1 00:00:01,068 --> 00:00:03,269 Viewers like you make this program possible. 2 00:00:03,270 --> 00:00:05,339 Support your local PBS station. 3 00:00:18,285 --> 00:00:20,053 Have you ever wanted to know 4 00:00:20,054 --> 00:00:22,922 what your dog is thinking? 5 00:00:22,923 --> 00:00:24,157 {\an8} Water. 6 00:00:24,158 --> 00:00:26,359 {\an8} Now some dog owners say 7 00:00:26,360 --> 00:00:27,560 {\an8}they do know. Walk. 8 00:00:27,561 --> 00:00:28,762 {\an8} Thanks to buttons that seem 9 00:00:28,763 --> 00:00:31,464 {\an8}to give dogs the ability to talk. 10 00:00:31,465 --> 00:00:34,134 {\an8} Snuffle mat. 11 00:00:34,135 --> 00:00:36,970 {\an8}It's, like, how is this even possible? 12 00:00:36,971 --> 00:00:39,706 {\an8} Come outside. 13 00:00:39,707 --> 00:00:42,108 {\an8} Some dogs are saying so much 14 00:00:42,109 --> 00:00:44,344 {\an8}that scientists have decided to study them. 15 00:00:44,345 --> 00:00:45,345 {\an8}Button. 16 00:00:45,346 --> 00:00:46,346 {\an8} Treat. 17 00:00:46,347 --> 00:00:47,381 {\an8}Yeah, so good-- go get it. 18 00:00:48,649 --> 00:00:50,717 {\an8}We need to look into this. 19 00:00:51,919 --> 00:00:53,319 {\an8} They've set up the largest 20 00:00:53,320 --> 00:00:55,622 {\an8}animal communication study in history... 21 00:00:55,623 --> 00:00:57,524 {\an8} Play, play, play. 22 00:00:57,525 --> 00:00:59,592 {\an8} ...with thousands of dogs from around the world. 23 00:00:59,593 --> 00:01:00,593 {\an8} Later. 24 00:01:00,594 --> 00:01:02,562 {\an8}Come. Come. 25 00:01:02,563 --> 00:01:03,763 {\an8}Play. Play. Play. 26 00:01:03,764 --> 00:01:05,031 {\an8} What we want to see is, 27 00:01:05,032 --> 00:01:07,367 where can we take this and how far can dogs go? 28 00:01:07,368 --> 00:01:10,437 Dog on site! 29 00:01:10,438 --> 00:01:12,840 {\an8} We know we can train dogs to do incredible things. 30 00:01:15,042 --> 00:01:16,843 But could we really teach them 31 00:01:16,844 --> 00:01:18,611 to speak our language? 32 00:01:18,612 --> 00:01:20,947 Snug, snuggle. 33 00:01:20,948 --> 00:01:22,549 You want a snuggle? Come here. 34 00:01:22,550 --> 00:01:24,617 She knows what she's saying. 35 00:01:24,618 --> 00:01:25,785 She's communicating with me. 36 00:01:25,786 --> 00:01:27,086 {\an8} Treat. Fridge. 37 00:01:27,087 --> 00:01:28,822 {\an8}Treat. Treat. Treat. 38 00:01:28,823 --> 00:01:30,690 {\an8} I know what dogs are capable of 39 00:01:30,691 --> 00:01:33,193 in terms of their linguistic ability. 40 00:01:34,528 --> 00:01:36,129 But I think that people 41 00:01:36,130 --> 00:01:38,097 read a little bit too far into things. 42 00:01:38,098 --> 00:01:40,099 Hungry. 43 00:01:40,100 --> 00:01:41,100 So, what's really going on here? 44 00:01:41,101 --> 00:01:43,436 Outside. 45 00:01:43,437 --> 00:01:44,604 Are they just pushing our buttons? 46 00:01:44,605 --> 00:01:46,706 I love you. 47 00:01:46,707 --> 00:01:49,309 Or can dogs talk? 48 00:01:49,310 --> 00:01:50,343 Food. 49 00:01:50,344 --> 00:01:52,679 Right now, on "NOVA." 50 00:02:13,033 --> 00:02:15,802 This is Stella, 51 00:02:15,803 --> 00:02:19,606 an Australian cattle dog/ Catahoula mix. 52 00:02:22,276 --> 00:02:24,778 It's late, but Stella isn't ready for bed. 53 00:02:26,847 --> 00:02:27,882 Yes. 54 00:02:31,852 --> 00:02:32,887 Park. 55 00:02:34,555 --> 00:02:35,955 Want. 56 00:02:35,956 --> 00:02:39,158 Stella appears to be talking. 57 00:02:39,159 --> 00:02:40,593 You want to go to the park? 58 00:02:40,594 --> 00:02:43,029 It's a little late to go to the park. 59 00:02:43,030 --> 00:02:46,666 This video has been sped up a little, 60 00:02:46,667 --> 00:02:49,802 but nothing has been edited in or out. 61 00:02:49,803 --> 00:02:51,906 Do you want to go for a walk instead? 62 00:02:55,109 --> 00:02:56,143 Yes. 63 00:02:59,647 --> 00:03:00,681 Want. 64 00:03:01,849 --> 00:03:03,683 Outside. 65 00:03:03,684 --> 00:03:07,587 Okay, we'll go for a walk. 66 00:03:07,588 --> 00:03:10,189 Let's get your leash on. 67 00:03:10,190 --> 00:03:11,257 Leash. 68 00:03:11,258 --> 00:03:12,892 On. 69 00:03:12,893 --> 00:03:15,062 Stella has been taught to use word buttons. 70 00:03:16,030 --> 00:03:17,196 {\an8} Walk. 71 00:03:17,197 --> 00:03:19,265 {\an8} Her owner, Christina Hunger, 72 00:03:19,266 --> 00:03:20,733 {\an8}can record different words on each of the buttons. 73 00:03:20,734 --> 00:03:22,201 {\an8} Stella. 74 00:03:22,202 --> 00:03:25,171 Stella is the world's first talking dog. 75 00:03:25,172 --> 00:03:27,840 It's a big claim. 76 00:03:27,841 --> 00:03:30,209 So what makes Christina think it's true? 77 00:03:30,210 --> 00:03:32,178 {\an8} Christina. 78 00:03:32,179 --> 00:03:35,248 Christina worked as a speech therapist 79 00:03:35,249 --> 00:03:37,750 and used communication devices with children 80 00:03:37,751 --> 00:03:39,152 who struggled to speak. 81 00:03:39,153 --> 00:03:40,620 What? 82 00:03:40,621 --> 00:03:42,322 It was my job as a speech therapist 83 00:03:42,323 --> 00:03:44,190 to find other ways for kids to talk 84 00:03:44,191 --> 00:03:45,825 if they couldn't use their verbal speech. 85 00:03:45,826 --> 00:03:49,429 When I first brought Stella home, 86 00:03:49,430 --> 00:03:50,530 I had so many observations 87 00:03:50,531 --> 00:03:53,299 about how well she was communicating 88 00:03:53,300 --> 00:03:55,101 with her gestures and her body language. 89 00:03:57,638 --> 00:03:59,840 She would whine to get our attention. 90 00:04:03,577 --> 00:04:06,613 She would paw at her water dish when it was empty. 91 00:04:06,614 --> 00:04:10,917 She would look at me, then look to an object that she wanted. 92 00:04:14,588 --> 00:04:19,859 This is all so similar to how kids start learning language. 93 00:04:19,860 --> 00:04:22,662 And it got me thinking, if she's already communicating 94 00:04:22,663 --> 00:04:26,899 so similarly to kids right before they start talking, 95 00:04:26,900 --> 00:04:29,335 what would happen if she had another way to talk? 96 00:04:29,336 --> 00:04:32,739 Christina got some buttons for Stella 97 00:04:32,740 --> 00:04:33,740 to see what would happen. 98 00:04:34,908 --> 00:04:37,276 I was pushing the button as I was saying the word. 99 00:04:37,277 --> 00:04:38,544 {\an8}Outside? 100 00:04:38,545 --> 00:04:39,979 {\an8} And for the first couple of weeks, 101 00:04:39,980 --> 00:04:42,949 {\an8}she didn't pay attention to the buttons at all. 102 00:04:42,950 --> 00:04:47,220 {\an8} Even when Stella became interested in the buttons, 103 00:04:47,221 --> 00:04:49,456 {\an8}it took a while before she managed to press one. 104 00:04:52,026 --> 00:04:54,661 {\an8} You're so close, Stella. 105 00:04:54,662 --> 00:04:57,296 {\an8} But finally, she cracked it. 106 00:04:57,297 --> 00:05:00,333 {\an8}And the next morning, Christina caught it on camera. 107 00:05:00,334 --> 00:05:01,401 {\an8} Outside! 108 00:05:01,402 --> 00:05:02,468 {\an8} Let's go outside! 109 00:05:02,469 --> 00:05:04,738 I was ecstatic. 110 00:05:06,306 --> 00:05:07,974 And then when she went outside, 111 00:05:07,975 --> 00:05:11,444 she went to the bathroom right away. 112 00:05:11,445 --> 00:05:13,146 So, to me, the most important part 113 00:05:13,147 --> 00:05:16,149 was seeing the communication intent behind it. 114 00:05:16,150 --> 00:05:18,818 Outside. 115 00:05:18,819 --> 00:05:20,987 She was using a word appropriately and functionally. 116 00:05:26,193 --> 00:05:27,827 {\an8} Stella is now six years old. 117 00:05:27,828 --> 00:05:29,395 {\an8}What do you want, girl? 118 00:05:29,396 --> 00:05:33,766 {\an8} And uses far more buttons than the three she started with. 119 00:05:33,767 --> 00:05:35,034 On Stella's main board here, 120 00:05:35,035 --> 00:05:37,270 we have some places that she likes to go. 121 00:05:37,271 --> 00:05:38,471 Park. 122 00:05:38,472 --> 00:05:40,173 And... Beach. 123 00:05:40,174 --> 00:05:41,174 The beach. 124 00:05:41,175 --> 00:05:43,009 {\an8} Where. 125 00:05:43,010 --> 00:05:44,945 {\an8}If she wants to know where we're heading out. 126 00:05:46,947 --> 00:05:48,548 {\an8} Happy. 127 00:05:48,549 --> 00:05:52,251 As Stella started to use her buttons more, 128 00:05:52,252 --> 00:05:55,521 Christina began sharing her videos online. 129 00:05:55,522 --> 00:06:00,560 We left our apartment one day and we set up a camera 130 00:06:00,561 --> 00:06:02,996 just to see, you know, what she would do while we were gone. 131 00:06:04,598 --> 00:06:05,998 Where? 132 00:06:10,671 --> 00:06:12,338 I had never heard her made a sound like that. 133 00:06:16,243 --> 00:06:17,243 It was heartbreaking. 134 00:06:21,415 --> 00:06:24,517 {\an8} Stella's videos went viral 135 00:06:24,518 --> 00:06:27,954 {\an8}and have turned her into a social media phenomenon. 136 00:06:27,955 --> 00:06:29,956 {\an8}We're joined now by Christina Hunger, 137 00:06:29,957 --> 00:06:32,125 {\an8}a speech pathologist in San Diego, 138 00:06:32,126 --> 00:06:34,026 owner of the very clever Stella. 139 00:06:34,027 --> 00:06:36,028 {\an8}Outside. 140 00:06:36,029 --> 00:06:37,864 {\an8}And so, Christina, what does Stella say the most? 141 00:06:37,865 --> 00:06:39,932 {\an8}She definitely says "outside" the most. 142 00:06:39,933 --> 00:06:42,168 {\an8}She absolutely loves being outside. 143 00:06:42,169 --> 00:06:43,770 {\an8} Outside. 144 00:06:43,771 --> 00:06:45,938 {\an8} As news of Stella and her word buttons has spread... 145 00:06:45,939 --> 00:06:47,573 No, stand up. 146 00:06:47,574 --> 00:06:50,810 ...other so-called talking dogs have flooded social media 147 00:06:50,811 --> 00:06:55,848 with their seemingly amazing feats of communication. 148 00:06:55,849 --> 00:06:57,416 Mommy. 149 00:06:57,417 --> 00:07:01,287 These dogs seem to be using our words. 150 00:07:01,288 --> 00:07:02,588 Box. 151 00:07:02,589 --> 00:07:05,391 So is that enough to say they're talking? 152 00:07:05,392 --> 00:07:06,993 Paw. 153 00:07:06,994 --> 00:07:08,461 Do they really understand what they're saying? 