Tyler Marghetis: The Leaps of Human Imagination

We're excited to welcome Tyler Marghetis, Assistant Professor of Cognitive & Information Sciences at the University of California, Merced, to the show today. Tyler studies what he calls the "lulls and leaps" or "ruts and ruptures" of human imagination and experience. He's fascinated by how we as humans can get stuck in certain patterns of thinking and acting, but then also occasionally experience radical transformations in our perspectives.

Listen on Apple, Spotify, or YouTube.

In our conversation, Tyler shares with us some of his lab's fascinating research into understanding and even predicting these creative breakthroughs and paradigm shifts. You'll hear about how he's using AI tools to analyze patterns in things like Picasso's entire body of work over his career. Tyler explains why he believes isolation and slowness are actually key ingredients for enabling many of history's greatest creative leaps. And he shares with us how his backgrounds in high-performance sports and in the LGBTQ community shape his inclusive approach to running his university research lab.

It's a wide-ranging and insightful discussion about the complexity of human creativity and innovation. Let's dive in to our interview with Tyler Marghetis.

Transcript (from Apple Podcasts):

Welcome to Artificiality, Where Minds Meet Machines.

We founded Artificiality to help people make sense of artificial intelligence.

Every week, we publish essays, podcasts, and research to help you be smarter about AI.

Please check out all of Artificiality at www.artificiality.world.

We are excited to welcome Tyler Marghetis, Assistant Professor of Cognitive and Information Sciences at the University of California, Merced, to the show today.

Tyler studies what he calls the lulls and leaps or ruts and ruptures of human imagination and experience.

He's fascinated by how we as humans can get stuck in certain patterns of thinking and acting, but then also occasionally experience radical transformations in our perspectives.

In our conversation, Tyler shares with us some of his lab's fascinating research into understanding and even predicting these creative breakthroughs and paradigm shifts.

You'll hear about how he's using AI tools to analyze patterns and things like Picasso's entire body of work over his career.

Tyler explains why he believes isolation and slowness are actually key ingredients for enabling many of history's greatest creative leaps.

And he shares with us how his backgrounds in high-performance sports and in the LGBTQ community shape his inclusive approach to running his university research lab.

It's a wide-ranging and insightful discussion about the complexity of human creativity and innovation.

Let's dive into our interview with Tyler Marghetis.

Tyler, thanks so much for joining us.

We're excited to talk to you today.

Thanks so much for having me.

You start off by telling us what inspired you to study the lulls and leaps of human imagination.

I'm a bit prone to alliteration, so sometimes I say that I study the lulls and leaps of human imagination.

Sometimes I study the ruts and ruptures of the human experience.

And, you know, so across those cutesy framings, the idea that really animates me and my fantastic team in my lab is the observation that humans can really get stuck.

We get stuck in a particular way of acting, of thinking about the world, of relating to each other, particular political ideologies or conceptual frameworks.

They're really sticky.

But then every once in a while, you see these examples of radical transformation.

The historian of science Thomas Kuhn talks about paradigm shifts in the history of science.

So at the scale of, you know, entire disciplinary orientations to truth, you see these radical reconfigurations.

But even at the mundane level of our day-to-day lives, we've all experienced this.

We've fallen in love and maybe fallen out of love just as suddenly.

Or we've had a sudden realization about how to solve a really, really vexing problem in our garden, where we couldn't quite figure out the right plant to place there.

And so across these time scales, from, you know, the many centuries of artistic or scientific creativity to the mundane moment-to-moment of our daily lives, humans are marked by the capacity for incredible fixation in our understanding and then radical transformation.

I just think that tension between the lulls and the leaps or the ruts and the ruptures is a really uniquely human thing that cuts to the core of what it means to be a thinking person.

I find that just a fascinating way to sort of re-

put some different names around things that other tensions that we talk about, like explore versus exploit, for example, which has a lot of technical baggage that comes along with it, especially if you're a computer scientist or an evolutionary biologist or something like that.

What have you found out so far about why that happens?

Is it like the way that you could look at optical illusions, the rabbit-duck optical illusion, where you can't do both at the same time, you actually have to only focus on one, that multi-stability, seeing two things at once?

