Hacker News new | past | comments | ask | show | jobs | submit login
When physics meets biology: a less known Feynman (arxiv.org)
67 points by sohkamyung on May 15, 2018 | hide | past | favorite | 60 comments

There was a really good chapter on his time in a biology lab in his book, "Surely you're joking Mr. Feynman". I have always remembered this quote:

When it came time for me to give my talk on the subject, I started off by drawing an outline of the cat and began to name the various muscles.

The other students in the class interrupt me: "We know all that!"

"Oh," I say, "you do? Then no wonder I can catch up with you so fast after you've had four years of biology." They had wasted all their time memorizing stuff like that, when it could be looked up in fifteen minutes.

Feynman was wrong on his view. The same thing could have been said of him and physicists. Oh, you spent your time memorizing formulas when it could be looked up in 15 minutes.

This is a nonsensical point of view for two reasons. People in areas of their expertise tend to know a lot in that area off the top of their head because they have encountered this stuff so much that it has become memorized. That memorized knowledge comes from experience not a memorization exercise. It’s also nonsensical because if you didn’t have a core set of knowledge memorized in your area of expertise and instead relied on spending 15 minutes looking up each fact necessary to understand a given situation you’d end up wasting your whole day just trying to understand the meaning of the problem.

I'm uncertain how it was in Feynman's day in particular, but in my physics course you are given essentially all relevant formulas during any exams. You have to understand what they represent, and may end up remembering some of the more useful formula off the top of your head from using them often, but there is no rote memorization required.

There are also dozens of constants involved that are possibly a better analog, very few of which end up memorized.

There is little rote memorisation required in physics because we have formulated it based on logical ideas, intuition, and fundamental principles. You can understand, not memorise.

Biology is immeasurably harder than physics, the systems so much more complex and opaque and non amenable to fundamental explanation. Rote memorisation is a must.

>Biology is immeasurably harder than physics

Statements of this form are essentially always false: basically everything has the same difficulty because the standards for what constitutes progress in a field are set according to what the people working in it can do on average.

It's easy to write down what you thought of Infinite Jest, but in order to achieve any special success as a critic you're going to have to beat out everybody else that's doing the same thing. That makes it very hard. Likewise in an alternate universe where physics was easy, the standards for how much you had to say in a paper would rise until the average paper was about as substantial as it is now.

Physical sciences have the benefit of being pitted against reality. A physicist a generation ahead of his time can make predictions about reality which will prove him correct, even if nobody else can understand what he's saying. A literary critic who truly understands more than everyone would have no recognition to show for it if everyone else were simply incapable of understanding it.

>A literary critic who truly understands more than everyone would have no recognition to show for it if everyone else were simply incapable of understanding it.

A successful critic isn't a critic, they're an author that writes for whoever is making them successful. So the critic at the top of academia would be the best at writing criticims that people understood: no garuntee of understanding the most, but that's still a skill that, say, Feynman would find it hard to replicate. The heirarchal nature of almost everything garuntees that there will always be "the best," and that they will be hard to beat, regardless of how much it means in the grand scheme of things. Finally, the scarcity of grant money means that participating in any field means beating the best: a regulatory effect that's at the center of the discussion.

>"the standards for what constitutes progress in a field are set according to what the people working in it can do on average."

For example, in physics they require p < 3e-7 for a discovery. While in biology it is p < 0.05 and sometimes even p < .1 is allowed.

But very different things are being measured--comparing them this way is very flawed.

Not sure what you mean but my post was a sarcastic comment about how people choose an arbitrary level of significance so that the "right number" of "discoveries" can be published each year.

If there is a lot of data they would have too many "discoveries" (it would seem too easy) so they need a more stringent threshold. If the data is very expensive they would have too few "discoveries" so the threshold gets relaxed.

Pardon, I have a weak ability to detect sarcasm before finishing my first coffee. I fully agree that p-value limits are arbitrary, and what one is examining can greatly impact the interpretation of 'significance' or not. As you imply, data can be inflationary , incrementally increasing in volume and decreasing in value.

