Posted by: CJ | April 6, 2009

Is it a problem to question Newton?

The single most useful college class I took, in retrospect, was an introductory philosophy class on epistemology. Technically, epistemology is the theory of knowledge. I dislike that definition because it’s not particularly informative. We spent a semester talking about what it meant to know something. Such as knowing in a Platonic sense, an Aristotelian sense, Cartesian sense, Humean sense, etc.. This class was so useful because I’m an academic, and as an academic it’s a REALLY GOOD IDEA to spend time seriously thinking about what knowledge is and what it means to know something.

Over the semester (and afterwards) I realized I have some strong desires for mechanisms. I want to know how things interact with each other before I really believe something is known, or is even really science. I want to know the method of transmission. I want to know nitty-gritty details. I don’t just want a mass of statistics with light interpretation. I want a model built from the ground up. And in very complex cases, as in biology or the social sciences, a model is also good to test your ideas. It’s good to avoid confusing causes and effects just because of some nice correlations in the data.

And I’m not the only person who wants that. Economists spent quite awhile dealing with “micro-foundations” of their models. Essentially that meant figuring out if there were individual behaviors that, when aggregated to all of society, would lead to the behaviors that the macro-level models predicted. And even before that the physicists were thinking about microscopic models of matter that lead to observed macroscopic behaviors. This was the impetus behind the statistical mechanical foundations of thermodynamics. (I.e. physicists constructed the idea of atoms and molecules bumping into each other and the associated ideas of energy in order to explain relations between temperature, pressure, and volume for gases.)

(Of course the outcomes of these two endeavors were completely different. Statistical mechanics has been a very good theory, but all the pioneering statistical mechanists committed suicide due to massive depression. The micro-foundations of economics has not been quite as successful, but not enough of the economists had massive depression or have committed suicide.)

All of the above is well and good, but there is a serious concern: Isaac Newton. When Newton first published his ideas about gravity, where two objects were attracted by a force proportional to the product of their masses divided by the inverse square of their distance, he had no model to base that on. He just had a bunch of data which happened to fit with his equation. So, in essence, he had a correlation and thought he was smart enough to see how it implied causation. To someone who wants models, this seems unreasonable.

So it seems like I’m left with the uncomfortable conclusion that wanting models would lead me to be one of those Cartesians who attacked Newton’s work. This does not amuse me…



  1. Seems to me it’s a matter of the difference between our aspirational goals and what our data can achieve right now. If the data show some very strong relationship, even in the absence of microfoundations and models, we ought to take that relationship into account — if every time we pass gas, butterflies appear around our heads, there’s a pretty good reason to believe something’s going on there. (Yes, counfounding causes, yadda yadda yadda, but the risk of that goes down as the number of observations and their consistency goes through the roof — when we’re talking about something like gravity that appears in every single observable interaction it’s a little idle to really worry about some other kind of causation going on.) But that doesn’t exempt us from the need to figure out precisely what that is. Newton didn’t err, he just had more work to do…

  2. Paul–it’s true that’s one response to this dilemma. But your response only emphasizes the universality of the relation, that it applies in all observed instances. I think that only gets at part of it. I think it’s not just the universality, but also the determinism and the underlying belief that the laws of nature are unchanging.

    The determinism matters because it’s easy to construct theories that seem to apply everywhere and explain everything. The catch is most of them aren’t deterministic, so there’s some interpretation for how to apply them.

    The unchanging laws of nature is, I guess, something to avoid some version of the Lucas critique. As I understand it, the Lucas critique basically says we can’t use historical data under one policy regime to predict behavior under another policy regime because people can change how they behave when the policy regime changes. But in physical law there’s no worry about things ignoring the laws when you try to exploit them. So when we use gravity to make a catapult, we don’t have to worry that the weights will decide they don’t like falling and won’t do it this time.

    The combination of the above two additions means it’s easy to test the laws once you formulate them. Which is, perhaps, the big difference between physical law and everything else.

  3. […] A few weeks ago I was wondering about whether wanting a mechanism meant rejecting Newton’s research. Paul Gowder pointed out […]

  4. […] for the science aspect. In an older post I spent awhile wondering how I could always want a mechanism and yet not reject Newton. The answer […]

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Google+ photo

You are commenting using your Google+ account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )


Connecting to %s


%d bloggers like this: