1 00:00:00,000 --> 00:00:04,259 What if I told you that the word dog isn't just 2 00:00:04,259 --> 00:00:06,759 a noun? Right. It's really not. It's not just 3 00:00:06,759 --> 00:00:09,019 a picture floating around in your head. And it 4 00:00:09,019 --> 00:00:11,599 is definitely not just some dictionary definition 5 00:00:11,599 --> 00:00:13,820 you memorized in kindergarten. It is actually 6 00:00:13,820 --> 00:00:16,260 a highly advanced, totally invisible software 7 00:00:16,260 --> 00:00:19,480 program running in your brain right now. And, 8 00:00:19,579 --> 00:00:21,579 well, until today, you probably had absolutely 9 00:00:21,579 --> 00:00:23,960 no idea it was even operating. Yeah, because 10 00:00:23,960 --> 00:00:26,879 it is entirely invisible to us. I mean, we just 11 00:00:26,879 --> 00:00:29,320 move through our daily lives operating under 12 00:00:29,320 --> 00:00:32,719 this incredibly persistent illusion, the illusion 13 00:00:32,719 --> 00:00:34,939 that our thoughts are just a transparent window. 14 00:00:35,079 --> 00:00:36,700 Like we're just looking at the world exactly 15 00:00:36,700 --> 00:00:38,820 as it is. Exactly. We think we look out into 16 00:00:38,820 --> 00:00:41,299 the world, we see a tree, our brain takes a little 17 00:00:41,299 --> 00:00:43,859 photograph of the tree, and we just slap a label 18 00:00:43,859 --> 00:00:47,539 on it. We assume categorization is just a passive 19 00:00:47,539 --> 00:00:50,200 recording of reality. But the truth is infinitely 20 00:00:50,200 --> 00:00:53,299 more complex. And honestly, It's a lot more fascinating. 21 00:00:53,979 --> 00:00:56,299 So that is our mission for this deep dive today. 22 00:00:56,799 --> 00:00:59,240 We are going to completely tear down that illusion 23 00:00:59,240 --> 00:01:01,920 of the transparent window. We are decoding a 24 00:01:01,920 --> 00:01:05,620 genuinely paradigm shifting text today, conceptual 25 00:01:05,620 --> 00:01:09,079 form, the hidden structure of common sense concepts 26 00:01:09,079 --> 00:01:12,400 by Sandy Prasada. It is such a profound piece 27 00:01:12,400 --> 00:01:15,599 of fork. Prasada is operating at this incredibly 28 00:01:15,599 --> 00:01:18,640 rich intersection of psychology, linguistics, 29 00:01:18,840 --> 00:01:20,739 and philosophy. Yeah, he's really tying it all 30 00:01:20,739 --> 00:01:23,060 together. He is. And he's taking aim at these 31 00:01:23,060 --> 00:01:25,420 mysteries regarding human cognition that researchers 32 00:01:25,420 --> 00:01:28,319 have been basically banging their heads against 33 00:01:28,319 --> 00:01:31,340 for decades. Mysteries that the standard models 34 00:01:31,340 --> 00:01:34,260 of how we think have just utterly failed to resolve. 35 00:01:34,719 --> 00:01:37,079 Because when you or I think of everyday objects 36 00:01:37,079 --> 00:01:41,459 like a table, a cup, a dog, we only ever really 37 00:01:41,459 --> 00:01:43,260 think about their features, right? Right. The 38 00:01:43,260 --> 00:01:46,219 wood, the handle, the fur, the material stuff. 39 00:01:46,280 --> 00:01:49,409 The what? But Prasad is arguing that our concepts 40 00:01:49,409 --> 00:01:52,569 are actually these intricate mental lenses. They 41 00:01:52,569 --> 00:01:55,170 aren't just passively holding information. They 42 00:01:55,170 --> 00:01:57,790 are actually issuing these covert instructions 43 00:01:57,790 --> 00:02:00,489 for how you are allowed to think, reason, and 44 00:02:00,489 --> 00:02:03,909 even speak about that object. Which is wild when 45 00:02:03,909 --> 00:02:06,170 you sit with it. Realizing your own mind has 46 00:02:06,170 --> 00:02:08,810 this hidden operating system dictating the rules 47 00:02:08,810 --> 00:02:11,610 of reality, it completely changes how you experience 48 00:02:11,610 --> 00:02:14,789 your own thoughts. It really does. So let's unpack 49 00:02:14,789 --> 00:02:17,349 this. Because to really appreciate the magnitude 50 00:02:17,349 --> 00:02:19,909 of this breakthrough, I think we have to travel 51 00:02:19,909 --> 00:02:22,310 back in time a bit. Yeah, we have to look at 52 00:02:22,310 --> 00:02:25,250 the scientific blind spot. Right. We need to 53 00:02:25,250 --> 00:02:26,969 understand how the smartest people in cognitive 54 00:02:26,969 --> 00:02:29,210 science used to think you and I categorized the 55 00:02:29,210 --> 00:02:31,930 world and why those old models eventually just 56 00:02:31,930 --> 00:02:34,569 hit a massive brick wall. Let's do it. Because 57 00:02:34,569 --> 00:02:36,530 for the longest time, the dominant idea was what 58 00:02:36,530 --> 00:02:40,069 they call the classical theory, or the definitions 59 00:02:40,069 --> 00:02:42,789 approach. Which makes sense on the surface. If 60 00:02:42,789 --> 00:02:44,430 you think about how we naturally want to organize 61 00:02:44,430 --> 00:02:46,770 the world, we want clean lines. We want neat 62 00:02:46,770 --> 00:02:49,770 little boxes. So the theory was that your concept 63 00:02:49,770 --> 00:02:52,750 of a dog is basically a strict mental checklist. 64 00:02:53,409 --> 00:02:55,650 In order for something to be categorized as a 65 00:02:55,650 --> 00:02:58,449 dog, it has to possess properties that are individually 66 00:02:58,449 --> 00:03:01,169 necessary and jointly sufficient. So like, it 67 00:03:01,169 --> 00:03:04,169 must have four legs. It must bark. It must have 68 00:03:04,169 --> 00:03:06,669 fur, a tail. You check all the boxes and boom, 69 00:03:06,870 --> 00:03:09,310 you have a dog. Which sounds wonderfully logical 70 00:03:09,310 --> 00:03:12,689 if you are building a database or, you know, 71 00:03:12,689 --> 00:03:15,490 it works perfectly for mathematics. A triangle 72 00:03:15,490 --> 00:03:17,830 is a three -sided polygon where the interior 73 00:03:17,830 --> 00:03:21,050 angles add up to 180 degrees. That is a strict, 74 00:03:21,469 --> 00:03:24,150 perfect definition. But everyday life isn't math. 75 00:03:24,389 --> 00:03:26,770 No, it's not. And when researchers try to map 76 00:03:26,770 --> 00:03:29,490 that strict checklist onto everyday human concepts, 77 00:03:29,650 --> 00:03:32,210 it collapsed immediately. Because the world just 78 00:03:32,210 --> 00:03:34,330 doesn't operate in strict definitions. Right. 79 00:03:34,449 --> 00:03:36,740 Think about your own neighborhood. You walk outside 80 00:03:36,740 --> 00:03:39,340 and you see a dog that unfortunately lost a leg 81 00:03:39,340 --> 00:03:41,699 in an accident. So it has three legs. Right. 82 00:03:41,979 --> 00:03:44,919 And maybe it had its tail docked. And maybe it's 83 00:03:44,919 --> 00:03:47,199 a Basenji, which is a reed that physically cannot 84 00:03:47,199 --> 00:03:49,960 bark. Yeah, they just sort of yodel. Exactly. 85 00:03:50,439 --> 00:03:52,599 And let's say it has a terrible skin condition, 86 00:03:52,659 --> 00:03:55,319 so it has absolutely no fur. According to that 87 00:03:55,319 --> 00:03:57,639 classical checklist, this creature is missing 88 00:03:57,639 --> 00:04:00,689 almost every single necessary feature. It fails 89 00:04:00,689 --> 00:04:03,250 a test completely. It fails. And yet your brain 90 00:04:03,250 --> 00:04:05,650 doesn't glitch. You don't stand on the sidewalk 91 00:04:05,650 --> 00:04:08,650 wondering if it's a weird cat or a mutated squirrel. 92 00:04:09,289 --> 00:04:12,530 You know instantly and undeniably that it is 93 00:04:12,530 --> 00:04:15,229 a dog. There is zero logical contradiction in 94 00:04:15,229 --> 00:04:17,790 your mind when you encounter a barkless, hairless, 95 00:04:17,850 --> 00:04:19,670 three -legged dog. And that immediately proves 96 00:04:19,670 --> 00:04:22,490 we aren't using strict definitions. So cognitive 97 00:04:22,490 --> 00:04:24,949 science had to pivot. Right. If we aren't using 98 00:04:24,949 --> 00:04:27,290 strict checklists, what are we doing? And this 99 00:04:27,290 --> 00:04:30,329 led to the era of prototypes and exemplars. This 100 00:04:30,329 --> 00:04:33,949 was a massive shift toward statistical similarity. 101 00:04:34,470 --> 00:04:37,170 Okay, so the idea here is that instead of a checklist, 102 00:04:37,870 --> 00:04:41,779 your brain operates more like a blurry composite 103 00:04:41,779 --> 00:04:43,980 photograph. You've seen hundreds of dogs in your 104 00:04:43,980 --> 00:04:46,439 life, right? Yeah, thousands probably. So your 105 00:04:46,439 --> 00:04:49,279 brain averages them all out to create the prototype 106 00:04:49,279 --> 00:04:52,740 dog. Maybe it looks like a generic golden retriever 107 00:04:52,740 --> 00:04:55,769 or something. Alternatively, the exemplar theory 108 00:04:55,769 --> 00:04:58,569 says you just have a massive mental filing cabinet 109 00:04:58,569 --> 00:05:01,410 holding specific memories of every single dog 110 00:05:01,410 --> 00:05:04,230 you've ever actually encountered. Exactly. So 111 00:05:04,230 --> 00:05:05,889 when you see a new animal on the street, you 112 00:05:05,889 --> 00:05:08,509 just do a quick statistical comparison. Does 113 00:05:08,509 --> 00:05:10,930 this new thing resemble my prototype or does 114 00:05:10,930 --> 00:05:14,170 it match enough of my exemplar memories? Which 115 00:05:14,170 --> 00:05:16,750 was a huge step forward. I mean, it finally accounted 116 00:05:16,750 --> 00:05:19,829 for fuzzy boundaries. Oh, absolutely. It explains 117 00:05:19,829 --> 00:05:22,430 why we think a robin is a better or more typical 118 00:05:22,430 --> 00:05:25,490 bird than, say, a penguin. A robin matches the 119 00:05:25,490 --> 00:05:28,009 statistical prototype of feathers, flying, and 120 00:05:28,009 --> 00:05:30,430 small size much better than a flightless bird 121 00:05:30,430 --> 00:05:33,910 in a tuxedo does. Right. But even this statistical 122 00:05:33,910 --> 00:05:36,230 approach ran into a wall eventually, didn't it? 123 00:05:36,449 --> 00:05:39,470 It did. Similarity just isn't enough to explain 124 00:05:39,470 --> 00:05:42,050 how we categorize things, which eventually force 125 00:05:42,050 --> 00:05:44,430 the field into what's called the theory approach, 126 00:05:44,730 --> 00:05:47,199 or psychological essentialism. I find this one 127 00:05:47,199 --> 00:05:49,980 so fascinating because it feels incredibly intuitive. 