This is true at the surface level. However, biological systems are organized by principles. Source: degrees in both fields.

Memorizing large numbers of facts give way to understanding and intuition in my experience. It’s quite similar to how a chess player’s experience regarding strategy starts out by exhaustively enumerating potential futures and eventually turns into an intuitive understanding of future plays and how training a neural network leads to a high dimensional manifold representation of a concept.

Isn't it at least literally true that biology encompasses all of physics, plus some added complexity/emergence on top of that?

Biology is a huge field so I don't mean to speak for all of them. My wife is doing a PhD in Immunology so I know that at least for her, she's thinking in a whole different abstraction level. She's not that concerned about the physics in her day to day experiments.

Good point. It really depends on what area in biology you're in. For immunology, pathology, and, honestly, most subfields, you're closer to phenomenology than emergence.

I will say that I'm skeptical of a lot of the work done in biology in relation to networks and complexity. It's often either done by those without sufficient mathematical background (and, typically, wholly statistically invalid) or a bunch of hand-wavy, feel-good nonsense for a sexy tagline by those who should know better.

Better work is done by researchers in complexity who analyze biological systems.

And by the time you are ready to get a degree in physics it will no longer be necessary to give you those formulas in a test. You have used them so often and understand them so well that a formula sheet isn’t necessary.

You have missed the point of my comment. Do you think Feynman lacked a vast amount of physics knowledge that he memorized (to use his phrasing)? Part of acquiring expertise in any area is the fact that you end up storing lots of information about that area so that you don’t need to look up every little detail or fact.

If you are going to have a serious discussion with biologists about cell processes you need to have lots of knowledge, names, etc. at the ready. Can't go looking up every little thing as you go along. It would be impossible communicate if you did that.

Feynman was being a jerk with that comment. His dismissiveness reminds me of comment that a well respected mathematician once made. “Quantumn mechanics is just such and such theory in dimesion 1.” Only ignorance allows one to hold such a belief.

In addition to that, I think it’s generally not a good teaching technique to belittle your students, even if they show a bit of immaturity.

> “Quantumn mechanics is just such and such theory in dimesion 1.”

Well, quantum mechanics is "just" 0+1 quantum field theory which is actually a good insight and not at all dismissive.

It is dismissive because if it was just that then knowing the mathematics would be sufficient to do research in theoretical quantumn mechanics but it isn't. One needs to know the physics and have physical intuition. To say that theoretical quantumn mechanics is just a small branch of mathematics is ignorance. It is not insightful at all.

Is commutative algebra just a branch of set of theory? Are set theorists generally equipped to understand research papers on Galois Cohomology? No. Set theory is used in these areas but to suggest that these areas "are just aspects of set theory" is not all insightful.

> It is dismissive because if it was just that then knowing the mathematics would be sufficient to do research in theoretical quantumn mechanics but it isn't.

This not at all the same. QM is a special case of QFT.

> To say that theoretical quantumn mechanics is just a small branch of mathematics is ignorance.

It’s not. That statement is objectively true.

> It is not insightful at all.

It is.


> Is commutative algebra just a branch of set of theory? Are set theorists generally equipped to understand research papers on Galois Cohomology? No. Set theory is used in these areas but to suggest that these areas "are just aspects of set theory" is not all insightful.

Intersection (overlap) is not the same as containment, so I have no idea why you’re bringing this up.

Isn’t it true that commutative algebra is contained within set theory? Aren’t all the objects sets?

> Isn’t it true that commutative algebra is contained within set theory? Aren’t all the objects sets?

Sets are one way (among others) to encode such mathematical structures. This doesn’t mean these structures are sets. For example, is the empty set an element of pi? This question is meaningless because the answer would have to depend on the encoding being used, and not on the properties of pi itself. It’s imposing additional structure above and beyond that of the mathematical structure in question.

See here for more discussion: https://news.ycombinator.com/item?id=16080027.

