1 00:00:01,680 --> 00:00:04,080 The following content is provided under a Creative 2 00:00:04,080 --> 00:00:05,620 Commons license. 3 00:00:05,620 --> 00:00:07,920 Your support will help MIT OpenCourseWare 4 00:00:07,920 --> 00:00:12,310 continue to offer high-quality educational resources for free. 5 00:00:12,310 --> 00:00:14,910 To make a donation or view additional materials 6 00:00:14,910 --> 00:00:18,870 from hundreds of MIT courses, visit MIT OpenCourseWare 7 00:00:18,870 --> 00:00:21,350 at ocw.mit.edu. 8 00:00:21,350 --> 00:00:23,100 ALIA MARTIN: I'm going to be talking today 9 00:00:23,100 --> 00:00:25,215 about how infants and kids start to develop 10 00:00:25,215 --> 00:00:26,610 an understanding of communication 11 00:00:26,610 --> 00:00:28,776 and why this is an important topic for understanding 12 00:00:28,776 --> 00:00:31,210 human intelligence. 13 00:00:31,210 --> 00:00:32,980 So what is communication? 14 00:00:32,980 --> 00:00:35,790 So really, basically, communication 15 00:00:35,790 --> 00:00:38,354 is a transfer or exchange of information. 16 00:00:38,354 --> 00:00:40,020 And importantly, in human communication, 17 00:00:40,020 --> 00:00:41,754 this isn't just any kind of information, 18 00:00:41,754 --> 00:00:43,920 but it's specifically the kind of information that's 19 00:00:43,920 --> 00:00:46,114 in our heads and in our minds. 20 00:00:46,114 --> 00:00:48,030 So for example, there's information in my head 21 00:00:48,030 --> 00:00:50,113 right now that I'm going to be transferring to you 22 00:00:50,113 --> 00:00:51,420 over the course of this talk. 23 00:00:51,420 --> 00:00:53,460 And the reason we need communication to do this 24 00:00:53,460 --> 00:00:55,710 and that we use it all the time is obviously 25 00:00:55,710 --> 00:00:57,760 that everyone has minds of their own 26 00:00:57,760 --> 00:01:00,630 and that we don't have access to the mental states of others. 27 00:01:00,630 --> 00:01:03,120 And we can't automatically transfer our own mental states 28 00:01:03,120 --> 00:01:06,840 to theirs without some means of making those mental states 29 00:01:06,840 --> 00:01:10,110 observable in the form of communicating with them. 30 00:01:10,110 --> 00:01:13,719 So people are both cognitive beings-- 31 00:01:13,719 --> 00:01:16,260 we have to sort of navigate the thoughts in our heads and try 32 00:01:16,260 --> 00:01:18,240 to figure out the thoughts that are in other's heads-- 33 00:01:18,240 --> 00:01:19,448 but we're also social beings. 34 00:01:19,448 --> 00:01:21,150 We really benefit from gaining access 35 00:01:21,150 --> 00:01:22,470 to the thoughts in the minds of others 36 00:01:22,470 --> 00:01:24,261 and giving them access to our own thoughts. 37 00:01:24,261 --> 00:01:27,450 So for example, for cooperation, competition, learning, 38 00:01:27,450 --> 00:01:29,700 and teaching, and all the kinds of social interactions 39 00:01:29,700 --> 00:01:31,200 we do with each other, we have to be 40 00:01:31,200 --> 00:01:33,570 able to share the contents of our mental states. 41 00:01:33,570 --> 00:01:36,420 And it's this sort of joint, both cognitive and social 42 00:01:36,420 --> 00:01:37,920 nature of communication that I think 43 00:01:37,920 --> 00:01:39,878 makes it especially important for understanding 44 00:01:39,878 --> 00:01:41,550 the development of human intelligence. 45 00:01:41,550 --> 00:01:44,010 It sort of crosscuts a lot of these core knowledge 46 00:01:44,010 --> 00:01:47,010 domains that Liz talked about in her talk. 47 00:01:47,010 --> 00:01:48,834 So human communication requires reasoning 48 00:01:48,834 --> 00:01:50,250 about others' cognitive states, so 49 00:01:50,250 --> 00:01:52,249 understanding what others' beliefs, and desires, 50 00:01:52,249 --> 00:01:53,520 and intentions are. 51 00:01:53,520 --> 00:01:56,320 It also requires reasoning about social interaction, 52 00:01:56,320 --> 00:01:58,752 so understanding that typically these kinds of thoughts, 53 00:01:58,752 --> 00:02:00,210 and beliefs, and intentions are not 54 00:02:00,210 --> 00:02:01,620 shared among different people. 55 00:02:01,620 --> 00:02:03,630 Everyone has their own that are unobservable 56 00:02:03,630 --> 00:02:05,880 and housed in their own heads. 57 00:02:05,880 --> 00:02:07,770 But in the context of a particular kind 58 00:02:07,770 --> 00:02:10,830 of social interaction, these people 59 00:02:10,830 --> 00:02:13,440 can intentionally share their thoughts with each other. 60 00:02:13,440 --> 00:02:15,199 And communication often involves reasoning 61 00:02:15,199 --> 00:02:17,490 about a third thing, which is language or communicative 62 00:02:17,490 --> 00:02:20,800 signals outside of language as well. 63 00:02:20,800 --> 00:02:22,970 And so I'm going to back to this point at the end, 64 00:02:22,970 --> 00:02:24,470 but all these types of reasoning are 65 00:02:24,470 --> 00:02:27,540 going to come up in the studies about infant communication 66 00:02:27,540 --> 00:02:29,580 that I describe in the talk. 67 00:02:32,950 --> 00:02:34,831 And so Liz talked about how infants start out 68 00:02:34,831 --> 00:02:37,330 with these coherent, separate, principle, and core knowledge 69 00:02:37,330 --> 00:02:39,910 systems that contain some limited representations 70 00:02:39,910 --> 00:02:42,520 for reasoning about the world, and which kind of come together 71 00:02:42,520 --> 00:02:44,219 around the end of the first year. 72 00:02:44,219 --> 00:02:46,510 So if we want to figure out how to build a model of how 73 00:02:46,510 --> 00:02:49,030 social cognitive development works, 74 00:02:49,030 --> 00:02:52,090 we're going to need to understand how an infant starts 75 00:02:52,090 --> 00:02:54,610 to figure out, using the core knowledge 76 00:02:54,610 --> 00:02:58,000 that they have early in life, the components of communication 77 00:02:58,000 --> 00:03:00,940 and also how these components come together so that infants 78 00:03:00,940 --> 00:03:03,790 can build a more complex causal model of communication 79 00:03:03,790 --> 00:03:04,600 like adults have. 80 00:03:07,620 --> 00:03:10,830 So this is just to illustrate the points 81 00:03:10,830 --> 00:03:13,094 I'm making about the communicative situation. 82 00:03:13,094 --> 00:03:14,760 So in any given communicative situation, 83 00:03:14,760 --> 00:03:17,714 we can't just think about the words that are being said 84 00:03:17,714 --> 00:03:18,630 or what we're hearing. 85 00:03:18,630 --> 00:03:21,088 We need to think about it in this broader context of having 86 00:03:21,088 --> 00:03:24,120 a communicator or maybe multiple communicators, an addressee 87 00:03:24,120 --> 00:03:26,730 or an audience, and also the things that the communicator is 88 00:03:26,730 --> 00:03:28,177 saying in the broader context that 89 00:03:28,177 --> 00:03:29,760 allow us to figure out what's going on 90 00:03:29,760 --> 00:03:31,960 in this social interaction. 91 00:03:31,960 --> 00:03:33,540 So it really requires understanding 92 00:03:33,540 --> 00:03:35,250 that the communicator's mental states are 93 00:03:35,250 --> 00:03:37,320 being transferred in this causal way 94 00:03:37,320 --> 00:03:40,809 to the mind of the addressee. 95 00:03:40,809 --> 00:03:42,600 So I'm going to structure the talk in terms 96 00:03:42,600 --> 00:03:44,516 of some important insights from the philosophy 97 00:03:44,516 --> 00:03:46,350 of human language and communication 98 00:03:46,350 --> 00:03:49,830 that I think have really guided the way 99 00:03:49,830 --> 00:03:51,780 researchers in the last, say, 50 years have 100 00:03:51,780 --> 00:03:53,430 been thinking about the development 101 00:03:53,430 --> 00:03:56,045 of human communication as well. 102 00:03:56,045 --> 00:03:58,170 So I'm going to broadly illustrate the three points 103 00:03:58,170 --> 00:04:00,461 that I'm taking from each, and then I'll expand on them 104 00:04:00,461 --> 00:04:03,035 in each section. 105 00:04:03,035 --> 00:04:04,410 So the first insight that I think 106 00:04:04,410 --> 00:04:05,951 is really important for understanding 107 00:04:05,951 --> 00:04:10,629 how we think about communication comes from John Austin, who 108 00:04:10,629 --> 00:04:11,920 brought up the important point. 109 00:04:11,920 --> 00:04:13,586 So before Austin, people tended to think 110 00:04:13,586 --> 00:04:16,079 about language or study language a lot in terms of language 111 00:04:16,079 --> 00:04:19,440 itself, and the semantics, and syntax, and just the language 112 00:04:19,440 --> 00:04:21,240 signal. 113 00:04:21,240 --> 00:04:23,250 But Austin pointed out that language is actually 114 00:04:23,250 --> 00:04:26,670 not just about the content, but it's also an action. 115 00:04:26,670 --> 00:04:30,750 It's actually something that we use, and we do things with it, 116 00:04:30,750 --> 00:04:33,060 in the same way that we do things and accomplish things 117 00:04:33,060 --> 00:04:37,969 in the world using other kinds of actions that we engage in. 118 00:04:37,969 --> 00:04:40,260 So Austin sort of brought up this important distinction 119 00:04:40,260 --> 00:04:43,770 that language is not just about the content. 120 00:04:43,770 --> 00:04:45,150 So to illustrate with an example, 121 00:04:45,150 --> 00:04:47,094 in this communicative interaction, 122 00:04:47,094 --> 00:04:48,010 there's language here. 123 00:04:48,010 --> 00:04:50,160 This person is saying, is there any salt? 124 00:04:50,160 --> 00:04:51,780 And Austin analyzed this in terms 125 00:04:51,780 --> 00:04:54,180 of not only what the words meant, 126 00:04:54,180 --> 00:04:57,090 but in terms of three layers of intentionality. 127 00:04:57,090 --> 00:04:59,940 So in this simple interaction, he pointed out 128 00:04:59,940 --> 00:05:03,057 that there's a locutionary act, or the meaning of the sentence, 129 00:05:03,057 --> 00:05:04,890 which is that there's a question being asked 130 00:05:04,890 --> 00:05:07,290 about the presence of salt. But that there's 131 00:05:07,290 --> 00:05:08,920 actually something else going on here 132 00:05:08,920 --> 00:05:12,210 as well, which is that there's an illocutionary action, which 133 00:05:12,210 --> 00:05:14,910 is what he considers to be the intention underlying 134 00:05:14,910 --> 00:05:17,920 the action, which is to request the salt. 135 00:05:17,920 --> 00:05:19,830 So if you actually just read this sentence, 136 00:05:19,830 --> 00:05:21,270 it's kind of under-determined. 