Jay McClelland: Neural Networks and the Emergence of Cognition | Lex Fridman Podcast #222

1. 0:00 – 0:43

   Introduction

   1. LFLex Fridman0:00

      The following is a conversation with Jay McClelland, a cognitive scientist at Stanford and one of the seminal figures in the history of artificial intelligence, and specifically neural networks, having written the Parallel Distributed Processing book with David Rumelhart, who co-authored the Backpropagation paper with Geoff Hinton. In their collaborations, they've paved the way for many of the ideas at the center of the neural network-based machine learning revolution of the past 15 years. To support this podcast, please check out our sponsors in the description. This is the Lex Fridman podcast, and here is my conversation with Jay McClelland.

2. 0:43 – 5:02

   Beauty in neural networks

   1. LFLex Fridman0:43

      You are one of the seminal figures in the history of neural networks at the intersection of, uh, cognitive psychology and computer science. What to you has, over the decades, emerged as the most beautiful aspect about neural networks, both artificial and biological?

   2. JMJay McClelland1:00

      The fundamental thing I think about with neural networks is how they allow us to link biology with the mysteries of thought. And, um, you know, in the. When I was first entering the field myself in the late '60s, early '70s, cogni- cognitive psychology had just become a field. There was a book published in '67 called Cognitive Psychology. Um, and the author said that, you know, the study of the nervous system was only of peripheral interest. It wasn't going to tell us anything about the mind, and I didn't agree with that. I, I always felt, "Oh, look, I'm, I'm a physical being." I... From dust to dust, you know, ashes to ashes, and somehow I emerged from that. Um-

   3. LFLex Fridman2:06

      So, so that's really interesting. So there was a sense with cognitive psychology that in understanding the sort of neuronal structure of things, you're not going to be able to understand the mind? And then your sense is if we study these neural networks, we might be able to get at least very close to understanding the fundamentals of the human mind.

   4. JMJay McClelland2:28

      Yeah. I used to think, um, or I used to talk about the idea of awakening from the Cartesian dream.

   5. LFLex Fridman2:35

      (laughs)

   6. JMJay McClelland2:36

      So Descartes, um, you know, thought about these things, right? He, he was walking in the gardens of Versailles one day, and he stepped on a stone and a statue moved. And he walked a little further, he stepped on another stone and another statue moved, and he, like, "Why did the statue move when I stepped on the stone?" And he went and talked to the gardeners, and he found out that they had a hydraulic system that allowed the physical contact with the stone to cause water to flow in various directions, which caused water to flow under the statue and moved the statue. And he used this as the beginnings of a theory about how animals act, and he had this notion that these little fibers that people had identified that weren't carrying the blood, you know, were these little hydraulic tubes that if you touch something, there would be pressure and it would send a signal of pressure to the other parts of the system and that would cause action. So he had a mechanistic theory of animal behavior, and he thought that the human had this animal body, but that some divine something else had to have come down and been placed in him to give him the ability to think, right? So the physical world includes the body in action, but it doesn't include thought according to Descartes, right?

   7. LFLex Fridman4:18

      Right.

   8. JMJay McClelland4:19

      And so the study of physiology at that time was the study of sensory systems and motor systems and things that you could directly measure when you stimulated neurons and stuff like that. And, um, the study of cognition was something that, you know, was tied in with abstract computer algorithms and things like that, but whe- when I was an undergraduate, I learned about the physiological mechanisms. Uh, and so when I'm studying cognitive psychology as a first-year PhD student, I'm saying, "Wait a minute, the whole thing is biological," right? (laughs)

   9. LFLex Fridman4:57

      Mm-hmm.

   10. JMJay McClelland4:58

       You know?

   11. LFLex Fridman4:58

       You had that intuition right away. That was- seemed obvious to you.

   12. JMJay McClelland5:01

       Yeah, yeah.

3. 5:02 – 10:47

   Darwin and evolution

   1. JMJay McClelland5:02

   2. LFLex Fridman5:03

      Is- isn't that magical though, that from just the little bit of biology can emerge the full beauty of the human experience? Is that... Wh- why is it so obvious to you?

   3. JMJay McClelland5:13

      Well, I- it... Obvious and not obvious at the same time. Um, and I- I think about Darwin in this context too, because Darwin knew very early on that none of the ideas that anybody had ever offered gave him a sense of understanding how evolution could have worked, but he wanted to figure out how it could have worked. That was his goal.

   4. LFLex Fridman5:41

      Mm-hmm.

   5. JMJay McClelland5:42

      And he spent a lot of time working on this idea and coming, you know, reading about things that gave him hints and thinking they were interesting but not knowing why, drawing more and more pictures of different birds that differ slightly from each other and so on, you know, and, and then, then he figured it out.But after he figured it out, he had nightmares about it.

   6. LFLex Fridman6:06

      Mm-hmm.

   7. JMJay McClelland6:06

      He would dream about the complexity of the eye and the arguments that people had given about how ridiculous it was to imagine that that could've ever emerged from some sort of, you know, unguided process.

   8. LFLex Fridman6:24

      Right.

   9. JMJay McClelland6:24

      That it hadn't been the product of design. And, and, uh, so he, he didn't publish for a long time, in part because he was scared of his own ideas. He didn't think they could proba- possibly be true.

   10. LFLex Fridman6:38

       Yeah.

   11. JMJay McClelland6:39

       Um, but then, you know, by the time the 20th century rolls around, we all, um, you know, we understand that evolu- or many people understand (laughs) or believe that evolution, uh, produced, you know, the entire, uh, range of, uh, animals that there are. Uh, and, uh, you know, Descartes' idea starts to seem a little wonky after a while, right? Like, "Well, wait a minute, um, there's the apes and the chimpanzees and the bonobos and, you know, like, they're pretty smart in some ways, you know? So what... Oh," You know, somebody comes up, "Oh, there's a certain part of the brain that's still different. They don't, you know, there's no hippocampus in the monkey brain. It's only in the human brain." Uh, Huxley had to do a surgery in front of many, many people in the late 19th century to show to them there's actually a hippocampus-

   12. LFLex Fridman7:38

       (laughs)

   13. JMJay McClelland7:38

       ... (laughs) in the chimpanzee's brain, you know? (laughs) So, so the continuity of the species is another element, uh, that, you know, contributes to, um, this sort of, you know, idea that we are ourselves, uh, total product of nature. Um, and, uh, that to me is the, is the magic and the mystery how, how nature could actually, um, you know, give rise to, uh, organisms that have the, uh, capabilities that we have.

   14. LFLex Fridman8:20

       So, it's interesting because even the idea of evolution is hard for me to keep all together in my mind. So because we think of a human time scale-

   15. JMJay McClelland8:29

       Mm.

   16. LFLex Fridman8:30

       ... it's hard to imagine the, like, like the, the, the development of the human eye would give me nightmares too.

   17. JMJay McClelland8:35

       Hm.

   18. LFLex Fridman8:36

       Because you have to think across many, many, many generations, and it's very tempting to to think about kinda a growth of a complicated object, and it's like, "How is it possible for that such, such a thing to be built?" 'Cause also, I mean, from a robotics engineering perspective, it's very hard to build these systems. How can through an undirected process can a complex thing be designed?

   19. JMJay McClelland9:01

       Mm.

   20. LFLex Fridman9:01

       It seems not... It seems wrong.

   21. JMJay McClelland9:03

       Yeah. So, that's absolutely right, and I, you know, um, a slightly different career path that would've been equally interesting to me would have, would have been, um, to actually study the process of embryological development flowing on into brain development and-

   22. LFLex Fridman9:24

       Yeah.