A ring is a set. So is a group. At least this is how almost every working commutative algebraist views things. Category theory and its like are gaining traction but the objects are still viewed as sets.

> A ring is a set.

No. It is a set together with additional structure. Without this additional structure you just have a set, not a ring.

> So is a group.

Again, no. It is a set together with additional structure.

Note: The collection of all groups is not even a set!

I don't understand the complaint here, Quantum Field Theory is physics just as much as Quantum Mechanics?

The mathematician who made the quote I mentioned said that quantum mechanics was just <some branch of mathematics that I’ve forgotten> in dimension 1. I don’t know what quantumn field theory is. If it is a branch of physics then it’s not the theory the mathematician I referred to mentioned. Some mathematicians view mathematics as superior to physics and are dismissive about physics. They tend to believe that since they mastered mathematics then learning physics is easy.

It's true for formulas, but not for muscle names. You need to know what muscles do and their context, but names are just useful for communication, not analysis.

Also true of street names, you just need to know physically where your town's various businesses are and where your people live. But still you kind of have to know most of the street names off the top of your head to function properly.

Besides, I think most muscle names are literally just derived from their position on the body and their movement range, so their name and their function are closely linked.

> But still you kind of have to know most of the street names off the top of your head to function properly.

I know 10 street names in my city. The rest, I look up.

Well, I used to, now I use the GPS.

Note that I do think you are more efficient if you know the street names and where they are.

But I also do think the time and energy requires to learn 100 street names and position in my city can be better used for something else.

What if you were a cop? Would you criticize a cop who has learned the streets? Shouldn’t a biologist who is at a talk about felines understand the biology of felines without looking it up in a book?

Feynman's criticism shows an ignorance on his part.

It's not about knowing, it's about learning artificially.

> What if you were a cop? Would you criticize a cop who has learned the streets?

I would appreciate cops learning anything, but given they got a GPS and older colleagues knowing the streets name by habit, I'd prefer they spend time learning the law. The names will come naturally with practice.

> Shouldn’t a biologist who is at a talk about felines understand the biology of felines without looking it up in a book?

See that's the problem here. You assume you can't know the biology of felines without knowing all the names of all muscle by heart.

It's like saying I can't understand a program if I don't know all the functions of it's API. And while after years of programming I know a lot of things without needing to look them up, I never learned nor will learn any listing by heart. It's a waste of my time. I need to understand algo, complexity, interractions. Names are just an interface, useful, but not required for modeling.

Same things for you biologist. He must understand the interractions between the organs, the chemistry, the musculo-skeletic chain. Naming all the muscle is just a bonus, and now that we have books and computers, it's not a priority anymore.

Does knowing stuff by heart makes you more efficient ? Does it make you a better pro ? Yes, but only if the more important stuff are understood.

And if you are a student, mindlessly learning all names is completly a waste of time. Espacially since they are going to forget most of it after graduation, and learn it back again on the way.

You can’t converse with an expert on felines without knowing the names. There are simply too many parts of the body. You have to name them in order to communicate. There’s a reason for the names. It’s nit like someone sadistically decided to give a name to each part. Go ahead have try having a conversation with someone who knows the names and functions of each muscle where you don’t. Looking up each term is not efficient.

I’ll take a cop who knows the streets over one who has to look up on GPS. Cops chasing a suspect have report on their progress so that backup can reach them. What do you think they’ll do? Pause every 20 seconds and look up GPS coordinates?

Feynman's critique was born of ignorance. He criticizes people for knowing the parts of the cat. In the lecture they criticize him for drawing the map of the cat. They know this already. Feynman does not. He is not at their level and instead of admitting this he makes an ignorant quip.

Virtually all knowledge is an easy internet search these days. This has not diminished the need to know, have memorized, terminology.

Let S be an infinite dimensional ring extension of R with R a commutative Noetherian ring with identity. Under what conditions is S flat as an R-module?

If someone gives a talk on commutative ring theory and the above is the topic they best not waste everyone’s time by looking up the meaning of each word and writing it on the board. Such a person exposes that they aren’t learned enough in the field to be giving a talk.