154 00:07:08,462 --> 00:07:11,330 Walk, walk, walk. 155 00:07:11,331 --> 00:07:14,500 Or is their button-pushing just random? 156 00:07:14,501 --> 00:07:16,002 Love you. 157 00:07:16,003 --> 00:07:17,870 I love you, too, Cop. 158 00:07:17,871 --> 00:07:20,706 I love you, too. 159 00:07:32,052 --> 00:07:36,222 Federico Rossano is director of the Comparative Cognition Lab 160 00:07:36,223 --> 00:07:39,192 at the University of California, San Diego. 161 00:07:39,193 --> 00:07:41,194 As dogs using word buttons 162 00:07:41,195 --> 00:07:43,396 gained a following on social media, 163 00:07:43,397 --> 00:07:45,598 Federico was approached to study them. 164 00:07:45,599 --> 00:07:48,901 But he was skeptical. 165 00:07:48,902 --> 00:07:51,270 {\an8}I most certainly did not want to get involved in it. 166 00:07:51,271 --> 00:07:54,240 {\an8}I have said no so many times. 167 00:07:57,110 --> 00:07:59,779 One of the thing about social media 168 00:07:59,780 --> 00:08:00,880 is that we try to portray ourselves 169 00:08:00,881 --> 00:08:03,883 always happy and excited 170 00:08:03,884 --> 00:08:06,052 and at the best of our life. 171 00:08:06,053 --> 00:08:07,086 Let's go! 172 00:08:09,323 --> 00:08:10,656 And we usually don't post 173 00:08:10,657 --> 00:08:12,024 all the things that go wrong 174 00:08:12,025 --> 00:08:14,060 and all the accidents. 175 00:08:14,061 --> 00:08:17,997 If dog owners only post clips of their dogs 176 00:08:17,998 --> 00:08:20,900 pressing word buttons that seem to make sense 177 00:08:20,901 --> 00:08:23,169 and ignore all the random presses, 178 00:08:23,170 --> 00:08:26,205 it would be completely misleading. 179 00:08:26,206 --> 00:08:28,007 The risk of social media, of course, 180 00:08:28,008 --> 00:08:32,612 is that because there is, you know, money coming with it, 181 00:08:32,613 --> 00:08:34,213 there is attention coming with it, 182 00:08:34,214 --> 00:08:37,516 there might be a way of exploiting social media, 183 00:08:37,517 --> 00:08:41,121 so creating clips that you think would attract more attention. 184 00:08:42,756 --> 00:08:43,790 Treat. 185 00:08:45,759 --> 00:08:49,762 But despite Federico's concerns about social media, 186 00:08:49,763 --> 00:08:52,266 he saw some clips that stunned him. 187 00:08:54,134 --> 00:08:55,301 Mad. 188 00:08:55,302 --> 00:08:57,404 Why are you mad? 189 00:08:59,273 --> 00:09:00,307 Why mad? 190 00:09:08,815 --> 00:09:10,216 Ouch. 191 00:09:10,217 --> 00:09:13,519 Where is your ouch? 192 00:09:13,520 --> 00:09:15,021 Where ouch? 193 00:09:15,022 --> 00:09:18,057 Stranger. 194 00:09:18,058 --> 00:09:20,693 In your ear? 195 00:09:20,694 --> 00:09:22,095 Where stranger? 196 00:09:26,667 --> 00:09:27,867 Paw. 197 00:09:27,868 --> 00:09:28,969 In your paw? 198 00:09:33,407 --> 00:09:34,574 Let me see your paw. 199 00:09:38,145 --> 00:09:39,979 Okay, I'm gonna put this down and check. 200 00:09:39,980 --> 00:09:42,415 This was the stranger in her paw. 201 00:09:42,416 --> 00:09:44,551 She's got a mat between her-- ow!-- toes. 202 00:09:46,853 --> 00:09:50,389 It looked as if they were having a conversation. 203 00:09:50,390 --> 00:09:52,625 And I was, like, wow, if that is possible, 204 00:09:52,626 --> 00:09:53,893 where, like, a human can talk 205 00:09:53,894 --> 00:09:55,328 and the animal can just respond with the buttons, 206 00:09:55,329 --> 00:09:56,762 that would be amazing. 207 00:09:56,763 --> 00:09:58,998 What we want is to look at these anecdotes 208 00:09:58,999 --> 00:10:02,368 and think, is this even possible? 209 00:10:02,369 --> 00:10:05,404 And if it has not been trained extensively, 210 00:10:05,405 --> 00:10:07,206 if it seems to be spontaneous, 211 00:10:07,207 --> 00:10:08,407 then the next question is, 212 00:10:08,408 --> 00:10:11,010 can we elicit it in a more systematic way? 213 00:10:11,011 --> 00:10:12,645 {\an8}Okay. 214 00:10:12,646 --> 00:10:14,113 {\an8}Pour it, pour it there? 215 00:10:14,114 --> 00:10:17,016 {\an8} The trouble is, past efforts to teach animals 216 00:10:17,017 --> 00:10:20,786 {\an8}to use our language have been incredibly controversial. 217 00:10:20,787 --> 00:10:22,755 {\an8}Which color? 218 00:10:22,756 --> 00:10:25,424 {\an8}That one? The blue one. 219 00:10:25,425 --> 00:10:27,593 {\an8}The way it was done was basically, 220 00:10:27,594 --> 00:10:30,596 a researcher would get one animal, 221 00:10:30,597 --> 00:10:33,032 and then train this animal to learn, 222 00:10:33,033 --> 00:10:34,467 for example, sign language, 223 00:10:34,468 --> 00:10:36,769 or to learn how to use other devices. 224 00:10:40,207 --> 00:10:43,142 Sometimes it looked like something extraordinary, 225 00:10:43,143 --> 00:10:45,177 but if you only have one animal, it's very hard to tell 226 00:10:45,178 --> 00:10:46,445 how that came about 227 00:10:46,446 --> 00:10:47,680 or what that really means 228 00:10:47,681 --> 00:10:49,849 from the perspective of cognition 229 00:10:49,850 --> 00:10:51,517 and what that means about how they're thinking. 230 00:10:53,420 --> 00:10:56,122 After several studies, focused on one or two animals 231 00:10:56,123 --> 00:10:58,624 taken out of their natural habitats, 232 00:10:58,625 --> 00:11:00,693 and with concern about researchers 233 00:11:00,694 --> 00:11:03,662 over-interpreting results, 234 00:11:03,663 --> 00:11:06,565 the entire field of research was called into question. 235 00:11:06,566 --> 00:11:08,434 Eat. Eat. 236 00:11:08,435 --> 00:11:11,070 {\an8} So when the button dogs emerged on social media, 237 00:11:11,071 --> 00:11:12,271 {\an8}many of Federico's colleagues 238 00:11:12,272 --> 00:11:14,473 {\an8}thought he'd be unwise to get involved. 239 00:11:14,474 --> 00:11:17,543 I had scientists calling me 240 00:11:17,544 --> 00:11:20,012 at the beginning of this study, 241 00:11:20,013 --> 00:11:22,681 saying, "Do you know what you're getting yourself into? 242 00:11:22,682 --> 00:11:25,317 You might actually lose your job because of what you're doing." 243 00:11:27,554 --> 00:11:28,888 But this time, 244 00:11:28,889 --> 00:11:31,557 Federico realized that instead of all the research 245 00:11:31,558 --> 00:11:34,260 being based on one or two highly trained subjects... 246 00:11:34,261 --> 00:11:35,761 Ball. 247 00:11:35,762 --> 00:11:37,997 ...there was a huge pool of button-pressing dogs. 248 00:11:37,998 --> 00:11:42,601 So he decided to study them. 249 00:11:42,602 --> 00:11:44,603 What happens if you start tracking thousands of animals 250 00:11:44,604 --> 00:11:45,604 at the same time? 251 00:11:45,605 --> 00:11:48,074 He asked Amalia Bastos, 252 00:11:48,075 --> 00:11:49,642 a researcher in animal cognition, 253 00:11:49,643 --> 00:11:51,710 to join him. 254 00:11:51,711 --> 00:11:53,379 Where can we take this 255 00:11:53,380 --> 00:11:54,980 and how far can dogs go? 256 00:11:54,981 --> 00:11:56,949 We need to actually test things out empirically 257 00:11:56,950 --> 00:11:58,285 to find out what's going on. 258 00:11:59,719 --> 00:12:01,387 Federico and Amalia have been 259 00:12:01,388 --> 00:12:04,390 recruiting participants from around the world. 260 00:12:04,391 --> 00:12:07,026 WOMAN 261 00:12:07,027 --> 00:12:08,828 While early contributors in the study 262 00:12:08,829 --> 00:12:11,764 used simple spreadsheets to log their dogs' button presses, 263 00:12:11,765 --> 00:12:14,400 many owners are now using buttons 264 00:12:14,401 --> 00:12:18,771 {\an8}that automatically record which words are pressed and when. 265 00:12:18,772 --> 00:12:20,039 {\an8}Working. 266 00:12:20,040 --> 00:12:21,907 {\an8}Every time a button is pressed... 267 00:12:21,908 --> 00:12:23,209 {\an8}Snuggle. 268 00:12:23,210 --> 00:12:24,977 {\an8}...that goes straight into the app, 269 00:12:24,978 --> 00:12:26,812 {\an8}and then we can download through the app 270 00:12:26,813 --> 00:12:28,614 the kind of data that has been collected. 271 00:12:36,356 --> 00:12:37,790 And what we ask you to do 272 00:12:37,791 --> 00:12:40,059 is to basically tell us, was it you 273 00:12:40,060 --> 00:12:42,495 or was the dog? 274 00:12:42,496 --> 00:12:45,297 As of now, I think we have approximately 275 00:12:45,298 --> 00:12:47,501 6,500 people using the app. 276 00:12:48,869 --> 00:12:50,569 Play. 277 00:12:50,570 --> 00:12:53,439 We know that some dogs seem to make sense some of the time 278 00:12:53,440 --> 00:12:55,374 with their button presses. 279 00:12:55,375 --> 00:12:58,310 But the data that Federico is collecting 280 00:12:58,311 --> 00:12:59,645 {\an8}will help his team... Outside. 281 00:12:59,646 --> 00:13:01,981 {\an8}...understand the bigger picture: 282 00:13:01,982 --> 00:13:03,582 do the thousands of dogs taking part 283 00:13:03,583 --> 00:13:05,484 know what they're doing with buttons? 284 00:13:05,485 --> 00:13:08,187 Or is most of it just random? 285 00:13:08,188 --> 00:13:09,255 Later. 286 00:13:10,724 --> 00:13:12,958 First, Federico needs to check 287 00:13:12,959 --> 00:13:14,994 that the dogs actually understand 288 00:13:14,995 --> 00:13:18,097 the words on the buttons when they hear them. 289 00:13:18,098 --> 00:13:19,665 Good boy. 290 00:13:19,666 --> 00:13:22,034 His colleague Amalia has visited several dogs 291 00:13:22,035 --> 00:13:25,971 with a team of researchers to conduct an experiment. 292 00:13:25,972 --> 00:13:28,741 This is a clip from our first 293 00:13:28,742 --> 00:13:30,009 experimental study. 294 00:13:30,010 --> 00:13:32,344 This is Mokai. 295 00:13:32,345 --> 00:13:34,380 You're so calm, hm? 296 00:13:34,381 --> 00:13:37,416 Federico and Amalia need to test whether a word 297 00:13:37,417 --> 00:13:39,151 generated by a button 298 00:13:39,152 --> 00:13:41,687 means the same thing to a dog as when that same word 299 00:13:41,688 --> 00:13:43,889 is spoken by a person. 300 00:13:43,890 --> 00:13:45,991 {\an8}Who's having some water? 