   23. JMJay McClelland9:25

       ... the, the, um, exquisite sort of laying down of pathways and so on that occurs in the brain. And, uh, I know the slightest bit about that. It's not my field, but, um, there are, you know, fascinating aspects to this process that eventually result in the, you know, the complexity of, of, uh, various brains. At, at least, you know, one thing, um, we're, um, in the, in the field, I think people have felt for a long time, it- it- in the study of vision, the continuity between humans and non-human animals has been, has been second nature for a lot longer.

   24. LFLex Fridman10:12

       Mm-hmm.

   25. JMJay McClelland10:12

       I was having, I had this conversation, um, with somebody who's a vision scientist, and he was saying, "Oh, we, we don't have any problem with this. You know, the monkey's visual system and the human visual system, extremely similar," um, up to certain levels, of course. They, they diverge after a while, but, um, the first, the, the visual pathway from the eye to the brain and the first few, um, layers of cortex, um, or cortical areas, I guess one would say, uh, are, are extremely similar.

4. 10:47 – 17:29

   The origin of intelligence

   1. JMJay McClelland10:48

   2. LFLex Fridman10:49

      Yeah, so on the cognition side is where the leap seems to happen with humans, that it does seem we're kind of special, and that's a really interesting question when thinking about alien life or if there's other intelligent alien civilizations out there is, how special is this leap? So, one special thing seems to be the origin of life itself. However you define that, there's a gray area. And the other leap, this is very biased perspective of a human, is the, the origin of intelligence, and it, again, from an engineer perspective, it's a difficult question to ask, an important one, is, uh, how difficult is that leap? How special are we humans? Did, uh, did, uh, a monolith come down? Did aliens bring down a monolith and some, um, apes had to touch a monolith or to-

   3. JMJay McClelland11:36

      (laughs)

   4. LFLex Fridman11:36

      ... (laughs) to get it, to get-

   5. JMJay McClelland11:38

      It's a lot like Dec- Descartes', uh, you know-

   6. LFLex Fridman11:40

      (laughs) .

   7. JMJay McClelland11:40

      ... idea, right?

   8. LFLex Fridman11:41

      Exactly, I-

   9. JMJay McClelland11:42

      (laughs) .

   10. LFLex Fridman11:42

       ... it's, but it just seems-

   11. JMJay McClelland11:44

       (laughs) .

   12. LFLex Fridman11:44

       ... that it seems one heck of a leap-

   13. JMJay McClelland11:46

       Yeah.

   14. LFLex Fridman11:46

       ... to get to this level of intelligence.

   15. JMJay McClelland11:48

       Yeah, and, you know, so Chomsky, um, uh, argued, um, that, you know, some, uh, genetic fluke occurred 100,000 years ago.

   16. LFLex Fridman12:02

       (laughs)

   17. JMJay McClelland12:02

       And, you know, just happened that some-... human, some hominin predecessor of current humans had this one genetic tweak that resulted in language.

   18. LFLex Fridman12:19

       Yeah.

   19. JMJay McClelland12:20

       And language then provided this special thing that separates us from all other animals. Um, I'm, I think there's a lot of truth to the value and importance of language, but I think it comes along with, um, the evolution of a lot of other related things, related to sociality and mutual engagement with others, and, um, establishment of, um, I don't know, rich mechanisms for organizing and understanding of the world which language then plugs into.

   20. LFLex Fridman13:13

       Right. So it's, uh, language is a tool that allows you to do this kind of collective intelligence. And whatever is at the core of the thing that allows for this collective intelligence is the main thing. And it's interesting to think about that one fluke, one mutation can lead to the, like, the, the first crack open, opening of the door to human intelligence. Like all it takes is one. Like, evolution just kind of opens the door a little bit, and then, uh, time and selection takes care of the rest.

   21. JMJay McClelland13:45

       You know, there's so many fascinating aspects to these kinds of things. So w- we think of evolution as continuous, right? We think, "Oh, yes. Okay. Over 500 million years, there could have been this, you know, relatively continuous, uh, changes." And, um, but that's not what anthropologists, evolutionary biologists found from the fossil record. They found, you know, hundreds of years of s- uh, hundreds of millions (laughs) of years of stasis. (laughs)

   22. LFLex Fridman14:23

       Yeah.

   23. JMJay McClelland14:24

       And then s- you know, suddenly a change occurs. Well, suddenly on that scale is a million years-

   24. LFLex Fridman14:30

       Yeah.

   25. JMJay McClelland14:30

       ... or something, but, but s- or even 10 million years, but, but, um, the concept of punctuated equilibrium was a, a very important concept in evolutionary biology, uh, and, uh, that also feels somehow right about, you know, st- the stages of our mental abilities. We, we seem to have a certain kind of mindset at a certain age, and then at another, uh, another age we, like, look at that four-year-old and say, "Oh my God, how could they have thought that way?" So Piaget was known for this kind of stage theory of child development, right? And you look at it closely and suddenly those stages aren't so discrete, and the transitions, but the difference between the four-year-old and the seven-year-old is profound. And that's another thing that's always interested me is how we... Something happens over the course of several years of experience where at some point we reach the point where something like an insight or a transition or a new stage of development occurs, and, uh, uh... You know, these kinds of things can be understood, um, in complex systems, uh, research. And so, um, evolutionary biology, developmental biology, um, cognitive development are all things that have been approached in this kind of way.

   26. LFLex Fridman16:00

       Yeah. Just like you said, I find both fascinating those early years of human life, but also the early, like, minutes, days of from, the embryonic development to, like, how from embryos you get, like, the brain. That development, again, from an engineering perspective is fascinating. So it's not, so the early, um... When you deploy the brain to the human world and it gets to explore that world and learn, that's fascinating. But just, like, the assembly of the mechanism that is capable of learning, that's, like, amazing. The stuff they're doing with, like, brain organoids where you can build many brains and study that, um, self-assembly of a mechanism from, like, the DNA material, that, that's, like, what the heck?

   27. JMJay McClelland16:51

       (laughs)

   28. LFLex Fridman16:51

       You have literally like, uh, biological programs that just generate a system, this mushy thing that's able to be robust and learn in a very unpredictable world, and learn seemingly arbitrary things, or, like, a very large number of things that'll enable survival.

   29. JMJay McClelland17:14

       Yeah. Ultimately, um, that is a very important part of the whole process of, you know, understanding this sort of emergence of mind from brain kind of-

   30. LFLex Fridman17:27

       (laughing)
5. 17:29 – 23:33

   Explorations in cognition

   1. LFLex Fridman17:29

      continuous. So let me, uh, let me step back to neural networks for, for, for another brief minute. You wrote parallel distributed processing books that explored ideas on neural networks in the 1980s together with a few folks, but the books you wrote with, uh, David, uh, Rumelhart, who is the first author on the back propagation paper with Geoff Hinton. So these are just some figures at the time that were thinking about these big ideas. Uh, what are some memorable moments of discovery and beautiful ideas from those early days?

   2. JMJay McClelland18:04

      I'm gonna start, uh, sort of with my own...... process in the mid-70s, and then into the late 70s when I met Geoff Hinton and, uh, he came to San Diego, and we were all together. Um, in my time in graduate school, as I've already described to you, I had this sort of feeling of, "Okay, I'm really interested in human cognition, but this disembodied sort of way of thinking about it that I'm getting from the current mode of thought about it is, isn't working fully for me." And when I got my assistant professorship, I went to UCSD, and, um, that was in 1974. Something amazing had just happened, Dave Rumelhart had written a book together with another man named Don Norman, and the book was called Explorations in Cognition. And it was a- a series of chapters exploring interesting questions about cognition, but in a completely sort of abstract, you know, non-biological kind of way. And I'm saying, "Gee, this is amazing. I'm coming to this community where people can get together and feel like they've collectively exploring, you know, ideas." And, um, it was a book that had a lot of, I don't know, lightness to it, and, you know, the- the co- Don Norman, who was the- the more senior figure to Rumelhart at that time, who led that project, um, you know, cre- always created this spirit of playful exploration of ideas. And so I'm like, "Wow, this is great." But I was also, you know, still trying to get from the neurons to the- to the cognition, and I realized at one point, I- I- I got this opportunity to go to a conference where I heard a talk by a man named James Anderson, who was an engineer, but by then a professor in a psychology department, who had used linear algebra to create neural network models of perception and categorization, and memory. And I... Just blew me out of the water that one could, you know, create a model that was simulating neurons, not just kind of engaged in a stepwise algorithmic process that was construed abstractly. But it was simulating remembering and recalling, and, um, recognizing the prior occurrence of a stimulus or something like that. So for me this was a bridge between the mind and the brain, and I just like star- and I... I remember I was walking across campus one day in 1977, and I almost felt like St. Paul on the road to Damascus. I said to myself, "You know, if I think about the mind in terms of a neural network, it will help me answer the questions about the mind that I'm trying to answer," and that really excited me. So, I think that a lot of people were becoming excited about that, and one of those people was Jim Anderson, who I had mentioned. Another one was Steve Grossberg, who had been, uh, writing about neural networks since the '60s, and Geoff Hinton was yet another. And his PhD dissertation showed up, uh, in an applicant pool to a postdoctoral training program that Dave and Don, the two men I mentioned before, Rumelhart and- and Norman, were administering, and Rumelhart got really excited about Hinton's PhD dissertation. Um, and so, uh, Hinton was one of the first, um, people who came and joined this group of postdoctoral scholars, uh, that, uh, was funded by this- this wonderful grant that they got. Another one who is also well known in neural network circus- uh, circles is Paul Smolensky. He was another one of that group. Anyway, um, Geoff and Jim Anderson organized a conference at UCSD, uh, where we- we were, and, uh, it was called Parallel Models of Associative Memory, and it brought all the people together who had been thinking about these kinds of ideas in 1979 or 1980. And this- this began to kind of really resonate with some of Rumelhart's, um, own thinking, some of his reasons for wanting something other than the kinds of computation he'd been doing so far. So let me talk about Rumelhart now for a minute-

   3. LFLex Fridman23:32

      Mm-hmm.

   4. JMJay McClelland23:32

      ... okay, with that context.

6. 23:33 – 29:58

   Learning representations by back-propagating errors

   1. JMJay McClelland23:33

   2. LFLex Fridman23:33

      Well, let me also just pause, because you said so many interesting things, before we go to Rumelhart. So first of all, for people who are not familiar, uh, neural networks are at the core of the machine learning, deep learning revolution of today. Uh, Geoffrey Hinton, that we mentioned, is one of the figures that were important in the history, like yourself, in the development of these neural networks, artificial neural networks that are then used for the machine learning application. Like I mentioned, the back propagation paper is one of the optimization mechanisms by which these, uh, networks c- uh, can learn. And, uh, the word "parallel" is really interesting. So it's- it's almost like synonymous from a computational perspective what- how you thought at the time about...... neural networks as parallel computation.

   3. JMJay McClelland24:19

      Yeah.

   4. LFLex Fridman24:19

      Is that... Would that, would that be fair to say?

   5. JMJay McClelland24:21

      Well, yeah, the, the, the parallel... The word parallel in this, you know, comes from the idea that each neuron is an independent computational unit, right?

   6. LFLex Fridman24:33

      Mm-hmm.

   7. JMJay McClelland24:33

      It, it gathers data from other neurons, it integrates it in a certain way, and then it produces a result. And it's a very simple little computational unit, but it, it's autonomous in the sense that, (clears throat) you know, it does its thing, right? It's, it's in a biological medium where it's getting nutrients and various, uh, chemicals from that medium, um, but it's, uh... You know, you can think of it as almost like a little, little computer in and of itself. So, the idea is that each... You know, our brains have, oh, look, you know, 100 or hundreds, almost a billion of these little neurons, right?

   8. LFLex Fridman25:19

      Mm-hmm.

   9. JMJay McClelland25:20

      Um, (clicks tongue) and they're all capable of doing their work at the same time. So, it's like, instead of just a single central processor that's engaged in, you know, chug, chug, one step after another, (gasps) we have a billion of these little computational units working at the same time.

   10. LFLex Fridman25:42

       So, at the time, that's... I don't know, maybe you can comment, it seems to me, even still to me, uh, quite a revolutionary way to think about computation relative to the development of theoretical computer science alongside of that, where it's very much like sequential computer. You're analyzing algorithms that are running on a single computer.

   11. JMJay McClelland26:03

       That's right.

   12. LFLex Fridman26:04

       You're saying, "Wait a minute. Why, why, why don't we take a really dumb, very simple computer and just have a lot of them interconnected together? And they're all operating in their own little world, and they're communicating-"

   13. JMJay McClelland26:18

       Mm-hmm.

   14. LFLex Fridman26:18

       "... with each other." And-

   15. JMJay McClelland26:19

       Yeah.

   16. LFLex Fridman26:19

       ... and thinking of computation in that way, and from that kind of computation un- trying to understand how things like certain characteristics of the human mind can emerge.

   17. JMJay McClelland26:30

       Right.

   18. LFLex Fridman26:31

       That, that's quite a revolutionary way of thinking, I would say.

   19. JMJay McClelland26:36

       Well, yes, I agree with you. And, um, there's still this sort of sense of (sighs) not sort of knowing how we kind of get all the way there, um, I think, and-

   20. LFLex Fridman26:54

       Mm-hmm.

   21. JMJay McClelland26:54

       ... this very much remains at the core of the questions that everybody's asking about the capabilities of deep learning and all these kinds of things. But if I could just play this out a little bit-

   22. LFLex Fridman27:05

       Mm-hmm.

   23. JMJay McClelland27:08

       ... um, a, a convolutional neural network or a CNN, which, you know, many people may have heard of, is a set of... You could think of it biologically as a set of collections of neurons. Each one ha- each collection has maybe 10,000 neurons in it, but there's many layers, right? Some of these things are hundreds or even a thousand layers deep, but others are closer to the biological brain, and maybe they're like 20 layers deep or something like that. So, we have, within each layer, we have thousands of neurons or tens of thousands maybe. Well, in the brain, we probably have millions in each layer, so... But we're getting sort of similar in a certain way, right? Um, and then we think, okay, at the bottom level, there's an array of things that are like the photoreceptors in the, in the eye. They respond to the amount of light of a certain wavelength at a certain location on the, on the pixel array. So, that's like the biological eye. And then there's several further stages going up, layers of these neuron-like units. And, um, you go f- from that raw input, array of pixels, to a classification. You've actually built a system that could do the same kind of thing that you and I do when we open our eyes and we look around, and we see there's a cup, there's a cellphone, there's a water bottle. (laughs) And th- these systems are doing that now, right?

   24. LFLex Fridman28:54

       Mm-hmm.

   25. JMJay McClelland28:55

       So, they are, in, in terms of the parallel idea that we were talking about before, they are doing this massively parallel computation in the sense that each of the neurons in each of those layers is thought of as computing its little bit of something about the input, uh, simultaneously with all the other ones in the same layer.

   26. LFLex Fridman29:21

       Mm-hmm.

   27. JMJay McClelland29:22

       We get to the point of abstracting that away and thinking, "Oh, it's just one whole vector that's being computed," when one activation pattern's computed in a single step, and that, that, that abstraction is useful, uh, but it's still that parallel and distributed processing, right? Each one of these guys is just contributing a tiny bit to that whole thing.

   28. LFLex Fridman29:45

       And that's the excitement that you felt that from these simple things you c- it can emerge th- when you add these level of abstractions on it.

   29. JMJay McClelland29:53

       Yeah.

   30. LFLex Fridman29:53

       You, you can start getting all the beautiful things that we think about as cognition.
7. 29:58 – 43:01

   Dave Rumelhart and cognitive modeling

   1. JMJay McClelland29:58

   2. LFLex Fridman29:58

      And so, okay, so you have this, uh, conference, I forgot the name already, but it's parallel and something associative memory and so on. (laughs) Very exciting technical and exciting title, and, uh, you started talking about, uh, Dave Rumelhart. So, who is this person that was so... Uh, you've spoken very highly of him.

   3. JMJay McClelland30:17

      Yeah.

   4. LFLex Fridman30:17

      Can you tell me about him, his ideas, his mind, uh, who he was as a human being, as a scientist?

   5. JMJay McClelland30:25

      So, Dave came from a little tiny town in western South Dakota. And, uh, his mother was the librarian and his father was the editor of the newspaper. Um... And, uh, I know one of his brothers pretty well. Um, they grew up... There were four brothers, uh, and, uh, they grew up together, uh, and their father encouraged them to compete with each other a lot.

   6. LFLex Fridman30:58

      (laughs)

   7. JMJay McClelland30:58

      Um, they competed in sports and they competed in mind games.

   8. LFLex Fridman31:02

      Mm-hmm.

   9. JMJay McClelland31:02

      You know. Um, I don't know, things like sudoku and chess and-

   10. LFLex Fridman31:07

       Okay.

   11. JMJay McClelland31:07

       ... various things like that.

   12. LFLex Fridman31:08

       Mm-hmm.

   13. JMJay McClelland31:08

       And, uh, Dave, um, was a standout undergraduate. He went, uh, as... At a younger age than most people do to college at the University of South Dakota and majored in mathematics and I don't know how he got interested in psychology, but he, um, applied to the Mathematical Psychology program at Stanford and was accepted as a PhD student to study mathematical psychology at Stanford. So, mathematical psychology, uh, is the use of mathematics to model mental processes. Right.

   14. LFLex Fridman31:50

       So, something that I think these days might be called cognitive modeling, that whole space?

   15. JMJay McClelland31:55

       Yeah. It's mathematical in the sense that, um, you say, "If this is true and that is true, then I can derive that this should follow." Okay?

   16. LFLex Fridman32:08

       Mm-hmm.

   17. JMJay McClelland32:08

       And so you say, "These are my stipulations about the fundamental principles, and this is my prediction about behavior," and it's all done with equations. It's not done with a computer simulation.

   18. LFLex Fridman32:18

       Mm-hmm.

   19. JMJay McClelland32:19

       Right? So then you- you solve the equation and that tells you what the probability that the subject will be correct on the seventh trial of the experiment is, or something like that, right? So it's a- it's a- it's um... It's a use of mathematics to descriptively characterize, uh, aspects of- of behavior. And, uh, Stanford at that time was the place where, uh, there were several really, really strong mathematical thinkers who were also connected with three or four others around the country, who, um, you know, brought a lot of really exciting ideas, uh, onto the table. And it was a very, very prestigious part of the field of psychology at that time. So, Rumelhart comes into this, um, he was a very strong student within that program, uh, and uh, he got this job at this brand new university in San Diego in 1967, where he's one of the first assistant professors in the Department of Psychology, uh, at UCSD. So, I got there in '74, seven years later, and Rumelhart at that time was still doing mathematical modeling, but he had gotten interested in cognition. He'd gotten interested in understanding, and, you know, understanding, I think, remains... You know, what does it mean to understand anyway (laughs) , you know? Uh, it's- it's an interesting sort of curious, you know, like, how would we know if we really understood something? But- but he was interested in building machines that would, you know, hear a couple of sentences and have an insight about what was going on. So, for example, one of his favorite things at that time was, um, "Margie was sitting on the front step when she heard the familiar jingle of the Good Humor man. She remembered her birthday money and ran into the house." What is Margie doing?

   20. LFLex Fridman34:43

       (laughs)

   21. JMJay McClelland34:44

       (laughs) Why? (laughs) Well, there's a couple of ideas you could have, but the most natural one is that the Good Humor man brings ice cream, she likes ice cream. She's- she knows she needs money to buy ice cream, so she's gonna run into the house and get her money so she can buy herself an ice cream.

   22. LFLex Fridman35:03

       Mm-hmm.

   23. JMJay McClelland35:03

       It's a huge amount of inference that has to happen to get those things to link up with each other. And- and he was interested in how the hell that could happen, and he was trying to build, um, you know, good old-fashioned AI style, uh, models of representation of language and- and content of, you know, things like, "Has money." (laughs)

   24. LFLex Fridman35:32

       So, like, lo- like formal logic and like knowledge bases, like that kind of stuff?

   25. JMJay McClelland35:36

       Yeah.

   26. LFLex Fridman35:36

       So, he was integrating that with his thinking about cognition?

   27. JMJay McClelland35:40

       Yes.

   28. LFLex Fridman35:40

       The mechanisms of cognition, how can they, like, mechanistically be applied to build these knowledge, like, to actually build something that looks like a web of knowledge and thereby s- from- from there emerges something like understanding-

   29. JMJay McClelland35:56

       Yeah.

   30. LFLex Fridman35:56

       ... whatever the heck that is?
8. 43:01 – 1:05:54

   Connectionism

   1. LFLex Fridman43:01

      idea. So, do you like the term, uh, connectionism, uh, to describe this kind of set of ideas?

   2. JMJay McClelland43:07

      I think it's useful. It highlights the notion that the knowledge that the system exploits is in the connections between the units, right? There isn't a separate dictionary, there's just the connections between the units. So, I already, sort of laid that on the table with the connections from the letter units to the unit for the word time, right? The unit for the word time isn't a unit for the word time for any other reason than it's got the connections to the letters that make up the word time. Those are the units on the input that excite it when it's excited that it- it in a sense represents in the system that there's support for the hypothesis that the word time is present in the input. Um, but it's not ... there, there's, the word time isn't written anywhere (laughs) inside the model. It's solely written there in the picture we drew of the model to say, "That's the unit for the word time," right?

   3. LFLex Fridman44:14

      Yeah.

   4. JMJay McClelland44:14

      And, and, um, if, if you, if somebody wants to tell me, "Well, what are the ... how do you spell that word?" You have to use the connections from that out to, to, to then get those letters, for example.

   5. LFLex Fridman44:27

      That's such a ... that's a counterintuitive idea.

   6. JMJay McClelland44:31

      Mm-hmm.

   7. LFLex Fridman44:31

      We humans want to think in this logic way.

   8. JMJay McClelland44:35

      Mm-hmm.

   9. LFLex Fridman44:36

      This, this idea of, uh, connectionism, it doesn't ... it's weird. It's weird that this is how it all works.

   10. JMJay McClelland44:43

       Yeah. But let's go back to that CNN, right? That CNN with all those layers of neuron-like processing units that we were talking about before, it's gonna come out and say, "This is a cat. That's a dog." But it has no idea why it said that, it's just got all these connections between all these layers of neurons, like from the very first layer to the, you know, the, uh, like whatever these layers are, they just get numbered after a while because they, you know, they, they, they ... somehow further in you go, the more, the more abstract the features are. But it's a graded and continuous sort of process of abstraction anyway.

   11. LFLex Fridman45:21

       Mm-hmm.

   12. JMJay McClelland45:22

       And, you know, it goes from very local, very, very specific to much more sort of global, but it's still, you know, another sort of pattern of activation over an array of units. And then at the output side it says it's a cat or it's a dog. And when, when we, when I open my eyes and say, "Oh, that's Lex."

   13. LFLex Fridman45:41

       Mm-hmm.

   14. JMJay McClelland45:42

       Or, um, "Oh," you know, "there's my own dog" and I recognize my dog-

   15. LFLex Fridman45:49

       Mm-hmm.

   16. JMJay McClelland45:49

       ... uh, which is a member of the same species as many other dogs, but I know this one because of-

   17. LFLex Fridman45:55

       (laughs)

   18. JMJay McClelland45:55

       ... some slightly unique characteristics. I don't know how to describe-

   19. LFLex Fridman45:59

       Yeah.

   20. JMJay McClelland45:59

       ... what it is that makes me know that I'm looking at Lex or at my particular dog, right?

   21. LFLex Fridman46:04

       Yeah.

   22. JMJay McClelland46:04

       Or even that I'm looking at a particular brand of car. Like, I can say a few words about it, but if I g- I wrote you a paragraph about the car, you, you would have trouble figuring out, "Well, which car is he talking about?" (laughs) Right?

   23. LFLex Fridman46:16

       Yeah.

   24. JMJay McClelland46:16

       So, the idea that we have propositional knowledge of what it is that allows us to recognize that this is an actual instance of this particular natural kind is, um, has always been a m- you know, something that, uh, it, it never worked, right? You couldn't ever write down a set of propositions for, you know, visual recognition. And, and, and so it ... in that space, it sort of always seemed very natural that something more implicit, um, you know, you, you don't have access to what the details of the computation were in between, you just get the result. So, that's the other part of connectionism. You cannot ... you don't read the contents of the connections; the connections only cause outputs to occur based on inputs.

   25. LFLex Fridman47:09

       Yeah. It's, it's f- and f- for us that, like, final layer or some particular layer is very important, the one that tells us that it's our dog or, like, that it's a cat or a dog. But, you know, each layer's probably equally as important in the grand scheme of things. Like- (laughs)

   26. JMJay McClelland47:28

       (laughs)

   27. LFLex Fridman47:28

       ... there's no reason why the cat versus dog is more important than the lower level activations. It doesn't really matter. I mean, all of it is just this beautiful stacking on top of each other, and we humans live in this particular layer, so for us, for us it's useful to, to survive, to, to use those i- uh, cat versus dog, predator versus prey, all those kinds of things. It's fascinating that it's all continuous. But then you, you then ask, you know, the history of artificial intelligence, you ask, are we able to introspect and convert the very things that allow us to tell the difference between cat and dog into, uh, logic, into formal logic? That's been the dream. I would say that's still part of the dr- the dream of symbolic AI and, um, I've, I've recently talked to, uh, Doug, uh, Lenat who created CYC. And that's, that's a project that lasted for many decades and still carries a sort of dream in it.

   28. JMJay McClelland48:26

       Mm-hmm.

   29. LFLex Fridman48:26

       Right? Um, but we still don't know the answer, right? It seems like, uh, connectionism is really powerful, but it also seems like there's this building of knowledge.

   30. JMJay McClelland48:38

       Mm-hmm.
9. 1:05:54 – 1:07:49

   Geoffrey Hinton

   1. LFLex Fridman1:05:54

      I'm, I'm a big believer that the loss of, uh, a human being that you value is as powerful as, you know, first falling in love with that human being. I, I think it's all a celebration of the human being, so i- the disintegration itself too is a celebration-

   2. JMJay McClelland1:06:09

      Yeah.

   3. LFLex Fridman1:06:10

      ... in a way.

   4. JMJay McClelland1:06:10

      Yeah. Yeah, yeah. And but just to say something more about the scientist and, and the back propagation idea that you mentioned.

   5. LFLex Fridman1:06:20

      Mm-hmm.

   6. JMJay McClelland1:06:21

      Um, so in, in 1982 Hinton had been there as a post-doc and organized that conference. He'd actually gone away and gotten an assistant professorship and then, um, there was this opportunity to bring him back. So Geoff Hinton was back, uh, on a sabbatical.

   7. LFLex Fridman1:06:45

      San Diego.

   8. JMJay McClelland1:06:46

      In San Diego. And, uh, Rumelhart and I had decided we wanted to do this, you know, we thought it was really exciting and, um, our, the papers on the interactive activation model that I was telling you about had just been published and we both sort of saw huge potential for this work and, and, and Geoff was there. And so the three of us, uh, started a, a research group which we called the PDP Research Group.

   9. LFLex Fridman1:07:15

      Mm-hmm.

   10. JMJay McClelland1:07:15

       And, uh, several other people-... came. Um, Francis Crick, who was at the Salk Institute, heard about it from Geoff, um, and, uh, 'cause Geoff was known among Brits to be brilliant, and Francis was well-connected with his British con- friends. So, Francis Crick came and-

   11. LFLex Fridman1:07:33

       That's a heck of a group of people. Wow, okay.

   12. JMJay McClelland1:07:35

       ... and, uh, uh, several, uh, Paul Smolensky, um, was one of the other post-docs. He was still there as a post-doc. And, um, a few other people. But, uh, anyway,

10. 1:07:49 – 1:24:42

    Learning in a neural network

    1. JMJay McClelland1:07:50

       Geoff talked to us about learning and how we should think about how, you know, learning occurs in a neural network. And he said, "The problem with the way you guys have been approaching this is that you've been looking for inspiration from biology to tell you how, what the rules should be for how the synapses should change the strengths of their connections. How the connections should form." He said, "That's the wrong way to go about it. What you should do is you should think in terms of how you can adjust connection weights to solve a problem. So you define your problem, and then you figure out how the adjustment of the connection weights will solve the problem." And Rumelhart heard that and said to himself, "Okay, so I'm going to start thinking about it that way. I'm going to essentially, uh, imagine that I have some objective function, some goal of the computation. I want my machine to correctly classify all of these images. And I can score that, I can measure how well they're doing on each image, and get some measure of lo- error, or loss it's typically called in, in deep learning. And, um, I'm going to figure out how to adjust the connection weights so as to minimize my loss, or reduce the error." Uh, and that's called, you know, gradient descent. And, uh, engineers were already, uh, familiar with the concept of gradient descent.

    2. LFLex Fridman1:09:53

       Mm-hmm.

    3. JMJay McClelland1:09:53

       Uh, and in fact, um, there was an algorithm called the delta rule, um, that had been invented by, uh, a professor in the engineering dep- electrical engineering department at Stanford, uh, Widrow, Bernie Widrow, and a collaborator named Hoff. I don't... Never met him. Anyway, so, so gradient descent in continuous neural networks with multiple neuron-like processing units was already understood, um, uh, for a single layer of connection weights. We have some inputs over a set of neurons. We want the output to produce a certain pattern. We can define the difference between our target and what the neuro network is producing, and we can figure out how to change the connection weights to reduce that error. So what Rumelhart did was to generalize that so as to be able to change the connections from earlier layers of units to the ones at a hidden layer between the input and the output. And so he first called the algorithm the generalized delta rule because it's just an extension of-

    4. LFLex Fridman1:11:05

       Got it.