137 00:05:21,270 --> 00:05:24,000 It's not necessarily obvious that this person's 138 00:05:24,000 --> 00:05:28,759 asking the other person at this table to pass the salt. 139 00:05:28,759 --> 00:05:30,300 But if you take it within the broader 140 00:05:30,300 --> 00:05:32,216 context of the interaction, you can understand 141 00:05:32,216 --> 00:05:34,920 that what's really going on here is not just a sentence being 142 00:05:34,920 --> 00:05:38,800 produced, but rather one individual requesting something 143 00:05:38,800 --> 00:05:40,800 from another, who is then supposed to understand 144 00:05:40,800 --> 00:05:46,380 the request, and in the third layer of analysis, 145 00:05:46,380 --> 00:05:48,990 cause the addressee to provide the salt. 146 00:05:48,990 --> 00:05:52,246 So the idea is that there are these multiple things going on 147 00:05:52,246 --> 00:05:54,120 in the context of a communicative interaction 148 00:05:54,120 --> 00:05:58,079 involving language that go beyond the words that 149 00:05:58,079 --> 00:05:59,370 are actually being spoken here. 150 00:06:02,105 --> 00:06:06,310 And so if you're an infant, just like in philosophy of language, 151 00:06:06,310 --> 00:06:08,890 the focus in infant cognitive development-- 152 00:06:08,890 --> 00:06:10,660 also for a long time, and still is, 153 00:06:10,660 --> 00:06:13,060 we still need to understand these things-- 154 00:06:13,060 --> 00:06:15,610 was on understanding how infants learn words, 155 00:06:15,610 --> 00:06:17,889 or how infants figure out the meaning of language, 156 00:06:17,889 --> 00:06:19,930 and how it relates to objects in the environment, 157 00:06:19,930 --> 00:06:21,640 and later, to the abstract concepts 158 00:06:21,640 --> 00:06:23,410 that language indicates. 159 00:06:23,410 --> 00:06:25,990 But importantly, understanding communication for an infant 160 00:06:25,990 --> 00:06:27,989 is going to be more than about just figuring out 161 00:06:27,989 --> 00:06:29,140 the meaning of these words. 162 00:06:29,140 --> 00:06:31,769 And so following the philosophical tradition 163 00:06:31,769 --> 00:06:33,310 of people like Austin, who introduced 164 00:06:33,310 --> 00:06:35,452 the study of pragmatics to language, 165 00:06:35,452 --> 00:06:36,910 developmental psychologists as well 166 00:06:36,910 --> 00:06:39,730 started thinking about the importance of recognizing 167 00:06:39,730 --> 00:06:42,820 communication as this broader action in its context 168 00:06:42,820 --> 00:06:46,150 for understanding how infants come to be good 169 00:06:46,150 --> 00:06:47,812 communicators themselves. 170 00:06:51,510 --> 00:06:54,110 So if communication involves this whole context of action 171 00:06:54,110 --> 00:06:56,135 and interaction between people, if you're 172 00:06:56,135 --> 00:06:58,010 a baby who's in the business of understanding 173 00:06:58,010 --> 00:07:00,010 the social world, the agent world, and the world 174 00:07:00,010 --> 00:07:02,480 of language, you're going to be well-served by developing 175 00:07:02,480 --> 00:07:05,000 an ability to identify these kinds of communicative actions 176 00:07:05,000 --> 00:07:06,800 or situations and their components, 177 00:07:06,800 --> 00:07:09,690 and to begin to understand how they work. 178 00:07:09,690 --> 00:07:12,050 So how do infants start to figure out 179 00:07:12,050 --> 00:07:14,840 when communication is going on in the world around them, 180 00:07:14,840 --> 00:07:17,150 rather than just identifying words and associating them 181 00:07:17,150 --> 00:07:18,777 with objects? 182 00:07:18,777 --> 00:07:20,360 So actually, there's a lot of evidence 183 00:07:20,360 --> 00:07:24,080 that infants are identifying key features 184 00:07:24,080 --> 00:07:27,540 of communicative situations really early in life. 185 00:07:27,540 --> 00:07:30,140 So from the time they're born, newborn human infants 186 00:07:30,140 --> 00:07:33,350 prefer listening to speech over other sounds. 187 00:07:33,350 --> 00:07:35,060 So the typical method that's used 188 00:07:35,060 --> 00:07:36,980 to measure the preference of an infant who's 189 00:07:36,980 --> 00:07:39,080 only one to four days old is to have 190 00:07:39,080 --> 00:07:42,020 them suck on a pacifier that's connected 191 00:07:42,020 --> 00:07:45,380 to a machine that detects how strongly infants are sucking. 192 00:07:45,380 --> 00:07:48,230 And then the sucking is used as a measure of infants 193 00:07:48,230 --> 00:07:50,180 arousal upon hearing a particular sound 194 00:07:50,180 --> 00:07:52,370 or being exposed to a particular stimulus. 195 00:07:52,370 --> 00:07:56,210 And so what researchers found is that if you have infants suck 196 00:07:56,210 --> 00:07:58,100 on this pacifier while listening to speech 197 00:07:58,100 --> 00:08:01,110 sounds and nonspeech sounds in alternation, 198 00:08:01,110 --> 00:08:03,860 nonspeech sounds being sine wave-produced sounds 199 00:08:03,860 --> 00:08:05,962 that are very similar to speech in their features. 200 00:08:05,962 --> 00:08:08,420 If you listen to it, it has the prosodic contours of speech 201 00:08:08,420 --> 00:08:11,600 and sounds a lot like it, but it's not actually speech. 202 00:08:11,600 --> 00:08:15,770 You can't actually parse any words from it. 203 00:08:15,770 --> 00:08:18,140 Infants showed a preference for listening 204 00:08:18,140 --> 00:08:19,900 to speech over nonspeech, or more arousal 205 00:08:19,900 --> 00:08:21,108 when they listened to speech. 206 00:08:21,108 --> 00:08:24,234 So they maintained their arousal for speech 207 00:08:24,234 --> 00:08:25,650 over the course of the experiment, 208 00:08:25,650 --> 00:08:27,710 but for nonspeech, it declined. 209 00:08:27,710 --> 00:08:29,600 And so extremely early in life, infants 210 00:08:29,600 --> 00:08:31,910 seem to have this bias for paying attention 211 00:08:31,910 --> 00:08:34,820 to the primary communicative signal of our species, that is, 212 00:08:34,820 --> 00:08:35,559 human speech. 213 00:08:39,640 --> 00:08:43,027 Quite early in life as well, infants 214 00:08:43,027 --> 00:08:45,610 recognize some of the important features of the social context 215 00:08:45,610 --> 00:08:47,260 surrounding speech, so getting more 216 00:08:47,260 --> 00:08:50,300 into the domain of the important features for communication. 217 00:08:50,300 --> 00:08:52,471 So by six months, infants seem to recognize 218 00:08:52,471 --> 00:08:54,220 that speech is human-produced, or at least 219 00:08:54,220 --> 00:08:56,020 associate speech with other humans, 220 00:08:56,020 --> 00:08:59,110 and also human-directed. 221 00:08:59,110 --> 00:09:02,290 So in one study, infants saw pictures of human faces 222 00:09:02,290 --> 00:09:06,474 or faces of monkeys, and they heard different sounds. 223 00:09:06,474 --> 00:09:07,390 So they either heard-- 224 00:09:11,374 --> 00:09:14,830 oh, it seems like the sound is not on, but that's OK. 225 00:09:14,830 --> 00:09:17,332 So they either-- oops. 226 00:09:20,580 --> 00:09:22,820 So they either heard a speech sound in a language 227 00:09:22,820 --> 00:09:24,980 they never heard before repeating itself, 228 00:09:24,980 --> 00:09:27,470 or they heard a monkey call. 229 00:09:27,470 --> 00:09:30,335 And infants in one trial either saw-- so in the first trial, 230 00:09:30,335 --> 00:09:32,210 say, they'd see the human face, and then they 231 00:09:32,210 --> 00:09:36,080 could listen repeatedly to a speech sound 232 00:09:36,080 --> 00:09:37,977 until they looked away for two seconds. 233 00:09:37,977 --> 00:09:39,560 And then they would see a monkey face, 234 00:09:39,560 --> 00:09:41,390 and then they would hear either a speech sound or a monkey 235 00:09:41,390 --> 00:09:43,010 sound and be able to look at this image 236 00:09:43,010 --> 00:09:44,780 while listening to that sound until they looked away 237 00:09:44,780 --> 00:09:45,800 for two seconds. 238 00:09:45,800 --> 00:09:47,690 And so they got all possible combinations. 239 00:09:47,690 --> 00:09:50,060 Sometimes they saw a human face and listened to speech. 240 00:09:50,060 --> 00:09:51,893 Sometimes they saw a human face and listened 241 00:09:51,893 --> 00:09:52,790 to the monkey sounds. 242 00:09:52,790 --> 00:09:54,373 And sometimes they saw the monkey face 243 00:09:54,373 --> 00:09:57,666 and listened to either human speech or the monkey sound. 244 00:09:57,666 --> 00:09:59,540 And the question was whether infants actually 245 00:09:59,540 --> 00:10:02,507 could match speech sounds to humans 246 00:10:02,507 --> 00:10:05,090 and recognize that they should be produced by the human rather 247 00:10:05,090 --> 00:10:06,665 than the monkey. 248 00:10:06,665 --> 00:10:07,790 And that's what they found. 249 00:10:07,790 --> 00:10:11,270 So you can see that when infants heard speech, 250 00:10:11,270 --> 00:10:13,790 they were much more likely to look at the-- 251 00:10:13,790 --> 00:10:16,654 yeah, they were looking at the human when they heard speech, 252 00:10:16,654 --> 00:10:18,320 and then they were looking at the monkey 253 00:10:18,320 --> 00:10:19,695 when they heard the monkey calls. 254 00:10:22,116 --> 00:10:24,240 Similarly, infants seem to understand by six months 255 00:10:24,240 --> 00:10:26,370 that speech is directed at other humans. 256 00:10:26,370 --> 00:10:29,370 So when they saw a person talking 257 00:10:29,370 --> 00:10:32,130 behind a barrier versus acting, so 258 00:10:32,130 --> 00:10:34,080 swiping with her hand behind a barrier, 259 00:10:34,080 --> 00:10:35,460 and then the barrier was revealed 260 00:10:35,460 --> 00:10:39,060 to reveal either a person or an object, 261 00:10:39,060 --> 00:10:41,975 infants looked longer for the speaking familiarization 262 00:10:41,975 --> 00:10:43,350 when they saw that there had been 263 00:10:43,350 --> 00:10:45,270 an object behind the barrier than a person, 264 00:10:45,270 --> 00:10:46,860 suggesting that they expected speech 265 00:10:46,860 --> 00:10:49,487 to be directed toward another person. 266 00:10:49,487 --> 00:10:51,570 And in contrast, when they saw the person swiping, 267 00:10:51,570 --> 00:10:53,190 they looked much longer when they saw a person 268 00:10:53,190 --> 00:10:54,856 behind the barrier because typically, we 269 00:10:54,856 --> 00:10:55,860 don't swipe at people. 270 00:10:55,860 --> 00:10:56,960 We tend to speak to them. 271 00:11:06,080 --> 00:11:07,910 So this suggests that infants understand 272 00:11:07,910 --> 00:11:10,490 some of the features of the context 273 00:11:10,490 --> 00:11:13,950 in which human communication is produced. 274 00:11:13,950 --> 00:11:16,095 Importantly, infants do still seem 275 00:11:16,095 --> 00:11:18,470 to be developing this ability over the first year of life 276 00:11:18,470 --> 00:11:20,511 because it's not until 10 months that they expect 277 00:11:20,511 --> 00:11:22,250 mutual gaze between speakers, which 278 00:11:22,250 --> 00:11:24,410 tends to be an important part of understanding 279 00:11:24,410 --> 00:11:27,750 the social context of communication. 280 00:11:27,750 --> 00:11:31,165 So in this study, infants saw two people facing each other. 281 00:11:31,165 --> 00:11:33,290 This is just one of the experiments from the study. 282 00:11:33,290 --> 00:11:35,600 But you can see that they saw the two people looking 283 00:11:35,600 --> 00:11:37,308 at each other and speaking to each other, 284 00:11:37,308 --> 00:11:41,540 or the two people looking apart and speaking to each other. 