301 00:13:45,992 --> 00:13:47,560 Because when we say words, 302 00:13:47,561 --> 00:13:49,728 we add all sorts of extra details: 303 00:13:49,729 --> 00:13:52,464 pitch, tone, body language. 304 00:13:52,465 --> 00:13:55,367 {\an8}You are the messiest drinker. 305 00:13:55,368 --> 00:13:59,371 But a word from a button is the same every time. 306 00:13:59,372 --> 00:14:00,906 The first study had to be: 307 00:14:00,907 --> 00:14:03,342 do the dogs that are trained with these buttons 308 00:14:03,343 --> 00:14:05,711 understand the meaning of the words 309 00:14:05,712 --> 00:14:07,379 that are associated with it? 310 00:14:07,380 --> 00:14:09,181 And can we show that they would respond 311 00:14:09,182 --> 00:14:10,849 not just to their owners, 312 00:14:10,850 --> 00:14:13,285 but they will respond to others pushing those buttons? 313 00:14:13,286 --> 00:14:15,054 All right. 314 00:14:15,055 --> 00:14:17,256 This is Kai, trial number four. 315 00:14:17,257 --> 00:14:19,825 The experimenter is wearing headphones, 316 00:14:19,826 --> 00:14:22,595 so that she could not hear the word that is produced 317 00:14:22,596 --> 00:14:24,330 once you push that button. 318 00:14:24,331 --> 00:14:27,266 And the instruction that she has in this case 319 00:14:27,267 --> 00:14:29,235 will be something like, 320 00:14:29,236 --> 00:14:30,536 {\an8}push the red button. 321 00:14:30,537 --> 00:14:31,837 {\an8} Play. 322 00:14:31,838 --> 00:14:34,039 {\an8}The research team wore headphones 323 00:14:34,040 --> 00:14:35,708 {\an8}and owners were blindfolded, 324 00:14:35,709 --> 00:14:38,677 {\an8}to make sure neither could inadvertently give clues 325 00:14:38,678 --> 00:14:40,446 {\an8}about which button had been pressed. 326 00:14:40,447 --> 00:14:44,850 {\an8} And then you see the dog goes straight for the toy basket, 327 00:14:44,851 --> 00:14:46,919 {\an8}retrieves the ball, 328 00:14:46,920 --> 00:14:48,688 {\an8}like, "Let's play." 329 00:14:50,523 --> 00:14:53,325 {\an8}And so it's contextually appropriate 330 00:14:53,326 --> 00:14:55,562 {\an8}with respect to what button had just been pressed. 331 00:14:58,265 --> 00:15:00,966 {\an8} It's the simplest possible test. 332 00:15:00,967 --> 00:15:04,770 58 other dogs were also tested. 333 00:15:04,771 --> 00:15:06,171 Chrissy! 334 00:15:06,172 --> 00:15:07,373 Chris! 335 00:15:07,374 --> 00:15:09,108 Outside. 336 00:15:09,109 --> 00:15:11,210 And the results showed that the dogs responded appropriately 337 00:15:11,211 --> 00:15:13,612 to the "play" and "outside" buttons... 338 00:15:13,613 --> 00:15:14,880 Outside. 339 00:15:14,881 --> 00:15:16,115 ...with no other cues, 340 00:15:16,116 --> 00:15:17,650 just as they'd respond 341 00:15:17,651 --> 00:15:20,786 {\an8}to those words spoken directly by a human. 342 00:15:20,787 --> 00:15:22,554 {\an8}Play. 343 00:15:22,555 --> 00:15:24,089 {\an8} That was critical for us, 344 00:15:24,090 --> 00:15:26,091 {\an8}because if they did not understand 345 00:15:26,092 --> 00:15:27,726 {\an8}the meaning of words... 346 00:15:27,727 --> 00:15:29,628 {\an8}Play. ...if they did not respond in a way 347 00:15:29,629 --> 00:15:31,397 that would suggest they understand it, 348 00:15:31,398 --> 00:15:33,732 then there's no point in moving forward with any other research. 349 00:15:35,802 --> 00:15:37,803 The dogs didn't react appropriately 350 00:15:37,804 --> 00:15:39,204 to hearing a "food" button. 351 00:15:39,205 --> 00:15:40,739 Scout! 352 00:15:40,740 --> 00:15:42,741 But it's clear that they can respond to at least 353 00:15:42,742 --> 00:15:45,678 some of their buttons in a useful, communicative way. 354 00:15:45,679 --> 00:15:47,212 Play. 355 00:15:47,213 --> 00:15:49,448 But that doesn't tell us whether they could use them 356 00:15:49,449 --> 00:15:51,251 correctly themselves. 357 00:15:56,823 --> 00:15:58,257 We know that dogs can respond 358 00:15:58,258 --> 00:16:00,492 appropriately to words, 359 00:16:00,493 --> 00:16:02,094 but that could just be conditioning, 360 00:16:02,095 --> 00:16:03,495 that could just be training. 361 00:16:03,496 --> 00:16:04,897 So whether or not they understand 362 00:16:04,898 --> 00:16:07,032 what the word itself means 363 00:16:07,033 --> 00:16:08,734 is much harder to test. 364 00:16:08,735 --> 00:16:13,038 Arik Kershenbaum studies animal communication 365 00:16:13,039 --> 00:16:16,008 at the University of Cambridge in England. 366 00:16:16,009 --> 00:16:18,911 So this ability to hear the word "sit," 367 00:16:18,912 --> 00:16:21,780 and to know that if they sit, they will get a reward, 368 00:16:21,781 --> 00:16:25,651 is a very, very powerful mechanism. 369 00:16:25,652 --> 00:16:28,487 But just because an animal responds correctly 370 00:16:28,488 --> 00:16:31,824 to a particular command that is given in language 371 00:16:31,825 --> 00:16:35,260 doesn't mean that they understand the language of it. 372 00:16:39,599 --> 00:16:41,734 {\an8} Whether they're responding to commands 373 00:16:41,735 --> 00:16:43,469 {\an8}or pushing buttons themselves, 374 00:16:43,470 --> 00:16:45,304 {\an8}the point is, 375 00:16:45,305 --> 00:16:46,372 it's incredibly hard to know 376 00:16:46,373 --> 00:16:49,274 how much dogs actually understand 377 00:16:49,275 --> 00:16:51,543 {\an8}the words they're hearing or using. 378 00:16:51,544 --> 00:16:54,381 {\an8}Good boy-- you gonna try not to bark at stuff today? Yeah? 379 00:17:00,453 --> 00:17:01,987 {\an8}Can you go to bed? 380 00:17:01,988 --> 00:17:03,088 {\an8}This is Rohan. 381 00:17:03,089 --> 00:17:04,590 {\an8}Yeah, good boy. 382 00:17:04,591 --> 00:17:07,793 {\an8}Amalia's 23-month-old Australian shepherd. 383 00:17:07,794 --> 00:17:08,794 {\an8}Ready? 384 00:17:08,795 --> 00:17:10,062 {\an8}Break. 385 00:17:10,063 --> 00:17:11,063 {\an8}Button. 386 00:17:11,064 --> 00:17:12,064 {\an8} Treat. 387 00:17:12,065 --> 00:17:15,467 {\an8}Yeah, so good, go get it. 388 00:17:15,468 --> 00:17:16,769 {\an8} Thanks. 389 00:17:18,271 --> 00:17:19,304 {\an8} Love you. 390 00:17:19,305 --> 00:17:20,305 Good boy. 391 00:17:20,306 --> 00:17:21,507 Go to bed. 392 00:17:23,510 --> 00:17:25,778 So what we saw looked like a conversation, 393 00:17:25,779 --> 00:17:27,446 {\an8}where the dog understood what it was saying, 394 00:17:27,447 --> 00:17:28,847 {\an8}and we had a back-and-forth. 395 00:17:28,848 --> 00:17:30,749 {\an8}In actual fact, these were 396 00:17:30,750 --> 00:17:32,518 trained interactions, where the dog 397 00:17:32,519 --> 00:17:34,853 was actually just trained to press the buttons as a trick. 398 00:17:34,854 --> 00:17:35,889 Button. 399 00:17:40,360 --> 00:17:41,927 Treat. Treat. That's the one. 400 00:17:41,928 --> 00:17:43,662 He doesn't actually know what that button means. 401 00:17:43,663 --> 00:17:45,864 Button. 402 00:17:45,865 --> 00:17:48,567 It could have had a bird chirp or it could have had, 403 00:17:48,568 --> 00:17:49,802 you know, siren sound. 404 00:17:49,803 --> 00:17:51,336 It could have been literally anything, 405 00:17:51,337 --> 00:17:53,338 and he would have performed the behavior exactly the same way. 406 00:17:53,339 --> 00:17:55,808 {\an8}Love you, thanks, love you, love you. 407 00:17:55,809 --> 00:17:57,976 {\an8} You know, the fact that they're 408 00:17:57,977 --> 00:18:00,646 human-language words is completely arbitrary. 409 00:18:00,647 --> 00:18:02,314 Love you. Thanks. 410 00:18:02,315 --> 00:18:03,615 It's just associative learning. 411 00:18:03,616 --> 00:18:05,584 Rohan has learned 412 00:18:05,585 --> 00:18:07,886 to associate an action, like pressing a button, 413 00:18:07,887 --> 00:18:10,756 with an outcome, like getting a treat. 414 00:18:10,757 --> 00:18:11,857 Treat. Treat. Treat. 415 00:18:11,858 --> 00:18:13,425 That's so good, I know. 416 00:18:13,426 --> 00:18:15,994 He likes treats, so he keeps pressing the buttons. 417 00:18:15,995 --> 00:18:18,464 It's known as "associative learning" 418 00:18:18,465 --> 00:18:19,465 or "conditioning." Button. 419 00:18:19,466 --> 00:18:20,899 Treat. Yes. 420 00:18:20,900 --> 00:18:22,801 But it doesn't mean he knows what the words 421 00:18:22,802 --> 00:18:25,337 "treat" or "thanks" or "love you" 422 00:18:25,338 --> 00:18:27,372 actually mean. 423 00:18:27,373 --> 00:18:29,408 Everything in life starts through associative learning, 424 00:18:29,409 --> 00:18:30,742 and actually, humans learn through associative learning. 425 00:18:30,743 --> 00:18:32,211 Good boy. 426 00:18:32,212 --> 00:18:35,514 Associative learning is common to pretty much all animals. 427 00:18:35,515 --> 00:18:37,182 Break. 428 00:18:37,183 --> 00:18:38,717 And it doesn't require 429 00:18:38,718 --> 00:18:41,186 any sort of complex cognition that we should be impressed by. 430 00:18:41,187 --> 00:18:42,421 Treat. 431 00:18:42,422 --> 00:18:45,357 You could easily teach a mouse to press buttons. 432 00:18:45,358 --> 00:18:47,192 Okay, you got that one-- button. 433 00:18:47,193 --> 00:18:49,094 And that's the big problem. 434 00:18:49,095 --> 00:18:51,830 Just because they can press buttons that say words 435 00:18:51,831 --> 00:18:53,165 doesn't necessarily mean 436 00:18:53,166 --> 00:18:55,200 the dogs have any real understanding 437 00:18:55,201 --> 00:18:56,735 of what those words mean. 438 00:18:56,736 --> 00:18:59,171 Perhaps they just enjoy the reward they get, 439 00:18:59,172 --> 00:19:01,406 whether that's a treat or just some attention. 440 00:19:01,407 --> 00:19:02,407 Good boy. 441 00:19:02,408 --> 00:19:04,543 Down. 442 00:19:04,544 --> 00:19:07,045 So could dogs learn to use different buttons 443 00:19:07,046 --> 00:19:09,448 to get different things? 444 00:19:09,449 --> 00:19:11,650 Could they choose which ones to press 445 00:19:11,651 --> 00:19:16,054 because of the specific words attached to them? 446 00:19:17,991 --> 00:19:21,693 It's something Federico is finally starting to answer. 447 00:19:21,694 --> 00:19:25,130 He's gathered data on the word buttons being used 448 00:19:25,131 --> 00:19:30,502 by an astonishing 10,000 dogs in almost 50 countries. 