Photo-illustration of Andy Clark
Illustration by Fanny Luor

The mind at work

The Mind at Work: Andy Clark on our brain’s endless entanglement with technology


Published on March 22, 2021

What is the limit of your mind? Is it the biggest number you can calculate? The number of facts you can remember from school? The wildest thing you can imagine? Your hardest feat of concentration?

Let’s look at it a different way. Think about all of the tools you use, the places you inhabit, and the people you share your life with. Now imagine the sum of these as the edge of your mental perimeter. Welcome to the concept of the extended mind, as proposed by philosophers Andy Clark and David Chalmers in their 1998 essay of the same title. Coincidentally, that was the same year that Google was founded, itself a milestone of mind expansion.

Clark developed these ideas further in his 2008 book, Supersizing the Mind. In the forward, Chalmers writes, “Clark is a connoisseur of the myriad ways in which the mind relies on the world to get its work done.” Although at first glance his ideas may seem sensationally mind-blowing, they are indeed practical explorations of how we actually do what we do.

Clark himself is a bit of a punk rock philosopher, not in the sense of being angry or alienated, but of being a provocateur. In his words, he’s trying to “pull the rug out from under intellectualism.” He’s a fan, too, of cyborgs and jurassic tech, but his interest goes beyond sci-fi and kitsch. “The role of technologies as either mind-enhancing or failing to be mind-enhancing,” he tells me, “is clearly something that I’m very keen to understand better.” Although he’s been at it for more than 30 years, he admits, “the fundamental principles are remarkably elusive.”

Extending the mind all over the place

Philosophers are heady by definition, but what’s unique in Clark’s perspective is his interest in how actual bodies move through actual worlds. His books and talks are filled with motivating examples drawn from a wide range of research and everyday experience.

One of my favorite, which surfaced in his 1996 book, Being There, concerns the swimming prowess of the bluefin tuna. Bluefins can swim very fast, able to leap out of the water at more than 40 mph. But fluid dynamicists at MIT concluded that their performance outpaces their basic physiological power by a factor of seven. The secret, it turns out, is that bluefins have learned to find, or use their tails to create, little vortices they can swim into to blast off rapidly. They turbocharge their travel by subtly changing their environment.

Clark’s point is that we do much the same with our iPhones. We can travel faster by Googling our route or calling a Lyft. The information we look for on our phones doesn’t exist in our brains. And much of it doesn’t even exist on our phones until we request it through the internet. So our minds aren’t just extended, but actually distributed throughout the world.

Our model of the world is, in many cases, the world itself.

This goes beyond offloading our brain’s work into devices like phones or calculators. Clark cites road signs, architecture, sketchbooks, and crucially, other people as ways that we constantly recruit the world outside our skulls for our own purposes. Our model of the world is, in many cases, the world itself.

Still, this emphasis on individual examples, singular extensions of the mind, belies the recursive complexity of all of it happening at once. “An awful lot of my work has actually been rather individualistic,” Clark says, “it’s been very much about someone on their iPhone, as opposed to a larger collective.” But the bigger picture is far messier and ill-defined. “There is another kind of extended minds scenario,” he continues, “that because it’s transient, you don’t actually own all of it, other people own bits of it, and some of it is owned by no one.” 

Knowledge work organizations are obsessed with questions of ownership precisely because it’s so hard to define. Even novelists and software engineers—Cal Newports heroic “deep workers”—meld the mental inputs of extended teams. Hardly anyone writes code from scratch anymore. We adapt libraries and packages written by others to our present need.

Although Clark acknowledges that it’s easier to make the core arguments for extended minds using these simple tech examples, he admits that this approach “trades on our (mistaken) idea that mind somehow starts off as an internal thing that then gets augmented with a bit of tech, rather than the even more challenging (but I think correct) idea that brains are—from the get-go—organs for the creation of extended systems that get things done.”

