Lecture 25: Scientific Explanation -

Lecture 25: Scientific Explanation
Logic, Science, and Society
March 18, 2015
Science as an Explanatory Enterprise
The difference between science and other enterprises that seek
explanations of why things are the way they are can be found in the
sorts of standards that science sets itself for what will count as an
explanation, a good explanation, and a better explanation. (Alex
Inference to the Best Explanation
Very often, scientists will adopt a theory because it would, if true,
give us the best explanation for some confusing data we have.
Example: Darwin’s theory is the first one to properly explain why the
structure of life would take its familiar “groups-within-groups” form
Counterexample: Lavoisier’s chemistry was worse at explaining lots
of things than its predecessors
Quick terminology
explanans: The things that are doing the explaining (usually data
and theories)
explanandum: The thing that’s getting explained (the thing that
we’re “dumb” about, not that that’s valid Latin etymology)
The First Model: D-N
The covering law theory of explanation (or the
deductive-nomological theory):
A good scientific explanation is a deductive (logical) argument
with the explanans as its premises, invoking a law of nature,
and the explanandum as its conclusion.
Some Caveats
The explanation must be a valid deductive argument.
The explanans must contain at least one general law actually
needed in the deduction.
The explanans must be empirically testable.
The sentences in the explanans must be true.
Characteristics of the D-N Model
• Explanation and prediction are basically the same.
• Laws are really important. (And really difficult to explain.)
• Statistical or probabilistic arguments (remember causation in
lung cancer?) are hard to explain on this theory.
The Big Problem
Asymmetry in explanations
Symptoms of diseases can’t explain the diseases. (Your illness isn’t
explained by saying that you’re flushed and have a temperature of
But symptoms of diseases can predict other symptoms of diseases,
say. (If you have some very particular kind of fever, we can predict
you’ll probably have some other symptoms A, B, and C, since you
probably have disease X.)
Explanations are asymmetric. And the D-N theory doesn’t say so,
so it’s wrong.
An alternative: A good scientific explanation is one that explains
a wide diversity of phenomena using a small set of basic
premises and kinds of arguments. (That is, it unifies our
understanding of a wide variety of natural phenomena.)
Seems to work well for: Newtonian mechanics, evolution
Doesn’t seem to work well for: chemistry
What about this theory: A good scientific explanation is a
description of the causes of the phenomenon at issue.
Trouble: Causation is really confusing! So this describes something
confusing (explanation) in terms of something even more
confusing (causation).
Advantage: This does seem to be what lots of scientists are doing
every day. Testing hypotheses about what things cause what other
things by crafting specific kinds of experiments.
Why not both?
Maybe we can craft a theory that holds that some kinds of
explanations are used in some areas, and other kinds in others.
Example, from Kuhn: At first, people thought that Newton hadn’t
explained anything. But now, we think he did – because we think
just giving a mathematical law can be an explanation.