    5. JMJay McClelland1:11:06

       ... the gradient descent idea. And interestingly enough, Hinton was thinking that this wasn't going to work very well. So Hinton had his own alternative algorithm at the time-

    6. LFLex Fridman1:11:20

       Mm-hmm.

    7. JMJay McClelland1:11:21

       ... based on, uh, the concept of the Boltzman machine that he was pursuing. So the paper on the Boltzman machine came out in, learning and Boltzman machines came out in 1985. But it turned out that back prop worked better than the Boltzman machine learning algorithm, um-

    8. LFLex Fridman1:11:38

       So this generalized delta algorithm ended up being called back propagation, as you say, back prop?

    9. JMJay McClelland1:11:45

       Yeah. And the, you know, probably that name is opaque to m- maybe what, what does that mean? (clears throat) It, what it, what it meant was that in order to figure out what the changes you needed to make to the connections from the input to the hidden layer, you had to back propagate the error signals from the output layer through the connections from the hidden layer to the output, to get the signals that would be the error signals for the hidden layer. And that's how Rumelhart formulated it. It was like, "Well, we know what the error signals are at the output layer. Let's see if we can get a signal at the hidden layer that tells each hidden unit what its error signal is," essentially.

    10. LFLex Fridman1:12:32

        Mm-hmm.

    11. JMJay McClelland1:12:32

        So it's, it's back propagating through the connections, uh, from the hidden to the output to get the signals to tell the hidden units how to change their weights from the input. And that's why it's called back prop. Yeah, but, uh, so it came from Hinton having introduced the concept of, you know, define your objective function, figure out how to take the derivative so that you can, um, adjust the connection so that they make progress towards your goal.

    12. LFLex Fridman1:13:05

        So stop thinking about biology for a second-

    13. JMJay McClelland1:13:07

        Mm-hmm.

    14. LFLex Fridman1:13:07

        ... and let's start to think about optimization and computation-

    15. JMJay McClelland1:13:11

        Yeah.

    16. LFLex Fridman1:13:11

        ... uh, a little bit more. So what about Geoff Hinton? What, um... You've gotten a chance to work with him and that little... (laughs) The set of people involved there, uh, is, is quite incredible. The small set of people under the, the PDP flag, uh, is just, given the amount of impact those ideas have had over the years, it's kind of incredible to think about. But-You know, j- just like you said, uh, like yourself, Geoffrey Hinton is seen as one of the not just, like, seminal figure in AI, but e- just a brilliant person, just a, like the horsepower of the mind is pretty, uh, high up there for him be- 'cause he's just a great thinker. So, what kind of ideas have you, um, learned from him, have you influenced each other on, have you debated over? What stands out to you? In, in the, in the full space of ideas here at the intersection of computation and cognition?

    17. JMJay McClelland1:14:10

        Well, so, um, Geoff has said many things to me that had a profound impact on my thinking, um, and he's written several articles which, um, uh, were way ahead of their time. Um, he, uh, (silence) he had two papers in 1981, just to give one example. (laughs)

    18. LFLex Fridman1:14:36

        (laughs) Yeah.

    19. JMJay McClelland1:14:38

        Uh, one of which was essentially the, uh, idea of transformers, uh, and another of which, uh, was a, uh, early paper on semantic cognition, which inspired, uh, him and Rumel Hart and me, uh, throughout the '80s and, uh, um, you know, still, uh, I think sort of grounds my own thinking about, um, the semantic aspects of, of cognition. Um, he also (laughs) , uh, in a, in a small paper that was never published that he wrote in 1977, you know, before he actually arrived at UCSD, or maybe a couple of years even before that, I don't know, uh, when he was a PhD student, he, he, um, described how a n- neural network could, uh, do recursive computation.

    20. LFLex Fridman1:15:40

        Mm-hmm.

    21. JMJay McClelland1:15:41

        And, um, uh, it, it was a very clever idea that he's continued to explore over time, which was sort of the idea that, um, when you, when you call a subroutine, you need to save the state that you had when you called it so you can get back to where you were when you're finished with the subroutine.

    22. LFLex Fridman1:16:04

        Mm-hmm.

    23. JMJay McClelland1:16:05

        And, and the idea was that you would save the state of the calling routine by making fast changes to connection weights, and then when you finished with the subroutine call, those fast changes in the connection weights would allow you to go back to where you had been before and reinstate the previous context so that you could continue on with the, the, the top level of the computation. Anyway, that was part of the idea, and, um, I always thought, "Okay, that's really..." You know, he just, he had extremely creative ideas that were, uh, quite a lot ahead of his time, and many of them in the 1970s and early, early 1980s. Um, so, uh, another thing about Geoff Hinton's way of thinking which, um, has profoundly influenced my, um, uh, effort to understand human mathematical cognition is that he doesn't write too many equations. And, uh, people tell stories like, "Oh, in, in the Hinton lab meetings, you don't get up at the board and write equations like you do in everybody else's machine learning lab."

    24. LFLex Fridman1:17:22

        Mm-hmm.

    25. JMJay McClelland1:17:23

        What you do is you draw a picture. (laughs) And, and, you know, he, he explains aspects of the way deep learning works by putting his hands together and showing you the shape of a ravine and, um, using that as a geometrical metaphor for the what's happening as this gradient descent process. You're coming down the wall of a ravine. If you take too big a jump, you're gonna jump to the other side, and, um, so that's why we have to turn down the learning rate, uh, for example. Um, and it, it, um, speaks to me of the, uh, fundamentally intuitive character of, uh, deep insight together with, um, commitment to really understanding, um, in a way that's absolutely ultimately explicit and clear, uh, but also intuitive.

    26. LFLex Fridman1:18:31

        Yeah. The- there are certain people like that. He's an example, some kind of weird mix of, uh, visual and intuitive and all those kinds of things. Feynman is another example, different style of thinking, but very unique. And when you, when you're around those people, for me in the engineering realm, uh, there's a guy named Jim Keller who's a chip designer engineer, he, s- it's every time I talk to him, it doesn't matter what we're talking about, just having experienced that unique way of thinking transforms you and makes your work much better.

    27. JMJay McClelland1:19:04

        Mm-hmm.

    28. LFLex Fridman1:19:05

        And that's, that's the magic. You look at Daniel Kahneman, you look at the great collaborations throughout the history of science, that's the magic of that. It's not always the exact ideas that you talk about, but it's the process of generating those ideas, being around that, spending time with that human being, you can come up with some brilliant work, especially when it's cross-discipline as it was a little bit in your case-

    29. JMJay McClelland1:19:29

        Yeah.

    30. LFLex Fridman1:19:29

        ... with Geoff.
11. 1:24:42 – 1:31:50

    Mathematics & reality

    1. LFLex Fridman1:24:42

       You write about modeling of, uh, mathematical cognition, so let me first ask about mathematics in general. Um, I... The- there's a paper, uh, titled Parallel Distributed Processing Approach to Mathematical Cognition where, in the introduction, there's some beautiful dis- discussion of mathematics. And, uh, you reference there, uh, Tristan Needham, who criticizes a narrow form of your mathematics by likening the studying of mathematics as symbol manipulation to studying music without ever hearing a note. So, from that perspective, what do you think is mathematics? What is this world of mathematics like?

