WEBVTT - Ep67 "How did human brains get runaway intelligence? " 0:00:05.080 --> 0:00:10.440 Humans are really smart? But how did intelligence evolve? If 0:00:10.440 --> 0:00:13.760 we're trying to look back at the history of intelligent brains, 0:00:14.360 --> 0:00:16.520 do we have to look all the way back to 0:00:16.640 --> 0:00:20.880 our common ancestors with the apes, or all mammals or 0:00:21.000 --> 0:00:25.640 all reptiles, or can the origins of intelligence be traced 0:00:25.680 --> 0:00:30.000 back even further? And now that our species is good 0:00:30.040 --> 0:00:33.320 and smart, what does the knowledge of our past mean 0:00:33.479 --> 0:00:41.640 for us as we work to build intelligence artificially? Welcome 0:00:41.640 --> 0:00:45.640 to Inner Cosmos with me David Eagleman. I'm a neuroscientist 0:00:45.680 --> 0:00:48.919 and an author at Stanford and in these episodes, we 0:00:49.000 --> 0:00:53.440 sail deeply into our three pound universe to understand why 0:00:53.560 --> 0:01:05.160 and how our lives look the way they do. Today's 0:01:05.160 --> 0:01:10.200 episode is about intelligence and the history of intelligence. How 0:01:10.240 --> 0:01:13.360 did human intelligence arrive on the scene? 0:01:13.880 --> 0:01:14.080 Now? 0:01:14.120 --> 0:01:16.800 This is an important question because we seem to be 0:01:17.000 --> 0:01:21.120 operating at a different level than our neighbors in the 0:01:21.200 --> 0:01:24.520 animal kingdom. We are the only ones, as far as 0:01:24.520 --> 0:01:29.160 we can tell, who compose symphonies and launch mars rover 0:01:29.280 --> 0:01:34.520 missions and discover DNA and build courthouses and have congresses 0:01:34.560 --> 0:01:39.480 and construct windmills and write novels and build screws and 0:01:39.520 --> 0:01:42.880 screwdrivers to hold things together, and so on and so on, 0:01:43.200 --> 0:01:46.679 none of which any other animal does. And this is 0:01:46.720 --> 0:01:50.000 how we've taken over the whole planet. But how the 0:01:50.040 --> 0:01:56.559 heck did this happen? Why are humans such a runaway species? Well, 0:01:56.640 --> 0:02:00.120 traditionally the explanation has been something like this is a 0:02:00.200 --> 0:02:04.080 special gift from your deity, whichever deity your family believed 0:02:04.120 --> 0:02:09.200 in at whatever moment in history. But centuries of people 0:02:09.240 --> 0:02:12.480 looking at this carefully, sometimes in a microscope, sometimes in 0:02:12.520 --> 0:02:18.000 the brain scanner, sometimes at autopsy, careful examination has made 0:02:18.080 --> 0:02:21.600 something very clear. When you look at the brains of 0:02:21.720 --> 0:02:26.800 other animals, those brains are very similar to our own. Now, 0:02:26.880 --> 0:02:30.240 this shouldn't be surprising. It's the same when you look 0:02:30.280 --> 0:02:34.320 at other animals hearts or lungs or kidneys. It's the 0:02:34.440 --> 0:02:39.600 same good idea, and it's conserved throughout evolution, and so 0:02:39.800 --> 0:02:44.960 it goes with brains, with neurons and cerebellum and thalamus 0:02:44.960 --> 0:02:48.360 and hippocampus and cortex and blah blah blah. It looks 0:02:48.520 --> 0:02:52.320 pretty similar everywhere. And this leads to a point which 0:02:52.360 --> 0:02:56.160 should be fairly obvious when you look across the evolution 0:02:56.480 --> 0:03:00.480 of the vast Kingdom of animals. You don't find that 0:03:00.520 --> 0:03:05.120 there was no intelligence and suddenly humans popped up like 0:03:05.240 --> 0:03:09.440 hairless geniuses. That's not what happened. Instead, what you find 0:03:10.080 --> 0:03:14.000 is there are versions of intelligence all around us. As 0:03:14.040 --> 0:03:17.880 one example, I always admire the squirrels hopping in my 0:03:18.080 --> 0:03:22.040 tree in the garden. They perform these sophisticated acrobatics and 0:03:22.080 --> 0:03:25.280 do the kind of stuff that human gymnasts would never 0:03:25.320 --> 0:03:29.680 even attempt. And crows show intelligence that's closer to our own. 0:03:29.720 --> 0:03:33.959 They can solve really sophisticated puzzles, and dolphins have some 0:03:34.000 --> 0:03:37.440 sort of societies and language, though