Kevin T. Kelly
Department of Philosophy
Carnegie Mellon University
The paradigm is a set of concrete problem solutions.
Methodological rules are generalizations that must be abstracted from these.
Underdetermination: The paradigm can be agreed upon when the rules it exemplifies are not:
If the coherence of the research traditionis to be understood in terms of rules, some specification of common ground in the corresponding area is needed. As a result, the search for a body of rules competent to constitute a given normal research tradition becomes a source of continual and deep frustration (p. 44).
Scientists can agree that [someone] has produced an apparently permanent solution to a group of outsenanding problems and still disagree... about the particular abstract charateristics that make those solutions permament (p. 44).
Wittgenstein: you can recognize particular chairs as chairs without having a fully general definition of chairs. All that is necessary is a "family resemblance" based on typical but not necessary traits (e.g., 4 legs, made of wood, back support).
Family resemblance suffices to place solutions in a paradigm. There may be no rules to be found.
Epistemic priority arguments:
Hard to find methods that characterize particular paradigms: like finding a definition of "chair" or "game".
Scientists learn paradigm problem solutions, not theories and principles in the abstract. Rules are not explicitly transmitted.
If scientists are asked about scientific method, they do little better than laymen in saying what it is.
Any rule, being inferred from the paradigm, would be less trustworthy than the paradigm itself.
Scientific consensus can be explained by the paradigm without recourse to methods.
Rules become important only when the paradigm is in trouble or in the pre-paradigm, foundational level.
When methods are discussed, they distinguish rather than unify the different camps.
When methods exist, they apply much more broadly than in a single paradigm.
Different students can be in different paradigms in the same discipline if they read different texts with different examples.
A methodological response:
Who said that methods have to characterize particular paradigms? This is a very powerful premise provided with no argument.
Scientific method is supposed to explain how science finds the truth and who wins in scientific revolutions. It is not supposed to characterize particular paradigms. Surely, paradigms will reflect historical and empirical contingencies.
Similarly, logical rules of proof are not supposed to characterize different branches of mathematics (one logic for calculus, one logic for algebra, one logic for geometry). These result from adding different definitions and assumptions.
What scientists say about method during revolutions must be taken with a grain of salt. Obviously, there is a rhetorical purpose for everything said in the thick of a battle for survival. Thus, the fact that traditions differ in what they say at such times does not really count against the idea of scientific method. To be plausible, such pronouncements do have to stem from valid aspects of scientific method that perhaps arise in light of the particular metaphysical presuppositions of the paradigm.
Who said that the rules of science have to be obvious to scientists? Scientists merely have to recognize violations when they occur.
Who said it should be easy for scientists to state these rules? We all master English grammar without being able to say what the principles are. But linguists can tell you. And they are much more complicated than most proposals for the nature of scientific method!
If the paradigm itself does not have to be characterized by methodological rules, then perhaps characterizing the rules is easier than Kuhn thinks.
If the paradigm were chosen over its competitor in light of tacit methodological principles, then presumably these principles are more robust than confidence in the paradigm, so the paradigm is not necessarily prior in the order of justification (e.g., in the next revolution!)
Science is taught using paradigmatic problem solutions because (a) it is empirical, so the data matter and (b) because the teachers cannot state what scientific method is. Neither point implies that there are no underlying principles of theory assessment.
In the biological and social sciences, as opposed to antique physics, explicit statistical methodologies are used and can be explained in a theoretical manner. These methods are not intended to characterize particular paradigms, however.
When different methods are recommended, the different traditions could be emphasizing, in a self-serving way, different aspects of a single method. Or they may be tweaking parameters of a single method in different directions.
Maybe particular paradigm choices are not what methodology is supposed to explain. Maybe it constrains choices of scientific strategy (e.g., as being faster or more reliable at finding the truth) without constraining particular scientific beliefs (equally good strategies could differ in a particular theoretical preference).
Anomaly = paradigmatic expectations violated by nature.
Normal science shuns novelty and thereby inevitably produces it!
Novelty = paradigm alteration
Discovery = novelty of fact
Invention = novelty of theory
Fact/theory and discovery/invention are ultimately baseless distinctions.
Anatomy of discovery:
Awareness of anomaly.
Extended investigation of anomaly.
Paradigm adjustment.
Anomaly now expected.
Anomaly is new scientific fact.
Example I: Discovery of Oxygen
1774 Priestly: identifies sample from heating red oxide of mercury as nitrous oxide.
1775 Priestly: identifies sample as dephlogisticated air.
1774 Lavoisier: identifies as ordinary air.
1777 Lavoisier identifes as one of two main constitutents of the atmosphere. Result of combining caloric (heat fluid)with the principle of acidity.
Lavoisier recognized oxygen in light of his revisions of the phlogiston paradigm.
Discovery = recognizing that something is + recognizing what something is.
Discovery is an extended process, not an event since recognizing that precedes recognizing what.
Example II: Discovery of X rays
Roentgen noticed glowing of barium platino-cyanide screen during a cathod ray discharge across the room.
Cause of glow propagated in straight lines and cast shadows.
Indifferent to magnetism.
Not cathode rays. Something like light.
Example III: Discovery of Leyden jar
Bottling electrical fluid: touch water in glass to a static charge. Touching water yields shock.
Didn't realize the hand around the glass was crucial.
After a sequence of surprises, the Leyden jar and Franklin's theory emerged as first paradigm.
Bruner and Postman (1949) Anomalous playing card experiment.
People shown normal and impossible cards (black hearts).
For short times, all cards identified as normal.
For long durations, distress. "I'm not even sure what a spade looks like!"
Discovery cycle:
Paradigm ===> specialization and resistance to change ===> increased precision ===> unavoidable anomaly ===> new paradigm
Resistance to change causes change.
Normal science is even less happy about inventing new theories.
New theories preceded by crisis in old paradigm.
Crisis = consistent inability of old paradigm to deal with anomalies.
Example I: Copernicus!
Major cause: Complexity of Ptolemaic theory was increasing faster than its accuracy.
Minor social causes affecting timing of breakdown:
Social presure for calendar reform.
Medieval objections to Aristotle's physics.
Rise of Renaissance NeoPlatonism.
Example II: Lavoisier's oxygen theory
Pneumatic chemistry (air pump experiments show that gases react chemically).
Phlogiston theory couldn't keep up with new discoveries.
Many new versions of phlogiston theory. Classic sign of trouble.
Attention to weight relations (roasted things weigh more).
This meant plogiston must have negative weight--- didn't fit with Newton's gravitation theory.
Phlogiston paradigm failed to generate consensus among new versions.
Example III: Relativity theory
Early attacks on absolute space were logical. Nobody suspected observable effects.
Wave theory of optics: speed of earth through wave medium should be detectable.
Drift not detected (aberration, Michelson-Morley).
Ether drag theories to account for null result.
Hard to accommodate drag in Maxwell's theory.
Lorentz-Fitzgerald contraction hypothesis generates new puzzles.
Einstein's theory.
Common features
Prolonged crisis
Anomalous problems resembled those solved in the past (not strange or irrelevant)
Solutions had been suggested as possibilities and ignored earlier.
Aristarchus ---> Copernicus
Leibniz ---> Einstein
Hooke, Rey, Mayow ---> Lavoisier
There are always alternative theories, but they are ignored by the paradigm until it stumbles (depth-first search again).