285 00:11:41,540 --> 00:11:44,240 Infants at nine months didn't differentiate 286 00:11:44,240 --> 00:11:45,470 between these at all. 287 00:11:45,470 --> 00:11:49,280 So at nine months, they didn't necessarily 288 00:11:49,280 --> 00:11:51,530 think that the people were going to look at each other 289 00:11:51,530 --> 00:11:52,760 when they were speaking. 290 00:11:52,760 --> 00:11:55,340 But at 10 months, you can see here 291 00:11:55,340 --> 00:11:58,580 that infants looked longer for the averted gaze 292 00:11:58,580 --> 00:11:59,840 than for the mutual gaze. 293 00:12:04,766 --> 00:12:06,390 So infants seem to be able to recognize 294 00:12:06,390 --> 00:12:10,250 when a communicative context is going on early in life. 295 00:12:10,250 --> 00:12:12,250 But we don't necessarily know from these studies 296 00:12:12,250 --> 00:12:15,000 whether they really understand that communication is happening 297 00:12:15,000 --> 00:12:17,442 or whether they're just detecting 298 00:12:17,442 --> 00:12:19,650 some important features of communicative interactions 299 00:12:19,650 --> 00:12:22,290 that might help them to glom onto communication so that they 300 00:12:22,290 --> 00:12:24,324 can eventually develop an ability to figure out 301 00:12:24,324 --> 00:12:26,490 what's going on within the communicative interaction 302 00:12:26,490 --> 00:12:27,604 themselves. 303 00:12:27,604 --> 00:12:29,520 So, so far in the studies that I've mentioned, 304 00:12:29,520 --> 00:12:30,570 there's nothing cognitive here. 305 00:12:30,570 --> 00:12:31,770 There's nothing really about infants 306 00:12:31,770 --> 00:12:33,870 having to understand the intentions or the mental states 307 00:12:33,870 --> 00:12:36,036 of the speaker going into the mind of the addressee, 308 00:12:36,036 --> 00:12:38,170 like I was talking about before. 309 00:12:38,170 --> 00:12:40,195 So a question I've explored in my research 310 00:12:40,195 --> 00:12:42,570 is whether infants recognize that when these features are 311 00:12:42,570 --> 00:12:44,490 in place, when you have a communicative signal, which 312 00:12:44,490 --> 00:12:45,872 is something they recognize, when 313 00:12:45,872 --> 00:12:47,580 you have a communicator and addressee who 314 00:12:47,580 --> 00:12:49,430 are socially engaged with each other, 315 00:12:49,430 --> 00:12:52,050 do infants recognize that communication can actually lead 316 00:12:52,050 --> 00:12:53,954 to the transfer of information? 317 00:12:56,920 --> 00:12:59,857 So do infants understand that speech, using speech, 318 00:12:59,857 --> 00:13:02,190 because this is something that infants seem to recognize 319 00:13:02,190 --> 00:13:04,590 as a signal that's important in communicative context, 320 00:13:04,590 --> 00:13:07,107 can transfer information between individuals? 321 00:13:10,690 --> 00:13:14,160 So we can use speech to communicate with each other 322 00:13:14,160 --> 00:13:16,210 about what we're interested in. 323 00:13:16,210 --> 00:13:18,180 So if I'm observing this interaction 324 00:13:18,180 --> 00:13:20,040 between a communicator and an addressee, 325 00:13:20,040 --> 00:13:22,686 I can infer that if the communicator says, the cup, 326 00:13:22,686 --> 00:13:24,060 and there's only one cup present, 327 00:13:24,060 --> 00:13:25,560 the addressee's probably going to be 328 00:13:25,560 --> 00:13:28,470 able to figure out what it is that the communicator wants. 329 00:13:28,470 --> 00:13:31,392 And I can figure this out even from a third party perspective. 330 00:13:31,392 --> 00:13:33,600 And I can also understand that other kinds of sounds, 331 00:13:33,600 --> 00:13:35,760 like perhaps a positive emotional vocalization, 332 00:13:35,760 --> 00:13:38,430 are not going to be as effective in communicating 333 00:13:38,430 --> 00:13:42,060 to the addressee what the communicator is interested in. 334 00:13:42,060 --> 00:13:44,130 So speech and also other communicative 335 00:13:44,130 --> 00:13:46,590 signals as well that can specify this sort of information, 336 00:13:46,590 --> 00:13:49,410 perhaps like pointing or particular kinds of gestures, 337 00:13:49,410 --> 00:13:53,340 can transfer information from one individual to another. 338 00:13:53,340 --> 00:13:56,010 So speech is going to be more effective for communicating 339 00:13:56,010 --> 00:14:00,210 than other kinds of noncommunicative actions. 340 00:14:00,210 --> 00:14:02,700 And importantly, for the purposes of this study, 341 00:14:02,700 --> 00:14:05,520 you're not just going to know that communicative information 342 00:14:05,520 --> 00:14:09,990 transfer is going on because you know what the word cup means, 343 00:14:09,990 --> 00:14:13,150 even in a situation where you're listening 344 00:14:13,150 --> 00:14:15,150 to a foreign language, and you have no idea what 345 00:14:15,150 --> 00:14:16,501 the meaning of the words are. 346 00:14:16,501 --> 00:14:18,000 Even in this situation, you're going 347 00:14:18,000 --> 00:14:20,375 to understand that some kind of communicative information 348 00:14:20,375 --> 00:14:21,577 transfer can happen. 349 00:14:21,577 --> 00:14:24,160 So we've all had the experience of being in a foreign country, 350 00:14:24,160 --> 00:14:25,410 presumably, where you're listening 351 00:14:25,410 --> 00:14:26,650 to people speak to each other. 352 00:14:26,650 --> 00:14:28,858 You can't necessarily understand what they're saying, 353 00:14:28,858 --> 00:14:31,090 but you know that information transfer is happening. 354 00:14:31,090 --> 00:14:33,256 So the interesting thing about communicative actions 355 00:14:33,256 --> 00:14:35,040 like speech is that when we witness them, 356 00:14:35,040 --> 00:14:36,831 we can know that people are communicatively 357 00:14:36,831 --> 00:14:38,490 sharing information, even when we 358 00:14:38,490 --> 00:14:42,480 don't know what information is being communicated ourselves. 359 00:14:42,480 --> 00:14:45,450 And this insight suggests that maybe, 360 00:14:45,450 --> 00:14:47,790 if you're an infant, one really good way 361 00:14:47,790 --> 00:14:50,430 to jump into the communicative interactions around you 362 00:14:50,430 --> 00:14:55,440 and start to understand what's going on 363 00:14:55,440 --> 00:14:57,810 is to be able to identify when others are communicating. 364 00:14:57,810 --> 00:14:59,630 And hearing the sounds of speech exchanged 365 00:14:59,630 --> 00:15:02,490 between two people in the context of a social interaction 366 00:15:02,490 --> 00:15:05,380 might be a really good way to do this. 367 00:15:05,380 --> 00:15:07,800 So an ability to figure out when communication's going on 368 00:15:07,800 --> 00:15:10,740 like this might provide infants with this really powerful way 369 00:15:10,740 --> 00:15:13,080 of being able to track the information flow between two 370 00:15:13,080 --> 00:15:15,720 other people's minds and also to figure out, 371 00:15:15,720 --> 00:15:17,760 based on their responses to each other, what 372 00:15:17,760 --> 00:15:21,870 the actual content of the words might mean. 373 00:15:21,870 --> 00:15:24,210 So we asked whether infants recognize 374 00:15:24,210 --> 00:15:28,447 that speech is communicative at 12 months of age. 375 00:15:28,447 --> 00:15:30,030 And so in the procedure of this study, 376 00:15:30,030 --> 00:15:32,280 infants were privy to a third party 377 00:15:32,280 --> 00:15:34,950 interaction between a communicator and an addressee. 378 00:15:34,950 --> 00:15:38,440 And the infant is always just observing the interaction. 379 00:15:38,440 --> 00:15:41,120 This is so that we can isolate whether infants think 380 00:15:41,120 --> 00:15:43,620 that certain communicative vocalizations can transfer 381 00:15:43,620 --> 00:15:46,230 information from the communicator to an addressee, 382 00:15:46,230 --> 00:15:49,179 even when the infant themselves has all the information. 383 00:15:49,179 --> 00:15:50,970 So the third party nature of this procedure 384 00:15:50,970 --> 00:15:53,310 is important here for seeing whether infants really 385 00:15:53,310 --> 00:15:56,370 are thinking about the fact that thoughts are typically isolated 386 00:15:56,370 --> 00:15:59,040 in particular people's minds, but with communication, they 387 00:15:59,040 --> 00:15:59,700 can be shared. 388 00:16:02,417 --> 00:16:04,500 So infants in these studies were given a violation 389 00:16:04,500 --> 00:16:05,860 of expectation paradigm. 390 00:16:05,860 --> 00:16:08,340 In this kind of experiment, as many of you probably know, 391 00:16:08,340 --> 00:16:11,610 infants are shown a story through a series of scenes. 392 00:16:11,610 --> 00:16:14,910 And then in the last scene or series of scenes, 393 00:16:14,910 --> 00:16:16,800 they're shown different kinds of endings. 394 00:16:16,800 --> 00:16:19,800 And we're interested in whether infants are surprised or look 395 00:16:19,800 --> 00:16:23,610 longer at some endings more than they do at others. 396 00:16:23,610 --> 00:16:25,890 So infants in these studies were shown a live display 397 00:16:25,890 --> 00:16:28,152 where they had an actor in front of them in a stage. 398 00:16:28,152 --> 00:16:30,360 First, they were familiarized with this actor showing 399 00:16:30,360 --> 00:16:32,730 a preference for an object by picking it up and playing 400 00:16:32,730 --> 00:16:36,090 with it repeatedly in three separate scenes. 401 00:16:36,090 --> 00:16:38,670 And so, as in the studies by Amanda Woodward that Liz 402 00:16:38,670 --> 00:16:42,400 already mentioned, infants, by 12 months 403 00:16:42,400 --> 00:16:45,010 and as early as three months, will attribute a goal 404 00:16:45,010 --> 00:16:48,811 to this person for reaching for that particular target object. 405 00:16:48,811 --> 00:16:50,310 So after being familiarized to this, 406 00:16:50,310 --> 00:16:52,799 infants saw an addressee, who was a new person, 407 00:16:52,799 --> 00:16:54,840 present in a totally different part of the stage. 408 00:16:54,840 --> 00:16:56,464 They had never seen this person before, 409 00:16:56,464 --> 00:16:59,010 and they'd never seen the two people together before. 410 00:16:59,010 --> 00:17:01,230 This person reached for both objects in turn. 411 00:17:01,230 --> 00:17:03,180 So first, she grabbed this one, then this one, 412 00:17:03,180 --> 00:17:04,640 and then this one, and then this one again 413 00:17:04,640 --> 00:17:07,181 to show that she didn't have a preference between the objects 414 00:17:07,181 --> 00:17:10,069 and could reach both of them. 415 00:17:10,069 --> 00:17:12,680 Then in the test scene, so this is the ending 416 00:17:12,680 --> 00:17:15,167 that we showed infants, they saw the two people together 417 00:17:15,167 --> 00:17:17,000 for the first time, but now the communicator 418 00:17:17,000 --> 00:17:19,730 couldn't reach the objects because only her face had 419 00:17:19,730 --> 00:17:21,020 access to the stage. 