I thought that was entertaining before I learned biology, but after knowing more it's far less entertaining.

They didn't know a map of the cat, they knew a map of mammals. Evolution is really cool that way. Also perhaps it was easy to catch up because it wasn't that much knowledge.

I forget the exact context, but I remember my elementry school math teacher telling us "you wont have a calculator with you at all times" when memorizing the times tables.

While she may have been dead wrong on that one, i'm definitely glad I don't have to rely on looking up / deriving the basics for every domain.

I've been taking medical and biology related courses for the last five years or so, well over 1000 hours in lecture hours thus far - thanks to the Internet (a counterpoint to the "I Don’t Know How to Waste Time on the Internet Anymore" thread popular on HN right now). I concluded that while I find it all extremely interesting and exciting as far as anatomy, physiology, neuroscience, organic and biochemistry, genetics, statistics (which I had learned - and forgotten - before but now I actually found useful and therefore interesting for the first time), etc., I could never study medicine to become a doctor. All that rote learning!

For a CS graduate, it's like learning everything there is about Oracle 9i. For years. By heart. Not even about databases in general, you learn the basics of course, but then you spend most of your time learning by heart every command and every setting of Oracle 9i.

I'm sure it makes you an efficient doctor within the given system, and when trying to solve some example cases there sure is value in knowing pathways (real ones, e.g. neurological ones - where is the damage if the patient feels this and does not feel any of that?, as well as biochemical ones, but also organizational stuff like various ways invented to categorize broken bones, e.g. http://www.thieme.com/media/samples/pubid1252333231.pdf). Quite frankly, I find most of it a waste of time.

Right now I'm re-taking edX "Principles of Biochemistry" (https://www.edx.org/course/principles-biochemistry-harvardx-...), and while it's a lot of fun I also feel it's quite a bit of a waste: From part III on it feels more and more useless. The course is done very well, and every bit is interesting, but how - WHY - am I supposed to learn soooo many paths in such great detail? What's the point? If I need it I can always look it up! No wonder that while there are several "Hello I'm new here, excited to start this course" messages in the forum every day there are only a handful of forum messages for the later sections of the course - after many months of running. I suspect >90% of people never get past part 3 (of 5). Because the brain just does not see the point of all this rote learning. If there was a larger project as context, a problem to solve! But just "here learn this" for no other reason than "it's interesting" (so are a trillion other pieces of information!!)... it does not work very well.

Learning needs both a PUSH and a PULL. The teaching is mostly about push(ing knowledge into brains), but where is the pull (reasons other than abstract "you need to know")? Brain have their own ways of detecting actual need, and that comes from having to solve actual problems, not from being told "this is important". And learning millions of facts when you don't need them but know you can always learn them/look them up quickly at any time when you do actually need them is soooo demotivating.

Technical reasoning in the medical field involves composing many facts together to crystallize into an argument.

It's very useful to be capable to do that quickly if you can recall it off the top of your head, as you'd be able to more fluidly connect the information into the argument without all that task switching overhead from going back to Google every couple minutes.

Let alone the issue of the information needing to be available in the first place! How would you know what you don't know? Wouldn't you be a lot less likely to forget if you memorized?

I find medical textbooks have significantly improved over the years at their capacity for organizing their finds both functionally and hierarchically. But we can't really put these findings in terms of their compiler, the laws of physics; you don't get unity except by association of the lower level facts with their referrers.

Additionally, discovery in the modern medical field cycles from inductive to abductive to deductive reasoning ad nauseam, in particular by associating diseases together from common molecular components. Your brain will be incapable of having these insights spontaneously without having them internally to percolate.

Yeah, totally with you on that. I knew lots of pre-Med majors in college and I could never wrap my head around how someone could enjoy such a masochistic form of learning. But the common thread of successful pre-Med majors is that they get straight A’s and are very successful memorizing everything.