Is there something in common there, or am I just kind of imagining that?

Yeah, I love two parts of your comment.

One was the connection to the technical machinery of explore and exploit, which is totally one of the tools in our lab's toolbox for making sense of these really varied phenomena that show both stability and transformation.

And so you're right that actually one of the benefits of the strategy that we adopt, which is to be really agnostic about what falls under this big umbrella of lulls and leaps or ruts and ruptures, is that it lets us pick and choose between these different tools.

So we can draw from the mathematics of information theory and look at surprise and predictability, or the tools from computer science, looking at explore and exploit and sort of be a little catch as catch can in the tools that we use.

And then the other point that you made, and this is part of my happy spot, is you notice that we have a really similar phenomena across lots of different disciplines.

So vision scientists study the bias stability of illusions.

Scholars of religion study the experience of sudden religious conversions or deconversions.

And those folks aren't in conversation with each other.

And I think my sort of sweet spot, my special sauce that I add to the conversation, is drawing out those surprising connections despite the really radical superficial differences.

And then to start to ask, why do we see a really similar process, really similar dynamics showing up in these really, really different systems?

And that's one of the reasons why I really love adopting a complexity science perspective, because that invites me to think about the more structural or dynamical properties that are shared between, say, a research lab or an artistic movement and a natural ecosystem, where in all of those you have these periods of deep stability, these regimes, and possibilities for critical transitions or transformation, and sort of drawing out those maybe more abstract, dynamical, structural connections sort of gives me a special microscope to figure out, okay, why are illusions bistable?

Why do people have religious conversions?

Why did quantum physics replace, you know, the traditional perspectives that existed before?

Complexity science gives us a language, right?

It allows us to put some names on some of these common things that we can say we can take sort of network dynamics from social media and apply it to an economy or something like that.

What are the tools that you find the most productive?

Are they things like phase changes or feedback?

What are the complexity features, if you like, that you find are the most productive for going between these disciplines?

Yeah, I think all the ones you mentioned absolutely are technical friends of mine and my collaborators.

And a lot of those tools are drawn from ecology, where there's a rich set of formal tools for understanding the stability and transformation of natural ecosystems.

Also from physics, where there's a long history of looking at critical transitions in natural phenomena, sort of magnetization dynamics in simple models of magnetism.

So all of those are absolutely part of it.

And so one, I think, really exemplary instance of this kind of transformation shows up in human creativity.

And so even though creativity, I think, is just one small slice of this larger phenomena of ruts and ruptures, it's sufficiently bounded that it sort of gives us a bit more traction to think about, OK, what are the different ways that we could look at this phenomenon, say, creativity?

And with creativity, you can describe it using different complementary formal perspectives.

So you could look at it from the perspective of information theory.

And there, what it means to be creativity is to generate a new idea or solution that's surprising relative to the space or the range or the distribution of ideas or solutions that you had considered before.

So now that offers us formal tools for quantifying how surprising a new proposal might be.

If you think of creativity as a process of combining previously unassociated ideas or approaches, now you're in the realm of combinatorics.

And you can sort of use the formal tools of modeling the different possible combinations that are available to understand how really transformative a new creative insight really was.

Or you could draw from geometry.

And you can think of something like musical improvisation or artistic creativity or even mathematical breakthroughs as a movement to a new region in a space of possible ideas.

So now we're thinking in terms of high dimensional spaces, a latent space of ideas or images or solutions, and a career or a history as a trajectory through that high dimensional space.

And that's a very geometric or spatial perspective.

All of those are very formal, right?

So the nice thing about this complexity science perspective is it's not just a nomenclature.

It's not just a set of descriptive tools.

It's also a set of really, really powerful formal tools that allow us to go beyond just a mere naming to a much more deeper formal understanding.

When we talk to people about complexity, people just love the explanatory power of that naming.

The world just makes sense to them in a different way.

And I think it's just hugely powerful.

How do you go beyond that?

You've got these formal tools, the modeling tools, the mathematical tools.

You can move beyond just the metaphorical descriptions.