again not quite as 0:03:37.480 --> 0:03:41.440 sophisticated as ours. And in episode thirty four, I explored 0:03:41.480 --> 0:03:44.760 what it would be like to have different levels of intelligence, 0:03:45.160 --> 0:03:47.480 So please check out that episode if you're interested in that. 0:03:47.960 --> 0:03:52.120 So back to this question. When we ask how intelligence 0:03:52.200 --> 0:03:55.480 got here, it ends up being a question about an 0:03:55.520 --> 0:03:59.720 evolutionary journey, like when we ask how did Homo sapiens 0:03:59.760 --> 0:04:02.440 start walking on our rear legs? Or how did we 0:04:02.520 --> 0:04:06.040 become hairless? Or why do we get pimples and other 0:04:06.120 --> 0:04:09.440 primates don't, or even deeper things like how did any 0:04:09.480 --> 0:04:12.640 of us we and other land dwelling animals, how did 0:04:12.680 --> 0:04:16.159 we get kidneys or lungs? We can ask the same 0:04:16.279 --> 0:04:19.560 sort of questions about the brain. The brain has a 0:04:19.800 --> 0:04:25.159 very rich evolutionary history, a long and sometimes branching pathway 0:04:25.240 --> 0:04:28.719 that has led from early brains swimming around looking for 0:04:28.839 --> 0:04:33.520 food to brains now that build skyscrapers and launch rocket 0:04:33.520 --> 0:04:37.520 ships and try to figure themselves out. This is the 0:04:37.600 --> 0:04:39.560 kind of stuff that none of our neighbors in the 0:04:39.600 --> 0:04:41.919 animal kingdom do, as far as we can tell. And 0:04:41.960 --> 0:04:45.680 there's clearly something special about the human brain that allows 0:04:45.720 --> 0:04:49.560 that to happen. In other words, we find smarts all 0:04:49.600 --> 0:04:52.800 across the animal kingdom, but there is something very special 0:04:52.920 --> 0:04:58.320 about human intelligence. There's an evolutionary biologist named Theodosius Dubzanski, 0:04:58.560 --> 0:05:03.440 and he once said all species are unique, but humans 0:05:03.680 --> 0:05:06.920 are the uniquest. So I've just told you two things. 0:05:07.040 --> 0:05:10.640 On the one hand, we have very similar brains to 0:05:10.720 --> 0:05:14.039 all our animal cousins, and on the other hand, we 0:05:14.200 --> 0:05:19.000 have a runaway intelligence. So what has happened here? One 0:05:19.040 --> 0:05:23.080 person who has devoted himself to this question is Max Bennett, 0:05:23.080 --> 0:05:27.080 who wrote a wonderful book called A Brief History of Intelligence, 0:05:27.480 --> 0:05:30.800 And in this book, Max distills an enormous amount of 0:05:30.880 --> 0:05:34.960 data about the history of animal species to reveal a 0:05:35.160 --> 0:05:40.839 clear path that stretches from very ancient ancestors to us. 0:05:41.520 --> 0:05:45.839 He attributes the story of human intelligence not just to 0:05:45.920 --> 0:05:50.719 a single breakthrough, but to five breakthroughs. I really loved 0:05:50.720 --> 0:05:58.840 his books, so I called him to join us today. So, Max, 0:05:58.880 --> 0:06:02.240 when we're talking about the origins of intelligence, you might 0:06:02.320 --> 0:06:04.400 think that what we need to do is look all 0:06:04.400 --> 0:06:07.279 the way back to our common ancestors with the apes, 0:06:07.440 --> 0:06:10.640 or maybe farther back to mammals, or maybe even as 0:06:10.640 --> 0:06:13.320 far back as reptiles. But you suggest in your book 0:06:13.560 --> 0:06:16.560 that we have to look back much farther than that. 0:06:16.720 --> 0:06:19.479 Even so tell us where you think the sparks of 0:06:19.520 --> 0:06:20.600 intelligence began. 0:06:21.320 --> 0:06:24.320 So what's so interesting in trying to understand how the 0:06:24.360 --> 0:06:28.360 human brain works is not only how much we've learned, 0:06:28.400 --> 0:06:31.880 but also how much we've still failed to learn because 0:06:31.880 --> 0:06:34.240 of how complicated the human brain is. I mean, the 0:06:34.279 --> 0:06:36.600 human brain has eighty six billion neurons and one hundred 0:06:36.640 --> 0:06:40.360 trillion connections, and so one strategy for trying to understand 0:06:40.400 --> 0:06:43.760 the brain is to look at the series of steps 0:06:43.800 --> 0:06:46.200 by which it came to be. Even if we only 0:06:46.240 --> 0:06:50.280 go as far back as the first vertebrates, with whom 0:06:50.320 --> 0:06:53.039 our common ancestors are around five hundred million years ago. 0:06:53.560 --> 0:06:56.880 Our ancestors had brains somewhat akin to a modern fish, 0:06:57.279 --> 0:07:00.680 and even in a fish brain there are a lot 0:07:00.760 --> 0:07:04.400 of complicated structures and a lot of neurons. So I 0:07:04.400 --> 0:07:06.280 think it behooves us to go back all the way 0:07:06.320 --> 0:07:09.200 to the very first brains, which have brains akin to 0:07:09.840 --> 0:07:13.560 a modern nematode and a modern Some species of modern nematodes, 0:07:13.600 --> 0:07:16.120 like C. Elegans, only have three hundred two neurons, and 0:07:16.160 --> 0:07:18.119 we can learn a lot about what the very first 0:07:18.120 --> 0:07:22.040 brain did by understanding what a nematode brain does. 0:07:22.120 --> 0:07:25.400 So tell us what a nematode is for listeners who don't. 0:07:25.280 --> 0:07:27.960 Know, there's many different species of nematodes, but the most 0:07:27.960 --> 0:07:30.760 well studied is something called Cea elegans, and it is 0:07:30.800 --> 0:07:34.040 a small wormlike creature. You could fit a few on 0:07:34.080 --> 0:07:37.880 your fingertip. And they have no eyes, they have no ears, 0:07:37.920 --> 0:07:40.640 they can't render an image of the external world. They 0:07:40.680 --> 0:07:43.600 only have three hundred two neurons in its entire nervous system, 0:07:44.040 --> 0:07:46.360 and yet it can do some really impressive stuff and 0:07:46.360 --> 0:07:48.840 that teach us a lot about the foundations of the 0:07:48.920 --> 0:07:49.680 very first brains. 0:07:49.960 --> 0:07:52.680 Okay, so give us a sense of what C. Elegans 0:07:52.720 --> 0:07:53.000 can do. 0:07:53.480 --> 0:07:56.000 One thing that's really interesting about C. Elegans is how 0:07:56.080 --> 0:07:59.400 well it navigates the world and the absence of a 0:07:59.520 --> 0:08:03.640 complex sensory apparatus. So one might think that in order 0:08:03.680 --> 0:08:07.679 to find food or avoid predators, one needs to build 0:08:07.680 --> 0:08:11.280 a map of space, or have eyes that enable them 0:08:11.320 --> 0:08:15.120 to see into the distance, or have complex ears that 0:08:15.160 --> 0:08:18.640 allow them to detect things through sound. But the elegance 0:08:18.640 --> 0:08:21.040 has none of this. And yet if you put sea 0:08:21.040 --> 0:08:24.720 elegans in a peatrie dish, it finds food rapidly. And 0:08:24.760 --> 0:08:27.720 if you put them in the wild, they eminently find 0:08:27.800 --> 0:08:31.400 optimal temperatures, and they eminently find ways to avoid predators. 0:08:31.800 --> 0:08:34.720 And so the ways that their brain does this seems 0:08:34.720 --> 0:08:37.120 to be quite similar to the way that a rumba works. 0:08:38.280 --> 0:08:42.800 So a rumba, if folks aren't familiar, is the sort 0:08:42.800 --> 0:08:46.280 of classic vacuum cleaning robot, and it also has no 0:08:46.360 --> 0:08:49.840 eyes or ears, and yet somehow it cleans up everything 0:08:49.880 --> 0:08:53.240 in your house. And so what a rumba does is 0:08:53.880 --> 0:08:55.600 when it hits the wall, it sort of backs away 0:08:55.600 --> 0:08:58.079 and turns randomly, and it keeps doing this randomly enough 0:08:58.160 --> 0:09:00.880 until it reaches all the corners of your house. And 0:09:00.920 --> 0:09:04.640 what nematoad does in some ways actually more advanced, where 0:09:04.679 --> 0:09:07.640 it has sensory neurons around its head, and all these 0:09:07.679 --> 0:09:10.600 sensory neurons do is they get excited when a good 0:09:10.640 --> 0:09:13.320 thing like a smell, is increasing in concentration like a 0:09:13.320 --> 0:09:17.679 food smell, and those drive forward movements, or another set 0:09:17.720 --> 0:09:22.000 of neurons gets excited when something bad increases or something 0:09:22.000 --> 0:09:24.520 good decreases, in other words, a decreasing concentration of a 0:09:24.559 --> 0:09:28.920 food smell. And just by detecting these changes, a brain 0:09:29.040 --> 0:09:31.120 can decide I'm going to keep going forward if good 0:09:31.160 --> 0:09:33.959 things are increasing, or I'm going to turn randomly if 0:09:34.000 --> 0:09:37.800 good things are decreasing, And this is classically called taxis navigation. 