420 00:17:21,020 --> 00:17:24,470 But the addressee could still reach them just fine. 421 00:17:24,470 --> 00:17:27,560 At this point, the crucial manipulation 422 00:17:27,560 --> 00:17:30,750 was the vocalization uttered by the communicator. 423 00:17:30,750 --> 00:17:34,390 So she either produced a speech sound, the nonsense word koba, 424 00:17:34,390 --> 00:17:38,300 which infants had not heard before, she produced 425 00:17:38,300 --> 00:17:40,520 a coughing sound, so something that would typically 426 00:17:40,520 --> 00:17:42,920 be physiological and not intentional or communicative 427 00:17:42,920 --> 00:17:45,867 in any way, or an emotional vocalization, so a sound 428 00:17:45,867 --> 00:17:47,450 that perhaps could convey information. 429 00:17:47,450 --> 00:17:50,720 If I say, ooh, you might think I'm interested in something, 430 00:17:50,720 --> 00:17:52,814 I'm feeling positive, but you won't necessarily 431 00:17:52,814 --> 00:17:53,480 know what it is. 432 00:17:53,480 --> 00:17:57,570 So this one's sort of in the middle of the other two. 433 00:17:57,570 --> 00:18:00,025 And then, infants saw the addressee either 434 00:18:00,025 --> 00:18:02,150 provide the target object that the communicator had 435 00:18:02,150 --> 00:18:04,720 reached for before or the nontarget object. 436 00:18:04,720 --> 00:18:07,640 And the question is, do infants understand that speech 437 00:18:07,640 --> 00:18:09,980 as a communicative signal, do they have some expectation 438 00:18:09,980 --> 00:18:12,438 that even though they've never heard this particular speech 439 00:18:12,438 --> 00:18:15,230 sound used before, that speech can transfer information such 440 00:18:15,230 --> 00:18:18,320 that the addressee will now be able to select 441 00:18:18,320 --> 00:18:19,550 the correct object? 442 00:18:19,550 --> 00:18:22,310 Whereas nonspeech vocalizations like coughing and emotional 443 00:18:22,310 --> 00:18:23,900 vocalizations can't. 444 00:18:23,900 --> 00:18:25,970 So if so, infants in the speech condition 445 00:18:25,970 --> 00:18:30,020 should look longer to nontarget than target actions, responses, 446 00:18:30,020 --> 00:18:31,666 finding these unexpected. 447 00:18:31,666 --> 00:18:33,290 And infants in the other two conditions 448 00:18:33,290 --> 00:18:34,956 shouldn't differentiate between the two. 449 00:18:37,537 --> 00:18:38,620 And this is what we found. 450 00:18:38,620 --> 00:18:40,150 So in the speech condition, infants 451 00:18:40,150 --> 00:18:42,139 are looking longer to nontarget than target, 452 00:18:42,139 --> 00:18:43,930 suggesting that they understand that speech 453 00:18:43,930 --> 00:18:46,747 can transfer information about the communicator's goal. 454 00:18:46,747 --> 00:18:48,580 But when the communicator coughs or produces 455 00:18:48,580 --> 00:18:50,350 an emotional vocalization, infants 456 00:18:50,350 --> 00:18:54,280 don't show these same expectations. 457 00:18:54,280 --> 00:18:56,709 So this is some initial evidence that by 12 months, 458 00:18:56,709 --> 00:18:59,250 infants not only recognize the context in which communication 459 00:18:59,250 --> 00:19:02,130 occurs and attend to speech as a special signal 460 00:19:02,130 --> 00:19:06,240 for communication, but they also understand 461 00:19:06,240 --> 00:19:09,360 that speech is something that's able to transfer information 462 00:19:09,360 --> 00:19:12,360 between two people. 463 00:19:12,360 --> 00:19:14,190 So we also ran a few control conditions 464 00:19:14,190 --> 00:19:17,534 to rule out alternative explanations for this. 465 00:19:17,534 --> 00:19:19,950 So one important question is, are infants really reasoning 466 00:19:19,950 --> 00:19:26,230 about the addressee's access to information in this situation? 467 00:19:26,230 --> 00:19:30,092 So for example, in the case where the-- 468 00:19:30,092 --> 00:19:32,550 if infants are really reasoning with the addressee's access 469 00:19:32,550 --> 00:19:35,460 to information, then in a case where the communicator makes 470 00:19:35,460 --> 00:19:37,322 a positive emotional vocalization, 471 00:19:37,322 --> 00:19:39,780 and the addressee has previous information about what she's 472 00:19:39,780 --> 00:19:41,880 interested in, so for example, if she just 473 00:19:41,880 --> 00:19:44,820 saw that the communicator was trying to feed her child 474 00:19:44,820 --> 00:19:47,100 or wanted a drink of water, she might know now 475 00:19:47,100 --> 00:19:49,562 that a positive emotional vocalization is indicating 476 00:19:49,562 --> 00:19:50,520 something like the cup. 477 00:19:54,040 --> 00:19:56,514 So we set up a scenario like this as well. 478 00:19:56,514 --> 00:19:58,930 In this case, it's exactly the same as the previous study, 479 00:19:58,930 --> 00:20:01,150 except that the addressee had visual access 480 00:20:01,150 --> 00:20:04,351 to the communicators preference during the familiarization 481 00:20:04,351 --> 00:20:04,850 phase. 482 00:20:04,850 --> 00:20:07,540 So now she knows what it is that the communicator likes. 483 00:20:07,540 --> 00:20:10,000 And this is important for making sure the infants recognize 484 00:20:10,000 --> 00:20:11,770 that something like visual access 485 00:20:11,770 --> 00:20:13,630 can provide the addressee with information 486 00:20:13,630 --> 00:20:17,300 about the communicator's interest. 487 00:20:17,300 --> 00:20:20,072 So now infants saw the same addressee familiarization. 488 00:20:20,072 --> 00:20:22,030 And then in the test, they saw the communicator 489 00:20:22,030 --> 00:20:25,570 produce the vocalization ooh once again. 490 00:20:25,570 --> 00:20:28,930 So now even though ooh was not treated 491 00:20:28,930 --> 00:20:31,319 as a communicative vocalization in the previous studies, 492 00:20:31,319 --> 00:20:33,610 if infants are reasoning about the kinds of information 493 00:20:33,610 --> 00:20:35,290 that the addressee has access to, 494 00:20:35,290 --> 00:20:37,702 they should now expect that the addressee will provide 495 00:20:37,702 --> 00:20:39,160 the target object because she knows 496 00:20:39,160 --> 00:20:42,709 from a previous scene what the communicator was interested in. 497 00:20:42,709 --> 00:20:45,250 So this is important also for ruling out the possibility that 498 00:20:45,250 --> 00:20:47,770 maybe just hearing these noncommunicative vocalizations 499 00:20:47,770 --> 00:20:49,900 surprises or confuses infants or makes 500 00:20:49,900 --> 00:20:52,360 them unable to reason about the scenario anymore. 501 00:20:52,360 --> 00:20:54,670 Here, they're getting a noncommunicative vocalization, 502 00:20:54,670 --> 00:20:58,180 but the addressee has access to information. 503 00:20:58,180 --> 00:21:00,160 And here, as in the speech condition, 504 00:21:00,160 --> 00:21:02,650 infants looked longer at the nontarget outcome. 505 00:21:02,650 --> 00:21:06,550 So this is just some evidence that infants are really 506 00:21:06,550 --> 00:21:08,920 thinking about the idea that relevant sources 507 00:21:08,920 --> 00:21:11,530 of information, such as a communicative vocalization, 508 00:21:11,530 --> 00:21:13,450 but also prior visual access, can 509 00:21:13,450 --> 00:21:18,070 give the addressee information about the communicator's goal. 510 00:21:18,070 --> 00:21:20,519 So they are reasoning about information access. 511 00:21:20,519 --> 00:21:22,810 Another important question is, are they really thinking 512 00:21:22,810 --> 00:21:24,310 about the source of information, or do they 513 00:21:24,310 --> 00:21:26,710 have-- so what's really the mechanism behind what infants 514 00:21:26,710 --> 00:21:27,335 are doing here? 515 00:21:27,335 --> 00:21:29,980 Are they thinking something like, when I hear speech, 516 00:21:29,980 --> 00:21:31,869 everything will go well, or the speaker 517 00:21:31,869 --> 00:21:33,160 is going to get what she wants? 518 00:21:33,160 --> 00:21:35,876 Or are they really thinking about communication 519 00:21:35,876 --> 00:21:37,750 in this causal way, where the information has 520 00:21:37,750 --> 00:21:40,150 to come from the speaker and be delivered 521 00:21:40,150 --> 00:21:43,430 to the listener in order for it to work? 522 00:21:43,430 --> 00:21:46,000 So to test for this, infants saw another communicator 523 00:21:46,000 --> 00:21:46,807 familiarization. 524 00:21:46,807 --> 00:21:49,390 Here, the communicator is alone, so they've never seen the two 525 00:21:49,390 --> 00:21:51,130 people together before. 526 00:21:51,130 --> 00:21:52,750 She reaches for the target object. 527 00:21:52,750 --> 00:21:55,390 Then they see the addressee again. 528 00:21:55,390 --> 00:21:57,700 And then in the test scene, in this case, 529 00:21:57,700 --> 00:22:00,130 the addressee is the one who produces the speech. 530 00:22:00,130 --> 00:22:03,320 So speech is present in the scene as it was before, 531 00:22:03,320 --> 00:22:06,100 but it's not being produced by the communicator or the person 532 00:22:06,100 --> 00:22:08,470 who showed a preference. 533 00:22:08,470 --> 00:22:09,970 So if infants just think that speech 534 00:22:09,970 --> 00:22:12,850 leads to people obtaining their goals or magical outcomes 535 00:22:12,850 --> 00:22:15,280 happening, then they should expect the addressee 536 00:22:15,280 --> 00:22:16,960 to provide the target here as well. 537 00:22:16,960 --> 00:22:19,540 But if they understand that they don't know anything 538 00:22:19,540 --> 00:22:21,610 about the addressee, in which case, this is not 539 00:22:21,610 --> 00:22:23,500 really informative about either object, 540 00:22:23,500 --> 00:22:26,710 then they should look equally to both outcomes. 541 00:22:26,710 --> 00:22:29,740 And in fact, this is what they do at 12 months. 542 00:22:29,740 --> 00:22:32,200 So infants here don't expect information 543 00:22:32,200 --> 00:22:33,700 to be transferred from one person 544 00:22:33,700 --> 00:22:36,370 to another, unless the first person who communicated 545 00:22:36,370 --> 00:22:39,220 is the one who actually had the information to provide. 546 00:22:44,340 --> 00:22:46,267 So by 12 months, infants seem to recognize 547 00:22:46,267 --> 00:22:47,850 that speech is communicative, and they 548 00:22:47,850 --> 00:22:49,850 seem to have some of the parts of a causal model 549 00:22:49,850 --> 00:22:50,620 of communication. 550 00:22:50,620 --> 00:22:52,203 They're not just thinking about speech 551 00:22:52,203 --> 00:22:55,600 as something that can produce successful outcomes, 552 00:22:55,600 --> 00:22:57,690 but as something that can be used 553 00:22:57,690 --> 00:23:00,120 to have information move from the mind of one individual 554 00:23:00,120 --> 00:23:01,920 to another. 555 00:23:01,920 --> 00:23:03,510 So a 12-month-old seemed to recognize 556 00:23:03,510 --> 00:23:05,580 that speech is communicative. 557 00:23:05,580 --> 00:23:07,260 But we really wanted to get at the idea 558 00:23:07,260 --> 00:23:09,300 that understanding that speech is communicative 559 00:23:09,300 --> 00:23:11,610 might be something that drives and guides 560 00:23:11,610 --> 00:23:13,560 word learning and language acquisition, rather 561 00:23:13,560 --> 00:23:15,018 than something that is as a result. 