And how do you do that and how are you doing that in creativity?

Yeah, and this is one place where AI enters, not as a phenomena of interest for me, but as a tool, a sort of new tool for making progress on these problems that have been animating me, really for the last few decades since I was a teenager.

I've been sort of thinking about this process of radical transformation of human thought.

And so the sort of latest batch of really powerful AIs, one of the things they're really good at doing is taking what might seem to be a really intractable input or text that it's trying to grapple with, and it takes that text and it embeds it in some space, a space that it's learned to use to make sense of, say, all the possible texts that it can encounter.

And you have AI tools like variational auto encoders, or there are these various flavors of AI that take this approach of finding some nice abstract space where now every possible text or an entire book or maybe a painting is just one point in that space.

And what that allows you to do, what allows the AI to do, is to say, okay, what if we moved just a step over in that space and we sort of see what that image is?

And that's how you create these really, really beautiful visualizations that I'm sure you've seen online where you have one painting and another and then they morph from one into the other in this really fluid and deeply uncanny and to me unsettling way.

That's the strategy that's being used there.

Those two paintings are embedded in a space and then the AI just sort of wanders in that space from one point to the other.

We're using those tools to make sense of, for instance, the entire sweep of Pablo Picasso's career.

And so what we're doing is we're using an AI that was trained to just recognize objects in the world.

It's not trained on art.

It's not an art historical AI.

It's just a run-of-the-mill vision AI.

And we say, look at Picasso's paintings and just try to make sense of these.

And as a result, we can then take each of his paintings, each of his scribbles, each little napkin doodle that he's had, these have all been digitized by incredible art historians, and look at his entire career, not as a bunch of singular accomplishments, these great paintings, but as a trajectory, a path through this latent space of possible paintings.

And AI allows us in a really data-driven way that's agnostic theoretically, or agnostic about the history of art, to say, okay, what does that path look like?

And then we could say, is he doing exploration followed by exploitation?

That explore-exploit trade-off that computer scientists have identified as one fantastic way to get out of local minima, local solutions that aren't so great, and to find new possible solutions.

Or maybe he's doing something entirely different.

And AI gives us a tool to think about these really, really complex phenomena in more tractable ways by taking advantage of the kind of data reduction that AI is doing under the hood.

And so that's one way in which we go beyond just naming things to actually saying, okay, yes, critical transitions involve a change in the underlying basin of attraction.

Like, okay, no, let's actually look at what basin of attraction Picasso was in when he was painting everything blue for five years and then see whether there's a loss of stability when all of a sudden he tips over into something like early Cubism.

That is so cool.

That is just so cool.

I'm reading a book at the moment that's tracing Picasso and Einstein more verbally, not there's no AI.

This was written a long time ago.

But doing the same kind of thing.

Are you doing creativity outside of art in that respect and following the same kind of process?

Are you finding any patterns in creativity across vastly different domains as a result of being able to use these AI tools?

That's exactly the hope.

That's the thing that we're trying to do.

Although, I'll just say really quickly, just because they didn't use AI, they were probably using NI, natural intelligence, which in some ways is even more advanced, and it took many millions of years to evolve.

So let's not poo-poo on the author before using AI.

I'm curious if you have looked at innovation, say, in the technology industry.

We spent a lot of time in that space.

We're deeply imbedded in the AI space.

And I like looking at it sort of as this moment relates to other technological advances over history, the dawn of the Internet, thinking about the mobile and social eras, and these, what I think commonly would be considered leaps and changes, sort of total paradigm shifts.

Have you looked at those kinds of shifts, and what do you see, and or can you speculate a little bit about how you would apply your thought process to better understand those changes?

What you're really asking is, will this rabbit device actually be the iPhone killer?

I'm not asking about the rabbit device specifically, no.

But I am trying to understand how you can look back in history and understand and apply this to look at it, like you're saying in terms of Picasso, and then my interest then becomes in the predictive power of this kind of analysis, when you might be able to identify when a leap might be happening soon, that kind of thing.

Yeah, one of the strengths of these tools, of this language, is it allows you to recognize connections across art and technology, or technology on the time scale of decades, centuries, millennia, and the kinds of maybe formal breakthroughs that happen on the scale of minutes or hours by the mathematician or the engineer sort of working in their office.

