How Richard Feynman Thought

Photo: Jupiterimages, Brand X Pictures

I am fascinated by how we can improve our thinking and problem solving and enjoy learning about and from masters of those arts. My interest was therefore caught by the advice on thinking given in a review of Quantum Man: Richard Feynman’s Life in Science. The reviewer, George Johnson, writes:

This triumph came early in his [Feynman’s] career. His later thinking (about solid-state physics, for example, or quantum cosmology) was just as original. Maybe sometimes too original, Krauss suggests. Science usually proceeds by building on what came before. The maverick in Feynman kept him from accepting even the most established ideas until he had torn them apart and reassembled the pieces. That led to a deeper understanding, but his time might have been better spent at the cutting edge…“He continued to push physics forward as few modern scientists have,” Krauss [the biographer] writes, “but he tended to lead from the rear or, at best, from a side flank.”

To simplify my discussion, I will conflate the reviewer and biographer (Lawrence Krauss), because they seem to agree that Feynman should have worked differently, that he should have spent his effort at the “cutting edge.”

The advice is interesting but I think fundamentally flawed. Had Feynman followed it, it may have killed the magical spark of his originality. The reviewer says that Feynman could not accept “even the most established ideas until he had torn them apart and reassembled the pieces.” As a graduate student at Caltech, where I was a student of two of Feynman’s collaborators, I absorbed many stories and lore about Feynman.

It’s not quite true that Feynman could not accept an idea until he had torn it apart. Rather, the idea could not yet be part of his way of thinking and looking at the world. Before an idea could contribute to that worldview, Feynman wanted to turn over the idea, to see why it was true, from any angle that he could find. (I like to think that Feynman would have enjoyed my book on Street-Fighting Mathematics.) In other words, he wanted to connect a new idea to what he already understood and thereby extend his understanding.

Feynman’s approach is a wonderful recipe for becoming more intelligent. I believe that intelligence is not a fixed quality; rather, it is a process of growth. The best explanation of and support for this view is Carol Dweck’s research, wonderfully described in her book Mindset: The New Psychology of Success. Feynman’s approach to new ideas is a fantastic example of the growth mindset.

(This approach, when combined with Feynman’s aversion to publishing, often disconcerted others. At Caltech, there were stories of Feynman’s younger colleagues who had worked on an idea and then showed the result to Feynman, who said, “Oh, right, I’ve thought about that,” and pulled out his notes where he had already derived that result — often with extensive generalizations.)

Spending time “at the cutting edge” may have killed the creative goose. When Feynman was developing his approach to quantum mechanics, the work for which he eventually won a Nobel Prize, the cutting edge was a very difficult version of Heisenberg’s original quantum mechanics, extended to handle the complexity of how radiation (such as radio waves, light, or gamma rays) interacts with matter. But the subject was stymied by infinities in every calculation. The answer was to start with minus infinity so that adding positive infinity resulted in a reasonable value. Through great efforts, brilliant physicists made this crazy idea usable and practical. But it required heroic mathematical efforts.

Instead of working at the cutting edge by improving those methods, Feynman found an entirely new way: his famous method of path integrals.

If we were perfect reasoning machines, it would not harm us to first work at the cutting edge and then develop our own ways of thinking. But we are imperfect. As economists might say, there’s a huge path dependence in our thinking: Once you know something, it’s hard to unknow it. Once you learn a way of thinking, it’s hard not to keep falling into that way of thinking at the expense of finding new ways of thinking. I think Feynman had a healthy respect for how our minds actually work — as opposed to how they might work if they were ideal reasoning machines. That humility, a word infrequently associated with Feynman, made him wary of digging those mental paths before he had explored his own ways and had made his own, perhaps different, paths. That’s how he reached or, rather, created the cutting edge.