    2. JMJay McClelland1:25:23

       Well, I think of mathematics as, um, a set of tools for exploring idealized worlds that, um, often turn out to be, uh, extremely relevant to the real world, but need not. Um-But they're worlds in which objects exist with idealized properties, and in which the relationships among them can be characterized with precision so as to allow the implications of certain facts to then allow you to derive other facts with certainty. So, you know, if, uh, you have two triangles and you know that there is, um, uh, an angle in the first one that has the same measure as an angle in the second one, and you know that the lengths of the sides adjacent to that angle in each of the two triangles, uh, the corresponding sides adjacent to that angle are also... have the same measure, then you can then conclude that the triangles are congruent. That is to say, they have all of their properties in common, and, and that is something about triangles. It's not a f- matter of formulas. These are idealized objects. In fact, you know, we build bridges out of triangles and, uh, we understand, uh, how to measure the height of something we can't climb by, um, extending these ideas about triangles a little further. (laughs) And, um, uh, you know, all of the ability to, um, get a tiny speck of matter launched from, uh, the planet Earth to intersect with some tiny, tiny little body way out in way beyond Pluto somewhere at exactly a predicted time and date is, is, is something that depends on these ideas, right? So-

    3. LFLex Fridman1:28:09

       Mm-hmm.

    4. JMJay McClelland1:28:09

       But, and, and it's actually, uh, happening in the real physical world that these ideas make contact with it, uh, in those kinds of instances. Um, and, um, so but, you know, there are these idealized objects, these triangles or these distances or these points, whatever they are, that, um, uh, allow for this, um, set of tools to be created that then gives human beings the, uh... It's this incredible leverage that they didn't have without these concepts. And, uh, I think this is actually already true when we think about just, you know, the natural numbers. Um, I always like to include zero, so I'm gonna say the-

    5. LFLex Fridman1:29:10

       (laughs)

    6. JMJay McClelland1:29:10

       ... the non-negative integers, (laughs) but, uh, that's, that's a place where some people prefer not to include zero but, uh-

    7. LFLex Fridman1:29:18

       No, we like zero here.

    8. JMJay McClelland1:29:19

       (laughs)

    9. LFLex Fridman1:29:19

       So natural numbers, zero, one, two, three, four, five, six, seven, and so on.

    10. JMJay McClelland1:29:23

        Yeah. And, and, you know, because they give you the ability to, um, be exact about, um, like, how many sheep you have. Like, you know, I sent you out this morning, there were 23 sheep. You came back with only 22. What happened?

    11. LFLex Fridman1:29:44

        Yeah.

    12. JMJay McClelland1:29:44

        Right? (laughs)

    13. LFLex Fridman1:29:45

        The fundamental problem of physics, how many sheep you have.

    14. JMJay McClelland1:29:48

        (laughs)

    15. LFLex Fridman1:29:48

        Yeah.

    16. JMJay McClelland1:29:49

        It's a fundamental problem of-

    17. LFLex Fridman1:29:50

        Life.

    18. JMJay McClelland1:29:51

        ... of human, uh, society that you damn well better bring back the same number of sheep as you started with.

    19. LFLex Fridman1:29:57

        (laughs)

    20. JMJay McClelland1:29:57

        Uh, and, you know, it allows commerce, it allows, um, contracts, it allows the establishment of, uh, records and so on to have systems that allow these things to be notated. But they, they have, um, an inherent aboutness to them that's, that's one p- at the... One and the same time sort of abstract and idealized and generalizable while at the other, on the other hand, um, potentially very, very grounded and concrete. And one of the things that, uh, makes for the, um, incredible achievements of the human mind is the fact that humans invented these idealized systems that leverage the power of human thought in such a way as to allow all this kind of thing to happen. And, and so that's w- what... Mathematics to me is the development of systems for thinking about, uh, the properties and relations among, uh, sets of idealized objects and, um, uh, you know, the, the mathematical notation system that we unfortunately focus way too much on is, um, just our way of expressing, uh, propositions about these properties.

    21. LFLex Fridman1:31:37

        Right. It's, it's just, just like we were talking with Chomsky and language, it's the thing we've invented for the communication of those ideas. They're not necessarily the deep representation of those ideas.

    22. JMJay McClelland1:31:48

        Yeah.

    23. LFLex Fridman1:31:49

        So what, um,

12. 1:31:50 – 1:42:28

    Modeling intelligence

    1. LFLex Fridman1:31:51

       what's a, what's a good way to model...... such powerful mathematical reasoning, would you say? What, what are some ideas you have for capturing this in a model?

    2. JMJay McClelland1:32:02

       The insights that human mathematicians have had is a combination of the kind of the intuitive kind of connectionist like knowledge that makes it so that something is just, like, obviously true, so that you don't have to think about why it's true. That then makes it possible to then take the next step and ponder and reason and figure out something that you previously didn't have that intuition about. It then ultimately becomes a part of the intuition that the next generation of mathematical thinkers have to ground their own thinking on, so that they can extend the ideas even further. I came across this quotation, uh, from Henri Poincaré while I was, um, walking in the, in the woods with my wife in a, a state park in Northern California, um, late last summer. And what it said on the bench was, "It is by logic that we prove, but by intuition that we discover." And so what, what for me the, the essence of the, of the project is to understand how to bring the intuitive connectionist resources to bear on letting the intuitive discovery arise, uh, you know, from engagement in thinking with this formal system.

    3. LFLex Fridman1:33:59

       Mm-hmm.

    4. JMJay McClelland1:34:00

       So, I, I think of, you know, the ability of somebody like Hinton or Newton or Einstein or Rommel Hart or Poincaré to, um... Archimedes is another example, right? So, su- suddenly a flash of insight occurs. It's, it's like the constellation of all of these simultaneous constraints that somehow or other causes the mind to settle into a, a novel state that it never did before and, and give rise to a new idea, um, that, you know, then (laughs) you can say, "Okay, well now how can I prove this?" You know, "How do I write down the steps of that theorem that, that allow me to make it rigorous and certain?" And so, I feel like the, the kinds of things that we're beginning to see, um, deep learning systems do of their own accord kind of gives me this feeling of, of, um, I don't know, hope or encouragement that ultimately, um, it'll all, uh, happen. Um, so, i- in particular as, uh, many people now have, have, uh, become really interested in thinking about, you know, neural networks that have been trained with massive amounts of text-

    5. LFLex Fridman1:35:52

       Mm-hmm.

    6. JMJay McClelland1:35:53

       ... can be given a prompt and they can then sort of generate some really interesting fanciful creative story from that prompt. Um, and, uh, there's, there's kind of like a sense that they've somehow synthesized something like novel out of the, you know, all of the particulars of all of the billions and billions of experiences that went into the training data that, that gives rise to something like this sort of intuitive sense of what would be a, a fun and interesting little story to tell or something like that. It just sort of wells up out of the, out of the letting the thing play out its own imagining of what somebody might say given this prompt as a, as a input to, to get it to, to start to generate its own thoughts. And, and to me that, that sort of represents the potential of capturing this st- the intuitive side of this.

    7. LFLex Fridman1:37:02

       Y- Yeah, and there's other examples. I don't know if you find them as captivating is, you know, on the DeepMind side with AlphaZero, if you study chess, the kind of solutions that ha- has come up in terms of chess, it, it, it is, it, it... There's novel ideas there. It feels very, uh, like there's brilliant moments of insight. And the mechanism they use, uh, if you think of search as, as maybe more towards good old-fashioned AI and, and then there's the connectionist, uh, neural network that has the intuition of looking at a board, looking at a set of patterns and saying, "How good is this set of positions? And the next few positions, how good are those?" And that's it. And those, that's just an intuition. Um-

    8. JMJay McClelland1:37:48

       Yeah. Yeah.

    9. LFLex Fridman1:37:48

       Grand, grand masters have this, an understanding positionally, tactically-... how good the situation is, how, how can it be improved, without doing this full, br- like, deep search. Um, and then maybe doing a little bit of the, what, uh, human chess players call calculation, which is the search.

    10. JMJay McClelland1:38:06

        Mm-hmm.