And so then the question is, okay, are there regularities that we see across these time scales?

And the answer is absolutely.

We absolutely do.

And in some cases, they do give us some predictive power.

So here's an example.

So I mentioned this combinatoric perspective on creativity, or the idea that creative breakthrough involves combining together pieces of your toolbox that previously hadn't been connected.

And that's how this really fantastic physicist turned multidisciplinary complexity scientist Hye Jin Yoon at Northwestern has approached the study of the history of patents and technology.

And so what she's shown is that sort of breakthroughs in technology can be quantified as sort of low probability connections between previous patents or ideas.

So she explicitly adopts this combinatorical formalism, this framework, and is able to then quantify the history of technological advancement by using all the data that's available in patent filings.

Now, the cool thing is, in my lab, we've used a very, very similar approach to study creative breakthroughs not on the scale of formal technological innovations over decades or generations, but on the scale of minutes and hours in the seminar room of the mathematician.

And so what we've done is we've put video cameras in the natural habitat of the mathematician.

And I think of the mathematician really as the prototypical, the sort of canonical example of the creative thinker.

They're dealing with a realm of ideas that are often completely unmoored from their empirical experience and trying to forge new paths through this sort of imagined realm of ideas.

It's just fantastic.

And so to understand them in the natural environment, we put video cameras in their own home departments, and then we filmed them while they worked hard trying to sort of have a creative breakthrough on a peculiar vexing problem.

And the cool thing is, if you look in the minutes leading up to them expressing a breakthrough, the minutes leading up to them saying, aha, Eureka, or like, oh, I finally have it, minutes before they even express themselves that they've had the breakthrough, you start to see a gradual increase in surprising connections between inscriptions or ideas or diagrams.

And you actually see this in their bodies.

They start shifting their eyes from an equation on the top left of the blackboard to a diagram on the bottom right.

And they had never really connected those before, but you see that shifts in their eyes start making a connection between those ideas, suggesting that maybe there's more of an explore element coming in instead of an exploit of the same set of cycles of ideas that they were in before.

And that exploration ramps up.

You get these more and more surprising connections until they reach this tipping point and they have this complete radical reorganization of how they're thinking about the problem.

And so whether you're looking on the time scale of decades or generations and engineers making these technological breakthroughs that generate new patents or the time scale of minutes and hours and days of the mathematician working on maybe sort of a sort of middling problem, in both cases you see that great breakthroughs are sort of almost defined by these novel connections or even prefigured by those novel connections.

So that you can predict an impending radical transformation in formal understanding, either technology or mathematics, by the presence of these increasingly surprising connections across previously unconnected ideas or facts.

What do you get from that study about what their conscious experiences, how they feel as they go through that?

Like, where's the struggle?

Where's the bit where the difference between the breakthrough and not the breakthrough is actually the tenacity to kind of keep wrestling before there's this, essentially, what you almost described as a phase change?

Yeah, I mean, yeah, right, the phenomenology, right, the subjective experience of radical transformation in humans is really what makes these kinds of phase transitions different from, say, phase transition in physics between, you know, say, water and ice.

Or in natural ecosystems, like a lake ecosystem that goes from a sort of fish-dominant clear water regime to a really sort of turgid, algae-driven regime.

In the case of the lake and the magnet or the ice cube, those systems don't have this sense of struggle or hope or aspiration in the way that the mathematician or the jazz musician or the technological tinkerer does, right?

They get frustrated.

They get annoyed.

And we really don't have a good sense of how that fits into this larger picture of critical transitions, right?

This cross-domain theory of ruts and ruptures that we're trying to articulate, we don't really have a good sense of how to incorporate that human element.

And that's one that really excites me, because in some ways, I think that's the key ingredient that makes humans so much, so powerfully capable of that radical transformation, is that they can hope for that.

A lake as a whole doesn't hope to switch from one regime to the next.

An ice cube doesn't dream of becoming a puddle, but a mathematician can aspire to radically new reconfigured understanding of a realm of mathematical ideas.