449 00:19:30,503 --> 00:19:34,541 It's the largest animal communication study in history. 450 00:19:35,909 --> 00:19:37,309 Just to give you an example, 451 00:19:37,310 --> 00:19:39,177 in the last three months, 452 00:19:39,178 --> 00:19:40,479 we have received 453 00:19:40,480 --> 00:19:42,548 more than one million button presses 454 00:19:42,549 --> 00:19:44,317 from dogs in our pool. 455 00:19:46,386 --> 00:19:47,619 {\an8} All done. 456 00:19:47,620 --> 00:19:49,588 All that data can help to show 457 00:19:49,589 --> 00:19:52,992 whether dog button use is random or intentional. 458 00:19:54,794 --> 00:19:55,828 {\an8} I love you. 459 00:19:57,730 --> 00:19:58,765 {\an8}Look. 460 00:20:01,734 --> 00:20:03,902 In principle, the dog could just be spamming 461 00:20:03,903 --> 00:20:05,237 the, the soundboard, just coming there, 462 00:20:05,238 --> 00:20:06,539 touching buttons randomly. 463 00:20:07,874 --> 00:20:10,442 Love you. Treat. 464 00:20:10,443 --> 00:20:11,643 And if it's random, 465 00:20:11,644 --> 00:20:13,245 we should stop doing this research, right? 466 00:20:13,246 --> 00:20:14,613 Why are you do, looking into something 467 00:20:14,614 --> 00:20:16,081 that's completely random? 468 00:20:17,784 --> 00:20:20,118 And what if the dogs are just pressing buttons 469 00:20:20,119 --> 00:20:22,521 that are easy to reach? 470 00:20:24,424 --> 00:20:26,291 {\an8} What can happen, of course, 471 00:20:26,292 --> 00:20:28,994 {\an8}is that the dog might just end up pushing, for example, 472 00:20:28,995 --> 00:20:30,295 {\an8}these two buttons 473 00:20:30,296 --> 00:20:31,797 {\an8}that are close to each other 474 00:20:31,798 --> 00:20:34,566 just because they are adjacent to each other 475 00:20:34,567 --> 00:20:36,969 and it's just easy for them. 476 00:20:36,970 --> 00:20:39,204 And so, it is very important for us to know 477 00:20:39,205 --> 00:20:42,641 not just which buttons they have, participants have, 478 00:20:42,642 --> 00:20:45,677 but also how they're laid out, so that we can, for example, 479 00:20:45,678 --> 00:20:47,245 run analysis to see 480 00:20:47,246 --> 00:20:51,283 the dogs seem to be actually doing something intentionally. 481 00:20:53,553 --> 00:20:56,388 Federico's analysis of over ten million button presses 482 00:20:56,389 --> 00:20:59,725 by thousands of dogs over almost two years 483 00:20:59,726 --> 00:21:03,363 shows that these are the most popular words. 484 00:21:04,430 --> 00:21:06,665 Of course, all the words on soundboards 485 00:21:06,666 --> 00:21:09,835 are initially put there by humans. 486 00:21:09,836 --> 00:21:14,306 But these are the ones that the dogs are pressing most. 487 00:21:14,307 --> 00:21:15,841 And among these words, 488 00:21:15,842 --> 00:21:19,878 Federico has found that some are much more popular. 489 00:21:19,879 --> 00:21:23,448 The dogs are choosing to press them more than others. 490 00:21:23,449 --> 00:21:27,686 So here, the size of the word is representative of the number 491 00:21:27,687 --> 00:21:30,288 of presses across the population. 492 00:21:30,289 --> 00:21:33,492 And "play," "outside," "food" are the buttons that the dogs 493 00:21:33,493 --> 00:21:34,893 tend to press the most, 494 00:21:34,894 --> 00:21:36,795 which is very much in line with being a dog. 495 00:21:36,796 --> 00:21:42,034 You can also see "water," "potty," "treat," "all done." 496 00:21:42,035 --> 00:21:45,671 They don't use a lot, other words that might not be 497 00:21:45,672 --> 00:21:48,473 quite in line with what we expect, 498 00:21:48,474 --> 00:21:55,447 like "training" or "mad or "friend." 499 00:21:55,448 --> 00:21:57,949 The fact that some words are used more than others... 500 00:21:57,950 --> 00:21:59,284 Bed. 501 00:21:59,285 --> 00:22:01,553 ...shows that across the population, 502 00:22:01,554 --> 00:22:03,488 the button use isn't random. 503 00:22:03,489 --> 00:22:05,123 Help. 504 00:22:05,124 --> 00:22:06,958 If it was random, you would just have all the buttons 505 00:22:06,959 --> 00:22:08,393 that you have on the soundboard 506 00:22:08,394 --> 00:22:10,562 have an equal probability of being pressed. 507 00:22:10,563 --> 00:22:13,131 So across the data set, 508 00:22:13,132 --> 00:22:16,134 dogs definitely show preference for certain words. 509 00:22:16,135 --> 00:22:17,502 Outside. 510 00:22:17,503 --> 00:22:19,337 Seemingly, things they want: 511 00:22:19,338 --> 00:22:23,742 to go outside, to get food, to play. 512 00:22:23,743 --> 00:22:26,912 But should we be impressed or surprised by that? 513 00:22:26,913 --> 00:22:30,782 We know a lot of animal communication is asking others, 514 00:22:30,783 --> 00:22:33,552 can be humans or non-humans, to do things for them. 515 00:22:33,553 --> 00:22:35,520 In principle, there is no difference 516 00:22:35,521 --> 00:22:38,423 between pushing the buttons and saying, "Outside," 517 00:22:38,424 --> 00:22:41,093 and trying to tell you by scratching on the door 518 00:22:41,094 --> 00:22:42,794 that they want to go outside. 519 00:22:42,795 --> 00:22:44,396 It's another way of communicating. 520 00:22:44,397 --> 00:22:47,399 So could dogs actually use buttons in a way 521 00:22:47,400 --> 00:22:50,102 {\an8}that goes beyond what they can already communicate? 522 00:22:50,103 --> 00:22:51,570 {\an8} Outside. 523 00:22:51,571 --> 00:22:54,873 Could they start to use language? 524 00:22:54,874 --> 00:22:56,308 All animals communicate. 525 00:22:56,309 --> 00:22:57,976 All animals communicate 526 00:22:57,977 --> 00:22:59,945 and can understand communication, 527 00:22:59,946 --> 00:23:02,314 but not all communication is language. 528 00:23:05,184 --> 00:23:07,452 Humans communicate using language. 529 00:23:07,453 --> 00:23:09,554 Obviously, this is our big difference 530 00:23:09,555 --> 00:23:12,357 from all other animals. 531 00:23:12,358 --> 00:23:14,159 My definition of language, 532 00:23:14,160 --> 00:23:17,763 just a simple rule of thumb that seems to work really well, 533 00:23:17,764 --> 00:23:19,865 is, language is the ability or at least the potential 534 00:23:19,866 --> 00:23:23,468 to convey an unlimited number of concepts. 535 00:23:23,469 --> 00:23:25,237 Doesn't matter how you do it, 536 00:23:25,238 --> 00:23:27,672 but what makes humans special 537 00:23:27,673 --> 00:23:29,875 is that we can communicate to each other 538 00:23:29,876 --> 00:23:31,877 anything, any concept. 539 00:23:31,878 --> 00:23:34,246 There's no end to the books that will ever be written 540 00:23:34,247 --> 00:23:36,414 or the ideas that we can convey, 541 00:23:36,415 --> 00:23:39,584 and no animals that we know of can do that. 542 00:23:39,585 --> 00:23:42,687 No one believes dogs could communicate 543 00:23:42,688 --> 00:23:46,091 absolutely any concept to us using words. 544 00:23:46,092 --> 00:23:49,127 But how far could they go on the spectrum 545 00:23:49,128 --> 00:23:52,531 between communication and language? 546 00:23:52,532 --> 00:23:54,900 I think really, communication becomes language 547 00:23:54,901 --> 00:23:58,170 when you cross that boundary from a signal 548 00:23:58,171 --> 00:24:00,872 that manipulates other individuals 549 00:24:00,873 --> 00:24:06,812 to a system by which information is exchanged between individuals 550 00:24:06,813 --> 00:24:11,116 who understand that their conversation partner 551 00:24:11,117 --> 00:24:13,886 is also participating in that conversation. 552 00:24:15,521 --> 00:24:18,590 Could dogs use words to do more than what they already do 553 00:24:18,591 --> 00:24:21,359 with body language and vocalizations? 554 00:24:21,360 --> 00:24:25,630 Could they communicate about more than just what they want? 555 00:24:26,933 --> 00:24:30,235 The thing is, dogs could be uniquely positioned 556 00:24:30,236 --> 00:24:32,871 to learn our language because of how they evolved. 557 00:24:40,947 --> 00:24:44,482 All dogs evolved from wolves, their wild ancestors. 558 00:24:44,483 --> 00:24:46,918 More than 10,000 years ago, 559 00:24:46,919 --> 00:24:52,057 some wolves were taken in by humans and domesticated. 560 00:24:52,058 --> 00:24:55,861 Over generations spent living with and around us, 561 00:24:55,862 --> 00:25:00,432 they evolved in different ways, according to how we bred them. 562 00:25:02,869 --> 00:25:04,970 We selected traits that we valued, 563 00:25:04,971 --> 00:25:08,673 leading to breeds that were smaller and cuter, 564 00:25:08,674 --> 00:25:11,509 or faster and slimmer, 565 00:25:11,510 --> 00:25:14,946 or bigger and stronger. 566 00:25:14,947 --> 00:25:17,082 And of course, over time, we favored dogs 567 00:25:17,083 --> 00:25:20,385 that could communicate well with us. 568 00:25:20,386 --> 00:25:22,120 Dogs that were good at understanding 569 00:25:22,121 --> 00:25:26,191 what we're trying to say. 570 00:25:26,192 --> 00:25:28,393 There does seem to be something very special 571 00:25:28,394 --> 00:25:32,364 about what happened during that domestication process 572 00:25:32,365 --> 00:25:35,333 that led dogs to be excellent understanders of humans, 573 00:25:35,334 --> 00:25:38,838 and humans to be excellent understanders of dogs, as well. 574 00:25:45,411 --> 00:25:48,280 That communication between humans and dogs 575 00:25:48,281 --> 00:25:49,314 can be so clear... 576 00:25:50,783 --> 00:25:52,751 ...that we can form incredibly constructive 577 00:25:52,752 --> 00:25:55,453 working relationships with them. 578 00:25:58,291 --> 00:26:00,125 Yeah! 579 00:26:00,126 --> 00:26:02,894 Amritha Mallikarjun works at 580 00:26:02,895 --> 00:26:06,598 the University of Pennsylvania's Working Dog Center. 581 00:26:06,599 --> 00:26:12,671 I think the human-dog relationship is so special. 