562 00:23:15,018 --> 00:23:17,250 So what a 12-month-old could be doing in this study 563 00:23:17,250 --> 00:23:19,740 is hearing the word koba, associating it with the object 564 00:23:19,740 --> 00:23:21,420 that the communicator had reached for, 565 00:23:21,420 --> 00:23:23,260 and then thinking, OK, that one's the koba, 566 00:23:23,260 --> 00:23:26,280 so that's what the addressee should reach for. 567 00:23:26,280 --> 00:23:29,040 But in these studies, we were interested in whether infants 568 00:23:29,040 --> 00:23:31,770 at an even younger age, who would be very, very 569 00:23:31,770 --> 00:23:34,037 unlikely to be associating the word with the object 570 00:23:34,037 --> 00:23:36,120 or learning a label for the object over the course 571 00:23:36,120 --> 00:23:37,800 of the study, would also recognize 572 00:23:37,800 --> 00:23:40,470 that speech can transfer information about one person 573 00:23:40,470 --> 00:23:41,890 to another person. 574 00:23:41,890 --> 00:23:44,700 And so for this reason, we tested 6-month-olds, 575 00:23:44,700 --> 00:23:47,466 because 6-month-olds understand only some very, very 576 00:23:47,466 --> 00:23:48,840 common words in their environment 577 00:23:48,840 --> 00:23:50,214 and will look to the right object 578 00:23:50,214 --> 00:23:52,930 when they hear the label for it. 579 00:23:52,930 --> 00:23:54,841 But these are very, very limited words, 580 00:23:54,841 --> 00:23:56,340 and there's no evidence for learning 581 00:23:56,340 --> 00:23:58,020 a word in a single trial, which is what they 582 00:23:58,020 --> 00:23:59,200 would have to do in this study. 583 00:23:59,200 --> 00:24:01,241 They would have to hear koba, and then think back 584 00:24:01,241 --> 00:24:03,327 to what the communicator had reached for, 585 00:24:03,327 --> 00:24:05,910 and really learn that word over the course of the study, which 586 00:24:05,910 --> 00:24:08,910 we have no evidence that a 6-month-old can do. 587 00:24:08,910 --> 00:24:10,590 So the goal of looking at this age group 588 00:24:10,590 --> 00:24:12,131 was to see whether infants might have 589 00:24:12,131 --> 00:24:15,270 a more abstract understanding of the idea that when speech 590 00:24:15,270 --> 00:24:18,030 is produced, information can be transferred from one person 591 00:24:18,030 --> 00:24:21,240 to another, even when they themselves 592 00:24:21,240 --> 00:24:22,830 don't know what that information-- 593 00:24:22,830 --> 00:24:29,250 or what the link between the word and the object. 594 00:24:29,250 --> 00:24:32,460 So a 6-month-old saw the same scenes. 595 00:24:32,460 --> 00:24:34,839 We gave them the speech versus cough contrast. 596 00:24:34,839 --> 00:24:36,880 And they look the same as the 12-month-olds here. 597 00:24:36,880 --> 00:24:38,250 So you can see that in the speech case, 598 00:24:38,250 --> 00:24:39,840 they're looking longer for the nontarget outcome. 599 00:24:39,840 --> 00:24:41,423 And in the cough case, they're looking 600 00:24:41,423 --> 00:24:42,940 longer for the target outcome. 601 00:24:42,940 --> 00:24:45,539 And so we haven't done all of the same control conditions 602 00:24:45,539 --> 00:24:46,830 as with the 12-month-olds here. 603 00:24:46,830 --> 00:24:49,200 So I think there's a lot of room for questions about how 604 00:24:49,200 --> 00:24:51,300 a 6-month-old's understanding of communication 605 00:24:51,300 --> 00:24:53,610 in this causal way more limited than the understanding 606 00:24:53,610 --> 00:24:54,660 of 12-month-olds. 607 00:24:54,660 --> 00:24:57,180 Would a six-month-old think that if speech 608 00:24:57,180 --> 00:24:59,610 was produced by a loudspeaker or by the addressee, 609 00:24:59,610 --> 00:25:01,960 that the communicator would still get the right object? 610 00:25:01,960 --> 00:25:06,160 So there are a lot of open questions about-- 611 00:25:06,160 --> 00:25:08,520 these basic questions about how infants' understanding 612 00:25:08,520 --> 00:25:10,561 of communicative information transfer starts out. 613 00:25:12,840 --> 00:25:14,840 So these experiments suggest that by six months, 614 00:25:14,840 --> 00:25:17,260 infants seem to understand that speech is communicative, 615 00:25:17,260 --> 00:25:18,850 in addition to the other studies that suggest 616 00:25:18,850 --> 00:25:20,350 they understand some of the features 617 00:25:20,350 --> 00:25:21,847 of communicative interactions. 618 00:25:21,847 --> 00:25:23,680 They recognize that it transfers information 619 00:25:23,680 --> 00:25:25,109 from one person to another. 620 00:25:25,109 --> 00:25:26,650 And I'd like to argue that this might 621 00:25:26,650 --> 00:25:28,774 be something that provides a mechanism for language 622 00:25:28,774 --> 00:25:32,099 and knowledge acquisition, so sort of guides infants 623 00:25:32,099 --> 00:25:33,640 to the kinds of relevant interactions 624 00:25:33,640 --> 00:25:35,390 where they might want to learn things about people, 625 00:25:35,390 --> 00:25:37,930 and their mental states, and the words that they're using, 626 00:25:37,930 --> 00:25:41,770 rather than first learning those words through association 627 00:25:41,770 --> 00:25:44,692 and then later coming to this more abstract understanding. 628 00:25:47,059 --> 00:25:48,850 So the idea is that infants might start out 629 00:25:48,850 --> 00:25:50,590 with this understanding of what communication is 630 00:25:50,590 --> 00:25:52,330 and when it's happening, and then can sort of fill 631 00:25:52,330 --> 00:25:54,490 in the rest from there, and that this might be one 632 00:25:54,490 --> 00:25:57,530 of the earlier blocks of that. 633 00:25:57,530 --> 00:25:59,730 So I'll try to go fairly quickly through the rest. 634 00:25:59,730 --> 00:26:01,480 So the second insight I wanted to bring up 635 00:26:01,480 --> 00:26:02,710 that I think is especially important 636 00:26:02,710 --> 00:26:05,043 is that communication requires this focus on intentions. 637 00:26:05,043 --> 00:26:06,700 And this was really the work of Grice 638 00:26:06,700 --> 00:26:10,750 that highlighted this, and in particular, a special type 639 00:26:10,750 --> 00:26:12,560 of communicative intention. 640 00:26:12,560 --> 00:26:14,560 So going back to the example from the beginning. 641 00:26:14,560 --> 00:26:16,390 When someone asks, is there any salt, 642 00:26:16,390 --> 00:26:17,800 Austin proposed that there are these three 643 00:26:17,800 --> 00:26:19,660 levels on which we can think about this communicative 644 00:26:19,660 --> 00:26:20,440 action. 645 00:26:20,440 --> 00:26:22,540 And Grice was the one who really formalized this 646 00:26:22,540 --> 00:26:24,640 by talking about this special kind of intention 647 00:26:24,640 --> 00:26:27,310 that we see in the domain of communication, 648 00:26:27,310 --> 00:26:29,590 human communication in particular, which 649 00:26:29,590 --> 00:26:32,080 is the idea of speaker meaning. 650 00:26:32,080 --> 00:26:35,324 And so his idea is that there's this double-layered intention 651 00:26:35,324 --> 00:26:37,490 that comes when we're communicating with each other. 652 00:26:37,490 --> 00:26:40,030 So we have a communicator who intends the addressee 653 00:26:40,030 --> 00:26:41,510 to respond in a particular way. 654 00:26:41,510 --> 00:26:43,630 So here, the communicator wants the addressee 655 00:26:43,630 --> 00:26:47,290 to provide some salt. The communicator 656 00:26:47,290 --> 00:26:50,350 intends for the addressee to recognize that the communicator 657 00:26:50,350 --> 00:26:52,040 intends to have that response. 658 00:26:52,040 --> 00:26:57,820 So this person not only wants this person to pass the salt, 659 00:26:57,820 --> 00:26:59,560 but wants her to recognize that that 660 00:26:59,560 --> 00:27:02,080 is what he's asking for, that that is the request that he's 661 00:27:02,080 --> 00:27:02,860 making. 662 00:27:02,860 --> 00:27:04,520 And importantly, this third point, 663 00:27:04,520 --> 00:27:08,650 which is that the communicator intends the addressee 664 00:27:08,650 --> 00:27:11,920 to respond that way on the basis of the recognition 665 00:27:11,920 --> 00:27:13,120 of that intention. 666 00:27:13,120 --> 00:27:15,220 And so to make this a little more concrete, 667 00:27:15,220 --> 00:27:17,279 in this case, if he asks her for the salt, 668 00:27:17,279 --> 00:27:19,570 and she passes it because she heard him and understands 669 00:27:19,570 --> 00:27:21,850 that's what he's asking for, that's great. 670 00:27:21,850 --> 00:27:24,160 But the argument is that communication wouldn't really 671 00:27:24,160 --> 00:27:25,900 be happening if, for example, she 672 00:27:25,900 --> 00:27:27,850 was wearing headphones and listening to music, 673 00:27:27,850 --> 00:27:29,350 and she hadn't really been listening to him, 674 00:27:29,350 --> 00:27:32,140 and she just happened to pick up the salt and pass it to him. 675 00:27:32,140 --> 00:27:35,027 So communication, I mean, it would 676 00:27:35,027 --> 00:27:36,610 look like a communicative interaction. 677 00:27:36,610 --> 00:27:38,980 The right kind of outcome is still happening in response 678 00:27:38,980 --> 00:27:40,060 to what he said. 679 00:27:40,060 --> 00:27:43,270 But if she has no access to the actual message, 680 00:27:43,270 --> 00:27:46,300 if she doesn't produce the response by virtue 681 00:27:46,300 --> 00:27:50,911 of understanding what the communicator is trying to say, 682 00:27:50,911 --> 00:27:52,660 then communication hasn't really occurred. 683 00:27:52,660 --> 00:27:54,118 It's just sort of a lucky accident. 684 00:27:58,670 --> 00:28:01,709 So I'll skip this part. 685 00:28:01,709 --> 00:28:03,500 So this is this important kind of intention 686 00:28:03,500 --> 00:28:04,980 that we see in human communication, in which I'll 687 00:28:04,980 --> 00:28:06,563 later briefly mention, we don't really 688 00:28:06,563 --> 00:28:10,520 see in animal communication in the same sort of way. 689 00:28:10,520 --> 00:28:13,701 So how do infants start to get this idea about communication, 690 00:28:13,701 --> 00:28:15,200 that it's not just about identifying 691 00:28:15,200 --> 00:28:16,700 communicative interactions, but also 692 00:28:16,700 --> 00:28:20,270 about understanding these particular kinds of intentions? 693 00:28:20,270 --> 00:28:22,250 So in the '90s, there was this shift 694 00:28:22,250 --> 00:28:25,640 in developmental psychology to looking at word learning, 695 00:28:25,640 --> 00:28:30,440 not only in terms of infants' associations of words 696 00:28:30,440 --> 00:28:33,560 in the environment, but to also, with the work of Dare Baldwin, 697 00:28:33,560 --> 00:28:35,780 to thinking about other people's intentions 698 00:28:35,780 --> 00:28:37,190 when they're using words. 