The human brain certainly exhibits hierarchy, often conceived as the frontal cortex at the top and the brain stem at the bottom. Although there is structural depth reminiscent of a government bureaucracy, the brain’s actual operations are flatter and more like a startup tech company. The challenge of describing something so fluid and anarchic led Clark to collect a large set of proof points without landing on any kind of unifying theory. What did seem clear to him early on was that most prevailing theories of mind contained a central controller somewhere, and that eventually he would need to “eject the homunculus from the machine.”

Tuning up the engine of prediction

Clark’s working methods are very social. And like Clark himself, they are both ambitious and informal. He’s been able to secure substantial grants that support multi-year working groups devoted to exploring open-ended topics that span philosophy, neuroscience, robotics, cognitive science, and lots more.

After finishing Supersizing the Mind, Clark became interested in predictive processing, a theory in neuroscience that emphasizes how the brain constantly updates its model of the world to predict what its senses will perceive. As discussed in my conversations with Karl Friston and Lisa Feldman Barrett, the intuitive idea that our brains “take in” the world, making sense of it before we act, is certainly wrong. Neuroscientists refer to the incoming sensory stream as “bottom-up” processing and the prediction of those inputs as “top-down.” In the current view, the twin interests of survival and energy efficiency lead us to predict what we expect to encounter in the world and only pay attention to the difference.

Prediction, for Clark, became a vital clue that brought together the ever-extending mind with the mobile yet finite body. In the preface to his 2105 book, Surfing Uncertainty, he made this connection explicit: “A skilled surfer stays ‘in the pocket’: close to, yet just ahead of the place where the wave is breaking. This provides power and, when the wave breaks, it does not catch her. The brain’s task is not dissimilar.”

Technology turns out to be a big part of the prediction story as well. The brain’s predictive machinery is always trying to identify patterns and regularities in the world to make its job easier. As unpredictable as the technological world seems as a whole, on an individual level technology makes our lives much more regular and knowable. “The world thus revealed is a world tailored to human needs, tasks, and actions,” he writes. “It is a world built of affordances—opportunities for action and intervention.”

According to Clark, artificial intelligence and robotics are still catching up to neuroscience. A big stumbling block is that computer systems don’t learn the way human minds do. Most machine learning applications only use prediction in the training phase, which prevents them from adjusting future responses as context changes. “I think we’ll start to see real progress when we have systems that learn this top-down nuanced response in an embodied way, by interacting with their environments over extended periods of time,” Clark suggests. He believes all the pieces of the puzzle already exist, “but strangely enough, they’re not quite being put together in that embodied learning context that I think would really deliver results.”

Cyborgs in the material world

Clark seems comfortable living in a world of imperfect information. “It’s actually good to operate in ignorance sometimes,” he says, “because it enables you to drive your behavior with optimistic predictions.” This isn’t wishful thinking, he explains, but more like self-fulfilling prophecy. Through the lens of predictive processing, our brain’s predictions of our future actions—like taking a sip of coffee—are actually the impetus for those actions. The coffee cup sitting on the table is a prediction error rectified by bringing it to our lips. “Realistic yet optimistic predictions are thus tools for bringing about the goals we want to achieve,” he says. Although the world is uncertain, we base much of our action on optimism. We dare to speak because we believe we will be understood. We build a new tool because we believe it will prove useful. “Humanity does very, very well in the long term,” Clark reminds.

The effectiveness of human culture stems from our ability to change every aspect of our internal and external environments to suit our needs. The price we pay for “doing well” is our utter dependence on this world we’ve collectively created. So much of our conscious experience is based on layers and layers of expectations that our brains predict based on these regularities. Clark’s next big project concerns “the fundamental principles by which material culture influences the mind.”

“It’s still a kind of ongoing cyborgification that we’re living through.” —Andy Clark

Material culture, our food, clothing, shelter, our tools, media, art, and rituals, are really cognitive prosthetics. And like mechanical limbs, the fit is important, and swapping out new ones is hard and uncomfortable. “It’s still a kind of ongoing cyborgification that we’re living through,” Clark explains. His taste for cyborgs is both literal and metaphorical. “I am interested in stuff that is not me and isn’t exactly not me, either,” he says. Future personal AIs, Clark predicts, “will be intimate technologies that fall just short of becoming parts of my mind. I will be very much lost without it, without quite meeting the conditions for extended mind.”

Our technological experience of the near future, Clark believes, “will be much less like me with a bunch of tools, and much more like multiple cognitive ecosystems that overlap and get things done.” This messy merging of our biology with digital systems is really just a natural extension of what’s already been going on for thousands of years of human culture. He finds this story of cognitive ecology far more plausible and less concerning than the existential threat of superintelligence. “I suspect we’ll have an awful lot of this before we have very, very powerful autonomous artificial intelligences,” he predicts.

Still, his optimism isn’t unbounded. The use of social media to spread misinformation, for instance, has raised unmet educational needs. “I feel that there’s an opportunity here for a kind of 21st century epistemology,” he says, “a theory of knowledge for these sorts of systems needs to be taught in schools at every level.” Not only do we need to be alert to how false information can easily masquerade as truth on the web, we also need “training in how to manage our own and others’ expectations—for example, the expectations of extreme physical perfection currently bred by easy image manipulation.” Academic philosophy, he believes, must move beyond proving logical arguments and fallacies, to “reinvent itself as a tool that everybody actually needs to deal with these opportunities.”

Blind date of the messy mind and the elegant brain

It’s telling that even when confronted with threat, Clark finds his way back to opportunity. His native optimism is one of the personality traits he shares with neuroscientist Karl Friston. Their friendship and collaboration have helped give Clark’s work mathematical rigor and Friston’s a compelling storyline. In the British Invasion, Clark would be a Rocker and Friston a Mod. 

But unlike in 1964 when these youth subcultures bloodied each other during riots in Margate and Brighton, in 2021 these opposing tendencies are more positively engaged. It’s a hopeful sign that in the study of the brain, at least, the way forward is marked by an appreciation of differences. Despite thousands of years of Western philosophy to the contrary, the mind is not separate or opposed to the body. The notion of the embodied mind that Clark brings to life also lends to technology a more cooperative role in social evolution. Our tools don’t have minds of their own, like the broom in Fantasia, but they do affect us in ways much deeper than we understand.

This is because from the brain’s perspective, information resources located within the brain are really no different from those out in the world, Clark explains, “as long as access is fluent and the information reliable.” And in this regard, he doesn’t believe that the evolution of work from farms and factories to digital technology platforms has changed the story that much. “I think the idea that our minds are prediction machines is one that resonates,” he says. “In each case, experience is structured by expectations of how sensations will flow as we act and communicate.”

He does admit that today’s “intense digital home-working scenarios could feed a kind of isolationist user-illusion, enhancing our native tendencies to a kind of dualism.” Looking at a screen reinforces the very notion of an inner self separate from the world that Clark’s extended mind concept explodes. He says that some of his team’s recent work addresses the question of whether our concept of our own consciousness itself isn’t “just one more inference made by the predictive brain to capture patterns in how we respond to the world.” 

What we’re left with is more of a conversation between strong-minded strangers than an orderly psychic bureaucracy. To make this very point, Clark ended a 2019 talk with a send up of The Guardian’s Blind Date column, fixing up the Messy Mind with the Elegant Brain. To paraphrase the Sex Pistols, our messy minds don’t know what they want, but our elegant brains know how to get it. The actual lyrics of ‘Anarchy in the UK’ don’t make that much sense today, but the raw snarl of Johnny Rotten’s vocals still shake us up. Clark’s message is similar, but surprisingly grounded in science—don’t accept as a given what is actually an inference. “This is a big flip,” he concludes, “and we are only just getting to grips with it now.”

Messy Mind reports back on date with Elegant Brain

As told to Andy Clark

What were you hoping for? 
Someone with principles. 

First impressions? 
Very formal, not my type. 

What did you talk about? 
Life, the mind, everything. 

Any awkward moments? 
Asked me for some ground rules. I didn’t have any.

Describe Elegant Brain in three words. 
Nice, ambitious, cryptic. 

What do you think they made of you?
Kind of all over the place.