How does science progress: Kuhn

0.1 Introduction

Consider what we will call the “positivistic view of science”

Definition 1 (postivistic view of science) Science consists of the following key features:

  1. science a constellation (aggregation) of a set of facts, theories, and methods. Science is understood as a stockpile.
  2. science is accumulative: science is the piecemeal development of this set of facts, theories, and methods.
  3. science is realistic: scientific theories and facts describe the real world and each generation, each new theory moves us closer towards describing the real world. add to science, and we move closer and closer toward truth
  4. science is governed by a value-free, neutral, objective method that relies heavily on neutral observations, mathematics, and logic.

Kuhn views this picture of science as problematic. Kuhn starts The Structure of Scientific Revolutions (hereafter SSR) by claiming that history can effect a profound transformation in how we view science. According to Kuhn, the historian of science has two main tasks:

  1. who discovered what scientific fact, law, or theory and when
  2. what error kept scientists from a swifter discovery of this fact, law, or theory (what did they have to overcome to make their discovery)

Concerning the second of the two tasks, Kuhn notes that (i) historians have struggled with giving an account of what part of earlier theories was “myth” and what was “scientific” and (ii) this renders problematic our view of science as accumulative. Consider a theory (call it “theory X”) that was once accepted as scientific knowledge but is later replaced by a newer theory (call it “theory Y”). Concerning theory X, the historian faces a dilemma.

Definition 2 (Historian’s Dilemma) If theory X is a myth (non-science), then this myth was produced by the same general method and accepted by the same general reasons that produces scientific knowledge. If theory X is scientific knowledge, then X and not-X are both true.

There are three possible responses to the dilemma if it is genuine.

  1. reject that science is accumulative
  2. reject that there is anything like an objective, value-free scientific method
  3. reject that science is accumulative and there is anything like an objective, value-free scientific method.

Discussion 1 What is the correct answer to the historian’s dilemma to science?

Let’s assume, temporarily, that (3) is the correct route. If this is the case, then we need a different picture of how science has proceeded through history. Thomas Kuhn provides such an image. In contrast to an image of science that is the result of gradually accumulating knowledge through the use of an objective, unquestioned, value-free method, Kuhn provides the image of science as one marked by radical disruptions (revolutions) and making use of an often arbitrary, value-laden method.

0.2 Kuhn and the Structure of Scientific Revolutions

Before delving into the details of Kuhn’s account, it is helpful to (i) obtain a general picture of how Kuhn sees the development of science and (ii) get clear on some of the key terms.


SVG-Viewer needed.


Figure 1: Kuhn’s picture of the development of science

Note 1 (Science begins as pre-science.) According to Kuhn, the story of how science progresses begins with prescience. “Prescience”, in short, is characterized by there being no consensus about the general nature of some scientific phenomenon and debate about the fundamentals of science.1 In particular, “pre-science” has the following features:

  1. lack of consensus about fundamentals: entities, laws, concepts, how science should be done
  2. all the facts seems equally relevant, known seem more important than any other (SSR:15), e.g. the heat as a result of chemical reaction and the heat of manure
  3. casual observation about some phenomenon (a collection of things people have observed)
  4. continual constant competition between distinct, incompatible views of the world (nature)
  5. marked by disagreement
  6. each individual works in their own field, often with very different fundamental assumptions
  7. the existence of a number of different theories but each theory had its say by being able to account for some specific attribute, e.g. with respect to light, each theory is able to account for some aspect of light but it struggled with others.2
  8. no shared method

Note 2 (Pre-science evolves to normal science when individuals adopt a scientific paradigm.)