582 00:26:12,672 --> 00:26:13,838 The way that we 583 00:26:13,839 --> 00:26:16,875 work together as human-dog teams 584 00:26:16,876 --> 00:26:20,578 is just unlike any other 585 00:26:20,579 --> 00:26:22,747 relationship that you have. 586 00:26:22,748 --> 00:26:26,284 Amritha and her colleague Emma Gaalaas Mullaney 587 00:26:26,285 --> 00:26:28,219 help the dogs here to go on 588 00:26:28,220 --> 00:26:30,689 to careers in search and rescue. 589 00:26:30,690 --> 00:26:32,958 It's all about training, 590 00:26:32,959 --> 00:26:36,628 and training is all about communication. 591 00:26:36,629 --> 00:26:39,764 Dog on site! 592 00:26:39,765 --> 00:26:41,299 Come on, go find. 593 00:26:41,300 --> 00:26:43,168 {\an8} Today, they're working with Andy, 594 00:26:43,169 --> 00:26:46,204 {\an8}an 11-month-old black Labrador, 595 00:26:46,205 --> 00:26:48,206 who has to find someone hidden in the rubble. 596 00:26:48,207 --> 00:26:51,309 Andy, turn! Good boy! 597 00:26:51,310 --> 00:26:52,544 Go find! 598 00:26:52,545 --> 00:26:54,646 Because there are always people 599 00:26:54,647 --> 00:26:56,414 standing around on the rubble pile, 600 00:26:56,415 --> 00:26:58,616 he knows not to find the people that he can see, 601 00:26:58,617 --> 00:27:00,519 but to find hidden people, specifically. 602 00:27:06,993 --> 00:27:09,427 When Andy finds his target, 603 00:27:09,428 --> 00:27:11,363 he starts barking, 604 00:27:11,364 --> 00:27:13,431 and he won't stop until his team knows 605 00:27:13,432 --> 00:27:17,135 what he's done. 606 00:27:18,270 --> 00:27:19,938 Yes! 607 00:27:19,939 --> 00:27:22,073 What a good boy! 608 00:27:22,074 --> 00:27:25,310 That was so good, yeah, yeah! Whoo-hoo! 609 00:27:25,311 --> 00:27:28,480 There is a remarkable array of skills on display. 610 00:27:28,481 --> 00:27:30,215 Good job, buddy! 611 00:27:30,216 --> 00:27:32,417 {\an8} But they are only helpful to the team 612 00:27:32,418 --> 00:27:36,054 {\an8}because Andy has learned what's expected of him. 613 00:27:36,055 --> 00:27:39,257 {\an8}These dogs could not do the work they do 614 00:27:39,258 --> 00:27:43,195 {\an8}if we didn't have the strength of our communication. 615 00:27:47,466 --> 00:27:51,102 {\an8} The communication between dogs and humans is 616 00:27:51,103 --> 00:27:53,238 how we save lives. 617 00:27:53,239 --> 00:27:55,440 That was so good! That was so good! 618 00:27:56,842 --> 00:27:58,443 In addition to word forms, 619 00:27:58,444 --> 00:28:02,013 dogs are also paying attention to the way our body moves. 620 00:28:02,014 --> 00:28:03,982 They're paying attention to where we're looking. 621 00:28:03,983 --> 00:28:06,651 {\an8}They're paying attention if we're pointing at things 622 00:28:06,652 --> 00:28:08,586 {\an8}or if we're gesturing at things. 623 00:28:08,587 --> 00:28:12,223 {\an8}They're paying attention to the tone of our voice. 624 00:28:12,224 --> 00:28:15,093 So they're reading things off of us all the time, 625 00:28:15,094 --> 00:28:17,596 you know, even as subtle as an arm gesture. 626 00:28:19,398 --> 00:28:22,267 With this extraordinary ability and desire of dogs 627 00:28:22,268 --> 00:28:24,769 to communicate with us, 628 00:28:24,770 --> 00:28:27,572 could they be better-suited to learning our language 629 00:28:27,573 --> 00:28:31,143 than the animals that we've tried to teach in the past? 630 00:28:33,412 --> 00:28:35,980 Because there are some dogs 631 00:28:35,981 --> 00:28:38,183 that seem to have an incredible ability 632 00:28:38,184 --> 00:28:41,319 to recognize large numbers of words. 633 00:28:47,526 --> 00:28:51,963 This is Gaia, a six-year-old border collie. 634 00:28:51,964 --> 00:28:55,266 And this is Isabella Ruiz, Gaia's owner. 635 00:28:58,704 --> 00:29:01,506 While most dogs are good at learning action words 636 00:29:01,507 --> 00:29:03,908 like "sit" and "stay," 637 00:29:03,909 --> 00:29:06,911 very few are able to easily learn the names of objects. 638 00:29:06,912 --> 00:29:10,849 But that's not the case with Gaia. 639 00:29:10,850 --> 00:29:13,118 Gaia has 215 toys now. 640 00:29:13,119 --> 00:29:15,920 And she knows all of them by name. 641 00:29:15,921 --> 00:29:19,357 It's a lot of toys now for us to remember, so, 642 00:29:19,358 --> 00:29:23,895 we did a catalogue of the toys to remember their names, 643 00:29:23,896 --> 00:29:26,164 but Gaia always knows the toys. 644 00:29:26,165 --> 00:29:28,967 So yeah, I think her memory is better than ours. 645 00:29:30,903 --> 00:29:33,037 Isabella enrolled Gaia in a study 646 00:29:33,038 --> 00:29:35,107 called the Genius Dog Challenge. 647 00:29:36,442 --> 00:29:39,444 It's run by Claudia Fugazza, 648 00:29:39,445 --> 00:29:41,980 who studies canine cognition. 649 00:29:41,981 --> 00:29:43,948 She's traveled here all the way from Hungary, 650 00:29:43,949 --> 00:29:45,383 where she works 651 00:29:45,384 --> 00:29:48,419 with one of the world's leading canine research groups. 652 00:29:48,420 --> 00:29:52,790 {\an8}Gaia is one of the, what we call the gifted word learner dogs, 653 00:29:52,791 --> 00:29:56,928 {\an8}and these are dogs that have a very special talent 654 00:29:56,929 --> 00:30:00,666 for learning the name of toys, and they learn very fast. 655 00:30:01,667 --> 00:30:05,470 Only a few dogs around the world seem to have this capacity. 656 00:30:05,471 --> 00:30:06,905 Go, go! 657 00:30:06,906 --> 00:30:08,306 One of the most capable dogs 658 00:30:08,307 --> 00:30:09,307 at learning object names 659 00:30:09,308 --> 00:30:11,609 was Chaser, a border collie 660 00:30:11,610 --> 00:30:14,547 who knew the names of over 1,000 toys. 661 00:30:16,315 --> 00:30:19,417 So far, Claudia has found fewer than 50 dogs 662 00:30:19,418 --> 00:30:20,585 around the world 663 00:30:20,586 --> 00:30:23,521 who are good enough at learning object names 664 00:30:23,522 --> 00:30:24,822 to take part in her study. 665 00:30:27,426 --> 00:30:30,228 Hello! Hi, Gaia, hello, girl! 666 00:30:30,229 --> 00:30:33,631 She's traveled here to put Gaia to the test. 667 00:30:33,632 --> 00:30:35,066 Thank you. 668 00:30:35,067 --> 00:30:36,768 So I got something for Gaia. 669 00:30:36,769 --> 00:30:38,269 Gaia! 670 00:30:38,270 --> 00:30:40,905 Look at here-- maybe there's something for you. 671 00:30:40,906 --> 00:30:43,509 That's for you. 672 00:30:45,110 --> 00:30:47,145 Yeah. 673 00:30:47,146 --> 00:30:48,314 Coffee? 674 00:30:49,281 --> 00:30:50,748 Coffee. 675 00:30:50,749 --> 00:30:53,017 A little test shows just how quickly 676 00:30:53,018 --> 00:30:55,321 Gaia can learn the name of a new toy. 677 00:30:57,456 --> 00:31:00,559 RUIZ 678 00:31:09,435 --> 00:31:10,469 Hey? 679 00:31:11,670 --> 00:31:12,804 RUIZ 680 00:31:12,805 --> 00:31:14,772 Good girl, she got it. 681 00:31:14,773 --> 00:31:16,241 So she learns really quick. 682 00:31:16,242 --> 00:31:17,742 That was a matter of 683 00:31:17,743 --> 00:31:19,844 couple of minutes playing, 684 00:31:19,845 --> 00:31:22,513 and she already got this new name. 685 00:31:23,749 --> 00:31:25,450 But Claudia already knows 686 00:31:25,451 --> 00:31:27,752 that Gaia is exceptionally talented 687 00:31:27,753 --> 00:31:29,454 at learning the names of toys. 688 00:31:29,455 --> 00:31:31,089 Coffee! 689 00:31:31,090 --> 00:31:33,691 She wants to find out whether Gaia's cognitive skills 690 00:31:33,692 --> 00:31:35,927 go a step further. 691 00:31:35,928 --> 00:31:40,865 Gaia knows that "tug" and "fetch" are games to play. 692 00:31:40,866 --> 00:31:44,302 Last week, she was given two new toys, 693 00:31:44,303 --> 00:31:47,472 one to tug and one to fetch. 694 00:31:47,473 --> 00:31:49,641 New toys. 695 00:31:49,642 --> 00:31:52,677 {\an8}Can she figure out which toy goes with which game, 696 00:31:52,678 --> 00:31:55,246 {\an8}based on how she plays with it? 697 00:31:55,247 --> 00:31:58,516 {\an8}This one, I will play like this. 698 00:31:58,517 --> 00:32:00,785 {\an8}Isabella hasn't given the toys any names 699 00:32:00,786 --> 00:32:03,988 {\an8}and hasn't said the words "tug" or "fetch" 700 00:32:03,989 --> 00:32:05,623 {\an8}when playing with them. 701 00:32:05,624 --> 00:32:07,492 {\an8}This one, we've been playing like this, 702 00:32:07,493 --> 00:32:10,295 {\an8}but I, I don't say the name of the category, so, yeah. 703 00:32:10,296 --> 00:32:12,630 Is Gaia smart enough 704 00:32:12,631 --> 00:32:14,198 to figure out which is a "fetch" toy 705 00:32:14,199 --> 00:32:15,867 and which is a "tug" toy, 706 00:32:15,868 --> 00:32:20,072 even though these words haven't been used with these toys? 707 00:32:22,808 --> 00:32:25,276 So are they able to sort those items 708 00:32:25,277 --> 00:32:28,012 into the tug or the fetch category 709 00:32:28,013 --> 00:32:29,947 just based on the way you play? 710 00:32:29,948 --> 00:32:31,816 RUIZ 711 00:32:31,817 --> 00:32:32,985 ...Fetch? 712 00:32:44,229 --> 00:32:45,697 Come on, come on, come on. 713 00:32:45,698 --> 00:32:47,465 Good girl! Good girl! 714 00:32:47,466 --> 00:32:49,600 Good girl, you got the fetch. Good girl! 715 00:32:49,601 --> 00:32:53,037 Good girl! 716 00:32:53,038 --> 00:32:54,038 Gaia. 717 00:32:55,574 --> 00:32:56,941 ...Tug? 718 00:33:08,253 --> 00:33:09,954 Good doggy! Good girl! 719 00:33:09,955 --> 00:33:12,357 Good doggy. Yeah. 720 00:33:12,358 --> 00:33:14,225 Yeah. That's a good girl. 721 00:33:14,226 --> 00:33:15,460 Yeah! Go, go, go! 722 00:33:15,461 --> 00:33:17,628 Yeah! 723 00:33:17,629 --> 00:33:21,899 Well, I have to say, I had no idea how this would go. 724 00:33:21,900 --> 00:33:24,001 But that was really, really exciting. 725 00:33:24,002 --> 00:33:26,070 Totally, I believe this is a big step 726 00:33:26,071 --> 00:33:28,673 in our understanding of dogs' minds. 727 00:33:28,674 --> 00:33:32,276 So, she was able to assign this toy to the named category 728 00:33:32,277 --> 00:33:35,847 without ever having heard that name paired with that toy. 729 00:33:35,848 --> 00:33:38,249 What Gaia has done seems different 730 00:33:38,250 --> 00:33:41,819 from just simple conditioning or associative learning. 731 00:33:41,820 --> 00:33:43,154 For sure, 732 00:33:43,155 --> 00:33:48,593 Gaia could not have associated the name "fetch" or "tug" 733 00:33:48,594 --> 00:33:50,061 to the specific item, 734 00:33:50,062 --> 00:33:52,330 because she has never heard that name 735 00:33:52,331 --> 00:33:55,767 while seeing that toy or while playing with that toy. 736 00:33:55,768 --> 00:33:57,135 This is not possible. 737 00:33:57,136 --> 00:33:59,237 She could not have learned that this way. 738 00:33:59,238 --> 00:34:02,340 It seems some dogs can understand 739 00:34:02,341 --> 00:34:05,943 that one word can refer to a whole category of objects-- 740 00:34:05,944 --> 00:34:08,413 not just one individual thing. 741 00:34:09,848 --> 00:34:12,917 But that's just words that they're hearing. 742 00:34:12,918 --> 00:34:14,986 Could some dogs also understand 743 00:34:14,987 --> 00:34:17,021 that they could use a button word 744 00:34:17,022 --> 00:34:18,957 in different situations? 745 00:34:26,598 --> 00:34:28,166 Meet Parker. 746 00:34:28,167 --> 00:34:32,003 She's a beagle cross and a prolific user of word buttons. 747 00:34:32,004 --> 00:34:33,905 {\an8} Water. 748 00:34:33,906 --> 00:34:35,807 {\an8} So Parker is a bit atypical. 749 00:34:35,808 --> 00:34:38,009 {\an8}When we first got them, it took her 750 00:34:38,010 --> 00:34:42,046 {\an8}about 30 seconds, maybe, not even, to press her first button. 751 00:34:42,047 --> 00:34:44,749 {\an8}And by the, you know, end of that kind of evening, 752 00:34:44,750 --> 00:34:47,919 {\an8}she had pressed, at least experimentally, all six of them. 753 00:34:47,920 --> 00:34:49,787 Snug, snuggle. 754 00:34:49,788 --> 00:34:51,589 You want to snuggle? Come here. 755 00:34:51,590 --> 00:34:53,991 Her owner, Sascha Crasnow, 756 00:34:53,992 --> 00:34:56,761 enrolled Parker in Federico's study 757 00:34:56,762 --> 00:34:59,330 as soon as the buttons arrived. 758 00:34:59,331 --> 00:35:01,399 She definitely says things with the buttons 759 00:35:01,400 --> 00:35:04,503 that she would not be able to communicate to me otherwise. 760 00:35:07,606 --> 00:35:09,340 Snuffle mat. 761 00:35:09,341 --> 00:35:10,908 So a snuffle mat has, 762 00:35:10,909 --> 00:35:13,644 you know, sort of these layers or these little pockets, 763 00:35:13,645 --> 00:35:17,148 {\an8}little places you can kind of hide food, 764 00:35:17,149 --> 00:35:20,052 {\an8}and then a puzzle... 765 00:35:21,653 --> 00:35:22,820 {\an8}...is sort of what it sounds like. 766 00:35:22,821 --> 00:35:25,790 {\an8}It's another way to kind of do mental enrichment 767 00:35:25,791 --> 00:35:27,058 while they're eating. 768 00:35:27,059 --> 00:35:29,894 But if I give her food in a snuffle mat, 769 00:35:29,895 --> 00:35:31,729 and she wants a puzzle that day, 770 00:35:31,730 --> 00:35:33,831 she will refuse to eat that food. 771 00:35:33,832 --> 00:35:36,734 And if I take the exact same food and pour it into a puzzle, 772 00:35:36,735 --> 00:35:38,236 she will then eat that food. 773 00:35:38,237 --> 00:35:40,171 And if I didn't have the buttons, 774 00:35:40,172 --> 00:35:42,440 I would think she was just off her food, right? 775 00:35:42,441 --> 00:35:44,242 That maybe there was something wrong with it. 776 00:35:44,243 --> 00:35:46,177 But no, she's just very particular 777 00:35:46,178 --> 00:35:49,248 about the experience she wants while she's eating. 778 00:35:51,250 --> 00:35:53,985 Of course, there may always be a temptation 779 00:35:53,986 --> 00:35:58,122 for owners to over-interpret their pets' use of word buttons. 780 00:35:58,123 --> 00:35:59,891 Food. 781 00:35:59,892 --> 00:36:01,859 Ball. 782 00:36:01,860 --> 00:36:05,263 But Parker certainly seems more capable than most. 783 00:36:05,264 --> 00:36:06,464 One time, when I was 784 00:36:06,465 --> 00:36:09,166 kind of just out for maybe an hour or so, 785 00:36:09,167 --> 00:36:12,236 she pressed "potty." 786 00:36:12,237 --> 00:36:14,472 Potty, potty. 787 00:36:14,473 --> 00:36:17,508 And then, on the video, you can see that she throws up. 788 00:36:20,445 --> 00:36:24,515 And then she goes and presses "help." 789 00:36:24,516 --> 00:36:25,817 Help. 790 00:36:27,185 --> 00:36:30,087 Amalia, Federico's colleague, 791 00:36:30,088 --> 00:36:32,189 felt that use of the word "help" 792 00:36:32,190 --> 00:36:34,725 suggested greater cognitive abilities 793 00:36:34,726 --> 00:36:37,362 than most of the button words used by dogs. 794 00:36:38,430 --> 00:36:40,298 {\an8}So with most of the buttons that dogs have, 795 00:36:40,299 --> 00:36:42,166 {\an8}like "food" or "play" or "walk," 796 00:36:42,167 --> 00:36:43,701 {\an8}these are simple associations 797 00:36:43,702 --> 00:36:45,636 {\an8}that are related to one particular context. 798 00:36:45,637 --> 00:36:46,971 Help. 799 00:36:46,972 --> 00:36:48,439 "Help" is interesting, 800 00:36:48,440 --> 00:36:49,674 because it's a little bit more general 801 00:36:49,675 --> 00:36:51,576 than a lot of the simpler buttons 802 00:36:51,577 --> 00:36:52,610 that these dogs might have. 803 00:36:54,479 --> 00:36:56,147 She decided to set up an experiment 804 00:36:56,148 --> 00:36:58,015 called The Impossible Task. 805 00:36:58,016 --> 00:36:59,250 Come on! 806 00:36:59,251 --> 00:37:00,952 Sascha and Parker took part. 807 00:37:00,953 --> 00:37:02,621 Sit-- wait. 808 00:37:03,722 --> 00:37:06,357 {\an8}So the way that the study is set up is, 809 00:37:06,358 --> 00:37:08,125 {\an8}we have this mat on the floor here, 810 00:37:08,126 --> 00:37:10,094 {\an8}and it's got a Tupperware in the middle. 811 00:37:10,095 --> 00:37:12,296 {\an8}First, the owners give the dogs opportunities 812 00:37:12,297 --> 00:37:13,798 {\an8}over four different trials 813 00:37:13,799 --> 00:37:15,466 {\an8}to obtain a food reward from inside the Tupperware, 814 00:37:15,467 --> 00:37:17,101 {\an8}and at that point, the lid is open. 815 00:37:17,102 --> 00:37:19,503 {\an8}And at all these stages, the dog can knock off the lid 816 00:37:19,504 --> 00:37:20,905 {\an8}and eat the food inside the Tupperware 817 00:37:20,906 --> 00:37:22,107 {\an8}without much difficulty. 818 00:37:23,275 --> 00:37:24,675 {\an8}Until, in the very final trial, 819 00:37:24,676 --> 00:37:26,944 {\an8}the fifth trial, the Tupperware is completely locked, 820 00:37:26,945 --> 00:37:29,280 {\an8}and that's the point at which the dog might need help 821 00:37:29,281 --> 00:37:31,015 {\an8}to access the food. 822 00:37:31,016 --> 00:37:33,150 {\an8}The owners are not allowed to speak to the dog, 823 00:37:33,151 --> 00:37:34,885 {\an8}point or gesture at the dog, 824 00:37:34,886 --> 00:37:36,655 {\an8}look at their dog-- nothing at all. 825 00:37:38,690 --> 00:37:40,958 {\an8}I didn't know what Parker would do 826 00:37:40,959 --> 00:37:42,159 {\an8}in that circumstance, 827 00:37:42,160 --> 00:37:43,295 {\an8}whether she would use the buttons or not. 828 00:37:46,131 --> 00:37:47,932 {\an8} We can see she's immediately gone to Sascha, 829 00:37:47,933 --> 00:37:49,333 {\an8}and she's climbed onto Sascha's back, 830 00:37:49,334 --> 00:37:50,801 {\an8}trying to get her attention. 831 00:37:50,802 --> 00:37:52,203 {\an8}But what we're interested in is, 832 00:37:52,204 --> 00:37:55,374 {\an8}can these dogs then make that leap and use a word? 833 00:37:57,142 --> 00:37:59,577 {\an8} Help. Look. 834 00:37:59,578 --> 00:38:00,845 {\an8} She presses "help," "look," 835 00:38:00,846 --> 00:38:02,780 {\an8}and then she looks to Sascha. 836 00:38:02,781 --> 00:38:06,550 {\an8} Parker had never before been presented with this scenario, 837 00:38:06,551 --> 00:38:09,620 {\an8}and so had never used the words "look" or "help" 838 00:38:09,621 --> 00:38:11,088 {\an8}in this context. 839 00:38:11,089 --> 00:38:13,491 {\an8} Help. Help. Look. Look. 840 00:38:13,492 --> 00:38:15,893 {\an8} So because it's the very first time that they've seen this, 841 00:38:15,894 --> 00:38:17,128 it tells us two things. 842 00:38:17,129 --> 00:38:18,696 It tells us the behavior wasn't trained 843 00:38:18,697 --> 00:38:23,234 and it tells us that the dog is having to think about what this 844 00:38:23,235 --> 00:38:24,702 requires from them, 845 00:38:24,703 --> 00:38:26,337 in terms of communication to their owner, 846 00:38:26,338 --> 00:38:27,972 and that they need to generalize a button 847 00:38:27,973 --> 00:38:29,573 that they've not used in this context before. 848 00:38:29,574 --> 00:38:31,242 {\an8} Help. Help. Help. 849 00:38:31,243 --> 00:38:32,810 {\an8} It seems from this experiment... Look. Look. 850 00:38:32,811 --> 00:38:35,713 {\an8} ...that Parker understands that the word "help" 851 00:38:35,714 --> 00:38:38,717 {\an8}can be used in a broad range of scenarios. 852 00:38:39,851 --> 00:38:42,454 Oh, hi! Do you need help? 853 00:38:45,223 --> 00:38:46,892 There you go. 854 00:38:48,760 --> 00:38:52,163 {\an8} As well as using words in new, untrained ways, 855 00:38:52,164 --> 00:38:55,633 {\an8}a huge step would be if dogs were using words 856 00:38:55,634 --> 00:38:58,135 {\an8}to do more than just ask for things. 857 00:38:58,136 --> 00:38:59,437 {\an8}What we're actually interested in 858 00:38:59,438 --> 00:39:02,139 {\an8}is whether dogs might be able to use 859 00:39:02,140 --> 00:39:04,408 the soundboards not just as a vending machine, 860 00:39:04,409 --> 00:39:06,243 where they're requesting things from humans, 861 00:39:06,244 --> 00:39:08,447 but actually understand it as a communicative device. 862 00:39:09,481 --> 00:39:11,816 {\an8} So humans do not just request things, 863 00:39:11,817 --> 00:39:14,351 {\an8}humans can inform. 864 00:39:14,352 --> 00:39:16,153 It's a thing that humans do all the time. 865 00:39:16,154 --> 00:39:18,355 Um, toddlers start doing it-- they will point at the moon 866 00:39:18,356 --> 00:39:20,558 and say "moon," and, you know, they're not requesting the moon, 867 00:39:20,559 --> 00:39:21,992 they're just letting you know the moon is there. 