Or Picasso can intentionally try to break out of whatever stylistic rut he's working within to land somewhere else.

Maybe they don't know where they want to land, but somewhere else.

And that seems like a really uniquely human feature of the creative experience.

And gosh, I wish we could have a consciousness scope that allows us to look inside the hopes and feelings of the mathematician and the artist.

We don't have that device yet.

And I think once we do have a better sense of what that experience is like, it really will advance not just our understanding of human creativity and critical transitions in humans, but our science of critical transitions more generally.

I'm so glad we're having this conversation because in the last hour, I've been wrestling with an article for our subscribers where I clearly have been in a rut.

And it's about some sort of healthy ways of using chat GPT, essentially.

And it sort of strikes me that one of the ways to recognize that you are in an unhealthy relationship with the chat GPT is that you're in a rut.

You keep asking the same thing over and over again.

You're not really getting what you want.

So being really conscious about, I'm in a rut now, I actually have to change my strategy, either disconnect altogether or change strategy with the tool itself.

But this sort of rut rupture is a nice way of thinking about something that you can choose.

You can choose to rupture things.

Yeah.

And that's one limitation.

I don't think chat GPT is particularly good at inducing those kinds of ruptures.

No, you have to do it yourself.

It would be nice if maybe one day it could.

But then of course, would we not like it?

And of course, that's not part of the commercial strategy.

Do you have tools to help understand why some novel combinations are successful breakthroughs and others end up being dopey ideas and duds?

Yeah, I don't think there's a solution that comes from within the world of the ideas themselves.

What makes them successful is actually the broader ecosystem of possibilities, the sort of larger cultural context that makes those things, makes a new technology sticky and attractive or a new idea particularly applicable.

And that's part of the challenge of studying these kinds of radical transformations is that to understand the transformation, you often need to look within the system that is undergoing the rupture.

But to make sense of whether that rupture is healthy or successful or lasting, you need to zoom out to the larger system within which the transforming system is embedded.

And this is the joy and the challenge of studying many complex systems is that you have structure and dynamics at all these nested scales.

And just as in one classic example of critical transition, religious conversion, people who suddenly go from one way of thinking about the nature of the divine to the other, that can be an incredibly positive social experience where all of a sudden they find a new community.

Or it can be fantastically and devastatingly isolating.

And in both cases, it might be the same kind of conceptual change within the person, but the larger social network or the larger cultural milieu in which they're embedded is going to make all the difference.

And so when we study the mathematicians, for instance, we don't actually even evaluate whether their breakthrough is good or not.

And sometimes they have this breakthrough and an hour down the road, they're like, and that was a false lead, darn.

And what I suspect is actually in both cases, the successful and the unsuccessful breakthroughs, you'll see really, really similar internal dynamics to the mathematician.

You'll see these same precursor, dynamical indicators of a loss of resilience or an increase in exploration.

And then whether it works or not, well, that's sort of left to the rest of the system.

That's an interesting way to start sort of separating parts of the innovation process and thinking about innovation as a more complex system than as a process per se.

I mean, you think about the best that we do in innovation at the moment is in many ways, we're either talking about the lone genius who has the brilliant idea, or we're putting diverse groups of people together and hoping something sort of magical happens with either data diversity, cognitive diversity, life experience diversity, whatever.

And when you think about what's really happening there, it's really a blunt way of saying either the individual that is, the lone genius also has a good nose for the overall system, or we're going to bring the overall system and proxy it with just a lot of people involved.

The way you described that previous example of the mathematician and separating out the system, that suddenly feels like such a crude way of thinking about innovation as a linear channel, and maybe flipping it on its side and saying, let's think about this more about the nested hierarchies and the multiple levels and scales, and thinking about it literally like that gives you different strategies for the broader system versus the smaller group.

Totally, and it invites you to think about the rightful role of different components of the system.

I'll give you an example.

One classic account of the nature of human creativity, this is from psychologist Campbell in the middle of the 20th century, who was inspired by Darwin.

He says, human creativity probably just involves, like evolution by natural selection, a process of random generation and selective retention.

Random generation is you throw in as many ideas as possible.

That's the equivalent to mutation in Darwin's theory of evolution by natural selection.

And then you have selective retention, which is where you keep the ideas that are good.

Or maybe your community keeps the ideas that are good.

Or maybe your discipline or your working partner, or maybe just sort of the nature of the problem tells you, yeah, nice try, but that's not actually going to get you to the moon or whatever you're trying to accomplish.

And so by separating that process of generation and then retention, you sort of begin to sort of think maybe more strategically about where different components of the creative process should actually be situated within the larger system.

That's often the motivation for a brainstorming session, right?

No wrong answers.

Let's get all the ideas out.

But we sometimes forget that for an ecosystem or a social system to be really productively innovative, to be really productively creative, you need to have different components of the system at different scales playing those different roles in a sustainable way.

This is why really, really successful institutions or universities have tea times or lab structures or talk series that sort of work on different timescales of bringing ideas at different stages of development to the prying critical eyes of their colleagues.

There's the nighttime science, as a great biologist once said, where you sort of generate your crazy ideas.

There's the sort of small group conversations where maybe you bring together a diverse group of people, you generate something really awesome.

And then there's the increasingly critical agonistic encounters with your colleagues or even your enemies.

And those happen on different spatial and temporal scales that are tough to reproduce in a completely artificial setting like a brainstorming session in a closed conference room.

One of the things you say about your lab is that, you know, very inclusive, anti-racist, which is just having it there on your homepage, the declaration of your values in that respect.

How are you seeing, how do you take the students that you have through a process where there's a lot of potential conflict, because you need a certain amount of conflict to argue out an idea?

Do you see something different that you're bringing to this that you don't see in other labs?

Gosh, yeah, I don't think I'm a great muse of social justice.

You know, I'm just inspired by the folks around me that I see doing really great work in the community.

And, you know, I bring tools to running my lab, really I think from two communities that I've had the great benefit of being part of.

And one is the community of high-performance sport, especially combat.

You know, in a past life, before deciding to become an academic, I was on the Canadian National Wrestling Team.

I was the alternate for the Beijing Olympics.

And devoted my life to, as I described it, to an academic colleague once, breaking the wills of men.

That was my job.

And so, you know, that sounds like an especially toxic hobby or profession or calling to have.

But that community was incredibly supportive while being devoted to really the highest level of performance.

And there you have folks coming from very different backgrounds training together to be the best in a really unforgiving endeavor.

So the worlds of mixed martial arts and wrestling, it's just you and your opponent.

There's no excuses.

You know whether you won or lost.

And so you have to combine both a supportive context where people can have these deeply uncomfortable training sessions or competitive encounters while still thriving as humans with the radical honesty of the nature of the encounter of combat.

So I think that has sort of inspired and formed in some ways my approach to running a lab and being a professor, this combination of radical candor of honesty with an understanding that the development of high performance takes a long time, right?

There's a cultivation required that's going to stretch over decades.

And so that's one place I sort of draw insights from there to sort of really support my really quite diverse lab community, diverse on lots of different dimensions.

And the other is my experience as a queer man.

So towards the end of my wrestling career, I came out as gay and had never really even really met gay people until then and suddenly found myself in this incredibly supportive, beautiful and justice-oriented community that really thought about how to support the folks within this community, community of gay, lesbian, trans, queer folks who often were incredibly marginalized or had been abandoned by their families.

And there's a real ethos of support within that community.

And I've tried to sort of take lessons from there about how we support ourselves in the queer community to try to create a supportive community in my own lab.

It's almost like either ends of the spectrum around sort of the rupture, repair.

That was one of the things that we learned in relationships, right?

And those communities, you're trying to be really high performance, but also really supportive at the same time.

There's that same tension.

Is there anywhere that we don't have these tensions?

And have you, what have you learned about, is there a way to get ideas out of one person's head and into another person's head faster than just sort of, is there something in that tension that allows us to actually transfer ideas faster between humans?

Because that always seems to be one of the kind of critical gating items with creativity and innovation is there's that tension too between what I know and what I sense and what I feel and what I see is the future versus what you see.