The transition from pre-science to normal science occurs when individuals become committed to what is known as a “paradigm”. 3

Definition 3 (paradigm) A set of general theoretical assumptions, laws, methods, techniques that are adopted by a community of individuals, but also:

  1. contains general (often unstated) metaphysical principles, explicitly stated theories, and general laws
  2. is codified in some exemplar theory, historical achievement, or classic text, e.g. Newtonian mechanics, etc. (even even though the achievement or text is rarely discussed)
  3. is articulated in scientific textbooks (namely, the textbooks articulate the theory that is accepted by scientific practitioners as well as a number of successful uses of the theory and (i) observations (facts) that the theory explains and/or (ii) experiments that support the theory).
  4. settles fundamental questions, no longer a need to build the field anew, no need for first principles (SSR:20): entities, laws, concepts, how science should be done
  5. suggests which experiments are worth performing, the types of puzzles and problems that can be solved, the types of questions that need answers, and what kinds of answers are acceptable
  6. focus on “the subtlest and most esoteric aspects of the natural phenomena” (SSR:20)
  7. no longer the need to publish books, but instead individuals focus on short articles in specialized journals (SSR:20).
  8. creates hyper-specialization so that the workings of the scientist cannot be understood by the lay person or other scientists working in different fields (SSR:20-21)
  9. attracts a large majority of the younger scientific practitioners while the older group of scientists gradually disappear (SSR:19)
  10. leads to the formation of specialized journals and academic societies (SSR:19)
  11. finds its way into the academic curriculum (SSR:19)

In short, a paradigm gives us a particular model of scientific activity that is usually marked by some significant scientific discovery or theory. This model of scientific activity has two characteristics:

  1. it needs to be attractive enough to bring about a consensus about fundamentals.4
  2. it is open-ended enough to leave various problems that can be taken up by its practitioners (SSR:10-11). That is, there is enough flexibility for scientists to investigate highly specific issues.

Example 1 (History of Light)

  1. In the 18th century, light was thought to be a collection of material corpuscles or little particles (Newton’s corpuscular theory of light). In this paradigm, you would try to investigate various properties of light under the assumption that light is a particle.
  2. In the 19th century, light was thought to simply be a wave (Fresnel, Young, et alia). In this paradigm, you would try to investigate various properties of light under the assumption that light is a wave.
  3. Today light is said to consist of photons that bear characteristics of waves and particles. This is known as wave-particle duality (Planck, Einstein, et alia). In this paradigm, you would try to investigate various properties of light under the assumption that light behaves like both a wave and a particle.

In each case, scientific practitioners are said to accept a paradigm. They accept certain fundamental assumptions about the nature of light and then proceed to investigate more esoteric, specific, highly-specialized problems relating to light. What they do not do is question its fundamental nature, viz., what light is.

In short, pre-science begins as a chaotic group of theories where there is little consensus about what a certain scientific phenomenon is. Each individual creatively tries to address what that phenomena is, how to investigate it, and various other issues from the ground up. This leaves us with a chaotic and diverse array of theories that are largely at odds with each other.

With the acceptance of a scientific paradigm, normal science begins. In normal science, there is a consensus about what light (or some other phenomenon) is, and this information is codified in textbooks, scientific journals, and societies. Scientists turn from fundamental questions to highly esoteric questions about that phenomena and attempt to answer these questions within the terms of the paradigm.

Note 3 (Once a paradigm is accepted, normal science is now practiced.) Once individuals accept a paradigm, these individuals are practicing “normal science”. Kuhn notes that “normal science” is the kind of activity that scientists do most of the time, that it involves certain shared assumptions about the fundamental entities that compose the world (and how they relate), what kinds of questions can be legitimately asked, and what are legitimate answers to these questions. For the most part, practitioners of normal science attempt to resist novelty. It is only when there is some problem (anomalies) that cannot be legitimately answered by the accepted methods and solutions that scientists will allow a hearing for more creative solutions.

Definition 4 (Normal science) Normal science is “research firmly based upon one or more past scientific achievements, achievements that some particular scientific community acknowledges for a time as supplying the foundation for its further practice” (SSR:10).