128 00:05:50,439 --> 00:05:53,139 It's this idea that we are all born as naive 129 00:05:53,139 --> 00:05:55,420 little scientists holding onto these intuitive 130 00:05:55,420 --> 00:05:57,879 theories about how the world works. Especially 131 00:05:57,879 --> 00:06:00,199 for biological things. Sure. We assume there's 132 00:06:00,199 --> 00:06:03,720 some hidden unchanging essence. Right. You might 133 00:06:03,720 --> 00:06:06,339 have no idea what canine DNA looks like. You 134 00:06:06,339 --> 00:06:07,899 might know literally nothing about genetics, 135 00:06:07,899 --> 00:06:10,860 but you have this conceptual placeholder, this 136 00:06:10,860 --> 00:06:13,579 deep seeded belief that there is an invisible 137 00:06:13,579 --> 00:06:16,569 dogness. inside the animal that makes it what 138 00:06:16,569 --> 00:06:19,490 it is. Yeah, you could shave a dog, paint it 139 00:06:19,490 --> 00:06:21,769 green with tiger stripes, teach it to walk on 140 00:06:21,769 --> 00:06:24,250 its hind legs. And your brain still says, underneath 141 00:06:24,250 --> 00:06:27,949 all that weird stuff, its essence is dog. Precisely. 142 00:06:28,310 --> 00:06:31,129 So we've gone from strict checklists to fuzzy 143 00:06:31,129 --> 00:06:34,370 statistical averages to mystical hidden essences. 144 00:06:34,629 --> 00:06:36,509 It sounds like science is getting closer to the 145 00:06:36,509 --> 00:06:39,329 truth, right? But then Prasada looks at this 146 00:06:39,329 --> 00:06:41,509 entire historical progression and says, you've 147 00:06:41,509 --> 00:06:44,360 all missed the actual machinery. He points out 148 00:06:44,360 --> 00:06:47,199 a massive glaring blind spot shared by every 149 00:06:47,199 --> 00:06:49,480 single one of those theories. He says they are 150 00:06:49,480 --> 00:06:53,240 all entirely non -perspectival. Non -perspectival. 151 00:06:53,660 --> 00:06:57,029 Okay. Let's really ground that, because it's 152 00:06:57,029 --> 00:06:58,769 basically the fulcrum of his whole argument. 153 00:06:59,230 --> 00:07:01,810 What does that mean in practice? It means that 154 00:07:01,810 --> 00:07:04,889 the classical theory, the prototype theory, the 155 00:07:04,889 --> 00:07:08,170 essentialist theory, they all focus 100 % of 156 00:07:08,170 --> 00:07:11,050 their attention on the content of the concept. 157 00:07:11,269 --> 00:07:13,750 The properties of the thing being represented. 158 00:07:14,189 --> 00:07:16,829 Yes. They are obsessed with the fur, the legs, 159 00:07:17,009 --> 00:07:19,449 the bark, or the hidden DNA, but they completely 160 00:07:19,449 --> 00:07:22,069 ignore the form. They ignore the structure of 161 00:07:22,069 --> 00:07:24,350 the representation itself. It's like standard 162 00:07:24,350 --> 00:07:26,930 cognitive science has spent 50 years obsessing 163 00:07:26,930 --> 00:07:29,649 over the slides we put under a microscope. That's 164 00:07:29,649 --> 00:07:31,629 a great way to look at it. Right. They've been 165 00:07:31,629 --> 00:07:34,740 meticulously categorizing the specimens. cataloging 166 00:07:34,740 --> 00:07:37,339 the dog fur in the dog DNA on the glass slide. 167 00:07:38,040 --> 00:07:40,120 And Prasada walks into the lab, taps on the metal 168 00:07:40,120 --> 00:07:43,040 barrel of the microscope, and says, have any 169 00:07:43,040 --> 00:07:45,279 of you bothered to look at the lenses? Because 170 00:07:45,279 --> 00:07:47,019 these lenses are bending the light before it 171 00:07:47,019 --> 00:07:49,920 even hits your eye. I love that analogy. Because 172 00:07:49,920 --> 00:07:52,480 past theories truly treated our concepts like 173 00:07:52,480 --> 00:07:54,959 passive buckets. You see a dog and your brain 174 00:07:54,959 --> 00:07:57,740 just drops dog facts into the dog bucket. But 175 00:07:57,740 --> 00:08:00,560 Prasada is arguing that concepts are active, 176 00:08:00,959 --> 00:08:03,199 refracting lenses. They don't just hold data. 177 00:08:03,360 --> 00:08:06,399 No, they generate a highly specific perspective. 178 00:08:07,360 --> 00:08:09,939 They dictate to your brain exactly how it is 179 00:08:09,939 --> 00:08:11,899 allowed to interpret the data it's receiving. 180 00:08:12,360 --> 00:08:14,019 Now, whenever we start talking about invisible 181 00:08:14,019 --> 00:08:16,800 lenses shaping how we think, we inevitably crash 182 00:08:16,800 --> 00:08:20,120 into the Worfian worry, linguistic relativity. 183 00:08:20,180 --> 00:08:22,620 Oh, inevitably. I feel like this goes up in almost 184 00:08:22,620 --> 00:08:25,019 every conversation about language. It's this 185 00:08:25,019 --> 00:08:26,939 creeping anxiety that if you speak a different 186 00:08:26,939 --> 00:08:28,899 language with a different grammatical structure, 187 00:08:29,360 --> 00:08:31,819 you are trapped in a completely different reality. 188 00:08:31,980 --> 00:08:34,860 Like your language is a prison. Right. If my 189 00:08:34,860 --> 00:08:36,860 language doesn't have a word for a concept, am 190 00:08:36,860 --> 00:08:39,320 I physically incapable of thinking that thought? 191 00:08:39,500 --> 00:08:42,259 It's a very common fear. And Prasada takes it 192 00:08:42,259 --> 00:08:44,799 head on right in the first chapter. He acknowledges 193 00:08:44,799 --> 00:08:47,139 that, yes, languages absolutely differ on the 194 00:08:47,139 --> 00:08:49,440 surface level. English and French, for example, 195 00:08:49,820 --> 00:08:51,799 treat the objects in your living room very differently. 196 00:08:52,220 --> 00:08:55,480 OK, how so? Well, in English, the word furniture 197 00:08:55,480 --> 00:08:57,940 is a mass noun. You cannot grammatically say, 198 00:08:58,200 --> 00:09:00,820 I bought one furniture today. It sounds wrong. 199 00:09:01,039 --> 00:09:03,399 Right. You have to say a piece of furniture. 200 00:09:03,620 --> 00:09:06,419 It's treated like water or sand. You can't say 201 00:09:06,419 --> 00:09:09,899 one sand. Exactly. But in French, the word is 202 00:09:09,899 --> 00:09:12,860 meubles and it is a count noun. You can just 203 00:09:12,860 --> 00:09:14,879 slap a number in front of it. You can count them 204 00:09:14,879 --> 00:09:17,710 directly. Oh, interesting. Or think about Hindi. 205 00:09:18,389 --> 00:09:21,289 The text mentions the word, by. It's used to 206 00:09:21,289 --> 00:09:23,710 describe both a biological brother and a male 207 00:09:23,710 --> 00:09:26,570 cousin. Right. English forces you to separate 208 00:09:26,570 --> 00:09:28,690 those two relationships into completely different 209 00:09:28,690 --> 00:09:32,230 words. Brother versus cousin. So the Warfian 210 00:09:32,230 --> 00:09:34,909 worry would say, wait, does the fact that Hindi 211 00:09:34,909 --> 00:09:38,509 uses one word mean a Hindi speaker is cognitively 212 00:09:38,509 --> 00:09:40,789 incapable of understanding the biological difference 213 00:09:40,789 --> 00:09:43,950 between their sibling and their aunt's son? But 214 00:09:43,950 --> 00:09:46,210 here is the critical distinction Prasada makes. 215 00:09:46,320 --> 00:09:49,519 Does a Hindi speaker actually get confused about 216 00:09:49,519 --> 00:09:51,659 who their mother gave birth to? Well, obviously 217 00:09:51,659 --> 00:09:54,220 not. They know exactly who their immediate family 218 00:09:54,220 --> 00:09:57,059 is. Exactly. And does an English speaker look 219 00:09:57,059 --> 00:09:59,360 at a chair and a sofa and genuinely think they 220 00:09:59,360 --> 00:10:02,220 are an undifferentiated puddle of material just 221 00:10:02,220 --> 00:10:04,679 because the word furniture is a mass noun? No, 222 00:10:04,679 --> 00:10:06,919 I clearly see a chair and a couch as two distinct 223 00:10:06,919 --> 00:10:09,139 things. Right. Because even when the surface 224 00:10:09,139 --> 00:10:11,399 of a language doesn't force a distinction or 225 00:10:11,399 --> 00:10:14,889 frames it oddly, the underlying abstract conceptual 226 00:10:14,889 --> 00:10:17,629 distinctions are still perfectly readily available 227 00:10:17,629 --> 00:10:22,549 to all human beings. So the Warfian worry basically 228 00:10:22,549 --> 00:10:25,110 overestimates the power of surface grammar, and 229 00:10:25,110 --> 00:10:27,809 it completely underestimates this universal deep 230 00:10:27,809 --> 00:10:31,230 architecture of conceptual form. Precisely. We 231 00:10:31,230 --> 00:10:33,629 all have the same microscope lenses even if we 232 00:10:33,629 --> 00:10:35,370 painted the outside of the microscope's different 233 00:10:35,370 --> 00:10:38,389 colors. The conceptual form is deep, abstract, 234 00:10:38,629 --> 00:10:41,929 and shared across humanity. Okay, I have to push 235 00:10:41,929 --> 00:10:44,600 back here. Go for it. If this conceptual form 236 00:10:44,600 --> 00:10:47,700 is completely invisible, and if we've been blind 237 00:10:47,700 --> 00:10:50,399 to it for decades, how do we even prove it's 238 00:10:50,399 --> 00:10:52,139 actually there? I mean, are we just inventing 239 00:10:52,139 --> 00:10:54,600 a new ghost in the machine? If I can only see 240 00:10:54,600 --> 00:10:57,340 the slide, how do you prove to me the lens exists? 241 00:10:57,860 --> 00:11:01,220 That is the perfect question. To prove the lens 242 00:11:01,220 --> 00:11:03,799 exists, we have to run a cognitive stress test. 243 00:11:04,100 --> 00:11:06,379 We have to take a single physical object, hold 244 00:11:06,379 --> 00:11:09,139 reality completely constant, and then deliberately 245 00:11:09,139 --> 00:11:11,919 swap out the concept we used to look at it. Swap 246 00:11:11,919 --> 00:11:14,620 the lens. Swap the lens. If changing the word 247 00:11:14,620 --> 00:11:16,960 in your head completely changes what you believe 248 00:11:16,960 --> 00:11:19,179 to be true about the physical object in front 249 00:11:19,179 --> 00:11:21,620 of you, then we have proven the concept is doing 250 00:11:21,620 --> 00:11:25,000 active, structural work. We have to dive into 251 00:11:25,000 --> 00:11:28,179 what Prasada calls the instance -of -kind perspective. 252 00:11:28,639 --> 00:11:30,879 Okay, let's set the scene for the stress test. 253 00:11:31,379 --> 00:11:33,740 Imagine you are standing in a carpenter's workshop 254 00:11:33,740 --> 00:11:36,240 and there is a wooden table in front of you. 255 00:11:36,240 --> 00:11:39,159 Got it. You can look at that exact same arrangement 256 00:11:39,159 --> 00:11:42,559 of atoms, that exact same physical object. and 257 00:11:42,559 --> 00:11:44,960 you can think about it using the concept table, 258 00:11:45,440 --> 00:11:47,620 or you can look at it and think about it using 259 00:11:47,620 --> 00:11:50,940 the concept wood. Right. The physical reality 260 00:11:50,940 --> 00:11:54,509 hasn't shifted a single millimeter, but your 261 00:11:54,509 --> 00:11:57,029 internal mental lens just clicked over. And notice 262 00:11:57,029 --> 00:11:59,009 what happens to your mind the moment it clicks. 