Yeah, so I actually want to push back on that.

I love that you brought up speed because I think actually one of the key ingredients for human creativity is slowness and isolation.

And you see this in a lot of different places.

So one, there's a long line of results in the formal modeling of innovation that show that if people start exchanging ideas early on, you very, very quickly get stuck in the consensus view that might be the majority approach but is often very uncreative.

On the other hand, if you let people develop their own perspective in isolation and only reconnect them towards the end, you're much more likely to get one of those peripheral outsiders to have some incredible breakthrough that they wouldn't have pursued if they had been rapidly coupled in the exchange of ideas with the rest of their community.

One of the fantastic PhD students in our department, Cody Moser, has been modeling this process, and one of his big findings is these peripheral agents, these folks that are on the outside, he calls them loser nodes.

The folks that are on the periphery of the social network, they're often core components for creative breakthroughs because they're pursuing unpopular ideas at a time when everyone else might be pursuing with a herd mentality what seems at first to be the most productive approach.

Unfortunately, that does lead to a certain amount of inequality in the system.

This is one of Cody Moser's results, that leads to an increase in extremes where you have some people that are getting all the benefits, the outsiders mostly not getting anything.

Every once in a while, they have the breakthrough.

And so again, you see this tension between wanting a kind of justice, you know, the tide lifts all boats approach with also the understanding that a certain amount of intellectual inequality, isolation and slowness can be really beneficial for these kinds of radical breakthroughs.

And actually, I think that's one of the things that makes us special as creative agents is that our brains are stuck in bodies and our bodies are slow.

Our bodies move slowly.

And I think that tension between the rapid time scale of thought, the millisecond time scale of firing neurons, and the second and minute and day time scale of changes in our bodies is really actually a core component of humans' ability for creativity.

Just as one example, the mathematicians we've studied, in addition to making these really novel connections, one thing they do before a creative breakthrough is they step back.

They wander away, and they gradually step back from what they're working on.

They literally take a walk.

And that is a slow process relative to the rapid fire pace of their thoughts.

And I think that tension between the slowness and the speed really creates the perfect environment for really kind of radical reconfiguration.

Is there any way to frame or quantify the isolation and the speed that drives success?

Is it different based on domain?

Is it different just based on person or the kind of breakthrough?

Or is there any sort of generalization that you can make there?

Yeah, I think the particular way that it's quantified is going to depend on the discipline or the practice.

One tool for quantifying this is using the tools of network science.

So you could quantify how central or peripheral a particular person is within the larger social network.

That's the approach that Cody Moser was using in his work with Paul Smaldino.

But you could also use informational measures to look at how much information is being exchanged between people.

So quantifying the rate of exchange and in a work context, you can sort of look at how many Slack messages or emails are being exchanged.

Use that to infer the network of informational exchange and then see, do you have a fully connected network?

If so, that's likely to be a very productive network, but not a transformative network.

You need these pockets.

You need that outsider that hasn't been part of the main conversation because they're off pursuing their own wild pursuits, only to find out, long forgotten, that the path they took is actually the most transformative one.

Is there a trade-off?

This makes so much sense.

I mean, it really does.

But you can see the next trade-off, which is we've all been in places where there's someone who is clearly slower, clearly isolated, doing really good work.

And in one scenario, people like them and trust them.

So when they come back to, if you like, to the group, the culture that they're bringing their idea to is more open to their ideas versus someone who actually no one really likes very much, for whatever reason.

And their ideas are no good, because they're not liked.

It's so unfair, but it's sort of such a factor of a social network.

How would you think about mitigating those two?

Because their lone genius that doesn't have a very good idea for whatever reason, but is liked, they can skew one way versus, you know, I'm thinking like this is a false positive, false negative kind of situation where you still want the good idea, but you recognize that it's coming in through this cultural frame.

Absolutely.

I think you're right that we all exist within multiple laminated, layered social networks, right?

We have the social network determined by how much we communicate with each other.

We have the social network determined by how much we like each other, maybe by shared interests.

In network science, they call these multiplex networks, where you have basically one network made up of nodes or people, and then you have another network, the same nodes and people, but they're connected in different ways.