Note 4 Normal science then is:

  1. marked by a paradigm (certain fundamental assumptions taken as given)
  2. sets the standard for the sort work that will be considered legitimate and acceptable
  3. the paradigm sets the stage for the types of puzzles and problems that can be solved
  4. also determines the types of questions that need answers and what answers are acceptable
  5. based upon some historical scientific achievement (even though the achievement is rarely discussed)
  6. articulated in scientific textbooks (namely, the textbooks articulate the theory that is accepted by scientific practitioners as well as a number of successful uses of the theory and (i) observations (facts) that the theory explains and/or (ii) experiments that support the theory).
  7. occupied by some problems defined as important and conducted according to a specific method deemed legitimate

Note 5 (When falsifying phenomena build up in normal science, there is a crisis.) Various falsifying observations or phenomena are observed, these build up to a point that individuals begin to question the paradigm. Normal science evolves into a state of crisis.

Definition 5 (Crisis) A crisis is a state of science where serious anomalies threaten the current paradigm.

Definition 6 (anomaly) An anomaly is any problem with a paradigm.

The presence of an anomaly becomes serious to a paradigm when

  1. it is a problem with some fundamental component of the paradigm that resists repeated resolution within the paradigm,
  2. it is a problem with the paradigm that is of pressing social concern
  3. there are a large number of them. See Chalmers What is this thing called Science, p.105).

When a paradigm falls into crisis, creative solutions emerge. These new solutions, in effect, challenge the fundamental assumptions of the old paradigm, and, in effect, propose new paradigms. When a new paradigm is accepted and the old paradigm rejected, a revolution has occurred. “The extraordinary episodes in which that shift of professional commitments occurs”, Kuhn writes, “are the ones known in this essay as scientific revolutions” (SSR:6).

Some revolutions include:


SVG-Viewer needed.


Figure 2: Kuhn’s picture of the development of science

0.2.1 Some Issues with Kuhn’s theory

The most radical part of Kuhn’s theory is the following:

  1. the history of science and theory-change does not reveal a linear progress to increasingly better theories
  2. science is not characterized by the use of an objective, value-free, theory-neutral method.

What argument does Kuhn have for these two claims? The short and sweet of Kuhn’s response is this: when a revolution occurs, and the scientific community replaces its current paradigm with a new paradigm, the new and the old paradigms are not comparable. They are incommensurable.


SVG-Viewer needed.


Figure 3: Kuhn’s argument that science does not progress and theory-change is not completely objective.

0.2.2 Incommensurability

What does it mean for two paradigms to be incommensurable and why would this imply that science is not progressive and theory change is not objective?

Definition 7 (incommensurability) X is incommensurable with Y if and only if X cannot be compared with Y in terms of whether X is better than Y.

If two paradigms are incommensurable, they cannot be compared. And, if they cannot be compared, one cannot be said to be better or worse than the other. And, if one cannot be said to be better or worse than the other, then this seems to imply that the choice to abandon one paradigm for another is not guided by any feature of the new paradigm that makes it any better.

In what ways are scientific paradigms incommensurable? There are two ways that two paradigms P1 and P2 can be said to be incommensurable.

The first is what we can call linguistic incommensurability. This type of incommensurability can be thought of in two different ways:

  1. there are terms in P1 that have no corresponding equivalent in P2 (and vice versa). If this is the case, then it might be said that P1 cannot be compared with P2 since P1 and P2 refer to differ parts of the world. Consider that one of the things that scientific paradigm does is determine the types of things that exist. Thus, if P1 had a term for the medium that light traveled through (e.g. lumineferous ether) but P2 had no corresponding term or notion for this entity, then it looks like P1 and P2 are linguistically incommensurable.
  2. there are terms in P1 that have mean something different in P2 (and vice versa). The underlying idea here involves what is known as “meaning holism”, the claim that the meanings of all of the terms in a scientific theory depending on the other parts. Thus, to understand the meaning of “light” in the Newtonian paradigm would be to understand it as a particle and obeying Newtonian mechanics. To understand it in the 19th century would be to understand it as a wave moving through an invisible substance.

SVG-Viewer needed.


Figure 4: Linguistic incommensurability

In short, if two paradigms are linguistically incommensurable, the claim is that they cannot be compared. And, if they cannot be compared, one cannot be said to be better or worse than the other. And, if one cannot be said to be better or worse than the other, then this seems to imply that the choice to abandon one paradigm for another is not guided by any feature of the new paradigm that makes it any better or worse.

Objection 1 (Incommensurability and Incompatibility.)

The problem with this notion of incommensurability is that it seems as though it makes it impossible to talk about differences between paradigms at all.

The way that Kuhn sets up the shift from one paradigm to the next seems to imply that the different paradigms are incompatible. That is, we shift from one paradigm that makes certain claims about the fundamentals of the universe (or some phenomena) or about laws to another. The problem though is that if two paradigms are incommensurable then they cannot also be incompatible.

If two paradigms are incompatible, then they both cannot be true at the same time. So, if X and Y are incompatible: if X is true than Y is false and if Y is true, than X is false. But, if X and Y are incommensurable, X cannot be compared with Y, and so we cannot say that X and Y are incompatible.

But this is absurd since we think that paradigms (scientific theories) do conflict with each other.

This brings us to the second kind of incommensurability, what we can call evaluative incommensurability. On this incommensurability, P1 and P2 make use of different standards for evaluating what counts as a good theory, argument, or explanation. That is, if only a paradigm sets the standards of evaluation for what counts as a good theory, then each paradigm will count its own theory and arguments as successful, while it will count all others as unsuccessful. That is, if T1 meets the standards of P1 but not P2, then P1 will count T1 as a good theory while P2 will count it as a bad theory. Similarly, if T2 meets the standards of P2 but not P1, it will be counted as good theory by P2 but not P1.

We can imagine simple examples where a paradigm sets the standard for what constitutes a good explanation by requiring that we take into account certain entities and relations that another paradigm might not allow, e.g. an explanation only counts as good iff it can take into account of how it fits with God’s goodness.

Example 2 (Causal/mechanical or mathematical/formal explanation) We can imagine two different paradigms. The first requiring a causal/mechanical explanation for certain phenomena while the other requiring a mathematical/formal explanation.

For example, Newton’s theory of gravity states that “The gravitational attraction force between two point masses is directly proportional to the product of their masses and inversely proportional to the square of their separation distance. The force is always attractive and acts along the line joining them.”

Note that this explanation of gravity is purely a mathematical / formal explanation and does not state what causes gravity. That is, it does not state the mechanism that causes two masses to attract. If a paradigm says that purely mathematical explanations are sufficient, then Newton’s theory is a good theory. If it says that it is necessary to explain what causes gravitation attraction, then this theory would be problematic.

This particular line of argument assumes that there is no paradigm-neutral (theory-neutral) data, information, evidence that we can appeal to that would allow us to say that one paradigm is better than another paradigm. And, to some extent, there is good reason to believe this given the following:

  1. perceptual beliefs are partially determined by background beliefs
  2. observations are often highly conditioned by theory

In short, given that each paradigm judges what is and isn’t a good argument and there not being any theory-neutral (paradigm-neutral) evidence to decide between different paradigms, it looks like there is no way of determining which paradigm is better or giving any account of scientific progress.


SVG-Viewer needed.


Figure 5: Kuhn’s argument for non-progress of science and non-objectivity of theory-change

Objection 2 (Kuhn’s theory makes scientific knowledge relative to a paradigm.) If what makes a theory successful is set by the paradigm, then a theory’s success is relative to the paradigm from which it is evaluated. Theories are not getting better and better as we transition from one theory to the next (since each paradigm evaluates its theory as sufficient and others as insufficient), but are being evaluated by different paradigms. In short, there is no measure of evaluation for evaluating paradigms.

This is counter-intuitive because:

  1. it cannot explain things like greater predictive success of science
  2. it ignores the development of technology
  3. it assumes there can be no paradigm-neutral account of information but presumably there is some information that cuts across some paradigms, e.g. observations accepted in both the Ptolemaic and Copernican systems.