263 00:11:59,669 --> 00:12:02,029 When you look at it through the lens of the concept 264 00:12:02,029 --> 00:12:05,370 table, you inherently understand that its shape 265 00:12:05,370 --> 00:12:08,210 and its structure are non -arbitrary. It has 266 00:12:08,210 --> 00:12:11,200 a flat top and four legs because It is a table. 267 00:12:11,460 --> 00:12:14,240 Exactly. If you were to take a sledgehammer and 268 00:12:14,240 --> 00:12:17,019 smash it into a thousand splinters, it would 269 00:12:17,019 --> 00:12:20,279 no longer be a table. The shape matters. But 270 00:12:20,279 --> 00:12:22,320 the second I shift my brain to look at it through 271 00:12:22,320 --> 00:12:25,360 the concept wood, the shape is suddenly entirely 272 00:12:25,360 --> 00:12:27,700 arbitrary. It doesn't matter at all. It doesn't 273 00:12:27,700 --> 00:12:29,460 matter if it's shaped like a table, shaped like 274 00:12:29,460 --> 00:12:32,080 a chair, or smashed into a pile of splinters 275 00:12:32,080 --> 00:12:35,379 on the floor. It is still wood. The object didn't 276 00:12:35,379 --> 00:12:38,990 change. The concept table assigns the object 277 00:12:38,990 --> 00:12:42,190 a proper orientation. It has an up and a down. 278 00:12:42,629 --> 00:12:44,529 Yeah, an upside down table is just a table that's 279 00:12:44,529 --> 00:12:47,529 the wrong way up. But the concept wood doesn't 280 00:12:47,529 --> 00:12:49,690 have an upside down. You can ask the carpenter, 281 00:12:49,870 --> 00:12:52,029 how many parts does this table have? You absolutely 282 00:12:52,029 --> 00:12:54,830 cannot ask, how many parts does this wood have? 283 00:12:55,210 --> 00:12:57,549 It doesn't make sense grammatically or conceptually. 284 00:12:58,049 --> 00:13:00,730 It is staggering when you realize that. The physical 285 00:13:00,730 --> 00:13:03,610 matter is identical. But the word you boot up 286 00:13:03,610 --> 00:13:05,990 in your mind dictates whether it has an orientation, 287 00:13:06,470 --> 00:13:08,070 whether it can be counted, whether its shape 288 00:13:08,070 --> 00:13:11,610 is important or irrelevant. It is absolute proof 289 00:13:11,610 --> 00:13:14,509 that the perspective isn't radiating out of the 290 00:13:14,509 --> 00:13:17,690 object. It is being beamed onto the object by 291 00:13:17,690 --> 00:13:20,309 your conceptual form. And Prasada breaks down 292 00:13:20,309 --> 00:13:23,289 the specific perspective, the instance of kind 293 00:13:23,289 --> 00:13:26,370 perspective, into six incredibly intricate abstract 294 00:13:26,370 --> 00:13:28,879 characteristics. And instead of just listing 295 00:13:28,879 --> 00:13:31,440 them out like a textbook, let's actually walk 296 00:13:31,440 --> 00:13:33,659 through them in real time. Let's do it. Okay. 297 00:13:34,100 --> 00:13:36,080 Imagine you wake up in the morning and you look 298 00:13:36,080 --> 00:13:38,740 at your family dog. Let's call him Fido. The 299 00:13:38,740 --> 00:13:40,820 very second your brain categorizes the animal 300 00:13:40,820 --> 00:13:44,240 in your kitchen as a dog, these six operations 301 00:13:44,240 --> 00:13:47,200 run simultaneously in the background. So, operation 302 00:13:47,200 --> 00:13:50,289 number one. My brain instantly recognizes that 303 00:13:50,289 --> 00:13:54,470 Fido is simultaneously unique and utterly non 304 00:13:54,470 --> 00:13:56,769 -unique. Right. Right. You look at Fido and you 305 00:13:56,769 --> 00:14:00,169 know there is only one Fido. He has his specific 306 00:14:00,169 --> 00:14:03,110 memories, his specific collar, that specific 307 00:14:03,110 --> 00:14:06,009 spot he likes to sleep on. He is a spatially 308 00:14:06,009 --> 00:14:08,590 and temporally bounded, utterly unique physical 309 00:14:08,590 --> 00:14:11,090 instance. But in the exact same millisecond? 310 00:14:11,309 --> 00:14:13,990 In the exact same millisecond, your conceptual 311 00:14:13,990 --> 00:14:16,990 lens categorizes him, which means you recognize 312 00:14:16,990 --> 00:14:19,830 he is just one token of a type. He is one of 313 00:14:19,830 --> 00:14:22,950 an infinite number of possible dogs. Your brain 314 00:14:22,950 --> 00:14:25,850 effortlessly holds this paradox. He is everything, 315 00:14:26,669 --> 00:14:28,629 and he is just another one of the many. Which 316 00:14:28,629 --> 00:14:30,850 leads right into operation number two, the member 317 00:14:30,850 --> 00:14:33,450 of relation. Right. So let's say I take Fido 318 00:14:33,450 --> 00:14:36,090 to the vet for his checkup. While we're sitting 319 00:14:36,090 --> 00:14:38,009 in the waiting room, I can look at Fido, the 320 00:14:38,009 --> 00:14:41,470 concrete instance, but my conceptual lens allows 321 00:14:41,470 --> 00:14:44,629 me to effortlessly zoom out to the abstract kind. 322 00:14:44,950 --> 00:14:46,850 Like, I can read a poster on the vet's wall that 323 00:14:46,850 --> 00:14:50,090 says, the dog evolved from wolves. And when you 324 00:14:50,090 --> 00:14:52,149 read that, your brain doesn't glitch. You don't 325 00:14:52,149 --> 00:14:54,690 look at Fido and think, wait, Fido didn't evolve 326 00:14:54,690 --> 00:14:57,070 from a wolf. Fido was born in Ohio three years 327 00:14:57,070 --> 00:14:59,070 ago. Exactly. Because I'm not talking about Fido. 328 00:14:59,129 --> 00:15:01,649 I'm not talking about any specific physical dog. 329 00:15:02,269 --> 00:15:05,029 My mind is toggled from the instance to the abstract 330 00:15:05,029 --> 00:15:07,429 kind itself. The lens allows you to navigate 331 00:15:07,429 --> 00:15:10,129 both levels of reality, the concrete physical 332 00:15:10,129 --> 00:15:12,549 thing in front of you and the abstract historical 333 00:15:12,549 --> 00:15:15,269 lineage of the entire species, without getting 334 00:15:15,269 --> 00:15:18,299 them confused for even a second. Okay, now the 335 00:15:18,299 --> 00:15:20,679 vet calls you into the room. This brings us to 336 00:15:20,679 --> 00:15:23,379 operation number three, which is honestly deeply 337 00:15:23,379 --> 00:15:25,960 profound for cognitive science. Prasada calls 338 00:15:25,960 --> 00:15:29,580 it being intrinsically typed. We do not engage 339 00:15:29,580 --> 00:15:32,340 in untyped reasoning. I really had to sit with 340 00:15:32,340 --> 00:15:34,740 this one because traditional psychology sometimes 341 00:15:34,740 --> 00:15:36,960 makes it sound like human vision is just a camera. 342 00:15:37,360 --> 00:15:40,740 We see a blob of matter, a thing, and then we 343 00:15:40,740 --> 00:15:42,799 search our memory to slap a label on it, like, 344 00:15:42,960 --> 00:15:46,259 huh, that thing is a dog. But Prasada shows that 345 00:15:46,259 --> 00:15:48,559 human reasoning physically cannot operate on 346 00:15:48,559 --> 00:15:51,639 an untyped thing. For your brain to even begin 347 00:15:51,639 --> 00:15:54,039 reasoning about an object in space, it must first 348 00:15:54,039 --> 00:15:56,759 categorize it as some kind of thing. Even if 349 00:15:56,759 --> 00:15:59,299 I don't know what it is. Even then. Let's say 350 00:15:59,299 --> 00:16:02,200 you walk into the dark vet examination room and 351 00:16:02,200 --> 00:16:04,779 see a terrifying shadowy lump on the table and 352 00:16:04,779 --> 00:16:07,320 you have absolutely no idea what it is. Your 353 00:16:07,320 --> 00:16:09,799 brain still demands that it be categorized as 354 00:16:09,799 --> 00:16:13,419 a kind of unknown physical object or a kind of 355 00:16:13,419 --> 00:16:15,720 unknown animal before you can even begin to investigate 356 00:16:15,720 --> 00:16:18,159 its properties. I can't just think about properties 357 00:16:18,159 --> 00:16:21,059 floating in a void. No, you cannot think about 358 00:16:21,059 --> 00:16:23,200 properties without first establishing a type. 359 00:16:23,799 --> 00:16:26,399 The kind is the foundation of the thought, not 360 00:16:26,399 --> 00:16:28,220 the decoration you add at the end once you've 361 00:16:28,220 --> 00:16:31,759 figured it out. OK, so the lights turn on. It's 362 00:16:31,759 --> 00:16:34,379 just Fido on the table. The vet starts examining 363 00:16:34,379 --> 00:16:37,799 him and the vet asks, hey, why does Fido have 364 00:16:37,799 --> 00:16:40,519 four legs? And I say, well, because he's a dog. 365 00:16:41,070 --> 00:16:43,169 And that brings us to operation number four, 366 00:16:43,509 --> 00:16:47,129 the explanatory link or formal explanation. Right. 367 00:16:48,120 --> 00:16:50,860 Now, let me play devil's advocate here. If I 368 00:16:50,860 --> 00:16:53,379 explain Fido's legs by just saying he's a dog, 369 00:16:53,860 --> 00:16:56,419 isn't that just a circular, entirely empty statement? 370 00:16:56,600 --> 00:16:59,059 It's a tautology. How is that actually doing 371 00:16:59,059 --> 00:17:02,500 any real cognitive work? It absolutely sounds 372 00:17:02,500 --> 00:17:05,140 like a rhetorical trick at first glance. It sounds 373 00:17:05,140 --> 00:17:08,970 totally circular. But empirical studies in psychology, 374 00:17:08,970 --> 00:17:11,430 they've run tests on both adults and very young 375 00:17:11,430 --> 00:17:14,210 children, and they consistently show that human 376 00:17:14,210 --> 00:17:18,569 beings treat these formal explanations as distinct, 377 00:17:18,930 --> 00:17:22,170 powerful, and genuinely explanatory. But why? 378 00:17:22,450 --> 00:17:24,869 Why does my brain accept that as a good answer? 379 00:17:25,250 --> 00:17:27,690 Because the conceptual form is linking the part 380 00:17:27,690 --> 00:17:31,170 to the whole. You are explaining a specific localized 381 00:17:31,170 --> 00:17:33,690 property, having four legs, by referencing the 382 00:17:33,690 --> 00:17:36,589 intrinsic nature of the kind. dog. Oh I see. 383 00:17:36,930 --> 00:17:39,529 You are telling the vet his legs aren't an accident 384 00:17:39,529 --> 00:17:41,390 of his environment. He didn't grow them because 385 00:17:41,390 --> 00:17:43,869 he eats a lot of protein. The property flows 386 00:17:43,869 --> 00:17:46,789 inevitably from the very core nature of what 387 00:17:46,789 --> 00:17:49,549 this creature is. It's an explanation of origin 388 00:17:49,549 --> 00:17:53,009 based on kind. Which perfectly sets up operation 389 00:17:53,009 --> 00:17:55,849 number five and honestly this is the one that 390 00:17:55,849 --> 00:17:59,150 feels like literal magic to me. It's the generation 391 00:17:59,150 --> 00:18:03,099 of normative expectations. So the vet continues 392 00:18:03,099 --> 00:18:06,539 the exam and let's say, unfortunately, they discover 393 00:18:06,539 --> 00:18:10,079 Fido only has three legs. Right. Think about 394 00:18:10,079 --> 00:18:12,400 your reaction to that. You do not just shrug 395 00:18:12,400 --> 00:18:15,460 and think, ah, a statistical outlier. Some dogs 396 00:18:15,460 --> 00:18:17,799 have four legs, some have three. Just like some 397 00:18:17,799 --> 00:18:21,000 cars are red and some are blue. No. My brain 398 00:18:21,000 --> 00:18:24,019 immediately generates a normative judgment. I 399 00:18:24,019 --> 00:18:26,339 think something happened to this animal. It is 400 00:18:26,339 --> 00:18:29,400 defective or it has been injured. But here is 401 00:18:29,400 --> 00:18:32,230 why that is so mind -bending. Raw experience 402 00:18:32,230 --> 00:18:35,710 only ever gives us what is. Exactly. I can only 403 00:18:35,710 --> 00:18:38,490 observe reality as it physically exists in front 404 00:18:38,490 --> 00:18:41,309 of me. I am observing a three -legged dog. That 405 00:18:41,309 --> 00:18:43,869 is the physical truth of the universe in that 406 00:18:43,869 --> 00:18:47,309 moment. But somehow, your invisible conceptual 407 00:18:47,309 --> 00:18:50,710 lens overrides the physical reality and imposes 408 00:18:50,710 --> 00:18:53,289 a rule on the universe, generating an idea of 409 00:18:53,289 --> 00:18:56,250 what ought to be. I look at what is, and my brain 410 00:18:56,250 --> 00:18:59,640 says, no, it's missing something. The concept 411 00:18:59,640 --> 00:19:03,049 dog. isn't just describing reality. It has an 412 00:19:03,049 --> 00:19:06,190 intrinsic structure dictating the rules for reality. 413 00:19:06,829 --> 00:19:09,089 It is the birth of the normative dimension in 414 00:19:09,089 --> 00:19:11,509 human thought. It cannot be derived simply from 415 00:19:11,509 --> 00:19:13,750 counting features in the physical world. You 416 00:19:13,750 --> 00:19:16,109 could count a million three -legged dogs and 417 00:19:16,109 --> 00:19:18,069 it wouldn't change the fact that the conceptual 418 00:19:18,069 --> 00:19:20,809 form dictates they are supposed to have four. 419 00:19:21,190 --> 00:19:23,049 The rule comes from the structure of the concept, 420 00:19:23,369 --> 00:19:25,910 not the statistics of the environment. Exactly. 421 00:19:26,890 --> 00:19:30,180 And finally, we leave the vet. Operation number 422 00:19:30,180 --> 00:19:34,079 six, similarity without being the basis of categorization. 423 00:19:34,700 --> 00:19:37,759 So I take Fido to the zoo after the vet. We walk 424 00:19:37,759 --> 00:19:41,220 past the dingo enclosure. You look at the dingo? 425 00:19:42,019 --> 00:19:44,880 Visually, physically, statistically, it looks 426 00:19:44,880 --> 00:19:48,079 incredibly similar to Fido. It has four legs, 427 00:19:48,319 --> 00:19:51,980 fur, a tail, pointy ears. But my conceptual lens 428 00:19:51,980 --> 00:19:55,339 knows it is a different kind of animal. It isn't 429 00:19:55,339 --> 00:19:57,779 a dog. And then you walk to the penguin enclosure. 430 00:19:58,140 --> 00:20:00,119 Right. A penguin looks absolutely nothing like 431 00:20:00,119 --> 00:20:03,710 a robin or an eagle. It doesn't fly. It swims. 432 00:20:04,029 --> 00:20:06,269 It stands upright. It looks like it's wearing 433 00:20:06,269 --> 00:20:08,950 a tuxedo. The physical similarity is basically 434 00:20:08,950 --> 00:20:12,589 zero. But my conceptual lens says that is a bird. 435 00:20:12,950 --> 00:20:15,369 Because the lens dictates what the thing is first 436 00:20:15,369 --> 00:20:18,750 at a deep structural level. And once it establishes 437 00:20:18,750 --> 00:20:21,289 what the thing is, then it tells you how to expect 438 00:20:21,289 --> 00:20:23,529 it to be similar to other things rather than 439 00:20:23,529 --> 00:20:26,589 letting raw visual similarity dictate the category. 440 00:20:26,880 --> 00:20:29,539 The Penguin and the Robin share an underlying 441 00:20:29,539 --> 00:20:32,700 avian kind. Even if the surface details diverge 442 00:20:32,700 --> 00:20:36,059 wildly, it's all about internal sorting rules. 443 00:20:36,599 --> 00:20:38,799 Right. And Prasada gives this brilliant illustration 444 00:20:38,799 --> 00:20:41,880 in the text of how deep these sortle principles 445 00:20:41,880 --> 00:20:46,299 go. And it involves a cognitive stress test using 446 00:20:46,299 --> 00:20:49,140 a fairy tale. Ah, the Cinderella test. Yeah. 447 00:20:49,259 --> 00:20:51,420 It perfectly reveals the limits of our internal 448 00:20:51,420 --> 00:20:54,579 software. So think about reading Cinderella to 449 00:20:54,579 --> 00:20:57,359 a child. You get to the part where the fairy 450 00:20:57,359 --> 00:20:59,579 godmother waves her wand and the pumpkin magically 451 00:20:59,579 --> 00:21:02,740 turns into a magnificent carriage. Kids accept 452 00:21:02,740 --> 00:21:05,619 this instantly. Adults accept it. It's a fairy 453 00:21:05,619 --> 00:21:09,099 tale. Magic exists. We move on. But imagine a 454 00:21:09,099 --> 00:21:11,640 different fairy tale. Imagine the fairy godmother 455 00:21:11,640 --> 00:21:13,720 waves her wand and the pumpkin magically turns 456 00:21:13,720 --> 00:21:17,440 into a totally different pumpkin. It immediately 457 00:21:17,440 --> 00:21:19,720 breaks your brain. It really does. It feels like 458 00:21:19,720 --> 00:21:22,759 a continuity error in a movie. It's deeply unsettling. 459 00:21:22,779 --> 00:21:26,009 But why? Why does my brain accept a vegetable 460 00:21:26,009 --> 00:21:29,009 turning into a vehicle, but completely reject 461 00:21:29,009 --> 00:21:31,230 a vegetable turning into another vegetable? Because 462 00:21:31,230 --> 00:21:34,309 turning into a carriage changes the kind of thing 463 00:21:34,309 --> 00:21:37,069 it is. Your conceptual software for physical 464 00:21:37,069 --> 00:21:40,990 objects tracks the matter. It says, OK, the physical 465 00:21:40,990 --> 00:21:43,029 stuff of the pumpkin was magically rearranged 466 00:21:43,029 --> 00:21:45,589 into a new kind of thing, a carriage. I can track 467 00:21:45,589 --> 00:21:47,650 the continuity of the matter. Right. This stuff 468 00:21:47,650 --> 00:21:50,670 got reorganized. But turning a pumpkin into a 469 00:21:50,670 --> 00:21:53,329 different pumpkin violates the internal sorting 470 00:21:53,329 --> 00:21:56,380 rules of identity. Because you can't have one 471 00:21:56,380 --> 00:21:58,880 instance of a kind become a different instance 472 00:21:58,880 --> 00:22:02,420 of the exact same kind. Exactly. It violates 473 00:22:02,420 --> 00:22:04,559 the numerical identity of the first instance. 474 00:22:05,180 --> 00:22:07,539 Pumpkin A cannot become pumpkin B because they 475 00:22:07,539 --> 00:22:09,819 are both already defined by the same physical 476 00:22:09,819 --> 00:22:12,660 matter rules. Your conceptual lens for physical 477 00:22:12,660 --> 00:22:15,359 objects simply rejects it as computationally 478 00:22:15,359 --> 00:22:17,500 impossible. Even in a world where magic wands 479 00:22:17,500 --> 00:22:20,200 exist. So we firmly established that these lenses 480 00:22:20,200 --> 00:22:23,420 exist and they are doing incredibly heavy lifting. 481 00:22:23,759 --> 00:22:25,880 They are dictating uniqueness. They are writing 482 00:22:25,880 --> 00:22:29,039 normative rules about what ought to be. They 483 00:22:29,039 --> 00:22:31,279 are policing identity in fairy tales. They're 484 00:22:31,279 --> 00:22:34,000 doing a lot. So the natural next question is, 485 00:22:34,380 --> 00:22:36,880 what is the scope of this? Does every single 486 00:22:36,880 --> 00:22:39,599 word in the English language have a totally unique 487 00:22:39,599 --> 00:22:42,539 bespoke lens built from scratch or is there an 488 00:22:42,539 --> 00:22:45,230 underlying system? This is where we move into 489 00:22:45,230 --> 00:22:47,750 chapters four and five, where Prasada maps out 490 00:22:47,750 --> 00:22:50,789 the architecture. He completely rejects the idea 491 00:22:50,789 --> 00:22:54,130 of a uniform conceptual form. Meaning? Meaning 492 00:22:54,130 --> 00:22:56,589 he doesn't believe there is just one master template 493 00:22:56,589 --> 00:23:00,369 that every single concept uses, but he also rejects 494 00:23:00,369 --> 00:23:02,589 the idea that it's total chaos, with every word 495 00:23:02,589 --> 00:23:05,430 having its own random rules. Instead, he argues 496 00:23:05,430 --> 00:23:08,170 we have a highly constrained generative system. 497 00:23:08,670 --> 00:23:11,349 I kept picturing it like downloading software 498 00:23:11,349 --> 00:23:13,589 packages to my phone. That's a good way to look 499 00:23:13,589 --> 00:23:16,569 at it. If I download the concept lion, I get 500 00:23:16,569 --> 00:23:19,170 a very basic straightforward counting app. But 501 00:23:19,170 --> 00:23:22,430 if I download the concept book, I get this incredibly 502 00:23:22,430 --> 00:23:25,990 complex dual ledger accounting software. Let's 503 00:23:25,990 --> 00:23:27,789 actually look at the intricate variety of how 504 00:23:27,789 --> 00:23:30,930 our brains count and individuate reality. Let's 505 00:23:30,930 --> 00:23:34,430 start simple with lion or fork. When you acquire 506 00:23:34,430 --> 00:23:38,000 the concept of a lion or a fork, That specific 507 00:23:38,000 --> 00:23:40,420 conceptual form provides your brain with exactly 508 00:23:40,420 --> 00:23:42,940 one way of numerically distinguishing instances. 509 00:23:43,619 --> 00:23:45,420 You distinguish them by their physical matter. 510 00:23:46,220 --> 00:23:48,599 Line A is not line B because they are made of 511 00:23:48,599 --> 00:23:51,279 completely different physical stuff. Exactly. 512 00:23:51,700 --> 00:23:53,920 Fork A is not fork B because they are two distinct 513 00:23:53,920 --> 00:23:56,099 pieces of metal. It's straightforward one -to 514 00:23:56,099 --> 00:23:58,960 -one counting. But then you acquire a concept 515 00:23:58,960 --> 00:24:01,740 like necklace and suddenly the rules change. 