    11. LFLex Fridman1:38:07

        They're taking a particular set of steps down the line to see how they unroll. But there, there is moments of genius in those systems too. So that's another hopeful illustration that from neural networks can emerge this novel creation of an idea.

    12. JMJay McClelland1:38:26

        Yes. And I think that... You know, I think Demis Hassabis is, um... You know, he's spoken about those things. He, uh... I heard him, uh, describe a, a move that was made in, in one of the Go matches against Lee Sedol in this very- in a very similar way. And, and, um, it caused me to become really excited to (laughs) k- collaborate with some of those guys at, at DeepMind. Um, so I think though that what, what I like to really emphasize here is... One part of what I like to emphasize about mathematical cognition at least, is that philosophers and logicians going back three or even a little more than 3,000 years ago, began to develop these formal systems, and gradually the whole idea about thinking formally got constructed. Um, and, you know, it's preceded Euclid, um, certainly present in the work of Thales and others. And I'm not, uh, the world's leading expert in all the details of that history, but Euclid's Elements were the, the kind of the touch point of a, of a coherent document that sort of laid out this idea of a actual formal system within which these objects were characterized, and the, um, the system of, uh, inference that, um, allowed new truths to be derived from others was sort of like established as a paradigm. And, um, what, what I find interesting is the idea that the ability to become a person who is capable of thinking in this abstract formal way is, you know, a result of the same kind of immersion, uh, in, in experience thinking in that way that, you know, we now begin to think of our understanding of language as being, right? So, we immerse ourselves in, in a particular language, in a particular world of objects and their relationships, and we learn to talk about that, and we develop intuitive understanding of the real world. In, in a similar way, we can think that what academia has created for us, what, you know, those early philosophers in their academies in Athens and Alexandria and others, other places, uh, allowed was the development of these, uh, schools of thought, modes of thought, that, that then become deeply ingrained. And, you know, it, it becomes what it is that makes it so that somebody like Jerry Fodor would think that, um, systematic thought is the essential characteristic of the human mind, as opposed to a derived and, and an acquired characteristic that results from acculturation in a certain mode that's been invented by

13. 1:42:28 – 1:56:49

    Noam Chomsky and linguistic cognition

    1. JMJay McClelland1:42:28

       humans.

    2. LFLex Fridman1:42:28

       Would you say it's more fundamental than, like, language if we start dancing? If we, if we bring Chomsky back into the conversation, w- first of all, is it unfair to draw a line between mathematical cognition and language, linguistic cognition?

    3. JMJay McClelland1:42:48

       I think that's a very interesting question, and I think, um, it's one of the ones that I'm actually very interested in right now. Um, but I, I think the answer is, in important ways, it is important to draw that line, but then to come back and look at it again and see, uh, some of the subtleties and interesting aspects of the difference. Um, so if we think about Chomsky himself, um, he, uh, was born into an academic family. His father was a professor of rabbinical studies at a small rabbinical college in Philadelphia. Um, and he was deeply enculturated in, uh, you know, a culture of thought and reason.And brought to the effort to understand natural language, this profound engagement with these formal systems. And, um, you know, I think that there was tremendous power in that, and that Chomsky had some amazing insights into the structure of natural language. But that... (laughs)

    4. LFLex Fridman1:44:29

       (laughs)

    5. JMJay McClelland1:44:29

       I'm gonna use the word but there, the actual intuitive knowledge of these things only goes so far, and does not go as far as it does in people like Chomsky himself. And this was something that was discovered in the PhD dissertation of Lila Gleitman, who was actually trained in the same linguistics department with Chomsky.

    6. LFLex Fridman1:44:52

       Mm-hmm.

    7. JMJay McClelland1:44:54

       So, what Lila discovered was that the intuitions that linguists had about even the meaning of a phrase, not just about its grammar, but about what they thought a phrase must mean, were very different from the intuitions of an ordinary person who wasn't a formally trained thinker. And, well, it recently has become much more salient. I happened to have learned about this when I myself was a PhD student at the University of Pennsylvania, but, um, I never knew how to put it together with all of my other thinking about these things. So, so I actually currently have the hypothesis that formally trained linguists and other formally trained, um, academics, uh, whether it be linguistics, philosophy, cognitive science, computer science, machine learning, mathematics, um, have a mode of engagement with experience that is intuitively deeply structured to be more organized around, uh, the systematicity, uh, and, um, ability to be conformant with the principles of a system than, um, than is actually true of the natural human mind without that immersion.

    8. LFLex Fridman1:46:42

       That's fascinating. So, the different fields and approaches with which you start to study the mind actually take you away from the natural operation of the mind. So, it makes it very difficult for you to, (laughs) to be somebody who introspects.

    9. JMJay McClelland1:46:59

       Yes. And, you know, this is where, um, uh, things about human belief and so-called knowledge, um, that we consider, um, private, um, not our business to manipulate in others, we are not entitled to tell somebody else what to believe about certain kinds of things. Um, what are those beliefs? Well, they are the product of this sort of immersion and inculturation, uh, that is what I believe. (laughs) So-

    10. LFLex Fridman1:47:51

        And that's limiting?

    11. JMJay McClelland1:47:53

        It's, it's something to be aware of. (laughs)

    12. LFLex Fridman1:47:58

        Does that limit you from, uh, having a good model of some... oh, of cognition? I mean-

    13. JMJay McClelland1:48:04

        It can.

    14. LFLex Fridman1:48:05

        So, when you look at mathematical or linguistics stuff, I mean, what, what is that line then? What, um... So, so is Chomsky unable to sneak up to the full picture of cognition? Are you, when you're focusing on mathematical, uh, thinking, are you also unable to do so?

    15. JMJay McClelland1:48:23

        I think you're, you're right. I think that's a great way of characterizing it. And, um, I also think that, um, it's related to, um, the concept of beginner's mind. Uh, and, um, a-another concept called the expert blind spot. So, the expert blind spot is much more prosaic seeming than, than this point that you were just making.

    16. LFLex Fridman1:48:54

        Mm-hmm.

    17. JMJay McClelland1:48:54

        But it's, it's something that plagues experts when they try to communicate their understanding to non-experts. And that is that things are self-evident to them that they, they can't begin to even think about how they could explain it to somebody else, because it, it... like, well, it's just, like, so patently obvious that it must be true. And, um, you know, like, um, when Kronecker said, "God made the natural numbers. All else is the work of man," he was expressing that, that intuition that, um, somehow or other, you know, the basic fundamentals of discrete quantities being countable and enumerable and, you know, indefinite in number, um, was, was not something that had to be...... had discovered, um-

    18. LFLex Fridman1:50:15

        Hmm.

    19. JMJay McClelland1:50:16

        But he was wrong. It turns out that, uh, many cognitive scientists agreed with him for a time. There was a long period of time where there were, where, um, you know, the natural numbers were considered to be a part of the innate endowment of, you know, core knowledge or, you know, to use the kind of phrases that, uh, Spelke and, and Carey used to talk about what they believe are the innate primitives of the human mind. And, um, they no longer believe that. They, i- i- it's actually, um, been more or less accepted by almost everyone that the natural numbers are actually a cultural construction. And it's, it's so interesting to go back and sort of, like, study those few people who still exist, who, you know, don't have those systems.

    20. LFLex Fridman1:51:05

        Mm-hmm.

    21. JMJay McClelland1:51:06

        So, so this is just an example to me and where, you know, a certain mode of thinking about language itself or a certain mode of thinking about geometry and those kinds of relations so becomes so second nature that you don't know what it is that you need to teach.

Episode duration: 2:31:57