699 00:28:37,190 --> 00:28:41,030 And so she had these really elegant studies where-- 700 00:28:41,030 --> 00:28:43,280 with this funny image-- 701 00:28:43,280 --> 00:28:46,670 where she pointed out that if infants really 702 00:28:46,670 --> 00:28:48,830 were learning words through association, 703 00:28:48,830 --> 00:28:51,160 then it wouldn't be very efficient for them 704 00:28:51,160 --> 00:28:52,910 because they would make a lot of mistakes. 705 00:28:52,910 --> 00:28:54,680 So for example, in this kind of situation, 706 00:28:54,680 --> 00:28:56,340 here's a dad and his baby. 707 00:28:56,340 --> 00:28:58,910 The baby is looking at this lizard. 708 00:28:58,910 --> 00:29:00,500 The dad is looking at this rooster. 709 00:29:00,500 --> 00:29:03,200 And the dad says, what a cheeky rooster. 710 00:29:03,200 --> 00:29:05,840 So if you as the infant are only learning words 711 00:29:05,840 --> 00:29:09,110 on the basis of associating what you hear with what you see, 712 00:29:09,110 --> 00:29:11,674 you're going to learn that the word rooster refers 713 00:29:11,674 --> 00:29:13,340 to this thing rather than to this thing. 714 00:29:13,340 --> 00:29:15,660 And you're going to get it wrong. 715 00:29:15,660 --> 00:29:18,500 So Baldwin did a whole host of studies 716 00:29:18,500 --> 00:29:24,020 in the second year of life where she set up situations 717 00:29:24,020 --> 00:29:26,510 where infants were looking at particular objects like here. 718 00:29:26,510 --> 00:29:29,000 And then she had their parents or an experimenter look 719 00:29:29,000 --> 00:29:30,680 at a different object and label that one 720 00:29:30,680 --> 00:29:32,720 to see what infants would do. 721 00:29:32,720 --> 00:29:36,380 And what she found is that infants actually 722 00:29:36,380 --> 00:29:39,440 would consult their parent or consult 723 00:29:39,440 --> 00:29:41,840 the other person for cues to reference, to what they're 724 00:29:41,840 --> 00:29:43,197 intending to label. 725 00:29:43,197 --> 00:29:44,780 So if, for example, in this situation, 726 00:29:44,780 --> 00:29:47,309 the infant, when hearing this, instead of just assuming 727 00:29:47,309 --> 00:29:49,100 the word refers to what they're looking at, 728 00:29:49,100 --> 00:29:51,890 would actually look up to dad to see what he's looking at 729 00:29:51,890 --> 00:29:53,956 and then follow his line of gaze to understand 730 00:29:53,956 --> 00:29:56,330 that this is the object that he's actually talking about. 731 00:29:56,330 --> 00:29:58,490 AUDIENCE: At this age? 732 00:29:58,490 --> 00:29:58,990 At this age? 733 00:29:58,990 --> 00:30:00,281 ALIA MARTIN: Yeah, at this age. 734 00:30:00,281 --> 00:30:02,270 Yeah. 735 00:30:02,270 --> 00:30:04,730 So the idea is that infants in the second year of life, 736 00:30:04,730 --> 00:30:07,880 at least, are understanding that understanding what someone's 737 00:30:07,880 --> 00:30:10,415 referring to involves consulting cues to their attention 738 00:30:10,415 --> 00:30:13,700 and intentions rather than just the infant's own. 739 00:30:17,260 --> 00:30:19,091 And I think I'll skip this one. 740 00:30:19,091 --> 00:30:20,840 There are other studies showing this, too. 741 00:30:20,840 --> 00:30:22,790 So in this one, there's an experimenter 742 00:30:22,790 --> 00:30:24,420 who puts these two objects in a box. 743 00:30:24,420 --> 00:30:27,200 The objects switch while she's absent. 744 00:30:27,200 --> 00:30:30,230 And then she looks in one box and says-- 745 00:30:30,230 --> 00:30:32,850 the boxes are closed, and she says, there's a sefo in here. 746 00:30:32,850 --> 00:30:35,350 So if the infant understands that they have to pay attention 747 00:30:35,350 --> 00:30:38,046 to what the person knows about or what the person thinks 748 00:30:38,046 --> 00:30:39,920 rather than their own knowledge to figure out 749 00:30:39,920 --> 00:30:42,000 what she's labeling, then now when she says, 750 00:30:42,000 --> 00:30:44,060 can you get the sefo, the infant should actually 751 00:30:44,060 --> 00:30:46,670 pull the object out of here, assuming 752 00:30:46,670 --> 00:30:48,170 that this is the one she's labeling, 753 00:30:48,170 --> 00:30:51,890 because that's the one that she was intending to label. 754 00:30:51,890 --> 00:30:56,150 And in fact, that's what infants do at 17 months. 755 00:30:56,150 --> 00:31:01,040 In the case where the experimenter had a false belief 756 00:31:01,040 --> 00:31:02,600 about the location of the object, 757 00:31:02,600 --> 00:31:05,302 infants tend to go to the non-referred box 758 00:31:05,302 --> 00:31:07,010 more often because they think that that's 759 00:31:07,010 --> 00:31:08,176 the one that she's labeling. 760 00:31:08,176 --> 00:31:10,797 But in the case where she saw the switch happen, so now she 761 00:31:10,797 --> 00:31:13,130 knows what's everywhere, infants will just go to the box 762 00:31:13,130 --> 00:31:15,710 that she's referring to figure out the location of the object 763 00:31:15,710 --> 00:31:17,574 that she's naming. 764 00:31:17,574 --> 00:31:19,490 So this is all in the second year of life, not 765 00:31:19,490 --> 00:31:21,485 those younger core knowledge ages 766 00:31:21,485 --> 00:31:22,610 that Liz was talking about. 767 00:31:25,530 --> 00:31:28,320 So this is evidence that infants consult speaker cues 768 00:31:28,320 --> 00:31:30,780 to intentions, but not of this special kind of intention 769 00:31:30,780 --> 00:31:32,484 that Grice was pointing out. 770 00:31:32,484 --> 00:31:34,650 So I'm just going to tell you about a couple-- well, 771 00:31:34,650 --> 00:31:37,149 one of my favorite studies, I just think that this is really 772 00:31:37,149 --> 00:31:37,680 neat-- 773 00:31:37,680 --> 00:31:41,130 where Ellen Markman's lab and then later 774 00:31:41,130 --> 00:31:43,500 replicated by Tomasello's lab, showed 775 00:31:43,500 --> 00:31:45,870 that it seems like children, by 30 months 776 00:31:45,870 --> 00:31:49,830 and then gotten down to 18 months by Grosse et al, 777 00:31:49,830 --> 00:31:54,450 actually do seem to care about not only the effects 778 00:31:54,450 --> 00:31:56,790 of communication, but also that the effects 779 00:31:56,790 --> 00:32:00,870 are produced by virtue of the message being understood. 780 00:32:00,870 --> 00:32:02,580 So you're probably wondering why there's 781 00:32:02,580 --> 00:32:03,970 a dirty sock on the screen. 782 00:32:03,970 --> 00:32:10,122 So in this study, children were presented with two objects. 783 00:32:10,122 --> 00:32:11,580 So they saw, on a table, there were 784 00:32:11,580 --> 00:32:13,621 two objects that were out of their reach, a truck 785 00:32:13,621 --> 00:32:15,210 and a dirty sock. 786 00:32:15,210 --> 00:32:18,259 And the idea was to elicit children to request an object. 787 00:32:18,259 --> 00:32:19,800 Now obviously, the children are going 788 00:32:19,800 --> 00:32:22,174 to request this object, which is what the researchers had 789 00:32:22,174 --> 00:32:23,490 intended in this case. 790 00:32:23,490 --> 00:32:25,549 So children tended to point at this object. 791 00:32:25,549 --> 00:32:27,840 So they had children producing their own communication, 792 00:32:27,840 --> 00:32:29,334 and then the experimenter responded 793 00:32:29,334 --> 00:32:30,750 to the communication of the infant 794 00:32:30,750 --> 00:32:32,430 in one of four different ways. 795 00:32:32,430 --> 00:32:34,847 So she either said, you asked for the truck? 796 00:32:34,847 --> 00:32:36,180 I'm going to give you the truck. 797 00:32:36,180 --> 00:32:38,220 So she expressed understanding of the request, 798 00:32:38,220 --> 00:32:41,124 and she provided the requested object. 799 00:32:41,124 --> 00:32:43,290 In a second case, she said, you asked for the truck? 800 00:32:43,290 --> 00:32:44,581 I'm going to give you the sock. 801 00:32:44,581 --> 00:32:46,290 So here, she expressed understanding, 802 00:32:46,290 --> 00:32:48,810 but refused to commit to the request, 803 00:32:48,810 --> 00:32:51,590 to give the thing the child had asked for. 804 00:32:51,590 --> 00:32:53,490 In a third case, the experimenter said, 805 00:32:53,490 --> 00:32:54,840 you asked for the sock? 806 00:32:54,840 --> 00:32:56,770 I'm going to give you the truck. 807 00:32:56,770 --> 00:32:59,827 So this is kind of a strange pragmatic situation where 808 00:32:59,827 --> 00:33:02,160 the experimenter actually gave the child the object they 809 00:33:02,160 --> 00:33:05,040 wanted, but she expressed a misunderstanding of the child's 810 00:33:05,040 --> 00:33:05,790 request. 811 00:33:05,790 --> 00:33:07,530 And this is the crucial case here. 812 00:33:07,530 --> 00:33:10,740 So the question is, obviously, in the case where 813 00:33:10,740 --> 00:33:12,270 the experimenter provides the thing 814 00:33:12,270 --> 00:33:15,960 and expresses understanding, children should be quite happy. 815 00:33:15,960 --> 00:33:17,610 And in this situation, they probably 816 00:33:17,610 --> 00:33:20,172 shouldn't be very happy. 817 00:33:20,172 --> 00:33:21,630 And what the experimenters measured 818 00:33:21,630 --> 00:33:23,940 was the amount of times that the child 819 00:33:23,940 --> 00:33:26,820 repeated the label of the object that they had asked for. 820 00:33:26,820 --> 00:33:30,150 And they also measured other behaviors as well. 821 00:33:30,150 --> 00:33:34,155 It's this situation that's important for understanding 822 00:33:34,155 --> 00:33:35,970 what kids care about in the context 823 00:33:35,970 --> 00:33:37,560 of this communicative interaction. 824 00:33:37,560 --> 00:33:40,342 So the question is, are infants perfectly happy here 825 00:33:40,342 --> 00:33:42,300 to take the truck and not complain because they 826 00:33:42,300 --> 00:33:43,410 got what they wanted? 827 00:33:43,410 --> 00:33:45,810 Or do they care about the experimenter giving them 828 00:33:45,810 --> 00:33:49,080 what they wanted because their communicated message had 829 00:33:49,080 --> 00:33:53,820 the proper effect on the addressee? 830 00:33:53,820 --> 00:34:00,390 And what they find is that when the correct object is provided, 831 00:34:00,390 --> 00:34:02,520 and the experimenter does not express understanding 832 00:34:02,520 --> 00:34:04,144 of the request, infants actually showed 833 00:34:04,144 --> 00:34:06,660 the most repetitions of the name of the requested object 834 00:34:06,660 --> 00:34:07,260 in this case. 835 00:34:10,686 --> 00:34:12,475 The researchers concluded from this 836 00:34:12,475 --> 00:34:14,850 that infants don't just care about getting what they want 837 00:34:14,850 --> 00:34:19,710 or using communication as a sort of instrumental 838 00:34:19,710 --> 00:34:21,690 means of getting people to act in certain ways, 839 00:34:21,690 --> 00:34:26,699 but rather as an intention to have their messages understood 840 00:34:26,699 --> 00:34:29,811 and delivered to the other person. 