Week 2 Lecture Notes
Hempel: The Received View continued
Requirements for scientific explanations:
Explanatory relevance: explanatory information (laws and conditions) affords
good grounds for believing that the phenomenon to be explained did or will
occur.
Testability: the statements making up a purported explanation must be
capable of empirical test.
D-N model: the logical structure of one kind of scientific explanation
L1,L2, … Ln Explanans sentences
C1, C2, … Cn Explanans sentences
____________
E Explanandum sentence
E, which is derived deductively from covering laws and initial conditions,
is a phenomenon.
E can also be an empirical generalization that follows deductively from a
set of “higher level” laws.
Relevance is met because E is derived deductively.
But recall that p v q follows deductively from p, and that q can be anything
at all.
Also, if the premises of an argument or line of reasoning contain a contradiction,
any statement whatsoever will follow deductively.
Testability: in principle. Predictions of relativity theory confirmed
by Eddington’s experiment.
Explanations can yield
The discovery of a fact
The discovery of a covering law and a new theory
How some phenomenon can be accounted for by reference
to laws already available
Universal statements: (Universal qualifier) (Mx > Wx)
UD: living things
Mx: x is a mammal
Wx: x is warmblooded
Many so-called laws (some would argue all such laws) hold only approximately
or are idealizations.
Laws vs. accidental generalizations:
(Universal quantifier) (Rx > Ix)
UD: everything
Rx: x is a rock in this box
Ix: x contains iron
Laws can support counterfactual conditionals and subjective conditionals,
and serve as a basis for explanation
Whether a universal statement counts as a law depends in part on accepted
scientific theories.
Sun spot activity and Wall St. crashes
Coffee drinking and sex
Probabilistic laws: the explanans implies the explanandum only with high
probability, yield probabilistic explanations.
Like D-N explanations, an event or regularity is explained by reference to
laws (but probabilistic ones) and specific conditions.
p(O, R) = r
In a long series of performances of random experiment R, the probability
of outcome O is almost certain to be close to r.
Statistical probability statements are tested by examining the long-run relative
frequencies of the outcomes confirmed; confirmation is judged in terms of
the closeness of the agreement between hypothetical possibilities and observed
frequencies.
Standards for accepting or rejecting such hypotheses:
What deviations of observed frequencies from the probability
stated by H are to count as grounds for rejecting H?
How close an agreement between observed frequencies and
hypothetical probability is to be required for accepting H?
Such standards are generally developed in light of what
a given context suggests is the worse of 2 possible errors:
Accepting a hypothesis that could be false
Rejecting a hypothesis that could be true
Probabilistic laws extend to an infinite # of cases:
“The radioactive decay of radium226 is a random process with an associated
half-life of 1,620 years.”
Hanson and the theory-ladenness of observation
Examples from contemporary science:
The Hubble
The Standard Model in physics:
12 building blocks: 6 quarks and 6 leptons
Everyday world is made up of just 3: the up quark, the
down quark, and the electron.
The electron neutrino, observed in the decay of other
particles, completes the first set of 4 building blocks.
Although there are reasons to believe there are no more
quarks and electrons, theorists think there may be other kinds of building
blocks, which may account for the dark matter implied by astronomical observations.
The Bottom Quark experiment: Interpreting the Results
“Backgrounds to observations are a general problem in science. Think of trying
to hear a friend speak in a noisy place. You use your expectation of how
your friend’s voice should sound to separate the voice from the background
noise. To find the upsilon in the noisy background, physicists used their
expectation that a new particle would appear as a bump in the plot of the
mass of all muon pairs.”
Evidence for theory-ladenness
Gestalt experiments, cross-cultural and trans-historic
examples, and scientific revolutions indicate that what an individual observes
is not solely a function of the object being viewed or their sensory receptors.
Background knowledge, i.e., conceptual schemes and prior experience,
together with the expectations that result also determine what is seen.
So, too, education in a scientific discipline impacts what
one is able to observe.
Observations are made possible by, and observation statements
are couched in, language.
A language incorporates a conceptual scheme: an overall ontology,
notions of relationships (identity, causal, transitive, and so forth), numbers,
colors, and so forth.
Thus observation statements presuppose conceptual schemes and
theories, and as fallible as those they presuppose.
Many observations, and certainly many made by scientists, involve
inferences -- not “instantaneous seeing” -- and theories have an inextricable
role in the inferential process
Science often proceeds on the assumption that what we “normally
observe” is explained by what we do not observe: that is, reality and appearance
need not be (and often are not) co-extensive.
Kuhn’s The Structure of Scientific Revolutions
“A role for history”
History demonstrates that science does not proceed by accumulation, with
one theory being replaced by a larger, more comprehensible, but nevertheless
compatible theory. It reveals something much more like this pattern:
“Pre-science” or pre-normal science
to
Normal (paradigm-driven) science
to
Emergence of anomalies
to
Extraordinary (or crisis) science
to
Scientific revolution: new paradigm, new normal science tradition
“Pre-science”:
Disagreements over fundamentals: what phenomena are the most important and
in need of explanation; over metaphysical commitments; over what will count
as an answer to the questions raised; many schools, rather than consensus
Normal science: research guided by a paradigm
A paradigm: A solution/model/theory that seems most promising in its approach
to important issues and gains adherents.
It must also be sufficiently open-ended to enable lots of research questions
and directions to be pursued.
Natural selection: from pre-science to normal science
Copernican hypothesis: from old paradigm to new paradigm (revolution)
Work of a normal science tradition:
Further articulating the paradigm (rather than replicating
it)
Fitting nature in the box (a relatively inflexible one)
the paradigm supplies
A paradigm is like an accepted judicial decision (Brown vs. Board of Education)
Normal science as puzzle solving
Puzzles are “supplied by” the paradigm, which “guarantees”
they have solutions
The solution may be known in detail in advance; it’s bringing
it about, according to standards abstracted from the paradigm, which constitutes
the work of n.s.
Allows for detailed work impossible before agreement over fundamentals, correct
questions, appropriate answers, relevant technologies
It is such detailed work that will eventually produce the anomalies that
lead to crisis and, in some cases, revolution
Some of the results (though not the paradigm if there is a revolution) will
prove to be permanent.
Emergence of professional societies and journals
PhD programs
Short journal articles, addressed to other specialists, replace books for
a broader educated public