The Orangutan Theory of Intelligence

Why Thinking In Pairs Is Powerful and What It Tells Us About AI

Male orangutan, Borneo. He’s listening. Photo: David Evison/Getty

Catch-up service:
Introducing ‘John & Paul’
Why Britain’s Elites Shun Highbrow Culture
How the World Failed Greta
Is Culture Dying?
How To Be Charles Darwin

The late Charlie Munger had a hack for making yourself instantly smarter: explain yourself to an orangutan. Munger liked to say that if a person goes into a room with an orangutan and explains whatever his or her idea is, and the orangutan just sits there eating a banana, then at the end of the conversation the person explaining will come out smarter.

Katherine Graham related Munger’s idea in her memoir Personal History. Graham had been the publisher of the Washington Post for ten years in 1973, when Warren Buffett, Munger’s business partner, bought a large stake in the business. Graham, who had been forced to take over the Post after her husband’s suicide, still felt very uncertain about financial matters. Buffett, then largely unknown, became her business mentor and confidante; she called him her “personal business psychiatrist”.¹ It was Buffett who told her about Charlie’s theory.

“Warren claimed to be my orangutan,” said Graham. “And in a way he was. I heard myself talk when I was with him and I always got a better idea of what I was saying.”

In a recent article for Nature, two scientists, Itai Yanai and Martin Lercher, described their favourite trick for generating new ideas: “talk to someone.” When you talk, they say, you don’t just share information and insights: you improvise new thoughts into being - thoughts which wouldn’t otherwise take flight. By speaking, you give form to what is inchoate. Or as they put it: “Language imposes structures on our thinking and forces us to project a tangled network of thoughts into a linear, logical string of words and ideas.” They point to a motivational dividend from talking, too: the partners gain momentum and morale from the process.

Yanai and Lercher argue that the ideal group size is two: large enough to be more generative than working alone, but not so large that thoughts become diffuse or members start showing off. Yanai says: “Doing good science is 90% finding a science buddy to constantly talk to about the project.”

Munger’s orangutan theory implies that to some extent it doesn’t matter who your interlocutor is, how smart they are, and so on. But that doesn’t mean you shouldn’t seek out smart orangutans. On at least one occasion, Munger implied that it at least helps if you believe your orangutan is smart, or just that they’re listening to you.

The Indian-American investor Mohnish Pabrai told this story about Munger on a podcast. He was at a business brunch in Omaha, when he spotted Munger sitting alone. So he went and sat with him. He brought his friend Guy, whom he introduced as his business confidante, which sparked Munger’s interest (lightly edited):

And Guy says to him, "I don't know why Mohnish cares to talk to me because I really have nothing to add to anything he usually comes up with." Charlie says, "Well, going through the process of talking to someone else about your ideas requires you to put them together in a certain kind of format and manner that can be articulated to the other person. And that process is useful in seeing some flaws in your argument." Guy says to Charlie, "You know, Mohnish could actually just do that with a monkey." And Charlie says, "Yes, but the problem is that Mohnish would know it's a monkey. And so it wouldn't work."

In Conflicted/How To Disagree I relate a few stories of creative partnerships where both members relished engaging in argument. The Wright brothers spent most of their waking hours arguing with each other about how to build a flying machine. Their debates were often heated but never hostile, and they listened as hard as they argued. “Honest argument is merely a process of mutually picking the beams and motes out of each other’s eyes so both can see clearly,” said Wilbur Wright. Francis Crick and James Watson believed they were first to crack the structure of DNA because they argued more vigorously and candidly than rival teams. The enemy of true collaboration, said Crick, is “politeness”.

But orangutans don’t argue back, and productive disagreement is just one version of the process Munger is describing. What he’s getting at is, I think, something broader and even more fundamental: that we do our best thinking in collaboration with others. Intelligence is a social process, something that Socrates cottoned on to. This is how Susan Sontag put it: 'I don't care about someone being intelligent; any situation between people, when they are really human with each other, produces "intelligence."'

Being ‘really human with each other’ is, with no shade to orangutans, a lovely phrase. I imagine it means things like being tuned into what the other person is saying or trying to say, and feeling relaxed about being wrong or seeming stupid. I don’t take Sontag’s remark entirely at face value: I’m pretty sure she did care about her interlocutors being intelligent, just as long as they didn’t come across as more intelligent than her, which of course was almost never the case. But the idea that intelligence is an emergent phenomenon arising out of social interaction is a profound one.

Thinking is social even when there’s nobody around, because good thinkers have internalised the voices of others (indeed some evolutionary psychologists believe that’s what consciousness is). They can hear counterpoints and opposing arguments and alternative expressions in their head, as a composer can hear the different sections of an orchestra in harmony before a note has been played. That’s why it’s so important to absorb a diversity of facts, arguments and opinions into your mind; to generate the ‘wisdom of crowds’ inside your head.

In lieu of actual person with whom you can “be really human” in person, just speaking your ideas aloud - or indeed writing them down - can help you work out the best version of those thoughts. Software developers talking about “rubber ducking”: articulating a technical problem in spoken or written ‘natural’ language. The phrase comes from a classic book on programming which cited a programmer who carried around a rubber duck and debugged his code by forcing himself to explain his work to it, line by line. The duck was a harsh judge.

Although advanced AI has its downsides, like the potential destruction of the human race, it does offer the possibility of unlocking a vast amount of human intelligence simply by providing very advanced rubber ducks, or very smart orangutans, to people who don’t have good conversation partners for whatever reason. Already, no matter where you are, you can ask a chatbot to test your thinking. Just the process of formulating your question or idea makes a person smarter, and the chatbot’s responses are often informed and smart enough to improve your thinking further.

This is the big difference between even a very good search engine and a good AI chatbot: the latter is highly responsive to dialogue. The first answer a bot gives can be rather lame, boring or error-strewn, but once you give it a push, pointing out its flaws, criticising its arguments, it quickly raises its game. (While I’m not a heavy user of AI bots this is the one tip I give to people who ask me how to get the most out of them). As the bots get smarter, and once they are able to ‘know’ the user, their ability to unlock our intelligence will increase significantly.

But let’s return to purely human creativity. What is it about the pair, or dyad, that works so well, and what makes them work best?