Science and experimental confirmation

Keith B Miller (
Thu, 13 Aug 1998 10:22:06 -0600

The following is taken from an essay by Kenneth R. Foster in the
July/August 1998 issue of Bioelectromagnetics Newsletter (p. 11). I
thought it raised some interesting points relative to the nature of science
and criteria for identifying "good scientific theories."


"Philosophers and sociologists of science have long debated what it means
to "confirm" an observation, and have pretty much concluded that the issue
is fruitless. It is fundamentally impossible to replicate any experiment.
And arguments whether an experiment has "confirmed" another one can go on
forever without resolution, becasue of what philosophers call
experimenters' regress."

"In concrete terms, suppose Investigator A claims to have observed an
effect, and Investigator B claims to have repeated the experiment, finding
nothing of the sort. Investigator A can always reply that the follow-up
experiment was done improperly, and B can retort that the original
experiment was poorly designed in the first place. Sound familiar?"

"A broader perspective is needed. Eminent philosopher of science Philip
Kitcher, in his book "The Advancement of Science" (Oxford Univ. Press
1993), proposes several criteria for a good scientific theory. These
include independent testability ("the ability to test hypotheses
independently of the particular cases for which they are introduced"),
unification ("the result of applying a small family of problem-solving
strategies to a broad class of cases") and fecundity ("when a theory opens
up new and profitable lines of investigation")."

"The history of science abounds in newly reported effects and subsequent
theories that meet, or fail to meet, these criteria. Small bumps in the
current-voltage function in a superconductor become the Josephson effect,
now the subject of thousands of scientific papers. Anomalies in the
thermophysical properties of water in capillaries led to polywater, a major
scientific embarrassment."

"In short, the important thing is not whether an effect has been
"confirmed" but what a scientist makes of it. If you discover an effect
that is real and significant, the world will beat a path to your door. If
not, science will soon forget the discovery, and nobody will care whether
it has been confirmed or not."

Kenneth R. Foster
Department of Bioengineering
University of Pennsylvania

Keith B. Miller
Department of Geology
Kansas State University
Manhattan, KS 66506