516 00:24:02,500 --> 00:24:04,440 Necklace provides your brain with two distinct 517 00:24:04,440 --> 00:24:06,960 ways of counting. Think about a scenario where 518 00:24:06,960 --> 00:24:09,240 two friends walk into a party. You look at them 519 00:24:09,240 --> 00:24:11,839 and say, wow, Sally and Sue are wearing the exact 520 00:24:11,839 --> 00:24:14,759 same necklace. Now, when I say that, nobody in 521 00:24:14,759 --> 00:24:16,720 the room thinks Sally and Sue have their heads 522 00:24:16,720 --> 00:24:19,940 squeezed together into one giant singular piece 523 00:24:19,940 --> 00:24:22,700 of jewelry. Right. We all know they are physically 524 00:24:22,700 --> 00:24:25,599 wearing two distinct objects. But linguistically 525 00:24:25,599 --> 00:24:28,519 and conceptually, we effortlessly count them 526 00:24:28,519 --> 00:24:31,839 as the same necklace. Why? Because they share 527 00:24:31,839 --> 00:24:35,299 an abstract perceptual similarity. The conceptual 528 00:24:35,299 --> 00:24:37,720 form of necklace is structurally different from 529 00:24:37,720 --> 00:24:41,240 fork. It dictates that instances can be distinguished 530 00:24:41,240 --> 00:24:43,519 by physical matter, the two distinct chains on 531 00:24:43,519 --> 00:24:46,980 their necks, and it can be distinguished by perceptual 532 00:24:46,980 --> 00:24:49,799 design, the singular pattern created by the jeweler. 533 00:24:49,960 --> 00:24:53,319 The software allows for dual tracking. Yes. And 534 00:24:53,319 --> 00:24:55,400 then the software gets even wilder when we look 535 00:24:55,400 --> 00:24:57,980 at the concept of a book. Book also gives you 536 00:24:57,980 --> 00:25:00,119 two ways to count, but it completely abandons 537 00:25:00,119 --> 00:25:02,819 visual perception, right? Imagine you and I are 538 00:25:02,819 --> 00:25:05,259 both reading War and Peace. I say, we are reading 539 00:25:05,259 --> 00:25:08,759 the same book. Your copy might be a massive red 540 00:25:08,759 --> 00:25:11,680 leather -bound hardcover. My copy might be an 541 00:25:11,680 --> 00:25:14,180 audiobook on my phone or a digital EPUB file 542 00:25:14,180 --> 00:25:16,660 in a completely different font on a screen. They 543 00:25:16,660 --> 00:25:19,859 share absolutely zero physical similarities. 544 00:25:20,619 --> 00:25:23,430 They share zero perceptual similarities. One 545 00:25:23,430 --> 00:25:26,410 is paper, one is digital code, but my brain firmly 546 00:25:26,410 --> 00:25:28,930 categorizes them as the exact same thing. Because 547 00:25:28,930 --> 00:25:31,630 the conceptual form of book allows instances 548 00:25:31,630 --> 00:25:34,190 to be individuated by an intelligible character, 549 00:25:34,890 --> 00:25:37,650 the abstract intellectual content, the story 550 00:25:37,650 --> 00:25:41,190 itself. The dual ledger software of the book 551 00:25:41,190 --> 00:25:43,930 concept tracks the physical paper object on one 552 00:25:43,930 --> 00:25:46,809 side and the abstract intellectual manuscript 553 00:25:46,809 --> 00:25:49,869 on the other, seamlessly. Which raises a massive 554 00:25:49,869 --> 00:25:52,269 question for me. If I am standing in a room, 555 00:25:52,509 --> 00:25:55,890 and there is a book on the table and a pile of 556 00:25:55,890 --> 00:25:58,430 sand on the floor. How does my brain instantly 557 00:25:58,430 --> 00:26:01,769 know the complex dual -ledger rules for counting 558 00:26:01,769 --> 00:26:04,210 the book versus the rules for counting the sand? 559 00:26:04,309 --> 00:26:06,690 I haven't touched either of them yet. It knows 560 00:26:06,690 --> 00:26:08,910 because the conceptual form dictates the unit 561 00:26:08,910 --> 00:26:11,470 of quantification before you even process the 562 00:26:11,470 --> 00:26:14,130 visual details of the object. The concept downloads 563 00:26:14,130 --> 00:26:16,329 the rules for engagement into your active memory. 564 00:26:16,670 --> 00:26:20,859 In fact, look at the concept of wood or sand, 565 00:26:21,460 --> 00:26:23,779 their conceptual form is structured so differently 566 00:26:23,779 --> 00:26:26,420 that it does not determine what counts as one 567 00:26:26,420 --> 00:26:29,220 instance on its own. Which is why it feels like 568 00:26:29,220 --> 00:26:32,640 my mouth is broken if I try to say, I have one 569 00:26:32,640 --> 00:26:37,339 wood or I poured one sand. It's not just a grammar 570 00:26:37,339 --> 00:26:40,940 mistake. It feels conceptually impossible. It 571 00:26:40,940 --> 00:26:43,890 is conceptually impossible. To individuate wood 572 00:26:43,890 --> 00:26:46,329 or sand, you need the architectural support of 573 00:26:46,329 --> 00:26:49,529 an entirely different concept. You need a piece 574 00:26:49,529 --> 00:26:53,190 of wood or a pile of sand. The conceptual form 575 00:26:53,190 --> 00:26:56,529 of wood only generates subkinds. It can generate 576 00:26:56,529 --> 00:26:59,789 oak, maple, pine, but it doesn't intrinsically 577 00:26:59,789 --> 00:27:02,470 generate discrete instances you can point to. 578 00:27:02,690 --> 00:27:05,490 It's the same with a concept animal. I can't 579 00:27:05,490 --> 00:27:07,069 point to something in the forest and say, look, 580 00:27:07,170 --> 00:27:09,670 a single animal without my brain immediately 581 00:27:09,670 --> 00:27:12,150 demanding to know what kind of animal. You can't 582 00:27:12,150 --> 00:27:15,109 just be an animal. You have to be a specific 583 00:27:15,109 --> 00:27:18,569 subkind, a deer, a bear, a bird. The concept 584 00:27:18,569 --> 00:27:21,650 animal only contains subkinds. It doesn't contain 585 00:27:21,650 --> 00:27:25,069 direct instances. This hidden machinery is working 586 00:27:25,069 --> 00:27:27,529 at lightning speed. Every time you open your 587 00:27:27,529 --> 00:27:29,890 mouth, every time your eyes sweep across a room, 588 00:27:30,529 --> 00:27:33,450 these forms are dictating what can be counted, 589 00:27:33,630 --> 00:27:35,849 how it can be counted, and what is physically 590 00:27:35,849 --> 00:27:38,569 possible, which brings us to perhaps the most 591 00:27:38,569 --> 00:27:42,019 complex operation they perform. how we make sweeping 592 00:27:42,019 --> 00:27:45,319 statements about the world. Yes. Section four, 593 00:27:45,900 --> 00:27:49,119 solving the mystery of generics. Because if these 594 00:27:49,119 --> 00:27:51,500 invisible lenses dictate how we count single 595 00:27:51,500 --> 00:27:55,690 objects, it makes me wonder, how do they handle 596 00:27:55,690 --> 00:27:59,109 it when we talk about all objects? Like, when 597 00:27:59,109 --> 00:28:01,589 I make a broad generalization, how does my brain 598 00:28:01,589 --> 00:28:04,890 know if I'm telling the truth? Generics are fascinating. 599 00:28:05,329 --> 00:28:07,190 Linguists have apparently been tearing their 600 00:28:07,190 --> 00:28:10,230 hair out over generics for decades. Because generics 601 00:28:10,230 --> 00:28:12,529 are statements that make broad generalizations 602 00:28:12,529 --> 00:28:16,089 without using quantifiers like all or some, right? 603 00:28:16,349 --> 00:28:19,069 Exactly. And they are a nightmare for traditional 604 00:28:19,069 --> 00:28:21,710 theories because they routinely defy real -world 605 00:28:21,710 --> 00:28:24,450 statistics. We accept generalizations that are 606 00:28:24,450 --> 00:28:27,049 statistically false, and we reject generalizations 607 00:28:27,049 --> 00:28:30,089 that are statistically true. Okay. Let's unpack 608 00:28:30,089 --> 00:28:32,269 this with a classic tire example from the text. 609 00:28:32,750 --> 00:28:35,809 I can say to you, tires are round. That feels 610 00:28:35,809 --> 00:28:38,730 true. Yes. I can also say to you, tires are black. 611 00:28:38,960 --> 00:28:42,660 That also feels entirely true. If we look at 612 00:28:42,660 --> 00:28:45,720 the statistical reality of the world, the vast 613 00:28:45,720 --> 00:28:47,519 majority of tires on the road are both round 614 00:28:47,519 --> 00:28:50,940 and black. They seem equal. But watch what happens 615 00:28:50,940 --> 00:28:53,680 when we tweak the linguistic framing. If we change 616 00:28:53,680 --> 00:28:56,779 it to the indefinite singular, the equality completely 617 00:28:56,779 --> 00:29:00,460 shatters. If you say, a tire is round, that sounds 618 00:29:00,460 --> 00:29:02,900 like a perfectly normal, sensible sentence. It 619 00:29:02,900 --> 00:29:05,799 feels like a law of nature. Yeah, totally. But 620 00:29:05,799 --> 00:29:07,940 if you stand on a street corner or point to a 621 00:29:07,940 --> 00:29:10,599 car driving by and declare, a tire is black, 622 00:29:11,559 --> 00:29:13,119 everyone around you will think you are speaking 623 00:29:13,119 --> 00:29:16,200 very weirdly. It feels profoundly wrong to say 624 00:29:16,200 --> 00:29:19,619 it like that. Why? How does the theory of conceptual 625 00:29:19,619 --> 00:29:23,220 form solve this? It solves it by revealing that 626 00:29:23,220 --> 00:29:26,400 our minds make a deep structural distinction 627 00:29:26,400 --> 00:29:29,319 between principled connections and statistical 628 00:29:29,319 --> 00:29:31,859 connections. So roundness is a principal connection. 629 00:29:32,240 --> 00:29:34,819 Yes. Roundness is inextricably linked to the 630 00:29:34,819 --> 00:29:37,740 aspect of being a tire. A conceptual form of 631 00:29:37,740 --> 00:29:40,460 a tire dictates its function, and it needs to 632 00:29:40,460 --> 00:29:43,440 be round to function. If it isn't round, it fails 633 00:29:43,440 --> 00:29:45,500 at being a tire. Right. A square tire is useless. 634 00:29:46,119 --> 00:29:48,880 But blackness, however, is merely a statistical 635 00:29:48,880 --> 00:29:52,410 connection. It is an environmental manufacturing 636 00:29:52,410 --> 00:29:55,210 correlation. Tires happen to be made of carbon 637 00:29:55,210 --> 00:29:57,990 black because it makes the rubber durable. But 638 00:29:57,990 --> 00:30:00,210 being black isn't an intrinsic philosophical 639 00:30:00,210 --> 00:30:03,109 aspect of what it means to be a tire. So the 640 00:30:03,109 --> 00:30:07,289 conceptual form tags are the the property of 641 00:30:07,289 --> 00:30:10,470 roundness as quote principle unquote and the 642 00:30:10,470 --> 00:30:13,309 property of blackness as statistical. And our 643 00:30:13,309 --> 00:30:15,670 spoken grammar treats those internal tags entirely 644 00:30:15,670 --> 00:30:18,309 differently. But the rabbit hole goes much deeper. 645 00:30:18,460 --> 00:30:20,799 This brings us to what cognitive scientists call 646 00:30:20,799 --> 00:30:24,059 minority characteristic generics. This is where 647 00:30:24,059 --> 00:30:27,710 it gets really wild. So let's talk about lions. 648 00:30:28,670 --> 00:30:31,250 So if I walk into a room of 100 people and state, 649 00:30:31,910 --> 00:30:34,869 lions have manes, every single person in that 650 00:30:34,869 --> 00:30:37,390 room will nod and agree that I have spoken a 651 00:30:37,390 --> 00:30:40,369 true sort of generic statement about the world. 652 00:30:40,710 --> 00:30:44,250 It is universally accepted, but it is a massive 653 00:30:44,480 --> 00:30:47,460 glaring paradox for any theory of mine based 654 00:30:47,460 --> 00:30:51,299 on statistics or prototypes. Why? Because if 655 00:30:51,299 --> 00:30:53,660 you go to the Serengeti and actually count them, 656 00:30:54,059 --> 00:30:57,680 over 50 % of lions do not have manes. All the 657 00:30:57,680 --> 00:31:00,460 adult females lack manes. All the cubs lack manes. 658 00:31:00,779 --> 00:31:03,339 The overwhelming majority of the species is maneless. 659 00:31:04,119 --> 00:31:06,599 So how can a statement claiming they do have 660 00:31:06,599 --> 00:31:08,619 manes be considered fundamentally true by the 661 00:31:08,619 --> 00:31:10,700 human brain? And to make it even more absurd? 662 00:31:10,920 --> 00:31:14,160 Look at the contrast with human beings. If we 663 00:31:14,160 --> 00:31:17,200 look at global statistics, roughly 90 % of the 664 00:31:17,200 --> 00:31:20,339 human population is right -handed. It is a massive 665 00:31:20,339 --> 00:31:22,759 statistical dominance. But if I stand on a stage 666 00:31:22,759 --> 00:31:25,759 and say, human beings are right -handed, the 667 00:31:25,759 --> 00:31:27,059 audience will immediately push back. They'll 668 00:31:27,059 --> 00:31:28,880 say, no, they aren't. What about left -handed 669 00:31:28,880 --> 00:31:31,180 people? That's a terrible generalization. This 670 00:31:31,180 --> 00:31:34,059 is the ultimate proof that we do not think in 671 00:31:34,059 --> 00:31:37,759 statistics. 90 % prevalence gets fiercely rejected 672 00:31:37,759 --> 00:31:41,380 as a generic truth, but less than 50 % prevalence 673 00:31:41,380 --> 00:31:44,359 gets universally accepted. The statistics are 674 00:31:44,359 --> 00:31:47,759 completely inverted. So how does the conceptual 675 00:31:47,759 --> 00:31:50,279 form rescue us from this paradox? The brilliance 676 00:31:50,279 --> 00:31:53,019 of the conceptual form lies in its structural 677 00:31:53,019 --> 00:31:56,819 map of a species. We accept lions have manes 678 00:31:56,819 --> 00:31:59,339 because the conceptual form we hold for animate 679 00:31:59,339 --> 00:32:02,099 kinds, like animals, has a specific structure 680 00:32:02,099 --> 00:32:05,000 that creates a principled connection to a specific 681 00:32:05,150 --> 00:32:07,549 necessary subtype, in this case adult males. 682 00:32:08,009 --> 00:32:10,970 Okay, I'm following. The concept represents the 683 00:32:10,970 --> 00:32:13,769 property, mains, as being principled and necessary 684 00:32:13,769 --> 00:32:16,509 for the male subtype while actively representing 685 00:32:16,509 --> 00:32:18,829 its absence in the females and cubs. So it's 686 00:32:18,829 --> 00:32:21,109 not taking a statistical average of the herd, 687 00:32:21,690 --> 00:32:24,809 it's looking at the biological blueprint. Exactly, 688 00:32:24,990 --> 00:32:27,519 it's a formal structural map. But when we look 689 00:32:27,519 --> 00:32:31,240 at handedness in humans, being right handed isn't 690 00:32:31,240 --> 00:32:33,359 a principled aspect of being human. There is 691 00:32:33,359 --> 00:32:36,359 no biological necessary subtype of humanity that 692 00:32:36,359 --> 00:32:39,019 is defined by being right handed in order to 693 00:32:39,019 --> 00:32:41,680 function as a human. It's just a trait with a 694 00:32:41,680 --> 00:32:44,059 high statistical prevalence. Let me make sure 695 00:32:44,059 --> 00:32:46,819 I'm locking this in. The text contrasts two more 696 00:32:46,819 --> 00:32:50,859 sentences. Dogs have four legs versus mosquitoes 697 00:32:50,859 --> 00:32:55,000 carry West Nile virus. A perfect contrast. Dogs 698 00:32:55,000 --> 00:32:57,240 have four legs is a principled connection, just 699 00:32:57,240 --> 00:33:00,440 like the tire being round, but carrying West 700 00:33:00,440 --> 00:33:02,839 Nile virus isn't a principled aspect of being 701 00:33:02,839 --> 00:33:05,339 a mosquito. Right, they don't need the virus 702 00:33:05,339 --> 00:33:08,700 to function as a mosquito, it's causal. They 703 00:33:08,700 --> 00:33:10,259 carry it because of their environment, their 704 00:33:10,259 --> 00:33:12,160 diet, their interaction with the material world. 705 00:33:12,579 --> 00:33:15,160 It's a highly dangerous statistical correlation, 706 00:33:15,220 --> 00:33:17,539 but it's not intrinsic to the concept of the 707 00:33:17,539 --> 00:33:20,619 mosquito itself. Spot on. Okay, but I have to 708 00:33:20,619 --> 00:33:23,220 push back on the lion example. I am trying to 709 00:33:23,220 --> 00:33:26,809 map this traffic cop software in my head. If 710 00:33:26,809 --> 00:33:29,509 the statement, lions have manes, is perfectly 711 00:33:29,509 --> 00:33:31,390 acceptable because the software routes it through 712 00:33:31,390 --> 00:33:34,730 the principled male subtype, why can't I just 713 00:33:34,730 --> 00:33:37,990 say lions are male? Ah. Right? If I say lions 714 00:33:37,990 --> 00:33:39,589 are male, everyone looks at me like I've lost 715 00:33:39,589 --> 00:33:42,349 my mind. That sounds like total gibberish. If 716 00:33:42,349 --> 00:33:44,970 the conceptual form allows the mane to pass the 717 00:33:44,970 --> 00:33:47,700 checkpoint, why does it block the gender? That 718 00:33:47,700 --> 00:33:50,079 is a brilliant pushback, and Prasada draws a 719 00:33:50,079 --> 00:33:52,859 very sharp, specific distinction for exactly 720 00:33:52,859 --> 00:33:56,220 this reason. The conceptual form creates a principled 721 00:33:56,220 --> 00:33:59,039 connection to the property, the main. It allows 722 00:33:59,039 --> 00:34:01,200 you to generalize the property because it is 723 00:34:01,200 --> 00:34:04,140 a principled feature of a necessary part of the 724 00:34:04,140 --> 00:34:07,940 species. But the conceptual form also contains 725 00:34:07,940 --> 00:34:10,619 a formal, structural distinction for the instances 726 00:34:10,619 --> 00:34:14,610 themselves, males versus females. It categorizes 727 00:34:14,610 --> 00:34:17,139 the differences between instances their gender, 728 00:34:17,539 --> 00:34:20,699 as accidental differences rather than principal 729 00:34:20,699 --> 00:34:24,000 differences of the kind itself. So you can generalize 730 00:34:24,000 --> 00:34:27,199 a property that belongs to a principled subtype. 731 00:34:27,800 --> 00:34:30,519 Yes, but you absolutely cannot generalize the 732 00:34:30,519 --> 00:34:33,480 subtype itself, being male, to the whole kind 733 00:34:33,480 --> 00:34:36,519 because the conceptual form explicitly holds 734 00:34:36,519 --> 00:34:39,000 structural space for the existence of the other 735 00:34:39,000 --> 00:34:42,760 subtype, females, for the species to exist. The 736 00:34:42,760 --> 00:34:45,119 structure blocks the generalization of the subtype 737 00:34:45,119 --> 00:34:48,000 to protect the definition of the kind. It literally 738 00:34:48,000 --> 00:34:50,840 is a built -in traffic cop. It's directing what 739 00:34:50,840 --> 00:34:52,699 generalizations are allowed to pass through into 740 00:34:52,699 --> 00:34:55,280 our speech and pulling over the ones that violate 741 00:34:55,280 --> 00:34:58,420 the structural map. Precisely. And once you understand 742 00:34:58,420 --> 00:35:00,840 how this traffic cop governs our generalizations, 743 00:35:01,400 --> 00:35:03,500 you can suddenly see how it allows us to bend 744 00:35:03,500 --> 00:35:05,460 language in incredible ways without breaking 745 00:35:05,460 --> 00:35:08,139 our understanding. Which brings us to Section 746 00:35:08,139 --> 00:35:12,960 5, Linguistic Quirks. systematic polysemy, and 747 00:35:12,960 --> 00:35:16,039 metaphor. Let's start with polysemy. This is 748 00:35:16,039 --> 00:35:18,900 the phenomenon where a single word has multiple 749 00:35:18,900 --> 00:35:21,400 systematically related meanings. And the theory 750 00:35:21,400 --> 00:35:23,960 of conceptual form gives us a strict rule for 751 00:35:23,960 --> 00:35:27,400 this called the individuation for polysemy principle. 752 00:35:28,099 --> 00:35:30,539 Let's use the newspaper example from the text. 753 00:35:31,119 --> 00:35:33,380 I am sitting at breakfast and I can say to my 754 00:35:33,380 --> 00:35:35,780 wife, I spilled coffee all over the newspaper. 755 00:35:36,440 --> 00:35:38,940 I'm clearly referring to the physical paper in 756 00:35:38,940 --> 00:35:41,710 ink on the table. Right. But five minutes later, 757 00:35:41,969 --> 00:35:44,530 I can read an article and say, wow, the newspaper 758 00:35:44,530 --> 00:35:48,110 fired its editor today. Now I am clearly referring 759 00:35:48,110 --> 00:35:51,170 to the corporate publishing organization. I use 760 00:35:51,170 --> 00:35:53,329 the exact same word for a physical object and 761 00:35:53,329 --> 00:35:55,449 a group of humans in a boardroom and nobody gets 762 00:35:55,449 --> 00:35:58,590 confused. The word handles both perfectly. But 763 00:35:58,590 --> 00:36:00,750 watch what happens when we try to stretch other 764 00:36:00,750 --> 00:36:03,170 words in the exact same way. Look at the word 765 00:36:03,170 --> 00:36:06,469 book. Right. I can easily say, I spilled coffee 766 00:36:06,469 --> 00:36:09,210 on the book. but I absolutely cannot say the 767 00:36:09,210 --> 00:36:12,400 book fired its editor. I can't use the word book 768 00:36:12,400 --> 00:36:15,199 to refer to the author who wrote it or the publisher 769 00:36:15,199 --> 00:36:18,800 who printed it. Why does the concept of newspaper 770 00:36:18,800 --> 00:36:21,679 stretch like a rubber band, but the concept of 771 00:36:21,679 --> 00:36:24,480 book snaps in half? Because it all comes down 772 00:36:24,480 --> 00:36:27,099 to the internal instruction set. The conceptual 773 00:36:27,099 --> 00:36:30,559 form of newspaper includes an intrinsic structural 774 00:36:30,559 --> 00:36:33,380 connection to its publication origin as a core 775 00:36:33,380 --> 00:36:36,019 way of individuating it. It is part of what makes 776 00:36:36,019 --> 00:36:40,000 a newspaper a newspaper. It is an ongoing organizational 777 00:36:40,000 --> 00:36:43,179 output. The organization individuates the physical 778 00:36:43,179 --> 00:36:46,920 copies. But book is structurally different. Exactly. 779 00:36:47,119 --> 00:36:49,619 A book is individuated by its intellectual content, 780 00:36:49,679 --> 00:36:51,940 not by the corporation that produced it. You 781 00:36:51,940 --> 00:36:54,760 can use the word magazine or journal to refer 782 00:36:54,760 --> 00:36:57,139 to the organization, just like newspaper. But 783 00:36:57,139 --> 00:37:00,579 words like pamphlet or book don't work that way. 784 00:37:00,940 --> 00:37:04,219 The conceptual form explicitly grants or denies 785 00:37:04,219 --> 00:37:06,639 these meaning extensions based on its internal 786 00:37:06,639 --> 00:37:09,539 rules of individuation. Which perfectly sets 787 00:37:09,539 --> 00:37:11,719 the stage for the ultimate stretching of language. 788 00:37:12,480 --> 00:37:15,579 Metaphor. How does our brain process metaphors 789 00:37:15,579 --> 00:37:18,000 without just throwing an error code? Why does 790 00:37:18,000 --> 00:37:21,079 the sentence, my lawyer is a shark, make perfect, 791 00:37:21,119 --> 00:37:23,659 vivid sense, but the sentence, dogs are cats, 792 00:37:23,920 --> 00:37:26,559 is processed as pure gibberish? Because traditional 793 00:37:26,559 --> 00:37:29,670 theories that rely on physical similarity fail 794 00:37:29,670 --> 00:37:32,730 completely here. A lawyer is not physically similar 795 00:37:32,730 --> 00:37:34,630 to a shark. They don't have gills. They don't 796 00:37:34,630 --> 00:37:36,829 have dorsal fins. They don't live in salt water. 797 00:37:37,449 --> 00:37:39,389 Unless you are dealing with a very specific, 798 00:37:39,789 --> 00:37:43,079 terrifying kind of maritime law firm. True, true. 799 00:37:43,739 --> 00:37:46,900 But cognitively, metaphor isn't about matching 800 00:37:46,900 --> 00:37:50,219 physical features at all. No, it is about perspective. 801 00:37:51,159 --> 00:37:54,500 Metaphor involves taking the perspective, the 802 00:37:54,500 --> 00:37:57,739 invisible conceptual form instructions of the 803 00:37:57,739 --> 00:38:00,500 vehicle, which is the shark. overlaying it onto 804 00:38:00,500 --> 00:38:03,099 the topic, which is the lawyer. So it's like 805 00:38:03,099 --> 00:38:06,179 putting a virtual reality headset on the concept 806 00:38:06,179 --> 00:38:09,300 of lawyer. I am not changing the data of the 807 00:38:09,300 --> 00:38:11,440 lawyer, he still wears a suit, he still goes 808 00:38:11,440 --> 00:38:15,039 to court, but I am booting up the shark operating 809 00:38:15,039 --> 00:38:17,599 system to navigate that data. That is the perfect 810 00:38:17,599 --> 00:38:20,650 analogy. You are imposing the structural map 811 00:38:20,650 --> 00:38:23,590 of the shark onto the environment of the lawyer. 812 00:38:24,289 --> 00:38:26,130 The text points out that we actually see the 813 00:38:26,130 --> 00:38:28,789 roots of this profound cognitive ability in children's 814 00:38:28,789 --> 00:38:32,710 pretend play. Yes, the stapler example. Imagine 815 00:38:32,710 --> 00:38:34,929 a child is playing at a desk and they pick up 816 00:38:34,929 --> 00:38:37,110 a stapler and they start pretending it is a shark. 817 00:38:37,630 --> 00:38:40,630 The child isn't hallucinating. They aren't confused 818 00:38:40,630 --> 00:38:42,809 about reality. They know perfectly well it's 819 00:38:42,809 --> 00:38:46,070 a stapler. But why do they snap the stapler open 820 00:38:46,070 --> 00:38:49,329 and shut like a jaw? Why do they make it swim 821 00:38:49,329 --> 00:38:53,409 through the air? Because they are using the conceptual 822 00:38:53,409 --> 00:38:57,010 form of shark to dictate how to think about and 823 00:38:57,010 --> 00:39:00,909 move the stapler. Exactly. The conceptual form 824 00:39:00,909 --> 00:39:03,869 of the shark has a part -whole structure that 825 00:39:03,869 --> 00:39:06,610 heavily emphasizes a biting mouth for survival. 826 00:39:07,159 --> 00:39:10,260 The child maps that structure onto the physical 827 00:39:10,260 --> 00:39:14,019 hinge of the stapler. Metaphor is just adult 828 00:39:14,019 --> 00:39:17,679 cognitive pretense. When I say my lawyer is a 829 00:39:17,679 --> 00:39:20,480 shark, I am overlaying the formal instructions 830 00:39:20,480 --> 00:39:23,760 of the shark, relentless forward motion, aggressive 831 00:39:23,760 --> 00:39:26,860 predatory behavior, cold calculation, onto the 832 00:39:26,860 --> 00:39:29,579 lawyer. And suddenly I know exactly how to expect 833 00:39:29,579 --> 00:39:33,019 him to behave during a deposition. But notice 834 00:39:33,019 --> 00:39:37,429 why dogs are cats. fails entirely as a metaphor. 835 00:39:37,869 --> 00:39:40,190 It fails because they both possess the exact 836 00:39:40,190 --> 00:39:43,869 same underlying conceptual form. They are both 837 00:39:43,869 --> 00:39:47,570 animate object kinds. Well, you can't boot up 838 00:39:47,570 --> 00:39:50,550 a Mac OS inside another Mac OS to get a new perspective. 839 00:39:51,130 --> 00:39:53,510 They are functionally on the exact same structural 840 00:39:53,510 --> 00:39:55,889 level, so they just crash into each other as 841 00:39:55,889 --> 00:39:59,309 an identity error. Exactly. It solves the paradox 842 00:39:59,309 --> 00:40:02,320 of why we compare unlike things. You can only 843 00:40:02,320 --> 00:40:05,019 map conceptual forms onto each other precisely 844 00:40:05,019 --> 00:40:07,940 because they are structurally different. The 845 00:40:07,940 --> 00:40:10,059 friction between the forms is what generates 846 00:40:10,059 --> 00:40:13,420 the insight. This entire system is breathtakingly 847 00:40:13,420 --> 00:40:16,239 complex. It is full of invisible rules about 848 00:40:16,239 --> 00:40:18,980 counting dual ledgers, policing generics through 849 00:40:18,980 --> 00:40:21,519 principled connections, normative expectations, 850 00:40:21,880 --> 00:40:25,360 polysemy, and VR headset metaphors. It's a massive 851 00:40:25,360 --> 00:40:28,119 unseen architecture. Which leads to a massive 852 00:40:28,119 --> 00:40:31,699 unavoidable question. How on earth to human babies 853 00:40:31,699 --> 00:40:35,920 learn this. Which brings us to section 6, the 854 00:40:35,920 --> 00:40:38,940 origin story, chapters 9 and 10. This is known 855 00:40:38,940 --> 00:40:42,059 in the field as the acquisition puzzle. And it's 856 00:40:42,059 --> 00:40:44,239 a puzzle because, as we discussed with the three 857 00:40:44,239 --> 00:40:47,460 -legged dog, experience only ever shows a child 858 00:40:47,460 --> 00:40:51,239 what is. A baby observes a physical dog with 859 00:40:51,239 --> 00:40:54,380 four legs in their living room. But visual experience 860 00:40:54,380 --> 00:40:57,480 can never ever show a baby what ought to be. 861 00:40:57,630 --> 00:41:00,369 You cannot observe a principled connection with 862 00:41:00,369 --> 00:41:03,710 your eyes. You cannot observe a normative expectation 863 00:41:03,710 --> 00:41:06,750 in the wild. You can't see the abstract kind 864 00:41:06,750 --> 00:41:09,670 dog walking down the street. You can only see 865 00:41:09,670 --> 00:41:12,440 Fido. So if you can't learn these structural 866 00:41:12,440 --> 00:41:14,639 rules from observing the physical environment, 867 00:41:14,840 --> 00:41:17,179 they must be hardwired into the brain, right? 868 00:41:17,679 --> 00:41:19,980 But they can't all be hardwired because babies 869 00:41:19,980 --> 00:41:22,099 obviously aren't born knowing what a newspaper 870 00:41:22,099 --> 00:41:24,599 is or what a carburetor is. And this touches 871 00:41:24,599 --> 00:41:26,980 on a massive war in cognitive science, doesn't 872 00:41:26,980 --> 00:41:30,030 it? The battle between empiricism and nativism? 873 00:41:30,590 --> 00:41:33,809 A centuries old war. The empiricists, following 874 00:41:33,809 --> 00:41:36,150 the tradition of John Locke, argue that the mind 875 00:41:36,150 --> 00:41:39,030 is a blank slate and absolutely everything is 876 00:41:39,030 --> 00:41:40,829 learned from experiencing the outside world. 877 00:41:41,010 --> 00:41:43,869 The nativists, following traditions from Descartes 878 00:41:43,869 --> 00:41:46,789 to Noam Chomsky, argue that because the environment 879 00:41:46,789 --> 00:41:49,409 is too chaotic, our fundamental concepts and 880 00:41:49,409 --> 00:41:52,010 grammar must be innate, pre -installed in our 881 00:41:52,010 --> 00:41:54,610 biology. And the text brings up philosopher Jerry 882 00:41:54,610 --> 00:41:57,949 Fodar's concept of the carburetor. To highlight 883 00:41:57,949 --> 00:42:01,349 how messy this gets, Fodar argued for atomic 884 00:42:01,349 --> 00:42:04,190 concepts, meaning the concept of a carburetor 885 00:42:04,190 --> 00:42:06,869 isn't built out of smaller concepts like metal 886 00:42:06,869 --> 00:42:10,570 plus engine plus air. It is an atomic, indivisible 887 00:42:10,570 --> 00:42:13,969 concept. Right. But if it's an indivisible primitive 888 00:42:13,969 --> 00:42:17,269 concept, how do we acquire it? We clearly aren't 889 00:42:17,269 --> 00:42:20,150 born with it. Cavemen didn't have a dormant carburetor 890 00:42:20,150 --> 00:42:21,989 concept waiting for Ford to invent the Model 891 00:42:21,989 --> 00:42:25,789 T. It was a massive headache for the field. But 892 00:42:25,789 --> 00:42:28,889 Prasada's theory offers a brilliant, elegant 893 00:42:28,889 --> 00:42:32,130 synthesis of the two extremes. He proposes conceptual 894 00:42:32,130 --> 00:42:35,190 bootstrapping via core knowledge systems. Okay, 895 00:42:35,190 --> 00:42:37,570 we need to take this step by step. What is a 896 00:42:37,570 --> 00:42:40,289 core knowledge system? Developmental psychologists 897 00:42:40,289 --> 00:42:42,710 have discovered that human infants are born with 898 00:42:42,710 --> 00:42:45,909 non -conceptual, hardwired core systems for tracking 899 00:42:45,909 --> 00:42:48,969 basic realities. The two big ones are objects 900 00:42:48,969 --> 00:42:52,650 and agents. A newborn baby intrinsically knows 901 00:42:52,650 --> 00:42:55,489 that solid physical objects cannot pass through 902 00:42:55,489 --> 00:42:58,150 each other. And they intrinsically know that 903 00:42:58,150 --> 00:43:01,170 agents, things that move on their own, act with 904 00:43:01,170 --> 00:43:03,929 purpose and goals. But these aren't full -blown 905 00:43:03,929 --> 00:43:06,909 concepts yet. A baby tracking a moving agent 906 00:43:06,909 --> 00:43:09,769 doesn't yet have the intricate conceptual form 907 00:43:09,769 --> 00:43:13,909 of an animal. Correct. So how do we cross the 908 00:43:13,909 --> 00:43:16,940 bridge? Prasada argues that while the specific 909 00:43:16,940 --> 00:43:19,340 content of concepts is acquired from the world, 910 00:43:19,800 --> 00:43:22,679 the formal structures are innate. We are born 911 00:43:22,679 --> 00:43:25,639 with parametric forms of conceptual understanding. 912 00:43:26,199 --> 00:43:29,719 Parametric forms? That's heavy jargon. Let's 913 00:43:29,719 --> 00:43:32,980 translate that. I'm picturing a character creation 914 00:43:32,980 --> 00:43:35,900 menu in a massive role -playing video game. That 915 00:43:35,900 --> 00:43:38,599 works perfectly. When you start the game, the 916 00:43:38,599 --> 00:43:40,679 fundamental code of the game provides an innate, 917 00:43:40,960 --> 00:43:44,750 unchangeable form. You must play as a bipedal 918 00:43:44,750 --> 00:43:47,989 humanoid. You cannot change that structure. But 919 00:43:47,989 --> 00:43:51,030 the menu gives you parameters that you get to 920 00:43:51,030 --> 00:43:54,349 fill in. You choose the eye color, the hair length, 921 00:43:54,550 --> 00:43:58,570 the voice. Okay, I see. The form is hardwired 922 00:43:58,570 --> 00:44:00,929 by the developer. The parameters are filled in 923 00:44:00,929 --> 00:44:03,690 by your choices. So in cognitive terms, the human 924 00:44:03,690 --> 00:44:06,269 brain comes with innate, unchangeable templates. 925 00:44:06,539 --> 00:44:09,480 We have a blank template for an animate kind, 926 00:44:09,980 --> 00:44:12,780 a blank template for an artifact kind, a blank 927 00:44:12,780 --> 00:44:15,900 template for a substance kind. We are basically 928 00:44:15,900 --> 00:44:18,559 born with the molds, but we aren't born with 929 00:44:18,559 --> 00:44:21,639 the clay. And the physical world provides the 930 00:44:21,639 --> 00:44:24,400 clay, but the final shape it takes is entirely 931 00:44:24,400 --> 00:44:27,340 dictated by the human mold. That is precisely 932 00:44:27,340 --> 00:44:30,659 it. Let's watch it happen. A toddler encounters 933 00:44:30,659 --> 00:44:33,309 a dog for the first time. The toddler's innate 934 00:44:33,309 --> 00:44:36,050 core knowledge system tracks the dog as a self 935 00:44:36,050 --> 00:44:39,130 -moving agent. This triggers the brain to boot 936 00:44:39,130 --> 00:44:42,469 up the innate animate kind conceptual form template. 937 00:44:43,329 --> 00:44:45,889 The mold opens up. The mold opens. And this mold 938 00:44:45,889 --> 00:44:48,789 comes preloaded with profound structural instructions. 939 00:44:49,230 --> 00:44:51,929 It tells the toddler's brain, expect this new 940 00:44:51,929 --> 00:44:53,789 thing to have a principled connection to its 941 00:44:53,789 --> 00:44:56,610 physical parts. Expect it to have a normative 942 00:44:56,610 --> 00:44:59,750 standard of health. Expect it to reproduce. Wow. 943 00:45:00,059 --> 00:45:02,860 The toddler's brain is now prepped. It then uses 944 00:45:02,860 --> 00:45:06,260 raw visual experience, the clay, to fill in the 945 00:45:06,260 --> 00:45:09,099 specific data parameters. Oh, the parts are fur 946 00:45:09,099 --> 00:45:12,920 and four legs. It barks. It eats meat. The mold 947 00:45:12,920 --> 00:45:15,739 dictates how to process the clay. It generates 948 00:45:15,739 --> 00:45:19,400 the concept dog, out of thin air, using the innate 949 00:45:19,400 --> 00:45:22,039 structural rules and the acquired environmental 950 00:45:22,039 --> 00:45:24,989 data. What's truly fascinating, and Prasada touches 951 00:45:24,989 --> 00:45:27,369 on this, is how beautifully this cutting -edge 952 00:45:27,369 --> 00:45:29,750 cognitive science maps onto ancient philosophy. 953 00:45:30,389 --> 00:45:33,829 It maps almost perfectly onto Aristotelian metaphysics. 954 00:45:34,090 --> 00:45:37,010 Thousands of years ago. Yes. Aristotle argued 955 00:45:37,010 --> 00:45:39,090 that human beings don't just passively look at 956 00:45:39,090 --> 00:45:41,369 the world. He said the mind comes preloaded with 957 00:45:41,369 --> 00:45:44,449 fundamental categories of being. Substance, quantity, 958 00:45:44,769 --> 00:45:47,710 quality, relation. We don't learn these categories 959 00:45:47,710 --> 00:45:50,710 from the dirt. We use these innate categories 960 00:45:50,710 --> 00:45:53,440 to organize the dirt. We use them to render the 961 00:45:53,440 --> 00:45:56,280 buzzing, blooming confusion of the universe intelligible. 962 00:45:56,659 --> 00:45:58,980 That's incredible. Brassata is essentially giving 963 00:45:58,980 --> 00:46:01,539 us the modern, empirical, cognitive science version 964 00:46:01,539 --> 00:46:04,599 of Aristotle's categories. The invisible conceptual 965 00:46:04,599 --> 00:46:07,360 forms in our brains are the direct modern equivalents 966 00:46:07,360 --> 00:46:09,800 of those ancient Aristotelian structures. That 967 00:46:09,800 --> 00:46:12,760 is genuinely mind -blowing. We are literally 968 00:46:12,760 --> 00:46:15,699 born with the architectural blueprints for reality. 969 00:46:16,280 --> 00:46:18,260 And we just spend our entire lives wandering 970 00:46:18,260 --> 00:46:20,739 around, populating the rooms with the things 971 00:46:20,739 --> 00:46:23,539 we find. And because these architectural blueprints, 972 00:46:23,780 --> 00:46:26,639 these conceptual forms, are innately shared across 973 00:46:26,639 --> 00:46:29,619 the entire human species, it explains everything. 974 00:46:30,300 --> 00:46:32,699 It explains why we can ultimately translate across 975 00:46:32,699 --> 00:46:35,179 wildly different languages. It explains why a 976 00:46:35,179 --> 00:46:37,460 businessman in Tokyo and a farmer in Ohio both 977 00:46:37,460 --> 00:46:40,659 understand the same underlying metaphor. It explains 978 00:46:40,659 --> 00:46:43,579 why we all implicitly agree on the bizarre statistical 979 00:46:43,579 --> 00:46:45,980 rules of generics, even if the surface grammar 980 00:46:45,980 --> 00:46:48,579 of our languages differ. We all have the same 981 00:46:48,579 --> 00:46:51,219 operating system. Okay. let's bring this all 982 00:46:51,219 --> 00:46:53,800 together and land the plane. We started this 983 00:46:53,800 --> 00:46:56,380 deep dive looking at a transparent window and 984 00:46:56,380 --> 00:46:59,139 we've shattered it. We've rejected the simple, 985 00:46:59,300 --> 00:47:02,199 comfortable idea that our concepts are just checklists 986 00:47:02,199 --> 00:47:04,900 or statistical averages. We discovered that the 987 00:47:04,900 --> 00:47:07,099 words we use every single day, table, dog, book, 988 00:47:07,320 --> 00:47:10,559 necklace, are actually intricate, invisible lenses. 989 00:47:11,460 --> 00:47:14,500 They are miniature operating systems that forcefully 990 00:47:14,500 --> 00:47:17,559 dictate what counts as one instance, what properties 991 00:47:17,559 --> 00:47:20,679 are normatively supposed to be there. and how 992 00:47:20,679 --> 00:47:23,059 we are allowed to stretch meanings through metaphors 993 00:47:23,059 --> 00:47:26,440 and polysemy. It's a humbling realization. It 994 00:47:26,440 --> 00:47:28,900 reminds us that every single time you open your 995 00:47:28,900 --> 00:47:31,599 mouth to speak, every time you categorize a simple 996 00:47:31,599 --> 00:47:33,840 object sitting on your desk, you are running 997 00:47:33,840 --> 00:47:37,380 highly advanced, abstract, generative philosophical 998 00:47:37,380 --> 00:47:39,780 programs without even realizing it. The sheer 999 00:47:39,780 --> 00:47:42,659 complexity of everyday, common -sense thought 1000 00:47:42,659 --> 00:47:46,320 is staggering. So what does this all mean for 1001 00:47:46,320 --> 00:47:49,559 you today? Why does this matter beyond a cognitive 1002 00:47:49,559 --> 00:47:53,239 science classroom? Consider this. Our conceptual 1003 00:47:53,239 --> 00:47:55,820 forms evolved over millennia. They were designed 1004 00:47:55,820 --> 00:47:58,179 to handle natural objects, artifacts, and living 1005 00:47:58,179 --> 00:48:01,239 things. But what happens when we try to conceptualize 1006 00:48:01,239 --> 00:48:04,539 completely novel, unprecedented digital entities? 1007 00:48:05,219 --> 00:48:07,559 What happens when we try to conceptualize artificial 1008 00:48:07,559 --> 00:48:10,659 intelligence? That is the ultimate question for 1009 00:48:10,659 --> 00:48:13,510 our era. Because whichever conceptual form we 1010 00:48:13,510 --> 00:48:16,690 subconsciously select to apply to AI will dictate 1011 00:48:16,690 --> 00:48:18,969 all of our normative expectations for it. If 1012 00:48:18,969 --> 00:48:21,269 we force an artificial intelligence into the 1013 00:48:21,269 --> 00:48:24,690 conceptual form template of an agent, if we treat 1014 00:48:24,690 --> 00:48:28,119 it like a mind, our hidden lens will automatically 1015 00:48:28,119 --> 00:48:30,739 expect it to have goals and we will feel it is 1016 00:48:30,739 --> 00:48:33,880 supposed to act a certain way ethically. Right. 1017 00:48:34,280 --> 00:48:36,699 But if we force it into the template of a tool 1018 00:48:36,699 --> 00:48:40,030 or an artifact like a hammer or a table, Our 1019 00:48:40,030 --> 00:48:42,090 lens will strip away those expectations and we 1020 00:48:42,090 --> 00:48:44,989 will assume its entire existence is dictated 1021 00:48:44,989 --> 00:48:47,650 solely by human use and orientation. And the 1022 00:48:47,650 --> 00:48:49,769 terrifying part is most people will make that 1023 00:48:49,769 --> 00:48:51,570 choice without ever knowing they made it. The 1024 00:48:51,570 --> 00:48:54,869 lens will just snap into place. How will that 1025 00:48:54,869 --> 00:48:57,710 hidden lens secretly control what you think AI 1026 00:48:57,710 --> 00:49:01,289 is supposed to do? That is something to deeply 1027 00:49:01,289 --> 00:49:03,530 mull over the next time you sit down and ask 1028 00:49:03,530 --> 00:49:06,269 a chatbot a question. Are you looking through 1029 00:49:06,269 --> 00:49:08,679 the lens of a person? or are you looking through 1030 00:49:08,679 --> 00:49:11,079 the lens of a calculator? Because your brain 1031 00:49:11,079 --> 00:49:13,480 is forcing one of those refractive windows over 1032 00:49:13,480 --> 00:49:16,539 your eyes right now, the window is definitely 1033 00:49:16,539 --> 00:49:19,320 not transparent. The lenses are always bending 1034 00:49:19,320 --> 00:49:21,739 the light. Thank you for taking this deep dive 1035 00:49:21,739 --> 00:49:22,340 with us today.