868 00:39:21,993 --> 00:39:24,028 {\an8} Right or wrong, 869 00:39:24,029 --> 00:39:26,030 {\an8}it's something that some dog owners believe 870 00:39:26,031 --> 00:39:27,766 {\an8}their dogs are capable of. 871 00:39:28,767 --> 00:39:31,235 Though of course, they may have a tendency 872 00:39:31,236 --> 00:39:34,106 to read too much into their dogs' button presses. 873 00:39:36,174 --> 00:39:38,509 Parker uses the buttons to narrate a decent amount. 874 00:39:38,510 --> 00:39:43,247 Um, she has narrated me watering the plants quite a bit. 875 00:39:43,248 --> 00:39:45,617 She'll narrate also things she does sometimes. 876 00:39:47,452 --> 00:39:49,086 Settle. 877 00:39:49,087 --> 00:39:51,122 She's pressed "settle" before, 878 00:39:51,123 --> 00:39:54,526 and then just gone into her bed and lie down. 879 00:39:56,094 --> 00:39:57,895 She didn't get anything from me for it. 880 00:39:57,896 --> 00:39:59,330 She just pressed "settle" 881 00:39:59,331 --> 00:40:00,599 and then she settled. 882 00:40:03,568 --> 00:40:06,003 It's impossible to know for sure whether Parker 883 00:40:06,004 --> 00:40:08,406 really was commenting on what she was doing. 884 00:40:09,441 --> 00:40:11,776 But other dogs, like Copper, 885 00:40:11,777 --> 00:40:14,011 a three-and-a-half-year-old Labrador, 886 00:40:14,012 --> 00:40:17,681 also seem to be using buttons to do more than just make requests. 887 00:40:17,682 --> 00:40:19,517 {\an8}Outside. Pool. 888 00:40:19,518 --> 00:40:20,718 {\an8} Outside pool. 889 00:40:20,719 --> 00:40:23,588 {\an8}Outside pool maybe tomorrow. 890 00:40:26,324 --> 00:40:27,359 {\an8}Soon. 891 00:40:28,727 --> 00:40:30,528 {\an8}Outside pool later. 892 00:40:30,529 --> 00:40:31,529 {\an8} Outside. Pool. 893 00:40:31,530 --> 00:40:32,798 {\an8}Later. Later. 894 00:40:35,433 --> 00:40:36,600 {\an8}Sad. 895 00:40:36,601 --> 00:40:38,903 {\an8}Oh, my goodness-- Copper. 896 00:40:38,904 --> 00:40:41,705 It looks like Copper was using her buttons 897 00:40:41,706 --> 00:40:43,440 to express emotion. 898 00:40:43,441 --> 00:40:46,911 But of course, we don't know for sure. 899 00:40:46,912 --> 00:40:49,547 {\an8}An animal could learn that pressing the "sad" button 900 00:40:49,548 --> 00:40:50,981 {\an8}gets attention. 901 00:40:50,982 --> 00:40:53,217 {\an8}And, and once they've learnt that, 902 00:40:53,218 --> 00:40:55,820 {\an8}they can manipulate these, these different concepts 903 00:40:55,821 --> 00:40:57,555 in, in quite complex ways, 904 00:40:57,556 --> 00:41:01,559 without understanding there's a semantic meaning behind it. 905 00:41:03,495 --> 00:41:06,397 Outside pool later. 906 00:41:10,035 --> 00:41:12,937 {\an8}From the data being submitted by dog owners, 907 00:41:12,938 --> 00:41:15,840 {\an8}Federico has found that many believe their dogs 908 00:41:15,841 --> 00:41:19,177 {\an8}are using buttons to do more than just ask for things. 909 00:41:21,580 --> 00:41:23,714 While 50% of button presses 910 00:41:23,715 --> 00:41:27,051 were categorized as requests, 911 00:41:27,052 --> 00:41:29,687 more than five percent of presses were believed by owners 912 00:41:29,688 --> 00:41:31,990 to be sharing thoughts and feelings. 913 00:41:34,159 --> 00:41:37,094 Six percent were informing, 914 00:41:37,095 --> 00:41:41,233 as well as 6.4 percent described by owners as "narrating." 915 00:41:43,001 --> 00:41:44,635 Crucially, these aren't 916 00:41:44,636 --> 00:41:47,171 scientifically validated reports. 917 00:41:47,172 --> 00:41:50,442 They're just what owners believe their dogs are doing. 918 00:41:53,044 --> 00:41:55,279 Knowing for certain what a dog means 919 00:41:55,280 --> 00:41:58,349 when it uses words, or thinks when it hears words, 920 00:41:58,350 --> 00:42:00,951 is much harder. 921 00:42:05,657 --> 00:42:07,458 But in Budapest, Hungary, 922 00:42:07,459 --> 00:42:10,561 Marianna Boros wants to find out what's happening 923 00:42:10,562 --> 00:42:11,929 inside a dog's brain 924 00:42:11,930 --> 00:42:14,599 when it hears words it knows. 925 00:42:16,568 --> 00:42:18,235 {\an8} For most people, we have 926 00:42:18,236 --> 00:42:19,937 {\an8}a mental representation 927 00:42:19,938 --> 00:42:22,006 {\an8}of things from the external world, 928 00:42:22,007 --> 00:42:24,341 {\an8}so we have a mental representation of an object, 929 00:42:24,342 --> 00:42:27,611 some kind of memory of that thing. 930 00:42:27,612 --> 00:42:31,782 We can imagine it in a way that, we hear the word "ball," 931 00:42:31,783 --> 00:42:34,152 we will see in our mental imagery a ball. 932 00:42:35,587 --> 00:42:37,054 Marianna wants to test 933 00:42:37,055 --> 00:42:40,357 whether dogs' brains might work similarly. 934 00:42:40,358 --> 00:42:42,259 When a dog hears a word, 935 00:42:42,260 --> 00:42:45,629 does it conjure up an image of that thing in its head? 936 00:42:45,630 --> 00:42:47,097 Hello. 937 00:42:47,098 --> 00:42:48,399 To find out, 938 00:42:48,400 --> 00:42:50,834 she's going to study the brain activity of Demi, 939 00:42:50,835 --> 00:42:53,104 an 11-year-old Swiss shepherd dog. 940 00:42:55,307 --> 00:42:57,341 So right now, I'm putting on 941 00:42:57,342 --> 00:42:59,343 the electrodes on Demi's head. 942 00:42:59,344 --> 00:43:01,546 These will measure the brain signal. 943 00:43:05,450 --> 00:43:06,617 Demi's owner, Timi, 944 00:43:06,618 --> 00:43:09,287 has brought in some of Demi's toys. 945 00:43:10,822 --> 00:43:13,657 Demi hears a message telling her what toy 946 00:43:13,658 --> 00:43:15,225 she's about to see. 947 00:43:15,226 --> 00:43:18,396 {\an8}RECORDED VOICE 948 00:43:19,898 --> 00:43:23,334 {\an8} Sometimes, Timi shows her that toy. 949 00:43:23,335 --> 00:43:25,502 {\an8}RECORDED VOICE 950 00:43:25,503 --> 00:43:26,870 But sometimes, 951 00:43:26,871 --> 00:43:28,706 Timi shows a different toy 952 00:43:28,707 --> 00:43:30,742 than what Demi is expecting. 953 00:43:31,943 --> 00:43:34,545 The idea is that if the dog sees a different object, 954 00:43:34,546 --> 00:43:36,413 it should be somehow surprised, 955 00:43:36,414 --> 00:43:38,782 and this surprise effect 956 00:43:38,783 --> 00:43:41,618 should be visible in the E.E.G. signal. 957 00:43:41,619 --> 00:43:44,856 {\an8}RECORDED VOICE 958 00:43:47,459 --> 00:43:50,260 What we can see here is that if a dog saw 959 00:43:50,261 --> 00:43:53,030 an object that matched the word that he just heard, 960 00:43:53,031 --> 00:43:55,199 this is the response that we get. 961 00:43:55,200 --> 00:43:57,034 But when we violate their expectations, 962 00:43:57,035 --> 00:43:59,703 so we show a different object, 963 00:43:59,704 --> 00:44:01,772 then we get a more positive response 964 00:44:01,773 --> 00:44:03,308 in the E.E.G. signal. 965 00:44:04,709 --> 00:44:05,909 Marianna believes 966 00:44:05,910 --> 00:44:08,145 that the difference in electrical activity 967 00:44:08,146 --> 00:44:09,813 indicates that when dogs hear a word, 968 00:44:09,814 --> 00:44:13,518 it can evoke a mental representation of an object. 969 00:44:15,453 --> 00:44:16,820 We can claim, based on this study, 970 00:44:16,821 --> 00:44:20,324 that when they hear the name of an object, 971 00:44:20,325 --> 00:44:24,261 they activate something, the so-called memory of this object. 972 00:44:24,262 --> 00:44:27,431 So we think that the dog is actually surprised 973 00:44:27,432 --> 00:44:29,299 when it sees a different thing, 974 00:44:29,300 --> 00:44:32,970 but it is surprised because it has an expectation 975 00:44:32,971 --> 00:44:35,873 and because it understands the meaning of that word. 976 00:44:37,208 --> 00:44:39,743 That's a qualitatively different way 977 00:44:39,744 --> 00:44:41,845 of using language, 978 00:44:41,846 --> 00:44:44,615 as opposed to associative learning. 979 00:44:46,184 --> 00:44:48,285 So yes, I would say 980 00:44:48,286 --> 00:44:50,821 there's a possibility that they understand words 981 00:44:50,822 --> 00:44:52,389 similarly to us. 982 00:44:54,592 --> 00:44:56,560 There are limitations in interpreting 983 00:44:56,561 --> 00:44:58,595 E.E.G. signals like these 984 00:44:58,596 --> 00:45:01,365 {\an8}and figuring out exactly what they tell us 985 00:45:01,366 --> 00:45:05,969 {\an8}about how much an animal really understands. 986 00:45:05,970 --> 00:45:08,038 {\an8}But if dogs do understand words 987 00:45:08,039 --> 00:45:10,040 {\an8}in similar ways to us, 988 00:45:10,041 --> 00:45:11,642 {\an8}could they start putting them together 989 00:45:11,643 --> 00:45:15,112 {\an8}and using them in more complex ways? 990 00:45:18,016 --> 00:45:19,416 To understand that, 991 00:45:19,417 --> 00:45:22,387 we need to look at the dogs that use a lot of buttons. 992 00:45:23,388 --> 00:45:25,255 {\an8} Come. Outside. 993 00:45:25,256 --> 00:45:27,324 Okay, we can go outside. 994 00:45:27,325 --> 00:45:28,726 Come on, girl. 995 00:45:30,862 --> 00:45:34,631 {\an8} Of all the dogs taking part in Federico's study, 996 00:45:34,632 --> 00:45:37,001 {\an8}most use just a handful of buttons. 997 00:45:39,671 --> 00:45:41,638 {\an8}But there are at least 30 dogs 998 00:45:41,639 --> 00:45:44,976 {\an8}who routinely use at least 50 different buttons. 999 00:45:46,611 --> 00:45:49,613 {\an8}The holy grail would be to find dogs that are capable 1000 00:45:49,614 --> 00:45:52,082 {\an8}of using a limited number of words 1001 00:45:52,083 --> 00:45:53,684 to express new ideas 1002 00:45:53,685 --> 00:45:56,353 that they don't have specific buttons for. 1003 00:45:56,354 --> 00:45:57,788 We humans have 1004 00:45:57,789 --> 00:45:59,256 a limited vocabulary. 1005 00:45:59,257 --> 00:46:02,159 But the reality is, there's no constraint 1006 00:46:02,160 --> 00:46:04,862 on how many different things we can express. 1007 00:46:04,863 --> 00:46:08,398 The only way that we can get that unlimited ability 1008 00:46:08,399 --> 00:46:10,334 from a limited vocabulary 1009 00:46:10,335 --> 00:46:12,769 is by combining those different terms, 1010 00:46:12,770 --> 00:46:15,305 those different words, that vocabulary, 1011 00:46:15,306 --> 00:46:17,942 in different ways to create new concepts. 1012 00:46:18,943 --> 00:46:21,545 This ability to say anything with limited words 1013 00:46:21,546 --> 00:46:23,780 is known as productivity. 1014 00:46:23,781 --> 00:46:25,349 I can give you an example. 1015 00:46:25,350 --> 00:46:26,517 My two-year-old kept telling me 1016 00:46:26,518 --> 00:46:30,487 that he wanted to be dressed as a green lion, 1017 00:46:30,488 --> 00:46:32,823 and I was, like, "What are you talking about? 1018 00:46:32,824 --> 00:46:34,324 There is no green lion." 1019 00:46:34,325 --> 00:46:36,860 And at some point, he sees somebody dressed as a dinosaur, 1020 00:46:36,861 --> 00:46:39,396 points to it, and says, "Green lion." 1021 00:46:39,397 --> 00:46:40,864 And I'm, like, "There you go. 1022 00:46:40,865 --> 00:46:42,366 "You don't have a word for dinosaur. 1023 00:46:42,367 --> 00:46:43,767 "But by saying 'green lion,' 1024 00:46:43,768 --> 00:46:46,303 "you're getting as close as you can with your vocabulary 1025 00:46:46,304 --> 00:46:49,573 to telling me, 'This is what I want.'" 1026 00:46:49,574 --> 00:46:51,875 Linguistic productivity is absolutely central 1027 00:46:51,876 --> 00:46:54,411 to the way that humans do language. 1028 00:46:54,412 --> 00:46:58,382 So that cognitive ability is very, very powerful 1029 00:46:58,383 --> 00:47:00,451 and very, very rare in the animal world. 1030 00:47:01,753 --> 00:47:05,856 It would be a huge step up in terms of cognitive ability. 1031 00:47:05,857 --> 00:47:08,158 So could some dogs be showing signs of it? 1032 00:47:11,029 --> 00:47:12,496 In order to look at productivity, 1033 00:47:12,497 --> 00:47:15,666 you need to have dogs that can combine at least two buttons. 1034 00:47:15,667 --> 00:47:19,336 As of yesterday, we have 790 dogs in our pool 1035 00:47:19,337 --> 00:47:21,805 that do multi-button combinations. 1036 00:47:21,806 --> 00:47:25,175 It is possible that out of this 700, 1037 00:47:25,176 --> 00:47:27,778 almost 800 individuals, 1038 00:47:27,779 --> 00:47:28,912 you can now look at some, 1039 00:47:28,913 --> 00:47:31,348 and maybe there's gonna be one, two, 1040 00:47:31,349 --> 00:47:34,184 three, five, ten, 20 that actually have 1041 00:47:34,185 --> 00:47:35,919 the ability to kind of refer to things 1042 00:47:35,920 --> 00:47:37,321 that you don't have a word for 1043 00:47:37,322 --> 00:47:39,690 or by combining other signals. 1044 00:47:39,691 --> 00:47:41,858 We have anecdotal reports from some of them 1045 00:47:41,859 --> 00:47:45,996 that they seem to be doing it. 1046 00:47:45,997 --> 00:47:47,931 {\an8}One time, when Stella's "beach" button broke, 1047 00:47:47,932 --> 00:47:50,200 {\an8}she went right over, 1048 00:47:50,201 --> 00:47:52,669 looked exactly where her "beach" button was, 1049 00:47:52,670 --> 00:47:56,506 she kind of sniffed it, looked at the empty space, 1050 00:47:56,507 --> 00:47:57,708 and then immediately after, 1051 00:47:57,709 --> 00:47:59,811 said, "Help," "Water," "Outside." 1052 00:48:01,179 --> 00:48:03,815 Help. Water. 1053 00:48:05,383 --> 00:48:06,917 Outside. 1054 00:48:06,918 --> 00:48:08,719 That was something that had never been taught. 1055 00:48:08,720 --> 00:48:12,589 I had never referred to the beach as "water outside." 1056 00:48:12,590 --> 00:48:14,591 But she put those together 1057 00:48:14,592 --> 00:48:16,493 in order to communicate that concept to me 1058 00:48:16,494 --> 00:48:19,230 when her "beach" button wasn't available anymore. 1059 00:48:20,298 --> 00:48:21,798 It's impossible to know 1060 00:48:21,799 --> 00:48:24,268 whether Christina's interpretation is correct. 1061 00:48:24,269 --> 00:48:27,137 But clips like these are helpful for Federico, 1062 00:48:27,138 --> 00:48:30,107 to highlight possible areas for future research. 1063 00:48:30,108 --> 00:48:32,944 Water. Outside. 1064 00:48:33,911 --> 00:48:35,312 This is the reason I am doing it, 1065 00:48:35,313 --> 00:48:37,314 even though people told me I might lose my job. 1066 00:48:37,315 --> 00:48:40,317 It's because I've seen things like that. 1067 00:48:40,318 --> 00:48:43,220 I've seen animals putting together two or three buttons 1068 00:48:43,221 --> 00:48:46,290 in ways that look like they were producing sentences. 1069 00:48:46,291 --> 00:48:48,492 But of course, you should not stop there, 1070 00:48:48,493 --> 00:48:49,893 because that thing could have been trained. 1071 00:48:49,894 --> 00:48:51,928 We'll go to the beach. 1072 00:48:51,929 --> 00:48:54,064 {\an8} It might look like productivity, 1073 00:48:54,065 --> 00:48:55,999 {\an8}but we don't know for sure that it is. 1074 00:48:56,000 --> 00:48:57,267 I do find interesting 1075 00:48:57,268 --> 00:48:59,569 that she pressed buttons that are near "beach." 1076 00:48:59,570 --> 00:49:01,438 Help. Water. 1077 00:49:01,439 --> 00:49:02,673 I would have loved 1078 00:49:02,674 --> 00:49:04,541 to know if she would have done the same 1079 00:49:04,542 --> 00:49:07,278 if, say, "beach" was on the other side of the soundboard. 1080 00:49:09,247 --> 00:49:12,916 Although Stella had never before used those buttons together, 1081 00:49:12,917 --> 00:49:17,421 testing what she meant by them after the event is impossible. 1082 00:49:17,422 --> 00:49:18,822 After that set of buttons is hit, 1083 00:49:18,823 --> 00:49:22,192 if Stella then goes to the beach, 1084 00:49:22,193 --> 00:49:24,227 now that set of buttons 1085 00:49:24,228 --> 00:49:26,263 definitely means "beach." 1086 00:49:26,264 --> 00:49:28,732 So then it's almost impossible for us to know 1087 00:49:28,733 --> 00:49:30,167 what, the first button press occurrence, 1088 00:49:30,168 --> 00:49:33,970 what Stella had intended, because now 1089 00:49:33,971 --> 00:49:35,640 it always means "beach." 1090 00:49:41,746 --> 00:49:44,848 So, can dogs talk? 1091 00:49:44,849 --> 00:49:47,951 I think Parker can communicate using 1092 00:49:47,952 --> 00:49:50,153 {\an8}English-language words. 1093 00:49:50,154 --> 00:49:52,756 {\an8}Do you want to do training? 1094 00:49:52,757 --> 00:49:53,757 {\an8} No. 1095 00:49:53,758 --> 00:49:55,659 {\an8} Okay! 1096 00:49:55,660 --> 00:49:57,795 We won't do training, then. 1097 00:49:59,230 --> 00:50:01,499 If you want to call that talking, great. 1098 00:50:02,867 --> 00:50:04,701 I say that Stella talks 1099 00:50:04,702 --> 00:50:07,471 because she's using words to communicate. 1100 00:50:07,472 --> 00:50:09,539 Outside. 1101 00:50:09,540 --> 00:50:12,042 The fact that it's not a trained behavior, 1102 00:50:12,043 --> 00:50:13,477 and that it's independent, 1103 00:50:13,478 --> 00:50:16,313 and that she can use words 1104 00:50:16,314 --> 00:50:17,681 to mean a lot of different things 1105 00:50:17,682 --> 00:50:22,687 indicates language, not just a conditioned trick. 1106 00:50:24,155 --> 00:50:25,689 It's not surprising that owners, 1107 00:50:25,690 --> 00:50:28,925 with their close relationships and strong communicative bonds, 1108 00:50:28,926 --> 00:50:33,096 might feel their dogs are talking. 1109 00:50:33,097 --> 00:50:34,598 Love you. 1110 00:50:34,599 --> 00:50:35,599 Stella. 1111 00:50:35,600 --> 00:50:36,666 Love you, Stella. 1112 00:50:36,667 --> 00:50:38,702 But scientific understanding 1113 00:50:38,703 --> 00:50:40,103 of their linguistic capabilities 1114 00:50:40,104 --> 00:50:42,973 is still in its early days. 1115 00:50:42,974 --> 00:50:45,008 Bedtime. 1116 00:50:45,009 --> 00:50:47,244 We're very much at the beginning of what we're doing 1117 00:50:47,245 --> 00:50:49,212 {\an8}and we need to be ready 1118 00:50:49,213 --> 00:50:51,415 {\an8}to collect data for years and years. 1119 00:50:51,416 --> 00:50:53,116 {\an8}Out, outside. 1120 00:50:53,117 --> 00:50:55,652 {\an8}And to me, the exciting part is 1121 00:50:55,653 --> 00:50:57,087 {\an8}that we now get a better sense 1122 00:50:57,088 --> 00:50:58,822 {\an8}of what might be happening out there. 1123 00:50:58,823 --> 00:51:00,290 {\an8} Outside. 1124 00:51:00,291 --> 00:51:02,959 Maybe you can see something that nobody else could see before, 1125 00:51:02,960 --> 00:51:05,596 because we just didn't have access to all this data. 1126 00:51:06,697 --> 00:51:08,298 Chewy. 1127 00:51:08,299 --> 00:51:10,267 We're really at the start of this journey 1128 00:51:10,268 --> 00:51:12,269 of understanding animal cognition, 1129 00:51:12,270 --> 00:51:15,071 animal communication, and where that lies 1130 00:51:15,072 --> 00:51:18,643 on that spectrum between language and non-language. 1131 00:51:20,278 --> 00:51:22,213 Where. Bye. 1132 00:51:23,481 --> 00:51:27,584 Where... 1133 00:51:27,585 --> 00:51:28,952 Bye. 1134 00:51:28,953 --> 00:51:31,054 Where bye? 1135 00:51:31,055 --> 00:51:32,689 Are you asking where you're going? 1136 00:51:32,690 --> 00:51:34,991 You're going to go see Natalie! 1137 00:51:34,992 --> 00:51:37,761 It's clear that dogs will never be capable 1138 00:51:37,762 --> 00:51:40,030 of using our language to generate the rich, 1139 00:51:40,031 --> 00:51:43,266 grammatically correct, infinite range of sentences 1140 00:51:43,267 --> 00:51:45,402 that humans can create. 1141 00:51:45,403 --> 00:51:46,937 Does it matter 1142 00:51:46,938 --> 00:51:48,738 if the sentence you're saying is grammatical or not, 1143 00:51:48,739 --> 00:51:50,774 as long as you're getting your point across, 1144 00:51:50,775 --> 00:51:53,210 and you're actually benefiting from that two-way communication? 1145 00:51:53,211 --> 00:51:54,611 I don't think so. 1146 00:51:56,013 --> 00:51:59,015 Whether dogs are using language or not, 1147 00:51:59,016 --> 00:52:01,284 {\an8}with a new tool in their communication kit 1148 00:52:01,285 --> 00:52:04,020 {\an8}and access to our words, 1149 00:52:04,021 --> 00:52:06,723 {\an8}perhaps the close bond between humans and dogs 1150 00:52:06,724 --> 00:52:09,560 {\an8}could become even closer.