Nice post. Let's put this in a context Feynman himself described: he solved a practical problem with renormalization and path integrals but that occurred and still exists today in a field that rests on mystery. He became famous to the general public for pointing out that we can calculate extremely well but have no understanding of the why underneath. To say he should have joined with the cutting edge ignores that essential mystery: his Nobel work literally represents a mind grappling with the unknowable and coming up with a pragmatic answer that fits. The unknowable is not in this case at the edge of what we can find but at the heart of the matter. He knew that. That's why he could acknowledge the mysteries with a bluntness few scientists can match.

To say he should have joined the cutting edge misplaces him. He was looking directly at the heart, at the essential mystery that underlies this physics, and his most famous work responds to that.



Just to let you know that the link to the book: "Mindset: The New Psychology of Success"
is wrong.

Should have been:


Sanjoy Mahajan

That was my mistake (another reminder to never type URLs by hand). Thanks for noticing it. The editors have fixed it.


The problem with the cutting edge is that everyone is looking there. You're less likely to find something new where everyone is looking. The upside of the cutting edge is that there are academic communities there. Breakthroughs made at the cutting edge have "impact," because many people care about those papers because many people are looking there. Much more important discoveries not at the cutting edge get no academic traction, simply because there's no community. In many cases, there isn't even a relevant journal where to publish such a breakthrough. Did you just make a fundamental discovery in classical mechanics? Oops. No one cares. String theory? Big deal. Circuit network theory? No one cares. Some new RF gizmo? Tenure-track position. Funding for non-cutting-edge research is also impossible to come by.

Bruce Casner

A better link for Street-Fighting Mathematics might be:


The key point I think is that knowing how to use your senses, and actually using them, is important. The typically human thing to do to any given field (given our crazy language ability and outsize brain structure devoted to it) is to develop the base terminology until it's just good enough, then forget about the real world and launch off into higher and higher abstraction, recombining and remixing like mad. Because it's such a human thing to do, that's where the jobs are. And no one really cares so much any more that we're not describing the real world, because everyone's job at that point depends on us all glossing over that fact. Eventually, someone whose job doesn't depend on all of this (not working yet, has rich parents, has a patron, works in another field, independently wealthy), will point it out. Then we're all screwed.

Who are the people who set aside standard verbal language and learn how to use their senses? The people we call artists (though everyone and their mother who have never drawn anything since kindergarten is an artist these days), athletes, people with some kinds of illness or who live in dangerous situations where their ability to keep living depends on paying attention to things that most can safely ignore. All those who, by the expensive process of trial and error, engineer some sort of strategy that looks totally nuts to people in standard situations, but solved whatever problem it was, that most of us don't have to deal with. And since most of us don't have to deal with it, no one is going to fund that person's time spent coming up with that solution. They're on their own, until they or someone else discovers that their strategy happens to be useful for something relevant to the rest of us. Or else steals their idea and makes a fortune off of it, while that person fades in obscurity. And maybe is hailed as a genius safely post mortem, when it's easier and cheaper to do so.

So how to identify who this might be? They're probably sitting there quietly, contending with the real world with their senses not by choice but out of necessity, building up to language from the most basic level. Drawing ability, musical talent, can be a good indicators: the capacity to shut up, observe, and mimic. I think Feynman drew, played music, and I would guess that he was a pretty good dancer too. This indicates reasoning ability in this dense, pre-verbal sensory space. Reasoning as much as possible here, closer to reality, reads as the child-like sensibility, or humility.



David said:

"Eventually, someone whose job doesn’t depend on all of this (not working yet, has rich parents, has a patron, works in another field, independently wealthy), will point it out."

I think this is pretty much what Nassim Taleb has been been doing.


I've always viewed Feynman's method as collecting alternative formulations, his path integral work is equivalent to the standard matrix formulation but produces a different mental view that can crack certain problems. Calculating a ball's flight can be done in different ways but viewing the flight as the ball seeking the minimum energy produces the same result along with a physical insight. Increasing one's mental toolbox with alternative formulations of the same physical phenomenon is a very powerful contribution to science.


Define "accept".


Feynman should have done what he did - just be Feynman. If there's one thing the man taught us is that to just be who you are is all that matters.

Alex McDermott

You lost me at "I like to think that Feynman would have enjoyed my book."

Awesome article thank you.