Medicine is basically all rote memorization, from pre-Med to the MCAT. It makes sense to select for doctors who are good at it, because they need to have a quickly queryable, but wide breadth of knowledge in their head at all times.

Personally I’m happy to know that my doctor is able to memorize and collate so much information. But I do worry that the way he learns may also restrict his creativity and problem solving.

Perhaps as AI moves into medicine, doctors won’t need to rely so much on their own memorization, and can focus their efforts on problem solving.

I think you underestimate the role of problem solving skill in medicine. It’s not all rote memorization. A good doctor is good at problem solving and has to have lots of information ready for instant recall.

EDIT/FOLLOW-UP (my edit right just disappeared a minute ago):

Here is a link to example patient cases, which may help to see what kind of knowledge is useful. I was not trying to say it is all useless, when I had to solve sample cases e.g. in Medical Neuroscience (Coursera, huge course, great teacher) it sure helps to know the major pathways, and it would take too much time to have to look them up when the patient is in front of you.


"I could never study medicine to become a doctor. All that rote learning!"

Having taught many med students, I say this attitude would have been a good start towards being a great doctor.

I think medicine, being both a very old, and having a very complex subject, is guilty of a lack of abstraction.

I tend to joke that if a doctor discovered a keyboard he would name it a negra quadrodepressus-hectomatrix. Accurate.. and semantically useless. That's how you name things when you have no clue about it's function.

And medicine likes its superbly tough anatomical projection/cut of internals. Honestly I find it's a miracle how doctors can look at organs in every angle and still find their way (same for radiologists).

Coming from CS with a taste for multistage compilation and the likes, I find no beauty in anatomy.. I need a little more principles and less description.

Not a very good abstract. It doesn't tell me anything about Feynman's contribution to molecular biology.

Very little. He did some work on what would now be called mutational "recovery" of frame shift insertion/deletion variants in phages, for about a year on sabbatical. Which at the time would have been enough for him to have predicted several aspects about the genetic code, such as DNA triplets forming codons. Which was funnily enough anticipated by George Gamow around the same time looking at the number of naturally occurring amino acids and the number of nucleobases.

OT, but if you like quantitative biology, there's http://bionumbers.hms.harvard.edu/ . The book draft "Cell Biology by the Numbers" there is also fun.

Even Alan Turing have done some biology-related work: https://en.wikipedia.org/wiki/The_Chemical_Basis_of_Morphoge...

Yeah which is waaaay more significant both in the field of chemistry and biology. The work Turing was doing at that time was seriously advanced thinking, and really on the right track for how things like transcriptional regulation works in embryogenesis. If he had lived longer its really depressing to imagine what he might have contributed in biology, and how much faster we would have got to where we are in molecular biology. Unlike this write up on Feynman, Turing genuinely provided a hypothesis with some insight and rigour that turned out to be important. It was first found to be a real thing in the world in chemistry with the discovery of the Belousov–Zhabotinsky reaction, which then directly tied the maths to an actual reactive process. The way biology works is a bit more digital and discrete, but the general idea and concept is similar. Turing couldn't have anticipated that his ideas on computation were probably even closer to biology than his differential rate equation models. If he had lived to see the evidence of the tie ins between his works, I think he would have dropped some seriously epic stuff we cant even imagine right now for dealing mathematically with molecular biology at various scales.

Having recently learnt that Feynman seems to have treated women quite badly [1]. I feel slightly uncomfortable about this article, it feels more like hero worship than anything else.

Feynman no doubt did some interesting work in Physics. But the work discussed here doesn’t appear to be particularly distinguished. It does not follow that someone did interesting work on X therefore everything they did must be valuable and interesting.

I suppose what I’m trying to say is that individuals are a complex mixture of good and bad. And that there’s a danger that an individual who has done interesting work in one area is seen as a universal genius, and an ideal to be emulated. This unfortunately is not the case.

[1] https://restructure.wordpress.com/2009/08/07/sexist-feynman-... for example...

