we found. About half the time they would be right, and half the time they would be wrong. But, if then I made a jester and I pointed to one of the places where the food might be. Then the story was really different. Oreo and Daisy were really good at finding the food. And more than that, they even could use my gaze. So if I look to one of the two locations, they were able to find the food as well. So that really blew Mike away. And is is what it looked like, pointing to the different cups. Gazing to the different cups. And then this is the control where you gave no social information. You just would look down at the floor and close your eyes and let them choose, and they chose 50/50. And here's the data. And this really blew away Mike, and also Josep Call who helped us get all this research started. And again this is where I really learned what science was. I thought what I was going to tell them look I found my dog is using gestures, you said they couldn't use gestures. They didn't say oh this is probably not true or whatever. They were so excited. They thought this was really cool. They might have discovered something completely new and unexpected, and really interesting. So, we did all whole other series of experiments because now that we had a phenomenon, and this is the key thing, we had a phenomenon. We had discovered that they could use the gestures, the question is how do you explain the phenomenon. And that's what good science is all about. Establish a phenomenon and then try to explain it. So the data that I'm showing you here, it's very simple. We gave 18 repetitions, where you would point to one of these two locations and if you get nine correct, that's chance. But if you get above 14 or 15 out of the 18 total chances well then you're above chance significantly, and that means it can't just be that you were guessing. It has to be you were using the gestures. And you see where it says no cue, that's the control there right at chance. So, they're not using their noses to solve this problem. And remember, that's really surprising to a lot of people. Even to us, when we first started this, but remember that we only count their first choice. So when you make a choice, if you, I have no doubt that they would always find the food if you let them go to both locations cuz they often would search both, but we only are interested in their first choice, can they go directly to the cup where the food is, okay? So, where you do that they just guess when there's no information. When you gesture they go right to where you point. One of the things that suggested to us that this might be something more sophisticated than just learning some cue, because remember Oreo and Daisy they'd grown up with me and spent years playing fetch especially with Oreo, so maybe this was just something he learned through thousands of interactions with me. And just like the chimpanzees who could learn this, they had learned it sort of in the same way. And it wasn't really that interesting or flexible like what we see in kids where it just sort of appears and they start solving all sorts of problems, and probably because kids understand communicative intention. But one thing that led us to think, hm, maybe this is different is, they were good on their first trial. And when we looked at the first half of the trials and the second half of the trials, there was no evidence for learning. And you can see in the data here that there's 18 trials, and they got all of them right, so there's no way they could have learned it. They really knew what they were doing when we showed up to play these games. And then, of course, the control ruled out the olfactory cues. But then also, we did a series of studies to rule out orienting responses or the effect of motion. So we would close Oreo's eyes before he made his choice, and then I would make a pointing gesture, while his eyes were closed, and then we would reveal that I was pointing in one direction or the other, and he didn't actually see the motion of my arm. And he didn't actually see the motion of my arm. Now the reason that's important is maybe when I point, he's just attracted to the motion of my arm. And now of course, he's just looking at one of the cups, and it happens to be the cup to where I'm pointing. He's not reasoning about, oh he's trying to help me find the food, he just said, oh, that moved. And what's over, oh, and now I'm gonna search, right? So that was very egocentric. No understanding of me. Just I was attracted to the motion. So we were able to rule that out as an explanation for this phenomenon because even when we closed his eyes and he couldn't see the motion or and Daisy still kept doing really well. The other thing we did to rule that out is I actually stepped towards the incorrect cup. I stepped away from the cup that had the food, and then I pointed so Oreo or Daisy's heads were actually orienting towards me moving away from the cup. And then I pointed and they still got it correct where the hidden food was. So it didn't seem to be emotion. So that got us really excited. The next thing we did we wondered is this just something about people. Or about the owner and a dog, or is this something that generalizes and lots of dogs can do this. So the next thing we did, is we did those same kind of games again with a bunch of dogs that I had never met before. And we found the same thing. They were really good at using human gestures. But then more than that, we actually were able to test whether they could use dog information and the orientation of dogs, and not really the gestures, but the direction that a dog was looking, to find food. And we found the same thing, that dogs were as good as using a human informant when looking for food in a situation, and so it's not just that dogs use your cues, they can use lots of different people's gestures. That's what allows trainers to be successful, or your friend to come over and communicate with your dog. But more than that, they're using the information or social cues of other dogs as well.