841 00:34:29,811 --> 00:34:32,310 So they care about the impact of their communicative signals 842 00:34:32,310 --> 00:34:33,929 on the understanding of the addressee, 843 00:34:33,929 --> 00:34:36,102 not just on the response of the addressee, 844 00:34:36,102 --> 00:34:38,310 which is sort of like the idea that Grice was talking 845 00:34:38,310 --> 00:34:40,320 about with speaker meaning. 846 00:34:40,320 --> 00:34:42,570 Similarly, when children are responding 847 00:34:42,570 --> 00:34:44,820 to other people's requests, they also care about this. 848 00:34:44,820 --> 00:34:47,280 So this is a study that I did in graduate school. 849 00:34:47,280 --> 00:34:50,230 It's sort of the opposite of the previous study, or the inverse. 850 00:34:50,230 --> 00:34:53,580 So we had an experimenter requesting from the child 851 00:34:53,580 --> 00:34:57,380 specific objects for doing specific tasks. 852 00:34:57,380 --> 00:34:58,530 This is 3-year-olds. 853 00:34:58,530 --> 00:35:00,960 So she would request something like a cup 854 00:35:00,960 --> 00:35:02,670 to pour a cup of water. 855 00:35:02,670 --> 00:35:05,340 The child had previous information from playing games 856 00:35:05,340 --> 00:35:07,110 with the objects with another experimenter 857 00:35:07,110 --> 00:35:09,037 that one of the cups was perfectly fine, 858 00:35:09,037 --> 00:35:10,620 and one of the cups was broken and had 859 00:35:10,620 --> 00:35:12,540 a big hole in the bottom. 860 00:35:12,540 --> 00:35:14,670 And so when the experimenter asked for a cup, 861 00:35:14,670 --> 00:35:17,440 the child could either give the cup the experimenter asked for, 862 00:35:17,440 --> 00:35:20,479 which was sometimes the perfectly good cup, 863 00:35:20,479 --> 00:35:22,770 or sometimes the experimenter requested the broken cup. 864 00:35:22,770 --> 00:35:26,400 And the question was, do children pay attention 865 00:35:26,400 --> 00:35:28,290 to the fit between the task and what 866 00:35:28,290 --> 00:35:30,420 the experimenter wanted when they're 867 00:35:30,420 --> 00:35:33,430 responding to her request? 868 00:35:33,430 --> 00:35:36,120 And what we found is that children were much more 869 00:35:36,120 --> 00:35:37,800 likely to give the requested object when 870 00:35:37,800 --> 00:35:40,320 the experimenter had requested a functional object than when 871 00:35:40,320 --> 00:35:41,920 she'd requested a dysfunctional object. 872 00:35:41,920 --> 00:35:43,711 So if I say, I need to pour a cup of water, 873 00:35:43,711 --> 00:35:45,870 can you give me that cup, and the cup is broken, 874 00:35:45,870 --> 00:35:49,770 children tended to go and get a better cup instead of the one 875 00:35:49,770 --> 00:35:51,570 that I had requested. 876 00:35:51,570 --> 00:35:54,570 Interestingly, though, even though children did this, 877 00:35:54,570 --> 00:35:57,240 it didn't seem to be enough for them to give the experimenter 878 00:35:57,240 --> 00:35:58,770 something good that she wanted. 879 00:35:58,770 --> 00:36:00,660 This is a graph of the comments that children 880 00:36:00,660 --> 00:36:02,460 made about the function of the objects 881 00:36:02,460 --> 00:36:05,300 depending on the kind of request that was made. 882 00:36:05,300 --> 00:36:07,800 And what you can see is that when a dysfunctional object was 883 00:36:07,800 --> 00:36:10,200 requested, when children tended to provide 884 00:36:10,200 --> 00:36:13,115 the functional object instead and not respond to the request, 885 00:36:13,115 --> 00:36:14,490 they were much more likely to try 886 00:36:14,490 --> 00:36:17,010 to explain their behavior to the experimenter 887 00:36:17,010 --> 00:36:20,190 and acknowledge what it was the experiment had originally 888 00:36:20,190 --> 00:36:21,060 wanted. 889 00:36:21,060 --> 00:36:23,521 So it seems like children in their own behavior, 890 00:36:23,521 --> 00:36:25,270 by at least three, which is a little older 891 00:36:25,270 --> 00:36:27,942 than the other studies, acknowledge 892 00:36:27,942 --> 00:36:30,400 what the speaker meant to ask for and explain while they're 893 00:36:30,400 --> 00:36:33,109 doing something else, even when they're not 894 00:36:33,109 --> 00:36:34,150 responding to that thing. 895 00:36:36,780 --> 00:36:39,005 And so, just briefly, in the last section, 896 00:36:39,005 --> 00:36:41,130 I'll talk about a third insight about communication 897 00:36:41,130 --> 00:36:42,810 that I think is really important, which 898 00:36:42,810 --> 00:36:45,210 comes from Clark, which is that communication 899 00:36:45,210 --> 00:36:49,480 is this joint action of accumulating common ground. 900 00:36:49,480 --> 00:36:52,530 And so in an example of an adult study about this, 901 00:36:52,530 --> 00:36:55,770 they showed adults pictures like this one of New York City, 902 00:36:55,770 --> 00:36:58,119 and they had people play this game in pairs. 903 00:36:58,119 --> 00:37:00,660 They had people who knew about New York City who were experts 904 00:37:00,660 --> 00:37:01,530 and who lived there, and then they 905 00:37:01,530 --> 00:37:04,113 had other people who didn't know anything about New York City. 906 00:37:04,113 --> 00:37:06,720 And they gave them a bunch of pictures. 907 00:37:06,720 --> 00:37:10,910 They told them that their goal was to-- 908 00:37:10,910 --> 00:37:12,630 the person who was the addressee had 909 00:37:12,630 --> 00:37:15,600 to sort the objects in the way that the communicator told them 910 00:37:15,600 --> 00:37:16,230 to. 911 00:37:16,230 --> 00:37:18,720 And so as a communicator, the communicator had to indicate, 912 00:37:18,720 --> 00:37:22,140 because they couldn't see the pictures, which picture 913 00:37:22,140 --> 00:37:23,940 the addressee should put where. 914 00:37:23,940 --> 00:37:26,880 And to do this, the communicator had 915 00:37:26,880 --> 00:37:30,330 to refer to things like the picture with the Empire State 916 00:37:30,330 --> 00:37:31,830 Building in it. 917 00:37:31,830 --> 00:37:35,430 And what they looked at was how people in these interactions 918 00:37:35,430 --> 00:37:38,490 accumulated shared knowledge or common ground over time 919 00:37:38,490 --> 00:37:40,650 and coordinated so their communication could 920 00:37:40,650 --> 00:37:41,709 become more efficient. 921 00:37:41,709 --> 00:37:44,250 And what they found is that when they had two New Yorkers who 922 00:37:44,250 --> 00:37:45,624 were interacting with each other, 923 00:37:45,624 --> 00:37:49,037 they tended to very quickly do the task because they could 924 00:37:49,037 --> 00:37:51,120 recognize immediately, that person's a New Yorker, 925 00:37:51,120 --> 00:37:52,380 and would just say things like, oh, it's 926 00:37:52,380 --> 00:37:54,030 the one with the Empire State Building. 927 00:37:54,030 --> 00:37:55,890 Whereas when they were talking to someone who wasn't a New 928 00:37:55,890 --> 00:37:58,200 Yorker, they had to sort of ground the conversation 929 00:37:58,200 --> 00:38:00,990 by establishing these common reference before getting 930 00:38:00,990 --> 00:38:01,774 to this point. 931 00:38:01,774 --> 00:38:03,940 And so they might start out with saying things like, 932 00:38:03,940 --> 00:38:07,280 oh, move the one with the building with the pointy top. 933 00:38:07,280 --> 00:38:09,030 And then eventually, they would coordinate 934 00:38:09,030 --> 00:38:13,450 on what the actual labels for these things were. 935 00:38:13,450 --> 00:38:16,260 And so Clark's idea is that communication 936 00:38:16,260 --> 00:38:18,210 is really efficient, and we're able to do it 937 00:38:18,210 --> 00:38:20,987 in the way we are because we're thinking 938 00:38:20,987 --> 00:38:23,070 about the common ground we have with other people. 939 00:38:23,070 --> 00:38:24,900 And we're able to figure out what kind of common ground 940 00:38:24,900 --> 00:38:26,610 we have with others fairly quickly. 941 00:38:26,610 --> 00:38:28,470 So if I'm interacting with one of you, 942 00:38:28,470 --> 00:38:31,290 I might assume, OK, we both know a lot about cognitive science 943 00:38:31,290 --> 00:38:31,920 already. 944 00:38:31,920 --> 00:38:33,570 So I can sort of start at a different level 945 00:38:33,570 --> 00:38:35,670 than I might start with, say, a child or someone who didn't 946 00:38:35,670 --> 00:38:36,990 know anything about this area. 947 00:38:39,560 --> 00:38:43,325 And so this is an important piece of human communication. 948 00:38:43,325 --> 00:38:45,450 It seems like-- and Liz talked a little about this, 949 00:38:45,450 --> 00:38:47,116 too-- it seems like infants are starting 950 00:38:47,116 --> 00:38:49,502 to show some signs of understanding 951 00:38:49,502 --> 00:38:51,210 the importance of common ground or shared 952 00:38:51,210 --> 00:38:54,180 knowledge in communication from a really early age. 953 00:38:54,180 --> 00:38:56,640 But importantly, this seems to come in around the same time 954 00:38:56,640 --> 00:38:58,556 that they recognize the importance of speakers 955 00:38:58,556 --> 00:39:00,325 facing each other in conversation 956 00:39:00,325 --> 00:39:02,700 and perhaps putting together some of their core knowledge 957 00:39:02,700 --> 00:39:07,200 domains, which is around nine months to a year. 958 00:39:07,200 --> 00:39:09,330 A lot of work in this domain has been 959 00:39:09,330 --> 00:39:12,411 done by Tomasello, who showed that around nine months, 960 00:39:12,411 --> 00:39:14,910 children were starting to do something different than they'd 961 00:39:14,910 --> 00:39:15,701 been doing earlier. 962 00:39:15,701 --> 00:39:19,610 So under nine months, children tend to-- 963 00:39:19,610 --> 00:39:22,230 they play with objects, they interact with people, 964 00:39:22,230 --> 00:39:25,480 but they don't seem to put these two things together. 965 00:39:25,480 --> 00:39:27,750 Whereas at nine months, what they start doing 966 00:39:27,750 --> 00:39:30,510 is paying attention, not just to objects or to people, 967 00:39:30,510 --> 00:39:33,360 but to objects and people at the same time in the context 968 00:39:33,360 --> 00:39:34,780 of a joint interaction. 969 00:39:34,780 --> 00:39:37,050 So they might do things like look at an object 970 00:39:37,050 --> 00:39:40,860 and then look at mom to make sure mom is also looking at it. 971 00:39:40,860 --> 00:39:42,780 Or they might point at things, not just 972 00:39:42,780 --> 00:39:45,030 because they want the things, as a younger child might 973 00:39:45,030 --> 00:39:48,750 do, but only to share attention with a parent 974 00:39:48,750 --> 00:39:50,239 or with someone else to point out 975 00:39:50,239 --> 00:39:51,780 that they're interested in something, 976 00:39:51,780 --> 00:39:53,863 and to make sure that the parent is looking and is 977 00:39:53,863 --> 00:39:56,070 interested in it as well. 978 00:39:56,070 --> 00:39:58,710 So Tomasello argues that this ability 979 00:39:58,710 --> 00:40:03,780 for engaging in joint attention, sharing attention with someone 980 00:40:03,780 --> 00:40:06,570 else, to an object or external referent in the world 981 00:40:06,570 --> 00:40:09,210 is the foundation of linguistic communication 982 00:40:09,210 --> 00:40:12,150 and also cooperation in other very important human 983 00:40:12,150 --> 00:40:13,980 activities. 