Defending Thomas Kuhn from Errol Morris

Errol Morris' series on Thomas Kuhn started amusingly: it was fun to hear that Morris -- the director of great documentary The Fog of War -- was actually a first-year graduate student in the History of Science program at Princeton (and rejected by Harvard's History of Science department!).  (I don't know, that gave me quite a kick.)  And it was funny to hear that he and Thomas Kuhn hadn't gotten along, that Kuhn had thrown an ash-tray at him because he refused to give up his Whiggish ways of doing the history of science.  (There's probably some poetic license here, but it's still funny!)

But as the story progressed (parts two, three, four, five), my enjoyment started to wane.  It wasn't the flowing, free-associative style, which was still fun, although I really didn't see what the digressions into the Pythogorean society had to do with the concept of incommensurability.  No, it was the intellectual portrayal of Thomas Kuhn.  If one reads Morris without reading Kuhn, then Kuhn comes across as an idiot.  And all of us who found The Structure of Scientific Revolutions useful come across as either idiots or fashionable post-modernists who like Structure because it fits in with the relativistic fashions of the day.

What is Morris' problem with Kuhn (other than that he clearly didn't like him)?  Two things.  First, he argues that the concept of incommensurability is incoherent.  And second, that it opens the doors to all sorts of relativism.  The second point is old hat.  Much ink has been shed on how if we give up the idea of Truth, of an unmediated reality out there that we are accountable to, then we are on a slippery slope to ethical relativism.  That it leaves us without a reply to totalitarian dictatorships, that it leaves the door open to the O'Briens of the world who want us to believe that 2+2=5.  Etc.  The first point is old hat too.  As John Holbo points out, incommensurability was criticized in just this way even when Structure was published.  The idea here is that if two paradigms are incommensurable, then we have no way of doing the history of science; so the fact that Kuhn himself could understand the older paradigms proves that paradigms are not really incommensurable.  

As fields, the history of science, and Science and Technology Studies (STS) have moved on: the idea of a linguistic conceptual framework that under-girds scientific theories has been replaced by a search for material practices that constitute science.  Kuhn is studied less as someone who offered fresh new insights but as one of  the oldies: routinely grouped together with Popper, Feyerabend and Lakatos.  (I also think that while Kuhn makes several missteps in Structure -- he tries to define incommensurability linguistically, he starts to talk about using computer programs to understand incommensurability -- the idea of science as being constituted by material practice is there in the book.  As are other things that we now look for as STS scholars: the practices of pedagogy and training in science, the incentive structure, etc.) 


Still.  I'd like to offer a defense of Kuhn and of Structure and why the criticisms that Morris offers are mistaken.  Reading Structure was a transforming experience for me -- and I'd like to bring out why. 

So without any further ado, here goes.

Kuhn's book is more heard about and talked about than read.  For example, I knew (or thought I knew) the fundamentals of its argument long before I read it.  That argument goes something like this:

• That scientific advances happen not cumulatively, but in bursts -- conceptual revolutions alternating with periods of "normal" science
• That  underlying all scientific theories is something called a "paradigm" (think of it as some kind of underlying conceptual scheme) and that one paradigm gets replaced by another during a scientific revolution.  
• And finally, when paradigms change, when a new paradigm replaces the old one, the two are incommensurable, so that when scientists argue for the merits of either paradigm, they are essentially talking past each other.  The paradigm that wins out wins not because it is true.  

The key thing to understand here is that most people would be fine with this argument if it wasn't for the incommensurability part.  Here, for instance, is Thomas Nagel, reviewing Alan Sokal's Fashionable Nonsense:

Much of what Kuhn says about great theoretical shifts, and the inertial role of long-established scientific paradigms and their cultural entrenchment in resisting recalcitrant evidence until it becomes overwhelming, is entirely reasonable, but it is also entirely compatible with the conception of science as seeking, and sometimes finding, objective truth about the world. What has made him a relativist hero is the addition of provocative remarks to the effect that Newton and Einstein, or Ptolemy and Galileo, live in "different worlds," that the paradigms of different scientific periods are "incommensurable," and that it is a mistake to think of the progress of science over time as bringing us closer to the truth about how the world really is.

Before I read Structure, I pretty much agreed with this.  (How could I not?)

When I did read Structure, however, I realized that while this summary was not wrong, the book said a great many other things.  I realized that: 

That Kuhn is not doing philosophy of science, despite the fact that he's almost universally known as a philosopher of science.  It's true that Structure makes a series of bold, philosophical claims.  But it also tries to back them up with historical evidence -- not by using anecdotes but with the kind of concrete, archival research that historians do.  Analytic philosophy, on the other hand, is not an empirical field.  The subject it resembles the most is pure mathematics: you solve "problems" (the problem of knowledge, of skepticism), you use symbolic logic, you use paper and pencil, you construct proofs and thought experiments.  But you do NOT collect data of any sort, you do not make hypotheses that you then hope to confirm by looking at the data.  Kuhn was grounded in the history of science, the theory of scientific change that he propounded came from his close study of early modern science. (Although, as Ian Hacking points out, very few historians of science mix history and philosophy in quite the way Kuhn did -- which is probably why he is more known as a philosopher than as a historian.)

That the biggest claim in Structure -- one that usually flies under the radar -- is not the concept of scientific paradigms, even though much space is spent talking about paradigm changes.  No.  Rather, the central aspect of science, according to Structure, is that the activity that doing science most resembles is solving puzzles.   Kuhn calls this puzzle-solving activity "normal science."  We need to understand the claim here: the point is not to denigrate scientific activity.  The point is that puzzle-solving is not usually understood as a search for knowledge.  So even if science, understood as a collective activity, is engaged in a search for knowledge, in actual concrete terms, this search for knowledge gets done by building and solving puzzles.  Now, to an extent, constructing and solving puzzles is part of all knowledge fields.  Kuhn's point though is that nowhere is this reduction of knowledge into puzzles as prevalent or as polished as in the sciences.  (This is related to how a scientific community is structured as well as its pedagogical practices.)

Finally, the paradigm is not just a linguistic conceptual scheme.  This is the mistake that interlocutors like Morris make.  They think that a paradigm and incommensurability are philosophical concepts.  They are not.  They are not even linguistic (although Kuhn sometimes talks about them that way).  They are -- and this is more important than anything else -- about practice.  My key to Structure, the moment when it all came together for me -- what a paradigm was and how it functioned.  I'm going to reproduce the full paragraph from Kuhn:

A phenomenon familiar to both students of science and historians of science provides a clue.  The former regularly report that they have read through a chapter of their text, understood it perfectly, but nonetheless had difficulty solving a number of the problems at the chapter's end.  Ordinarily, also, those difficulties dissolve in the same way.  The student dicovers, with or without the assistance of his instructor, a way to see a problem as like a problem he has already encountered.  Having seen the resemblance, grasped the analogy between two or more distinct problems, he can interrelate symbols and attach them to nature in the ways that have proved effective before.  The law-sketch, say f = ma, has functioned as a tool, informing the student what similarities to look for, signaling the gestalt in which the situation is to be seen.  The resultant ability to see a variety of situations as like each other, as subjects for f = ma or some other symbolic generalization, is, I think, the main thing a student acquires by doing exemplary problems, whether with a pencil and paper or in a well-designed laboratory.  After he has completed a certain number, which may vary widely from one individual to the next, he views the situations that confront him as a scientist in the same gestalt as other members of his specialists' group.  For him they are no longer the same situations he had encountered when his training began.  He has meanwhile assimilated a time-tested and group-licensed way of learning. (Structure, pg 189) [emphasis mine]

Believe it or not, this is exactly how it happened for me.  I studied how to draw free-body diagrams on my own, just after 10th grade.  So I had no teacher to take me step-by-step through a few solved examples, which is generally the case.  Instead, I had to read the solved examples in my textbook -- many of which struck me as incomprehensible.  I remember staring at diagrams of bodies moving on inclined planes in frustration, almost ready to cry because I didn't know what to do.  My frustration was accentuated, I think, because we had moved to a new town and I was just starting to adapt to it.

And then one day -- I am not sure how --  it went away, .  All I remember is that one fine day I found I could do free-body diagrams just fine, that in fact, I even enjoyed doing them.  Gestalt switch is probably a really good way of describing the change.  A diagram of a body on an inclined plane now means something to me that it did not before.  To use some of Kuhn's own expressions, it was as if, for me, the world had changed and while I could recall recall my frustration with how the world had looked before, I could never recapture my previous world-view; it was irretrievably lost.  Kuhn's description of paradigm shifts as gestalt switches and his talk of scientists with differnet paradigms living in "different worlds" often leads people like Morris and Nagel to call him an idealist and a relativist but I know exactly what he is talking about.

Personal remininsces aside, it's worth unpacking the paragraph in detail and making explicit all the points that Kuhn is making in it:

• A paradigm is not a set of rules.  It's more like practice, a skill, like knowing how to apply the equation f=ma to different types of scenarios (a body moving on an inclined plane, an oscillating pendulum, a body attached to a spring, etc.)   There is no way to describe in the form of rules what it means to solve problems using f=ma, you just have to learn to do it.  It's tacit knowledge, sort of like riding a bicycle.  At the same time, being able to solve problems is the only way to become a scientist.
• One has to literally go through hell to be initiated into a paradigm.  And as a matter of pedagogy, scientists have it down pat what it takes to create a competent member of their community: you teach him or her the concepts, and then you make them solve a bunch of problems.  Somewhere along the way, the student gains competence.  This authoritarian way of doing things turns away many people from doing science, but it works wonderfully well for those who like it. 
• Finally, it provides an explanation of incommensurability, of what Kuhn means when he asserts that when scientists argue over competing paradigms, they are essentially talking past each other.  Yes, if they wanted to, if they took time and effort, they could possibly understand each other.  Philosophically, incommensurability can be refuted (as Morris tries to do).  But practically, in practice, it exists.  Because learning a paradigm is a long back-breaking process, and is done usually when one is a student, established, practicing scientists have neither the time, nor the inclination, to imbibe and learn a new paradigm.   So they keep on arguing without truly understanding each other.

"But doesn't that open the door to relativism?" Morris might say.  Kuhn's answer is that the paradigm that wins out is one that the community perceives as more productive. The key word here is productive.  Scientists should feel that a certain paradigm allows them to create a rich set of problems (puzzles) that they can then solve -- which makes them prefer that paradigm.  So there is some sort of progress.  But what about the role of nature, you might ask.  Does nature play a role in the choice of a paradigm during a revolution?  The answer is yes.  And again, it goes back to the learning.  Paradigms need to be learned: by solving problems and by practicing doing experiments; nature plays a role in both of these. 

All in all, Structure taught me three things.  First, to pay attention to material practices and craft-work that are often the building blocks of any kind of scientific or engineering work.  And second, contrary to rhetoric, to pay attention to the values that often underlie scientific work.  By values, I mean things like what counts as a "good" problem, what counts as an "elegant" solution etc.  To be able to use these words successfully is to have mastered scientific practice.  Scientific knowledge can't be studied without a close understanding of the practices and values that produce it.  And finally, pedagogy is incredibly important.  Pedagogy is how a community licenses its practitioners.  How scientific practitioners are made is important if we want to understand scientific knowledge. 

I am not sure Morris will buy this defense of Kuhn.  He is an admirer of Saul Kripke, after all. 

Quote of the day

Anthony Grafton on the William Cronon affair:

Second, the Republicans seem remarkably fragile. A professor writing a blog post gives them the shivers. It’s a good thing they chose politics, and not the kind of career where the going can really get rough. Professors, for example, teach their hearts out to surly adolescents who call them boring in course evaluations and write their hearts out for colleagues who trash their books in snarky reviews. These Wisconsin Republicans may never have survived ordeals like that. Happily, Cronon has been toughened by decades of academic life. He’ll be blogging—and teaching and writing—long after Wisconsin voters have sent these Republicans back to obscurity.

See Cronon's original blog-post that sparked it all off here.  

Internal vs. External Incentives: The Case of Economics

[This is a random, off-the-cuff post, with some wild generalizations, so please let me know if you think any part of it is wrong.]

There's been an interesting debate in the blogosphere recently about what science is and whether economics is a science.  The debate is interesting not so much because it settles the issue but because it's a good data-point for looking at public understandings of what it means to do science. 

Tyler Cowen started it all off by saying:

Economics is most like a science when people do not care about the outcome of the argument.

To which Matthew Yglesias responded in agreement, adding:

In other words, social science is like science when it’s like science—disinterested. But when it’s like politics, then it’s like politics. I note that American economists are generally able to reach a much greater degree of consensus when they offer policy recommendations to foreign countries than when they offer recommendations to the US congress. That’s not because foreign countries have easier problems to solve. 

When Cowen talks about the "outcome" of the argument, he means the external outcome of the argument: in this case, specifically, the public policy that seems to be the logical consequence of a certain knowledge claim.  When a knowledge claim has no link to a public policy choice, he suggests, economics is most like a science -- a condition Yglesias calls being "disinterested."  In Ygelasias' view, when a economic knowledge claim has no obvious public policy consequences (especially for the country said economists live in), economists behave like scientists -- which is not to say that they agree, but rather, that they disagree, but in a disinterested way.

In STS, this is what I call the Internal vs. External debate on what drives science.  Do scientists accept theories because they are, in some sense, true?  Are they convinced of the validity of a certain knowledge claim because of "internal" reasons -- reasons deemed to be "proper" within the field?  Or are they influenced by what one might call "external" reasons, reasons deemed by the community to be properly out of bounds, such as the desire for money, support for a political program, etc.? 

But an interesting aspect of what it means to be a disinterested economist struck me (and correct me if I'm wrong).  As far as my reading goes, ideology doesn't count as an improper reason for being partial to a knowledge claim.  So economists will routinely proclaim that they are libertarians or progressives or egalitarians.  What they won't say is whether they are Democrats or Republicans -- and so this is often deployed in debunking someone's claim.  Thus one  is more likely to hear things like "X has nothing to say about this because he worked for a Republican president and so would rather not contradict said president's policies" rather than "So-and-so makes this claim because he is left-wing and is therefore biased"  One also hears things like (at least on blogs) "I am sympathetic to such-and-such claim because I believe in progressive causes" -- which means that ideology is arguably a more acceptable reason for being sympathetic/antagonistic to a knowledge claim, than, say, membership of a political party.

Which is, in a way, different from what counts as a proper reason for accepting a knowledge claim in the actual sciences, like physics or biology.  Here, I'd say, even ideology does not fly.  Membership of a political party, of course, is completely out of bounds.  (Note that this is an assertion about community standards and rhetoric rather than about the actual workings of the hard sciences.)

Which is to say economists should probably be comparing themselves to other social scientists rather than to the practitioners of the "hard" sciences.

Other links:

And here are two more posts about the "Is economics a science?" debate which bring out interesting rhetorical perspectives on what economists think being a science is: here and here.   Greg Mankiw's little paper on "The Macro-economist as a scientist and engineer" [PDF].  And finally, on Crooked Timber, Henry Farrell applied public choice theory to economics.