0:09:38.480 --> 0:09:41.000 In simpler terms, you call this just steering. And in 0:09:41.040 --> 0:09:44.160 the absence of any site, nematoads can find the origin 0:09:44.200 --> 0:09:48.040 of food smells because food creates this gradience in water, 0:09:48.240 --> 0:09:51.280 where the concentration of the smell is higher towards the source. 0:09:51.760 --> 0:09:55.200 So the very first brain, its core function, was just 0:09:55.240 --> 0:09:57.520 to categorize things in the world and too good and bad, 0:09:57.920 --> 0:09:59.760 such that it would turn towards good things in a 0:09:59.800 --> 0:10:00.720 way from bad things. 0:10:01.160 --> 0:10:03.760 Now bacteria do that too, yes they do. 0:10:03.920 --> 0:10:09.800 Clinokinesis absolutely what's almost mesmerizing about evolution is how this 0:10:10.080 --> 0:10:13.320 exact same algorithm seems to have been recapitulated in a 0:10:13.320 --> 0:10:17.679 completely different substrate. So single celled organisms do this exact 0:10:17.720 --> 0:10:20.920 same type of taxis navigation, but it's implemented in sort 0:10:20.920 --> 0:10:24.319 of the protein machinery of a single cell. And animatode 0:10:24.360 --> 0:10:27.400 does the exact same algorithm, but not implemented within a 0:10:27.440 --> 0:10:29.360 single cell, but through a web of neurons. 0:10:29.800 --> 0:10:32.240 And so what you've proposed in your book, which is 0:10:32.280 --> 0:10:38.679 an amazing book, is five breakthroughs that happened in evolutionary 0:10:38.760 --> 0:10:43.040 time scales that led to intelligence the way that we 0:10:43.840 --> 0:10:47.840 have and care about intelligence. So tell us about breakthrough 0:10:47.920 --> 0:10:48.400 number one. 0:10:49.040 --> 0:10:52.679 So breakthrough number one was this idea of steering. So 0:10:52.920 --> 0:10:57.000 the animals before the first animals with brains, which are 0:10:57.000 --> 0:11:01.800 classically called bileatrians because they have bilineateral symmetry, meaning they're 0:11:01.840 --> 0:11:04.959 symmetric across the central plane. It is interesting to real 0:11:05.040 --> 0:11:07.320 people don't realize this until they think about it, but 0:11:07.440 --> 0:11:09.800 all animals that we think of as animals are symmetric 0:11:09.920 --> 0:11:11.600 across the central line through their body. 0:11:11.840 --> 0:11:13.200 So you mean they have a left side on the 0:11:13.280 --> 0:11:15.880 right side, and they are a mirror image. 0:11:16.240 --> 0:11:19.439 Yeah, and so, but not all animals have that. So 0:11:19.600 --> 0:11:21.920 the very very first animals, we think, we don't have 0:11:21.960 --> 0:11:24.719 perfect evidence for this, but we think we're probably more 0:11:24.760 --> 0:11:28.280 akin to a coral polyp or a jellyfish, which has 0:11:28.400 --> 0:11:33.160 radial symmetry, so they're symmetric across a central axis. And 0:11:33.240 --> 0:11:37.080 so the transition from radial symmetry to bilateral symmetry seems 0:11:37.120 --> 0:11:39.960 to be in part driven by the need to navigate. So, 0:11:40.440 --> 0:11:43.280 although jellyfish are an interesting exception because some of them 0:11:43.360 --> 0:11:48.080 independently seem to have evolved relatively complex navigational systems, most 0:11:48.400 --> 0:11:52.559 evolutionary neuroscientists think the very first animals were more sensile, 0:11:52.800 --> 0:11:54.960 like a coral polyp, where they sit in place. They 0:11:55.000 --> 0:11:57.040 have tentacles and they just try to detect food that 0:11:57.080 --> 0:12:00.280 pass by the tentacles. But the very first animal with 0:12:00.320 --> 0:12:05.040 brains are bilateral ancestors. They use this brain to categorize 0:12:05.080 --> 0:12:07.040 the world and to good and bad. To implement this 0:12:07.160 --> 0:12:10.280 taxis navigation to find food and avoid predators. 0:12:10.679 --> 0:12:14.640 So the existence of a brain correlates with having this 0:12:14.760 --> 0:12:17.120 left right side. Is that correct? 0:12:17.480 --> 0:12:20.880 There are all animals with brains descend have bilateral symmetry, 0:12:20.960 --> 0:12:24.000 or descend from the bilaterally symmetric ancestor in which the 0:12:24.000 --> 0:12:27.000 first brains evolved. And so we also see a suite 0:12:27.000 --> 0:12:30.199 of other interesting things emerged with this first breakthrough of steering. 0:12:31.120 --> 0:12:34.440 One is classically called affect, which is sort of the 0:12:34.440 --> 0:12:40.160 first template of emotional states. And so a nematode actually 0:12:40.240 --> 0:12:44.400 has dopamine neurons, and what these dopamine neurons do is 0:12:44.400 --> 0:12:48.160 they detect the presence of bacteria outside of the nematode. 0:12:48.520 --> 0:12:51.240 And what it does is it changed their behavioral repertoire 0:12:51.320 --> 0:12:54.720 to search in their local area. And we see why 0:12:54.760 --> 0:12:56.840 this exists in the rumba. So a rumba has something 0:12:56.880 --> 0:13:00.200 called dirt detect and what dirt detect does is if 0:13:00.200 --> 0:13:03.840 it bumps into dirt, it starts turning randomly in that area. 0:13:04.679 --> 0:13:06.839 And the reason it does that is because the world 0:13:06.880 --> 0:13:09.600 is clumpy, So if you detect dirt, it's likely that 0:13:09.640 --> 0:13:12.880 there's other dirt nearby, even though you're not. Maybe detecting 0:13:12.960 --> 0:13:16.080 dirt in the moment. So what anematod does is the 0:13:16.080 --> 0:13:18.800 exact same thing. If it runs into food, even though 0:13:18.840 --> 0:13:21.480 it might not detect food a second later, it's probably 0:13:21.520 --> 0:13:24.160 the case there's other food nearby, and so this rush 0:13:24.160 --> 0:13:29.199 of dopamine drives this local search in these very early brains. Similarly, 0:13:29.640 --> 0:13:33.200 there are serotonin neurons, but they're in the throat, and 0:13:33.280 --> 0:13:36.640 so what serotonin signals is the consumption of food, and 0:13:36.720 --> 0:13:40.199 serotonin in these very early nematodes drives sort of satiation. 0:13:41.120 --> 0:13:44.800 And of course those chemicals do much more complicated things 0:13:44.800 --> 0:13:48.000 than human brains. That basic template of dopamine being the 0:13:48.040 --> 0:13:53.240 seeking exploitation nearby reward signal and serotonin being the sort 0:13:53.280 --> 0:13:58.680 of satiation consumption satisfaction signal. We do see hints of 0:13:58.760 --> 0:14:04.480 that basic even in human brains. So we see categorizing 0:14:04.480 --> 0:14:07.000 the world into good to bad, we see bilateral symmetry, 0:14:07.040 --> 0:14:09.800 we see these very basic behavioral states. And then the 0:14:09.880 --> 0:14:11.920 last thing we also see emerge in this breakthrough of 0:14:12.000 --> 0:14:15.520 steering is the foundation of associative learning, and this is 0:14:15.520 --> 0:14:18.280 the first form of real learning that we see emerge 0:14:18.440 --> 0:14:24.520 in animal evolution. And anematode can associate a stimulus with 0:14:24.840 --> 0:14:27.240 a good or bad thing. So if you put a 0:14:27.240 --> 0:14:30.720 nematode in a peach redish and put salt on one side, 0:14:31.000 --> 0:14:35.120 nematoads typically steer towards salts because salt tends to correlate 0:14:35.160 --> 0:14:37.720 with food. But if you leave them in a peach redition, 0:14:37.920 --> 0:14:39.760 starve them for a long period of time in the 0:14:39.760 --> 0:14:42.800 presence of salt water, they change their opinion and they 0:14:42.800 --> 0:14:45.160 will start steering away from salt in the future. And 0:14:45.200 --> 0:14:48.360 it makes sense why associative learning would emerge if the 0:14:48.440 --> 0:14:51.080 very first brain of steering, because you want to tweak 0:14:51.120 --> 0:14:54.000 the goodness and badness of things, because deciding what to 0:14:54.040 --> 0:14:55.840 turn towards and away from is a life or death 0:14:55.840 --> 0:14:58.720 decision for anema TOD. So this first breakthrough of steering, 0:14:58.760 --> 0:15:01.640 we see the suite of of new abilities from associative learning. 0:15:01.640 --> 0:15:05.440 Bilateral symmetry categorizing things in a good and bad emerged 0:15:05.480 --> 0:15:08.240 with the very first brain. So that was breakthrough number one. 0:15:08.080 --> 0:15:10.400 Okay, terrific. And what was break through number two? 0:15:11.360 --> 0:15:14.400 So if we fast forward about fifty million years or so, 0:15:14.840 --> 0:15:18.680 we enter what's famously known as the Cambrian Period and 0:15:18.720 --> 0:15:23.200 the Cambrian Explosion, is this huge diversification of life, which 0:15:23.280 --> 0:15:27.320 actually is all of the children of this first bilateral animal. 0:15:27.520 --> 0:15:30.240 So if you were to swim around the Cambrian Ocean, 0:15:30.560 --> 0:15:34.520 you would see many ancestors of this bilateral wormlike creature 0:15:34.800 --> 0:15:37.960 who had proliferated into what would look like crustaceans and 0:15:38.080 --> 0:15:42.240 arthropods of today. There were huge insect like creatures in 0:15:42.280 --> 0:15:45.400 the ocean, and then there were also our ancestors, which 0:15:45.440 --> 0:15:49.120 were much smaller, modest creatures, but they were most akin 0:15:49.200 --> 0:15:51.240 to a fish of today, and they were called the 0:15:51.280 --> 0:15:54.840 first vertebrates. And the reason they're called vertebrates is because 0:15:54.840 --> 0:15:58.640 in fossils, the most salient feature is the vertebral column, 0:15:58.760 --> 0:16:01.840 so they had a spine. And in these first vertebrates, 0:16:02.360 --> 0:16:04.920 we can get insight into what their brains did by 0:16:04.960 --> 0:16:09.160 looking into the brains of fish today, because there's many 0:16:09.240 --> 0:16:12.920 species of fish that evolutionary neuroscientists think have brains that 0:16:12.960 --> 0:16:15.480 were quite similar to the very first vertebrates. And what 0:16:15.560 --> 0:16:17.840 I found most surprising when I first started looking into 0:16:17.920 --> 0:16:20.880 this is how similar fish brains are to human brains. 0:16:21.000 --> 0:16:22.880 So I would have expected a fish brain to have 0:16:22.920 --> 0:16:25.600 almost none of the features of a human brain, but 0:16:26.480 --> 0:16:29.680 counter to that, intuition. Fish brain have all, with the 0:16:29.720 --> 0:16:32.040 exception of a few things, have all of the major 0:16:32.080 --> 0:16:36.200 brain structures that a human brain does. And also, counter 0:16:36.360 --> 0:16:39.400 to what my expectations would have been, there's sort of 0:16:39.400 --> 0:16:42.440 a stereotype that fish are really dumb, but the more 0:16:42.440 --> 0:16:45.520 you look into the comparative psychology work done on fish, 0:16:45.680 --> 0:16:48.320 fish are way smarter than we think. And for example, 0:16:48.720 --> 0:16:50.920 fish can learn how to navigate out of a maze 0:16:50.960 --> 0:16:53.000 and remember exactly how to do it a year later. 0:16:53.320 --> 0:16:55.680 You can go to YouTube and find really funny cute 0:16:55.760 --> 0:16:58.560 videos of people training fish to jump through hoops for treats, 0:16:59.040 --> 0:17:01.320 and you can train a to push levers for food 0:17:01.360 --> 0:17:03.600 and all of these sort of fun things. And so 0:17:04.040 --> 0:17:06.760 when we look at these brain structures that emerged, there's 0:17:06.800 --> 0:17:09.320 a lot of really good evidence that the key thing 0:17:09.400 --> 0:17:13.280 that happened was these early vertebrate brains enabled the ability 0:17:13.320 --> 0:17:16.720 to learn through reinforcement and AI. This is called reinforcement 0:17:16.840 --> 0:17:20.960 learning and behavioral psychology is typically called trial and error learning. 0:17:21.280 --> 0:17:24.360 So they could learn to perform arbitrary sequences of actions 0:17:24.440 --> 0:17:27.600 on the basis of whether or not it led to 0:17:27.640 --> 0:17:30.440 a reward at the end. So when we go into 0:17:30.440 --> 0:17:33.439 the fish brain. There are two key structures that are 0:17:33.520 --> 0:17:36.879 useful to know about because they will keep coming up 0:17:36.880 --> 0:17:39.400 through our story and the evolution of the human brain. 