The story you link to describes interactions with two women. One ends with

> "Just as we’re coming out of the bar, here comes Ann, running across Route 66 toward me. She puts her arm in mine, and says, “Come on, let’s go over to my place.”"

the other with

> "We went into the bar, and before I sat down, I said, “Listen, before I buy you a drink, I want to know one thing: Will you sleep with me tonight?”

> “Yes.”"

These interactions seem like voluntary, enthusiastic consent by rational adults. Maybe I'm seeing a contradiction where there is none, but if a woman who left and never had any need to see him again ran back to initiate physical interaction and invite him to her place, is it possible she knew more about the situation than you and didn't consider his behavior quite bad?

We don’t even know if any of it was true at all, it could all be a work of fiction. But we know at least he considered it not much of a problem to call a woman a whore for not having sex with him because he bought her a sandwich.

There are other allegations [1]. But that isn’t the central point really, the point is, why idolize someone who did some good work in one area and expect everything they do to be a work of genius? He was imperfect (or are you arguing otherwise?) his work in Biology is not very interesting...

[1] https://galileospendulum.org/2014/07/13/the-problem-of-richa...

I'm perfectly fine reading about Feynman's scientific pursuits in full knowledge of these anecdotes, because they are interesting and he was a brilliant man, and I don't see him as any more flawed than any other human. I especially couldn't care less about his thoughts relating to bars since they are by nature, back then and now, a warzone between nasty men and nasty women.

But, I suppose the point is that work on Biology and Physics is as far removed as his work on Physics and picking up women in bars...

The work discussed in the paper, is not particularly interesting in isolation.

To play devils advocate and give him the benefit of the doubt... wasn’t the 1950’s a much different time in terms of respect for women?

Reading that excerpt, I got the sense he was writing from the perspective of his past-self, in an almost self-deprecating way. He’s citing that incident of an extreme example of how literally he would follow instructions. By choosing it as an extreme example, he’s effectively acknowledging that the attitude was disrespectful / wrong.

Granted I haven’t read the whole book, but I wouldn’t be surprised if there is some crucial context missing from those excerpts.

Feynman did, after all, write the words himself.

Having read the book, I didn't get the impression he was criticizing himself. It read as just another story of an experiment, which how it seems he approached most of life (and which is what makes the book interesting, in my opinion).

Perhaps my usage of the words “good and bad” implied a moral judgement. But, that’s not really my intention.

What I’m saying, is that it’s unwise to idolize people because of their work in one area and then elevate (or more worryingly emulate) there work in other areas.

Like his treatment of women, Feynman’s work in Biology is not really very interesting. It wouldn’t stand up on its own merits... the only reason to talk about it is if you treat Feynman as some kind of Hero/ideal. Feynman is not really a wonderful person to idolize (not that anybody is)...

Individuals are not a complex mixture of good and bad. Almost nobody commits what might be considered a bad action just for the sake of being a bad person. We should not be so readily to apply black and white moral judgments on others.

Good and Bad are loosely defined terms. I guess in this case it might be better to say “currrently socially unacceptable behavior which is considered selfish and harmful to others”. It seems self apparent that calling a woman a whore for not having sex in exchange for a sandwich is socially unacceptable and harmful behavior. But I would guess it’s probably possible to find data to support this.

I guess it must be the time he lived in, but it has some tragic comedy to see him completley fail to understand women and put their behavior in binary terms.

Why do you feel this article is hero worship?

Would this article be of general interest if it wasn’t about Feynman? (I don’t believe so)

The work Feynman did in Biology that’s discussed here doesn’t seem to be particularly ground breaking, or interesting. It feels like it’s only being treated as such because Feynman did interesting work in other areas and is somewhat idolized for this.

Would this article be of general interest if it wasn’t about Feynman?

Probably not, but I don't think being interested in an individual necessarily means idolizing them. In fact, sometimes the actions that we consider abhorrent is what makes them interesting. That's not the case here, but still, I think there's potential value in studying how a very good physicist approached a foreign subject matter.

Applications are open for YC Winter 2022

Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact