Tuesday, April 10, 2012

Critique of Kuhn: Scientific Revolutions and Semantic Incommensurability

Preamble and Notes to the Reader 
People is always axin' me (jk, lol, #thatneverhappens), why don't you post your papers?  The answer is that while, for the most part, I enjoy writing in my blog, for some reason I don't like how my philosophy papers turn out.  They seem sterile.  Anyway, I'm posting this paper because it's the first paper I've written since undergrad that I feel ok about.  I don't think its content is genius or anything, but it is in the style and tone that I've been aiming for.  Also, lucky for you all, it's a short paper (by grad paper standards, anyway).

For my non-philosopher friends, before you read, don't be scur'd of the technical term "semantic incommensurability" in the title.  It just means that if two people hold different theories that share words, they cannot have meaningful debate over theory choice because the words they use refer to different things or are conceptually different.  

To use a very simple example, suppose there is a debate between a prescientific person and a Newtonian.  They might both use the word "weight".  Kuhn says they can't engage in any meaningful debate because "weight" for the prescientific refers to an object's intrinsic "nature" to fall toward the earth.  For the Newtonian, "weight" is a technical term which refers to an object's mass times its rate of acceleration.  Because their terms refer to different concepts, Kuhn argues they are talking past each other (unbeknownst to them).  I explain it more in the essay, but that's the main idea.

For my philosopher friends, you are probably already familiar with Kuhn's arguments so you can skip to mine, starting with the section entitled "Case Study: Paradigm Shift from Newtonian to Einsteinian Physics and Semantic Incommensurability".

Anyway, I hope you guys enjoy and/or learn something...

Zombie Newton is Havin' a Crisis:  A Critique of Kuhn's Account of Science and Semantic Incommensurability 

At the heart of Kuhn's view of revolutionary science is that paradigm shifts arise out of crises: A growing group within a community of experts will call for the overthrow of a theory because of its inability to satisfactorily account for observations. In such situations--when one paradigm is adopted over another--the choice is not one that can be made by appeal to logic or evidence: it is made by social persuasion. A consequence of this view is that science is not a cumulative enterprise, rather it is a series of extrarational choices of one paradigm over another. The purpose of this paper is to put pressure on Kuhn's argument that scientific revolutions necessarily involve a choice between logically mutually exclusive theories. Specifically, I will seek to undermine one of his central claims that supports his non-cumulative view: that theoretical terms have fundamentally different meanings between theories. 

Instead, I will propose that often enough meaning is sufficiently preserved in intra-theoretical terms to create a bridge between the old and new paradigm, thereby lending support to the “science-as-cumulative” hypothesis. Given the constraints on this paper's length, I will limit my investigation to Kuhn's example of the incommensurability of terms between the Newtonian and the Einsteinian paradigm. In the first section, I will give a general overview of Kuhn's argument for a revolutionary view of scientific paradigm shifts. Second, I will briefly outline Kuhn's argument for semantic incommensurability. Third, I will propose three counter-arguments. Finally, I will conclude that given the weakness of the incommensurability argument, the cumulative view is a plausible alternative account of scientific shifts between paradigms, including the move from Newtonian to Einsteinian physics.

Overview of Kuhn
Thomas Kuhn argues that when an area of science adopts one scientific theory over another, the process mirrors that of political revolution rather than that of evolution. Political and scientific revolution is a choice between two incompatible modes of thought--where adoption of the new “demands the destruction” of the old (Kuhn, p. 89). He relies on his politics/science analogy to more carefully spell out the criteria for what is constitutive of revolutionary change and the crises that precipitate it.

The first criteria is that a portion of the scientific community will be increasingly doubtful of the prevailing paradigm's ability to adequately explain the natural phenomena it is purported to explain. The inability of a theory to explain a type or types of phenomena will lead to “scientific unrest”--the precondition to revolutionary change and the breaking of test tubes. The second characteristic is that since competing camps reject the other's paradigm, any resolution

is not and cannot be determined merely by the evaluative procedures characteristic of normal science, for these depend in part upon a particular paradigm, and that paradigm is at issue. (Kuhn, p.89)

In short, disputes can only be settled by appeal elements that are external to both competing paradigm.

Kuhn's view is in direct opposition to the “science-as-a-cumulative-enterpise” view. Unlike proponents of this view, it is Kuhn's contention that adoption of new paradigms very rarely occurs through discovery of unanticipated novelties. It is only “normal science” that is cumulative1. When scientists run up against a growing amount phenomena that cannot be explained by or be made to conform to an existing theory, then a new incompatible theory can arise. Adopting this new theory entails the rejection of the previous theory and its concepts because paradigms can only be rejected or accepted as wholes. 

A central supporting argument to Kuhn's main thesis is that the theoretical terms of one theory cannot be understood in a rival theory because of the theory-dependence of meaning, or “semantic incommensurability”. The implication is that, although two paradigms can share a theoretical term nominally, the term has fundamentally different meanings since the terms derive their meaning from within a theory-specific paradigm. For example, 'mass' in Newtonian physics refers to force divided by acceleration, while 'mass' in Einsteinian physics refers to energy divided by the speed of light squared. 

Case Study: Paradigm Shift from Newtonian to Einsteinian Physics and Semantic Incommensurability

Kuhn appeals to the historical example of physicists' adoption of the Einsteinian over the Newtonian paradigm as support for his science-as-revolution hypothesis. Within this paradigm shift, he argues for the semantic incommensurability of Newtonian and Einsteinian concepts. The basic argument is that Newtonian concepts cannot be derived from Einsteinian concepts because they refer to fundamentally different things. (E.g. Each theory's concept of mass refers to different things, as mentioned in the previous section.) In response I will give three counter-arguments.

Argument From Zombie Newton and Inter-theory Communication
Is this idea of semantic incommensurability right? Suppose Newton himself came back from the dead and walked into Einstein's physics lab just after Einstein had figured out his new physics. Upon hearing it, poor zombie Newton goes into crisis. Is the Newtonian concept of mass so different from Einstein's that any discussion of mass between the two would entail that they talk past each other? Would not zombie Newton be able to understand when Einstein speaks of mass? I think that in a limited way he would. He would at least have a general notion of mass no different from the lay general public of today: laypeople can talk to their physicist friends about mass without too much head-scratching. We might argue that although the notions of mass are measured differently, they are not so different that zombie Newton and Einstein wouldn't find each other's use of 'mass' unintelligible.

We can look to other fields of study and see that intra-theoretic communication is quite common. Suppose there's a room with cognitive scientists, psychologists, behaviourists, cognitivist, and AI theorists. Although all of these people work within different but related theories, it seems odd to say they wouldn't understand each other if they were speaking about 'beliefs'. Concepts which are shared between different theories don't seem to be so radically different such that (necessarily) there is no overlap in meaning to the point where meaningful dialogue is impossible. It seems as though there are plenty of examples of people holding different theories with common terms who can engage in meaningful dialogue without speaking past each other.

Kuhn's Possible Reply
Kuhn might reply that the similarities between their concepts are are trivial. In the case of zombie Newton, he doesn't have sufficient knowledge of Einsteinian concepts to be able to understand important things that are implicit whenever Einstein refers to mass. The issue is not whether a 17th Century physicist shares a concept with the contemporary general population, rather we want to know if his concepts would allow him to engage in important scientific discussions with his Einsteinian peers. Once discussion turns to measuring the mass of bodies at high relative velocities, the important fundamental differences between the two concepts of mass will come into stark relief.

Perhaps Kuhn is right here. While it is true that Newton's notion of mass would yield similar measurements where slow moving objects are concerned, the fact that their respective theories produce similar measurements would mask the fact that their respective notions of mass measure fundamentally different things. And because they measure fundamentally different things, meaningful debate isn't possible because their terms refer to different things.

Despite this possible Kuhnian response, there are two more inter-related replies to Kuhn's response to show that meaningful dialogue is possible between competing theories at a time of crises. The first is to try to show that the similarities between the nominally identical concepts are more fundamental than the differences; and second, is to propose that rational meaningful debate is possible so long as it is between epistemic peers.

Argument from Fundamentally Similar Meaning
Before arguing that intra-theoretic meanings are commensurable in important ways, I need to lay some ground work2. First, I want to put forward a very basic picture of what science does: Science attempts to measure fundamental aspects of the natural world3. I do not mean to say this is all science does, but I think it fairly uncontroversial that this is a fundamental aspect of the activity of science. Now, for whatever reason, humans--and particularly scientists--have adopted certain conventions about how we “chop up” the natural world. Among some of these aspects, we measure space, time, energy, and mass.

There are several ways we go about measuring these qualities. Take 'space' as an example. Space is comprised of notions of length, width, and depth. To measure space we first had simple notions of “long” and “short”. Then we had comparative notions like “longer”,“shorter”, or “equal”. An incremental step above that is to employ an instrument, such as a string, to measure the differences. In the final step of measurement, we quantify. I might start out by choosing a piece of string and subsequently measure everything in terms of that string. I'd say, “this wall is 5 strings long”. Perhaps I go around measuring things for a year, then come back to the wall and find out that the wall now measures 5 ½ stings. Now, obviously the wall didn't grow; over the course of the year there was wear and tear on the ends of my string which caused it to shorten. The standard by which I chose to make measurements, in this case, didn't yield consistent measurements over time. So, later, I might choose a metal rod as my standard.

We can observe a similar progression in the measurement of time. Early humans quantified time according to the movement of a shadow on a sundial, or relative movement of the sun to the earth. Under this standard, time seemed to pass more quickly in winter than in summer. Eventually, other standards were adopted, such as the swings of Foucault's pendulum, the piezo-electric effect of quartz, all the way up to today's measurement of decay rates of radioactive elements.

Behind this simple explanation there is something very important going on. The first is that how I choose to quantify a particular aspect of the natural world is in part conventional. I could have chosen a string of any length and I could have chosen a measuring instrument of any material. I could also have chosen to measure everything in terms of my foot or thumb length. In the case of time, I could have quantified time according to any of the above standards and I could have divided up their cyclical rhythms an infinite number of ways.

But there is also an aspect of quantifying that isn't conventional—these are facts about the natural world and logical laws. No matter how I choose to measure length (and by extension, space) it is a fact of nature that certain materials will make better measuring instruments than others. In the case of time, I can compare two cyclical phenomena to determine which cycle is more regular relative to the other. It is by appealing to these facts of nature that I choose my convention. A logical law will make so if I add two lengths together, their sum will not be greater or less than the total of the two lengths.

Now lets relate this information about conventionality, facts of nature, and logical laws back to Kuhn's notions of semantic incommensurability. Suppose there are two populations that use different conventions for measuring time: One uses Foucault's pendulum swings and the other uses the piezo-electric effect of quartz. Would it make sense for the latter group to say to the former: “No, no, no! You've got it all wrong! You're measuring the wrong way! Time isn't measured by swings of a pendulum, time is measured by quartz' piezo-electric effect.” This type of claim seems no more odd than a modern physicist telling the quartz proponents that the proper way the measure time is according to atomic half-lives. Furthermore, it would be equally curious if one group were to claim that the other group was measuring the wrong thing. Time itself isn't pendulum swings or atomic decay rates any more than length is pieces of string.

How can we describe our intuitions that debates between these groups are kind of silly? It is because the different groups are disputing the conventional part of the concept of time—how it is measured. There is no objectively right answer, nevertheless, it is true that choosing the convention of quantifying time according to the intervals of radio-active decay will yield more consistent measurements. But it certainly seems that all groups are intending to measure the same fundamental aspect of nature. 

This brings us back to Kuhn. When I say that a Newtonian can meaningfully talk about mass with an Einsteinian what do I mean? I mean that, although there is a dispute between the pendulum people and the quartz people over what measurement convention should be adopted, they are both referring to the same fundamental aspect and facts of the world. It might be true that the Einsteinian convention of measuring mass as equal to E/c2 yields more consistent results, but both theories aim to quantify the same fundamental aspect of the natural world. 

So, contra Kuhn, I argue that semantic incommensurability between Newtonian and Einsteinians is not so great that no meaningful debate can occur between between then. Despite important theoretical differences, the fundamental meaning of their terms, which corresponds to how we fundamentally “chop up” the universe, allows for important mutual understanding.

I'd like to qualify my claim and acknowledge that the different ways in which Einsteinians and Newtonians measure mass arise out of very important theoretical differences, however, I still think that the fundamental notions of mass are sufficient for some meaningful debate to be possible. Furthermore, I will show Kuhn's incommensurability claim is further undermined when the argument from fundamentally similar meaning is coupled with the subsequent argument:

Argument from Epistemic Peers
Lets return the argument from zombie Newton. I suggested that Kuhn might reply that while zombie Newton might nominally share Einstein's notion of mass, his understanding wouldn't be sufficient to engage in important scientific debate. Such debate wouldn't be possible because zombie Newton's epistemic position would lack all the empirical and theoretic scientific advances since he last dawned his lab coat. Specifically, Einstein was only able to develop his theory because he had a sophisticated particle theory and non-Euclidian geometry, among other concepts foreign to zombie Newton. So, given the paucity of zombie Newton's current background knowledge, it is quite plain that he would not be able to understand everything that Einsteinian mass entails. Consequentially, Kuhn would be right to say they would be speaking past each other.

While semantic incommensurability may be true of zombie Newton, I don't think it would be true of Einstein's contemporaries; that is, his epistemic peers. When Kuhn argues that paradigm shifts are revolutionary, they only appear so because he is narrowly focussing his proverbial microscope on the theories and principles of Newtonian and Einsteinian physics. But in engaging in this narrow approach his is excluding from his account the important developments outside of and adjacent to Newtonian physics proper that allowed Einstein to develop his theory. After all, Einstein did not develop his theory ex nihilo. Quite certainly, Einstein's peers would also have been familiar with non-Euclidian geometry and have shared knowledge of the advanced particle theory.

From this fact we make several points against Kuhn. The first is that these adjacent fields of development could act as bridges for Einstein to explain to his Newtonian contemporaries how to derive his theory in terms of theirs. Presumably, these new developments are what allowed him to make his own leap from Newtonian physics; if Einstein used those concepts to move from Newton's model to his own, it seems reasonable that by sharing his reasoning with epistemic peers, others could too. Or at least they could intelligently debate it.

Second, that Einstein's epistemic peers had access to the same relevant advances weakens Kuhn's claim that choice between paradigms cannot be decided by rational debate. Einsteinian physics appeals to non-Euclidian geometry and advanced particle models which, although not part of Newtonian physics, are epistemic considerations from which one can make rational arguments. 

Finally, if we acknowledge that Einstein's theory was possible because of important relevant advances in adjacent fields, there is support for the idea that changes new paradigms can be described as cumulative4. While it is true that there was no direct leap from traditional Newtonian to Einsteinian physics, such a move was only possible because of advances in other scientific areas. So, the argument is that since there was a kind of synthesis between Newtonian physics and other areas of science, when viewed through a wide-angle lens, the picture of change looks evolutionary rather than revolutionary.

 Kuhn proposes that revolutionary science necessarily entails a non-rational choice between two mutually incompatible theories. Furthermore, because paradigms are mutually exclusive and the decision to adopt one over the other isn't rational, science isn't necessarily cumulative and/or progressive. One specific argument he makes for his view is that of semantic incommensurability whereby proponents of competing theories cannot engage in meaningful debate because the theory-ladenness of their respective terms implies they refer to fundamentally different things. They end up speaking past each other. 

In this paper, in the context of Newtonian vs Einsteinian physics, I have argued against there being semantic incommensurability to a degree such that important scientific debate isn't possible between proponents of competing paradigms. In the argument from zombie Newton I suggested it is an empirical fact that meaningful communication does occur between proponents of different paradigms. In the argument from fundamentally similar meaning I showed that the essential meaning of basic terms like time, space, energy, and mass are preserved in significant ways across theories. Finally, I advanced the idea that, in a community of epistemic peers, rational appeal to concepts and principles outside of a paradigm make possible meaningful intra-paradigm discussion.

If, as I suggest, meaningful debate is possible and there is sufficient shared relevant background knowledge, then we should acknowledge that there are bridge concepts and principles. If there are bridge concepts and principles linking an older theory to a new, then there is a case to be made for the evolutionary and cumulative view of science.

1By “normal science” Kuhn means the scientific activity of answering questions and solving problems within a particular paradigm, not resolving disputes between competing paradigms.
2The information on measurements is from my lecture notes from Dr. Richard Creath's class on Quine and Carnap.
3At this point I'm being intentionally vague by employing the term “aspect”. Depending on your metaphysics this could refer to properties, kinds, laws—fill in the blank as you wish. At this point, it is unimportant. I only seek to make the point that measurement is a key activity of science.
4My claim is not that every paradigm shift is evolutionary rather than revolutionary; only that, it is not as improbable as Kuhn makes it out to be. And the shift from Newtonian to Einsteinian physics might be one such instance.


  1. Where's your argument on semantic incommensurability? This should involve things like 'meaning holism' and 'Frege sense reference', which are central to Kuhn's subscription with the 'descriptive theory of meaning'. Semantic incommensurability involves the assessment of linguistic theories - not measurement. You see, Kuhn's incommensurability argument is backed up by 3 parts (which need to be kept separately): measurement incommensurability, perceptual/observational incommensurability (regarding theory laden influences) and semantic incommensurability.

    These terms were kind of mixed throughout your argument and were misused.
    Would be nice to know how to resolve the semantic issue.... its central to my undergrad paper - but I'm finding it too complex :(

    1. Hi Joshua,
      Thank you for you comment. I haven't done any phil sci since I wrote this piece a few years ago so the details are a bit hazy. When I get a chance I'll go through it with your points in mind and make the necessary clarifications. Thank you for pointing them out. In the meantime, if you're writing an undergrad paper I'd recommend you focus on just one aspect of incommensurability. As a general rule for papers, the narrower the focus, the deeper you can go, the better your prof will like it!

    2. Hey Amitabha,

      Yes I have been asked to focus on the one area of incommensurability. I chose to write about the semantic version which involves the linguistic arguments. I feel that I have reached my limit for comprehending this stuff!
      All the best,