0:17:39.680 --> 0:17:42.320 One is something called the basal ganglia, and the basil 0:17:42.320 --> 0:17:45.600 ganglia of a fish has almost exactly the same structure 0:17:46.240 --> 0:17:49.520 and network as the basil ganglia of a human, and 0:17:49.960 --> 0:17:53.920 computational neuroscientists have gone to painstaking efforts to show that 0:17:53.960 --> 0:17:58.000 the basil gangly is implementing a reinforcement learning algorithm almost 0:17:58.000 --> 0:18:01.240 identical to the reinforcement learning algorithm we use in AI 0:18:01.359 --> 0:18:05.040 system today. And the way that it works in principle 0:18:05.280 --> 0:18:08.920 is it trains itself based on the exciting the excitement 0:18:08.920 --> 0:18:12.600 of dopamine, and it learns to repeat behaviors that drive 0:18:12.640 --> 0:18:16.439 dopamine release and inhibit behaviors that drive dopamine decreasing. And 0:18:16.480 --> 0:18:18.680 what's so fascinating is if you look at how this 0:18:19.000 --> 0:18:21.920 system came to be, you can see how reinforcement learning 0:18:22.000 --> 0:18:25.560 is only possible if brains first had the foundation of steering. 0:18:26.040 --> 0:18:29.359 Because the foundation of steering gives us the categorization of 0:18:29.359 --> 0:18:31.600 things in the world and to good and bad, and 0:18:31.640 --> 0:18:34.720 that is repurposed to create this reward signal that the 0:18:34.720 --> 0:18:38.159 basal ganglia then can use to create arbitrary sequences of 0:18:38.200 --> 0:18:41.400 behavior on the basis of what leads to reward or none. 0:18:41.960 --> 0:18:44.120 And this is how a fish can learn really complex 0:18:44.160 --> 0:18:47.600 sequences of actions on the basis of what leads to reward. 0:18:47.640 --> 0:18:50.320 In the end, the second key structure in a fish 0:18:50.359 --> 0:18:53.480 brain is something called the cortex, and we do have 0:18:53.520 --> 0:18:55.919 a version of a cortex. There's a portion of our 0:18:55.960 --> 0:18:58.439 cortex that we'll talk about that's way more advanced. But 0:18:58.480 --> 0:19:01.800 a phish cortex can still do something incredible that the 0:19:01.800 --> 0:19:05.399 first nematodes could not, which is it recognizes things in 0:19:05.440 --> 0:19:08.679 the world on the basis of patterns. So in the 0:19:08.720 --> 0:19:11.919 first nema, in the first bilateral brain, it could not 0:19:12.000 --> 0:19:13.679 detect things in the world on the basis of a 0:19:13.680 --> 0:19:16.080 pattern of activation. So when you look at a horse, 0:19:16.440 --> 0:19:19.000 you recognize a horse not because of any single neuron 0:19:19.040 --> 0:19:21.760 in your brain, but because your brain is somehow decoding 0:19:21.800 --> 0:19:24.480 the pattern of activation on your retina, the neurons in 0:19:24.520 --> 0:19:27.760 your retina. And so the first brains could not do 0:19:27.840 --> 0:19:30.639 anything like this. They only detected things when a single 0:19:30.680 --> 0:19:34.040 neuron got excited in the presence of some stimulus. But fish, 0:19:34.400 --> 0:19:36.760 fish can even recognize human faces. There have been some 0:19:36.800 --> 0:19:39.040 amazing studies that show a fish can recognize a human 0:19:39.040 --> 0:19:41.520 face and learn which face leads to a reward in 0:19:41.560 --> 0:19:44.640 which face does not. Even when that face is rotated 0:19:44.680 --> 0:19:48.320 in space, they still recognize it. So the cortex somehow, 0:19:48.320 --> 0:19:51.600 this is still an outstanding mystery in the field of neuroscience. 0:19:51.920 --> 0:19:56.320 Somehow the cortex recognizes patterns and fish eminently well. And 0:19:56.359 --> 0:19:59.160 in some ways the cortex of a fish recognizes patterns 0:19:59.200 --> 0:20:01.359 better than even our our best vision systems in AI, 0:20:01.480 --> 0:20:03.680 because we can. They've done studies that show that a 0:20:03.760 --> 0:20:07.200 fish can recognize objects in one shots even though it's 0:20:07.240 --> 0:20:10.159 been rotated in space, and AI systems typically don't do that. 0:20:10.160 --> 0:20:11.600 You need a lot of data to get into that. 0:20:12.000 --> 0:20:15.560 So at the first fish brain, we see reinforcement learning emerge, 0:20:15.840 --> 0:20:19.000 which can recognize patterns in the world and can learn 0:20:19.040 --> 0:20:21.359 to take actions in the presence of those patterns based 0:20:21.400 --> 0:20:24.200 on rewards. We see reinforcement learning as breakthrough number two. 0:20:24.800 --> 0:20:26.879 Excellent, Okay, how about number three? 