984 00:40:13,980 --> 00:40:16,740 And it's certainly going to be important for an understanding 985 00:40:16,740 --> 00:40:21,352 of common ground, which is important for communication. 986 00:40:21,352 --> 00:40:23,310 So there's also evidence that around 12 months, 987 00:40:23,310 --> 00:40:25,245 infants start to use prior shared experience 988 00:40:25,245 --> 00:40:26,370 to interpret communication. 989 00:40:26,370 --> 00:40:27,900 I'm going to skip this, I think. 990 00:40:27,900 --> 00:40:28,995 But basically, the idea-- 991 00:40:28,995 --> 00:40:30,620 well, I'll just go through it quickly-- 992 00:40:30,620 --> 00:40:33,900 the idea is that infants will use the activities and objects 993 00:40:33,900 --> 00:40:36,632 that they've shared with people previously 994 00:40:36,632 --> 00:40:38,715 to figure out what the person means in a new case. 995 00:40:38,715 --> 00:40:41,550 So if the infant interacts with one communicator with one toy 996 00:40:41,550 --> 00:40:43,633 and with another communicator with a different toy 997 00:40:43,633 --> 00:40:47,220 separately, they will figure out the referent 998 00:40:47,220 --> 00:40:49,020 of a communicator's ambiguous request 999 00:40:49,020 --> 00:40:50,520 by thinking about what information 1000 00:40:50,520 --> 00:40:51,880 they've shared in the past. 1001 00:40:51,880 --> 00:40:52,860 So it seems like they're starting 1002 00:40:52,860 --> 00:40:54,610 to track what kinds of knowledge is shared 1003 00:40:54,610 --> 00:40:57,090 and what isn't in order to effectively communicate 1004 00:40:57,090 --> 00:40:57,600 with others. 1005 00:41:04,320 --> 00:41:05,535 OK. 1006 00:41:05,535 --> 00:41:07,410 So there's a lot of other important questions 1007 00:41:07,410 --> 00:41:09,618 about common ground, but I'm going to skip those now. 1008 00:41:09,618 --> 00:41:13,355 I just want to come back to the question of why 1009 00:41:13,355 --> 00:41:15,480 infants' understanding and children's understanding 1010 00:41:15,480 --> 00:41:17,438 of communication is important for understanding 1011 00:41:17,438 --> 00:41:19,480 human intelligence. 1012 00:41:19,480 --> 00:41:21,210 So I think that one reason is that when 1013 00:41:21,210 --> 00:41:25,080 you think about the insights these philosophers had 1014 00:41:25,080 --> 00:41:28,680 and how they seemed to be realized 1015 00:41:28,680 --> 00:41:30,720 in fairly young infants early-- 1016 00:41:30,720 --> 00:41:33,419 children and infants early in development, 1017 00:41:33,419 --> 00:41:35,460 when you look at these abilities that humans have 1018 00:41:35,460 --> 00:41:38,880 and even that very young humans have, and you compare them 1019 00:41:38,880 --> 00:41:40,650 to what nonhuman animals are doing, 1020 00:41:40,650 --> 00:41:44,220 things look really different. 1021 00:41:44,220 --> 00:41:48,180 And so, just briefly, in animal communication, 1022 00:41:48,180 --> 00:41:49,890 we see some of the same kinds of features 1023 00:41:49,890 --> 00:41:51,790 that we see in human communication. 1024 00:41:51,790 --> 00:41:53,850 So for one thing, animal communication clearly 1025 00:41:53,850 --> 00:41:55,470 has a social function in the same way 1026 00:41:55,470 --> 00:41:57,190 that human communication does. 1027 00:41:57,190 --> 00:41:59,815 So it's socially rich in a number of ways. 1028 00:41:59,815 --> 00:42:01,440 I'm not going to over specific species, 1029 00:42:01,440 --> 00:42:04,030 but just to gloss over it, most animal communication 1030 00:42:04,030 --> 00:42:05,994 is sensitive to the presence of an audience. 1031 00:42:05,994 --> 00:42:07,410 So it matters that someone's there 1032 00:42:07,410 --> 00:42:09,360 to hear your communication. 1033 00:42:09,360 --> 00:42:10,890 For example, species that produce 1034 00:42:10,890 --> 00:42:15,159 alarm calls to warn others in their group of predators 1035 00:42:15,159 --> 00:42:17,700 will rarely produce these calls if there are no other members 1036 00:42:17,700 --> 00:42:19,560 of their species present. 1037 00:42:19,560 --> 00:42:21,810 In many cases, the sensitivity to the audience 1038 00:42:21,810 --> 00:42:23,920 depends on the identity of the audience. 1039 00:42:23,920 --> 00:42:26,950 So for example, in some species such as ground squirrels, 1040 00:42:26,950 --> 00:42:28,950 they call much more in the presence of their kin 1041 00:42:28,950 --> 00:42:30,366 than when their kin are not around 1042 00:42:30,366 --> 00:42:32,654 or when there are other individuals there. 1043 00:42:32,654 --> 00:42:35,070 And in some cases, it seems like there might actually even 1044 00:42:35,070 --> 00:42:36,330 be a sensitivity to the knowledge 1045 00:42:36,330 --> 00:42:37,246 state of the audience. 1046 00:42:37,246 --> 00:42:39,180 So this is looking a little more like the kind 1047 00:42:39,180 --> 00:42:41,970 of sophisticated communication we see in humans. 1048 00:42:41,970 --> 00:42:44,010 So a wild chimpanzee, for example, 1049 00:42:44,010 --> 00:42:46,350 will produce an alarm call, will start to alarm call 1050 00:42:46,350 --> 00:42:48,210 more if other chimpanzees come over 1051 00:42:48,210 --> 00:42:50,670 who hadn't heard the original alarm call 1052 00:42:50,670 --> 00:42:52,290 or who hadn't seen the predator. 1053 00:42:52,290 --> 00:42:53,760 But if everyone around has already seen it, 1054 00:42:53,760 --> 00:42:55,218 they'll reduce their alarm calling. 1055 00:42:55,218 --> 00:42:57,690 So it seems like they tailor it to how much information 1056 00:42:57,690 --> 00:42:58,860 others around them have had. 1057 00:43:02,010 --> 00:43:04,020 However, despite these really interesting ways 1058 00:43:04,020 --> 00:43:07,170 in which animal communication is social and complex, 1059 00:43:07,170 --> 00:43:08,760 it's also limited in a number of ways 1060 00:43:08,760 --> 00:43:10,500 that human communication is not. 1061 00:43:10,500 --> 00:43:13,320 So the eliciting stimuli for these communicative signals 1062 00:43:13,320 --> 00:43:15,720 tend to be fairly limited, as do the signals themselves. 1063 00:43:15,720 --> 00:43:17,553 So there tends to be, say in vervet monkeys, 1064 00:43:17,553 --> 00:43:19,830 one cry for a hawk and one cry for a snake, 1065 00:43:19,830 --> 00:43:22,590 and they can't realize new signals 1066 00:43:22,590 --> 00:43:26,310 for new kinds of predators and situations. 1067 00:43:26,310 --> 00:43:28,980 As in humans, the receivers-- or in humans, 1068 00:43:28,980 --> 00:43:30,510 the addressees-- acquire information 1069 00:43:30,510 --> 00:43:32,820 from the signals of others, but there's no evidence 1070 00:43:32,820 --> 00:43:37,140 that this information tells the receivers in animal species 1071 00:43:37,140 --> 00:43:41,457 anything about the mental states of the communicator. 1072 00:43:41,457 --> 00:43:43,290 And additionally, the communicator's signals 1073 00:43:43,290 --> 00:43:45,470 can often cause a response in receivers that's 1074 00:43:45,470 --> 00:43:46,770 beneficial to the communicator. 1075 00:43:46,770 --> 00:43:49,080 For example, they get a bump to indirect fitness 1076 00:43:49,080 --> 00:43:53,160 if they're kin run away and survive predators. 1077 00:43:53,160 --> 00:43:56,130 So there can be benefits of communication, 1078 00:43:56,130 --> 00:43:58,380 but there's no evidence that the communicator 1079 00:43:58,380 --> 00:44:01,419 has any intention of changing the receiver's mental state. 1080 00:44:01,419 --> 00:44:03,960 So there's really no evidence of this sort of speaker meaning 1081 00:44:03,960 --> 00:44:06,570 or this special kind of communicative intention 1082 00:44:06,570 --> 00:44:09,600 we see in humans, which is that a communicator doesn't just 1083 00:44:09,600 --> 00:44:12,780 intend for the addressee to respond in a particular way, 1084 00:44:12,780 --> 00:44:16,830 but intends for the audience to respond in a particular way 1085 00:44:16,830 --> 00:44:19,170 by virtue of having understood the intention 1086 00:44:19,170 --> 00:44:21,480 of that communication. 1087 00:44:21,480 --> 00:44:25,110 And so this is just a, I think, particularly well-worded quote 1088 00:44:25,110 --> 00:44:27,330 from Seyfarth and Cheney who say basically 1089 00:44:27,330 --> 00:44:30,570 that listeners can acquire information from signallers, 1090 00:44:30,570 --> 00:44:33,030 but the signallers themselves don't, in the human sense, 1091 00:44:33,030 --> 00:44:37,064 intend to provide that information. 1092 00:44:37,064 --> 00:44:38,730 So the reason, really, for this contrast 1093 00:44:38,730 --> 00:44:40,470 between the human and animal cases 1094 00:44:40,470 --> 00:44:45,447 that I think if we want to build a model of human communication, 1095 00:44:45,447 --> 00:44:47,280 we need to differentiate it from other kinds 1096 00:44:47,280 --> 00:44:48,890 of communicative models we could have. 1097 00:44:48,890 --> 00:44:50,910 And we also need it to develop the types of abilities 1098 00:44:50,910 --> 00:44:52,650 that human infants have, but taking into account 1099 00:44:52,650 --> 00:44:54,358 the resources that infants start out with 1100 00:44:54,358 --> 00:44:58,396 and the developments that we see in the first few years of life. 1101 00:44:58,396 --> 00:45:00,270 So it's not going to be enough to have agents 1102 00:45:00,270 --> 00:45:01,978 that can influence each other's responses 1103 00:45:01,978 --> 00:45:04,034 or who can understand language, because it 1104 00:45:04,034 --> 00:45:06,450 seems like the recognition of these more abstract features 1105 00:45:06,450 --> 00:45:08,783 of communication and its causal effects on mental states 1106 00:45:08,783 --> 00:45:11,520 is actually present fairly early on as well. 1107 00:45:11,520 --> 00:45:15,720 And I think just relating this back to Liz's theory 1108 00:45:15,720 --> 00:45:18,330 that infants might have these different systems for agents, 1109 00:45:18,330 --> 00:45:19,740 understanding agents and their actions 1110 00:45:19,740 --> 00:45:22,240 on objects and then for social beings and their interactions 1111 00:45:22,240 --> 00:45:22,990 with each other. 1112 00:45:22,990 --> 00:45:24,570 It seems like communication and this type 1113 00:45:24,570 --> 00:45:25,986 of communicative intention that we 1114 00:45:25,986 --> 00:45:29,580 see combines these two kinds of things, where you have 1115 00:45:29,580 --> 00:45:32,370 an intention to produce an effect on someone else, 1116 00:45:32,370 --> 00:45:35,340 but by virtue of them understanding the mental states 1117 00:45:35,340 --> 00:45:39,000 that you have toward the world and as well as toward them. 1118 00:45:41,640 --> 00:45:44,820 And maybe it's the combining of these different domains that 1119 00:45:44,820 --> 00:45:47,550 helps infants put together their possibly 1120 00:45:47,550 --> 00:45:50,990 human-unique, but maybe not, communication skills.