And you're right to understand the breakthroughs that happen at the level of ideas, right?

The network of exchange of insights.

You have to take into account the other networks that the people are embedded in, right?

The network of care and interest, the network of confidence, how much you trust people.

And it's only recently, actually, that network scientists and social scientists that use these tools from network science have really been trying to take into account the multiple layers of connection that have these sort of cross-layer effects.

Now, how do you mitigate the possible disastrous effects of the cranky-faced person with all the great ideas and all the bad vibes?

And I think the practice of science has some lessons for us all to learn from.

Science has some really quirky characters, really fun people to hang out with, but also some people that are so smart that I do not want to have dinner with.

I don't want to sit down and share a beer with them.

But if you look at the way that ideas are transmitted in science, they're in this really stripped down format of the published paper.

Now, published paper and the entire industry of academic publishing is broken for so many reasons.

It's slow, it's exclusive, it's expensive, it's a racket.

But I do think there's a great insight, which is that when the presenter of the idea is reduced to only a name listed under the title, it allows the ideas to be evaluated on their own terms.

And that in some ways is one of the benefits of email and correspondence.

And so there's going to be these tradeoffs between the bandwidth of the exchange, how much information is being presented, but also the dangers of other aspects, other layers of that multi-layer cake of the social network sort of coming to bear.

Bezos' idea for his two pages or whatever that just looks smarter every day, really, doesn't it?

Yeah, well, I'm having reflections back to I worked at Apple 20 years ago, and there was much less communication.

I mean, we had email, obviously, but not the rapid fire slack.

And Steve ran the company in a way that there were many, many, many, many isolated teams.

And I'm sort of, this is a fascinating thread to pull to help understand why that was such a creative period of time and why that company in particular stood out.

You know, there's sort of the legend and the lore that Steve organized the company in a way that nobody else, no one but him really knew what everybody else was working on as a management technique and a power structure thing.

But it also clearly created these isolated teams that could go off and try something really creative and crazy without it getting squashed by whatever other narrative was happening throughout the organization.

And we could dig into that with Cody for hours and hours, but I find it fascinating to think about that and to have these moments of going, oh, I remember that one really extraordinary leap that happened, and that was one person or a small team that was off someplace behind a locked door.

And often people didn't even know these people existed until it was ready to be unveiled.

And that's a really interesting way of thinking about it, that this level of isolation sometimes can be the thing that drives great leaps.

Isolation is graded, right?

It's not all or nothing.

And if you look at the history of scientific or artistic innovation, very seldom was anyone actually truly isolated.

I mean, the myth of the lone genius is exactly that.

It's a myth.

But maybe folks were interacting with people in their town.

On the daily, they would have lunch.

They would go to each other's studios, go to each other's labs.

And then on a slightly slower time scale, they would interact with folks within maybe the same language community that they can exchange letters with.

And then on an even slower time scale, they were interacting in a really mediated way with, if you're talking about visual artists, with the paintings that finally made it to the great salons or the galleries.

Those are all forms of interaction and communication, but they operate on really different time scales and have increasing and decreasing as you go up and down that cline of slowness, isolation.

So tight coupling at the local levels, more isolation at the higher levels, and I think that mix of some people are highly connected, and then you're moderately connected with other folks, and then very, very slow contact with other folks, there's some optimal mix there that allows for both this incredible stability, these ruts, that I think are really healthy for, as a computer scientist would say, exploitation, you dig down to where you are, while also allowing for exploration at all these different nested time scales of other regions, other approaches, other possibilities.

Thank you so much for taking the time to talk to us.

This has been fascinating.

The time just flew.

Now we have a lot more things to talk about tonight over dinner.

But no, seriously, this has really been eye-opening, and appreciate you taking the time to talk to us.

Such a pleasure.

Great chat.

If you enjoy our podcasts, please subscribe and leave a positive rating or comment. Sharing your positive feedback helps us reach more people and connect them with the world’s great minds.

Subscribe to get Artificiality delivered to your email

Learn about our book Make Better Decisions and buy it on Amazon

Thanks to Jonathan Coulton for our music