0:20:27.160 --> 0:20:30.199 Then we're going to fast forward through a long period 0:20:30.240 --> 0:20:34.160 of evolutionary time, all the way until about one hundred 0:20:34.160 --> 0:20:36.240 and fifty million years ago. Between hundred and hundred million 0:20:36.280 --> 0:20:40.480 years ago. This is the era of dinosaurs. Our ancestors 0:20:40.480 --> 0:20:44.600 were very, very humble, tiny squirrel like creatures that lived underground, 0:20:45.080 --> 0:20:48.560 and we only came out at nights to hunt for insects. 0:20:49.240 --> 0:20:51.800 But these were the first mammals. We know a lot 0:20:51.800 --> 0:20:54.120 about mammal brains, way more than we actually know about 0:20:54.160 --> 0:20:57.959 fish brains, because the main stay of neuroscience research typically 0:20:58.000 --> 0:21:02.760 happens in rats and mice when we go into these brains. Interestingly, 0:21:03.000 --> 0:21:06.399 the fundamental difference between a mammal brain and a fish 0:21:06.440 --> 0:21:09.840 brain is the presence of one key new structure, which 0:21:09.920 --> 0:21:13.440 is a part of the cortex elaborates into what's famously 0:21:13.480 --> 0:21:17.560 called the neocortex NEO for new, and under a microscope 0:21:17.560 --> 0:21:20.800 there's some really interesting things. So we have remnants of 0:21:20.840 --> 0:21:24.800 the old cortex of fish they are called the olfactory cortex, 0:21:24.920 --> 0:21:28.440 and humans and mammals they're called the hippocampus, and they're 0:21:28.440 --> 0:21:32.000 called the cortical amygdala. These are all ancestral remnants of 0:21:32.040 --> 0:21:35.240 the very first cortex. But the neocortex is entirely new. 0:21:35.400 --> 0:21:38.560 This is something that only occurred within mammals, and it 0:21:38.600 --> 0:21:42.120 looks way more complicated under a microscope and so there's 0:21:42.119 --> 0:21:47.080 this grand question what did this neocortex do? And classically, 0:21:47.400 --> 0:21:50.240 when we study the neocortex, we look at a lot 0:21:50.280 --> 0:21:52.200 of humans, and when you look at a human brain, 0:21:52.880 --> 0:21:54.960 the whole thing seems to be neocortex. So when you 0:21:55.000 --> 0:21:57.160 look at human brain, all of this all is full, 0:21:57.240 --> 0:22:01.720 that's all neocortex bunched together. It's this sort of has 0:22:01.800 --> 0:22:06.879 this sort of surface area, and the neocortex seems to 0:22:06.920 --> 0:22:09.960 do everything, which is this funny perplexing thing in neuroscience. 0:22:10.040 --> 0:22:12.840 Because there's one region that seems to do vision. If 0:22:12.840 --> 0:22:15.960 it gets damaged, people can't see. There's another area that 0:22:16.000 --> 0:22:18.520 seems to do audition. If it gets damaged, people struggle 0:22:18.520 --> 0:22:20.919 to hear things. There's a region that seems to do attention. 0:22:20.960 --> 0:22:23.000 If it get damaged, you can't perceive things on one 0:22:23.000 --> 0:22:26.400 side or visual of view. There's an area for movements. 0:22:26.480 --> 0:22:28.560 If it gets damage, you get paralyzed, so on and 0:22:28.600 --> 0:22:30.840 so forth. So it's this grand sort of mystery of 0:22:30.840 --> 0:22:33.480 what the neocortex does, but most of it seems to 0:22:33.520 --> 0:22:35.679 have been based on this idea of perception. A lot 0:22:35.720 --> 0:22:38.040 of the neocortex seems to enable us to perceive things 0:22:38.040 --> 0:22:40.240 in the world, But what's odd is if we think 0:22:40.240 --> 0:22:44.639 about this from an evolutionary perspective, there's no clear perceptual 0:22:44.840 --> 0:22:49.040 improvements or very salient at least perceptual improvements, and a 0:22:49.080 --> 0:22:51.680 mammal relative to a fish, so a fish can recognize 0:22:51.680 --> 0:22:54.760 faces as well as a rat can. It recognizes them 0:22:54.800 --> 0:22:57.600 when rotated in space, So it's not so clear from 0:22:57.600 --> 0:23:01.680 an evolutionary perspective that the neocortech evolve for better perception. 0:23:02.119 --> 0:23:05.639