Should Scientific Method be X-Rated?
by Craig Rusbult, Ph.D.
This page has not been revised
since May 2001, but
the version on another website has been revised many times
since then, so I strongly recommend that you read
THE
REVISED VERSION.
Why should we wonder if scientific methods are X-Rated? The title for this page is borrowed from Stephen Brush (1974) who asks a serious question in a humorous title, "Should the History of Science Be Rated X?" Why is this a relevant question? Because, as Brush explains in a subtitle, "The way scientists behave (according to historians) might not be a good model for students."
Should our confidence in science be lessened by the limits of logic and the influence of culture? This question has sparked heated debates among scholars who hold contrasting views of science. Since these views seem irreconcilable, it would be futile to aim for a solution that is acceptable to everyone. Therefore, this page will just discuss issues and express opinions. I will also make modest recommendations, based on a simple principle (that if a good idea is taken to extremes without sufficient balance from rational critical thinking, there may be undesirable consequences) and an assumption that undesirable consequences should be avoided.
The following summaries explain what is in each section,
so you can decide whether you want to explore it more deeply.
To avoid time-wasting reloads when using your browser's BACK-button, click this link now.
You can go directly to the summary for each section, or to a simple Table of Contents:
Summary
for Section 1: Responsibility in Education
We should be deeply concerned about our
responsibilities as educators, about the effects that our educational policies
will have on students and society. One way to express this concern
is with a thoughtful evaluation of different ways to teach the nature of
science. We should ask, "What description of science is the most
accurate, and most beneficial for students?" But is the answer
to both questions the same, in all educational situations?
Because my model of Integrated Scientific
Method (ISM) claims that "cultural factors" affect the process
and content of science, ISM can be used to express a wide range of "culture
in science" views, including some that may not be accurate or beneficial.
Should this be a cause for concern?
Is "the way scientists behave (according
to historians)" the way scientists really behave? And if they
do, are students better off not knowing? Are any views of science
potentially dangerous? Should any views be x-rated (unsuitable for
young minds) because they may be harmful for students? Generally I
favor a "free marketplace of ideas" in the classroom, openly discussing
a wide range of perspectives. But if some scholars are advocating
views that seem to "cross over the line" of rationality and good
taste, moving into areas that seem foolish or dangerous, should educators
avoid these views? Or is it better to discuss them openly, exposing
them to the bright light of critical thinking?
These questions are discussed in Section 1.
Summary
for Section 2: The Limits of Logic
Yes, there are limits. It is impossible,
using any type of logic, to prove that any theory is either true or false.
Why? If observations agree with a theory's predictions, this does
not prove the theory is true, because another theory (maybe even one that
has not yet been invented) might also predict the same observations, and
might be a better explanation. But if there is disagreement between
observations and theory-based predictions, doesn't this prove a theory is
false? No, because the lack of agreement could be due to any of the
many elements (only one of these is the theory being "tested")
that are involved in making the observations and predictions, and in comparing
them.
Or the foundation of empirical science
can be attacked by claiming that observations are "theory laden"
and therefore involve circular logic, with theories being used to generate
and interpret the observations that are used to support theories.
This circularity makes the use of observation-based logic unreliable.
And when this shaky observational foundation is extended by inductive generalization,
the conclusions become even more uncertain.
Yes, these skeptical challenges are logically
valid. But a critical thinker should know, not just the limits of
logic, but also the sophisticated methods that scientists have developed
to cope with these limitations, to minimize their practical effects.
By using these methods, scientists can develop a rationally justified confidence
in their conclusions, despite the impossibility of proof or disproof.
We should challenge the rationality of
an implication made by skeptics -- that if we cannot claim certainty, we
can claim nothing. Modern science has given up the quest for certainty,
and has decided to aim for a high degree of plausibility, for a way to determine
"what is a good way to bet."
The question, "Can science cope
with the limits of logic?", is discussed in Section 2.
Summary
for Section 3: Radical Relativism
An extreme relativist claims that no
idea is more worthy of acceptance than any other idea. Usually, relativism
about science is defended by arguing that, when scientific theories are
being evaluated, observation-based logic is less important than cultural
factors. But if theories are determined mainly by culture, not logic,
in a different culture our scientific theories would be different.
And we have relativism.
As with many ideas that seem extreme,
radical relativism begins on solid ground. Most scholars agree with
its two basic premises: the limits of logic and the influence of culture.
But there is plenty of disagreement about balance, about the relative contributions
of logic and culture in science, about how far these "good ideas"
can be extended before they become harmful to rationality and society.
This section ends by asking, "Does
scientific knowledge improve over time?" Although a skeptic may
appeal to the impossibility of proof, the "best way to bet" seems
obvious. To illustrate, we'll imagine a million dollar wager involving
a "truth competition" between scientific theories from the past,
present, and future: from 1501, 2001, and 2101. Would a relativist
really be willing to bet on theories from 500 years ago?
The question, "Is one idea as good
as another?", is discussed in Section 3.
Summary for Section 4: Do scientists search for truth?
This summary will be completed in October 2001.
Summary for Section 5: Science and Unobservables
This summary will be completed in October 2001.
1. Responsibility in Education
When selecting a description of science to be used for education, we should ask two important questions: What is the most accurate description of science, and what educational approach is most beneficial for students?
ONE
FRAMEWORK, MANY VIEWS.
These questions -- asking whether a model of science is accurate and beneficial
-- are important if we want to use ISM in education because: 1) ISM
claims that cultural influences [thought styles and cultural-personal factors]
affect the process and content of science; 2) there is a wide
range of views about cultural influence; 3) ISM can be used to
express each of these views.
For example, a teacher
using ISM could claim that although some extremists emphasize the rare cases
where cultural factors exert significant influence on the evaluation of
scientific theories, for most evaluations the effect of cultural factors
is minimal. This teacher could explain how checks-and-balances occur
when a scientific community evaluates claims for knowledge, and how this
communal process tends to counteract individual biases. Or a community
could be the source of pressures that produce bias. This teacher might
think that cultural-personal bias is common, but is not a part of authentic
science, so it should be avoided. This could be the entry point for
a discussion about sources of bias, and for warnings about the dangers --
because bias is detrimental to objective critical thinking, yet is difficult
to avoid, tough to detect, and easy to rationalize -- along with practical strategies for detecting bias and minimizing its
effects.
Another teacher might criticize "scientific
objectivity" because it indicates a lack of cultural conscience.
This could lead to an activist stance with appeals for patriotism or populism,
by exhorting students to use science for the benefit of a nation or "the
people."
Or a teacher may prefer the extreme relativism of radical sociologists who propose that a central
characteristic of science is the cultural activity
of "creating objects and facts" in the laboratory. This
activity produces a relationship in which observations
-- the supposedly firm foundation for empirical evaluation -- are caused
by culture, not by nature. And there is a change in the balance
of criteria used for theory evaluation, with a dramatic shift toward cultural-personal
factors and away from empirical factors.
Comparing this radical view of science
with a more conventional view, such as my own, we see a sharp contrast.
But both views can be expressed using the framework of ISM. These
"alternative elaborations of ISM"
would use the same basic elements -- thought styles and cultural-personal
factors -- but they would propose different characteristics, relationships,
and balances, and would therefore propose a different view of science.
{ The GOALS-page looks at alternative elaborations,
and illustrates by showing how ISM could be used to describe an orthodox
view of science (with external consistency and empirically constrained retroduction)
or the anarchist ideas of Feyerabend (with external inconsistency and unconstrained
counterinduction). }
THE
RESPONSIBILITY OF ISM.
The existence of alternative elaborations is intentional; ISM is designed
to be flexible, so that by varying the characteristics, relationships, and
balances of its components, it can be used to describe a wide range of views
about science and scientists. In my opinion, this flexibility is a
strength, but it is also a cause for concern. If wild ideas are expressed
using ISM, should I feel responsible, despite my disclaimer that "the
opinions expressed using ISM are those of the expresser, and not necessarily
those of ISM"? Maybe. But in a classrooms the teacher (not
ISM or any other instructional tool) makes final decisions about the views
that are expressed. ISM can allow and encourage an accurate description,
but cannot guarantee it.
Is ISM intended to change behavior?
Philosophers of science make a distinction between a descriptive
model (that tries to describe science as it actually is done) and
a normative model (that tries to describe science
as it should be done). Is ISM descriptive or normative?
With respect to scientists, ISM is descriptive, with no intention of telling
scientists how to do science. But for students, even though the primary
function of ISM is descriptive -- to allow a complete, accurate description
of science as it really is -- when any model of science is used in the classroom
there will be a normative influence on students. In fact, promoting
a change in student thinking and behavior is the purpose of education, and
is the main reason to include a model of science (like ISM) in education.
Two areas of ISM, cultural-personal factors
and thought styles, are especially susceptible to being misunderstood and
abused. Both of these elements are a part of science, so I defend
their inclusion in ISM, and a strong case can be made for including them
in education. As usual, however, if this good idea is taken to extremes,
with exaggerated interpretations, the result will be a distorted picture
of science that is not an accurate description, and is not beneficial for
students.
WHAT IS BENEFICIAL? It is difficult
to answer the "accurate and beneficial" questions with confidence,
due to legitimate questions about what constitutes an accurate view of science,
and about the effects of what we do in the classroom. For example,
will the enthusiasm of future scientists be dimmed if their role models
are tarnished by portrayals of scientists as politically motivated, status-seeking
mercenaries? Or will some students want to become scientists because
they see the socially interactive aspects of science, and they realize that
scientists are real people, like themselves? Similar questions can
be asked about extreme skepticism. Will students stop doing experiments
if they are told that observations are inevitably biased and unreliable?
And will students stop trying to learn the theories in textbooks if they
are told that the justification for these theories is weak or (with anti-realist interpretations) that science does not claim
to describe the truth, and does not even try to search for truth?
Or will skepticism merely encourage healthy critical thinking? And
is there a danger when science education becomes politicized so that it
argues for certain metaphysical or ideological views? Or might certain
types of politicization be beneficial for students?
By combining an earlier question (Should
the History of Science Be Rated X?) with the goals of wanting education
to be accurate and beneficial, we can ask whether "the way scientists
behave (according to historians)" is the way scientists really behave;
and if they do, are students better off not knowing? Pedagogical considerations
of "what is beneficial" should be heavily influenced by, but not
totally determined by, what is regarded as most accurate. The possible
effects on students and society should also be considered. If something
is true, it may seem foolish to ask "Are students better off not knowing?",
but this question is worth asking for young students who are not well equipped
to cope with complex new ideas or to defend their own ideas. An essential
ingredient in the art of teaching is to judge the intellectual sophistication
of students, and then use this awareness to make adjustments so the demands
for thinking and learning will be at an appropriate level.
But should any perspectives really be
x-rated, in the sense that students should not be exposed to them?
In general I favor a "free marketplace of ideas" approach, with
an open-minded tolerance for a variety of viewpoints. In my opinion,
a wide range of views about science can be discussed in the classroom, with
effects on students that are mostly beneficial, especially if the discussion
is done wisely by adjusting the demands for critical thinking to an appropriate
level, as described above. But a responsible educator should avoid
the advocacy of views that "cross over the line" of good pedagogical
taste, moving into areas that are foolish and even dangerous. In my
opinion, some scholars in the "study of science" community have
crossed over this line. Way over. Especially with views such
as radical relativism and "creating
reality." I oppose these views mainly because I think they
are not accurate. But this inaccuracy can also produce effects that
are not beneficial.
A SUMMARY. In an effort to act wisely, motivated by an awareness of our responsibilities as educators (or as parents or citizens), we should be deeply concerned with the effects that our educational policies will have on students and society. This section began by asking, "What is the most accurate description of science, and what educational approach is most beneficial for students?" These questions are worth asking, even though (or because) there are no simple answers. Instead of seeking a solution that will satisfy everyone, which is impossible, the goal of our question-asking should be to stimulate a thoughtful evaluation of the merits of different approaches to teaching the nature of science. And while we're doing this, we can think about how our evaluations are being influenced by our individual and collective perspectives on the complex relationships between models of science, quality of education, and quality of life.
The limits of logic are summarized in a principle of underdetermination which states that it is impossible, using any type of logic, to prove that a theory is either true or false.
In a reversal of the usual pattern,
I'll begin with my conclusions (in Section 2A) before discussing the skeptical
challenges to hypothetico-deduction, observation, and induction.
2A. Potential Problems and Actual Problems
Logical skepticism is based on
sound principles. A critical thinker should be aware of the limitations
of observations, and of logic that is hypothetico-deductive, retroductive,
or inductive. But although some skepticism is good, too much of this
good thing -- without sufficient balance by thinking critically about the
claims of skeptics -- can be detrimental to science and rationality.
In extreme logical skepticism there is
a tendency to ignore the distinction between potential problems and actual
problems, and to therefore propose "cures for which there is no adequate
disease. (Fodor, 1986)" If extreme skeptics assume that modern
science aims for certainty, they are wrong. In response to claims
that nothing can be proved, most scientists would simply say "So what?",
because instead of asking "What can be proved using formal logic?"
it is more practical for scientists to ask "What is a good way to bet?"
Scientists have developed methods for coping with the concerns of skeptics,
so that in most situations the skeptics' potential problems do not
seem to be significant actual problems for science.
note: For access to references (such as "Fodor, 1986"),
check the end of this page.
2B. Limitations of Hypothetico-Deductive
Logic
If observations agree with a theory's
predictions, skeptics correctly point out that this does not prove the theory
is true, because another theory -- including one that has not yet been invented,
and maybe never will be invented -- might also predict the same observations,
and might be a better explanation. And when a theory is invented using
retroductive logic (which is a variation of H-D logic, subject to the same
limitations) an additional reason for caution is that this theory is being
constructed so it will fit known data, and empirical agreement can be obtained
by ad hoc patchwork.
The "Overview of Scientific Method"
describes one method for coping with this logical difficulty: "A
theory can be false even if its predictions agree with observations, so
it is necessary to supplement this 'agreement logic' with another criterion,
the degree of predictive contrast, by
asking "How much contrast exists between the predictions of this theory
and the predictions of plausible alternative theories?"
in an effort to consider the possibility that two or more theories could
make the same correct predictions for this system." { a detailed
explanation of predictive contrast -- and the "So What?" question
-- is on the "Details of Scientific Method"
page }
Compared with the
impossible task of proving a theory is true, it is generally considered
easier to gather evidence showing that a theory is inadequate. Popper
(1963) emphasizes the asymmetry between verification and falsification:
if a theory predicts "if T then O" and O occurs, this does not
prove T is true; but if O does not occur, this proves T is false.
Despite this valid logic, it still is
impossible to logically prove a theory is false, because if there is anomaly for a theory (due to a low degree of agreement
between predictions and observations) the disagreement could be due to any
of the many elements that contribute to the predictions, the observations,
and their comparison. Erroneous predictions could be caused by an
inadequate theory or supplementary theory, or by a characterization of the
experimental system that is inaccurate or incomplete, or by mis-applying
theories to construct a model, or using faulty deductive logic to make a
prediction. But perhaps it is the observations that are not reliable,
due to poor experimental design or sloppy technique; or maybe there was
defective equipment, such as an observation detector that did not function
as expected. Or the logic used in comparing the predictions and observations
may be deficient, and this has produced an estimated degree of agreement
that is inappropriately low.
There are many possible causes for anomaly,
and each can be illustrated with examples from the history of science.
A rigorous logical analysis (Duhem, 1906; Quine, 1953) leads to the skeptical
conclusion that anomaly cannot ever be localized to any of these possibilities.
But according to Shapere (1982, p. 516), "What this shows is that formal
logic does not exhaust what counts as reasoning in science."
Scientists are quite willing to use "reasoning that goes beyond formal
logic" to cope with a complex situation and to make educated estimates
-- based on their confidence in each factor that affects the predictions,
observations, and comparison -- about where the anomaly is likely to be
located.
Another reason for the impossibility
of proof or disproof comes from the statistical nature of some predictions
and observations. For example, Grinnell (1992) discusses the differences
in logic between three theoretical claims: "All X are Y"
can be falsified but not verified; "Some X are Y" can be
verified but not falsified; and "90% of X are Y" cannot
be verified or falsified.
Most scientists will agree with these
conclusions about what can and cannot be proved. But skeptics
will challenge the first two claims, which assert that a theory "can
be verified" or "can be falsified." And scientists
will challenge the pessimistic conclusion that the third claim "cannot
be verified or falsified" because a sophisticated statistical analysis
of data can lead to a rationally justified confidence about the truth or
falsity of a statistical claim such as "90% of X are Y."
But skeptics will question whether this confidence is justified.
According to formal logic, a theory
can never be proved true. But sometimes a theory correctly predicts
old and new data for a wide variety of experimental systems, even though
the combined empirical constraints (for all experiments) are so demanding
that it seems unlikely any alternative theory could also satisfy them.
This is why, for example, few scientists doubt the double-helix structure
of DNA, despite valid logical arguments that this theory is underdetermined
by the data.
Even though it is logically impossible
to prove that any theory is either true or false, scientists can have a
rationally justified confidence that a particular
theory is true, or at least approximately true. Or they may be confident
that it is false.
Most scientists will say that an extreme
skeptic is wrong in implying that if science cannot claim certainty, it
can claim nothing. Modern science has given up the quest for epistemological
certainty, and is willing to settle for a high degree of plausibility.
Scientists rarely worry about skeptical challenges such as "Can you
be certain the sun will rise tomorrow?" (argued by Hume), or "How
do you know it isn't all a dream?" (asked by Descartes), or "Can
you prove that scientific theories of today are closer to the truth than
theories of 500 years ago?" (a challenge by extreme relativists).
When it comes to theory evaluation, instead of asking "What can be
proved using formal logic?", it is more practical for scientists to
ask "What is a good way to bet?"
Consistent with the lack of certainty
in science, in ISM the concept of status uses
a continuum to estimate the degree of confidence in a theory. And
the definition (from Giere, 1991) of hypothesis
-- as a claim that a system and a theory-based model are similar in specified
respects and to a specified (or implied) degree of accuracy -- allows flexibility
in defining what is (and is not) being claimed for a theory. In addition
to status and variable-strength hypotheses,
other types of status (intrinsic and relative, for pursuit
and acceptance, for truth and utility) can be used to modify and thus to
more accurately describe the results of evaluation.
2C. Limitations of Observations
For skeptics, another option is
to attack the foundation of empirical science by claiming that observations
are biased and unreliable. Some challenges are described below, along
with the methods {in brackets} that scientists have developed in order to
cope with each potential difficulty.
2D. Limitations of Inductive Logic
Skeptics can claim that:
2E. A Summary
It is logically impossible to
prove a theory is either true or false. Why? Hypothetico-deduction
has logical limitations (because even when a prediction is "if A then
B" and we observe B, this does not prove A) and there can be suspicions
about ad hoc adjustments when (in retroductive inference) a theory is proposed
to fit known data. Other difficulties include biased data collection,
circularity between theories and observation-theories, and the logical limitations
of inductive generalization.
In an effort to cope with their own concerns
about these logical limitations, scientists have developed methods -- including
estimates for predictive contrast, and sophisticated techniques for logical
analysis -- that encourage them to claim a "rationally justified confidence"
for their scientific conclusions, despite the impossibility of proof or
disproof.
Taken to an extreme, relativism claims that no idea is more well founded, and deserving of acceptance, than any other idea.
3A. Logic and Culture
Relativism about scientific theories
is usually defended by combining two premises, by claiming that during theory
evaluation: 1) due to the limits of logic (discussed above), observation-based logic exerts
only a weak influence, but 2) a strong influence is exerted by
cultural-personal factors. If logical
input is weak and cultural influence is strong, with ideas determined mainly
by culture, then in a different culture the results of theory evaluations
would be different.
While there is a correlation between
a heavy emphasis on cultural factors (in science process)
and relativism (in science content), there
is no necessary link. For example, Hull (1988) thinks that reliable
content can emerge from a chaotic process. So does Bauer (1992), who
claims that during a communal "filtering" process the non-objective
behavior of individuals (or small groups) tends to cancel, thus producing
a result that is more objective than the objectivity of individual scientists.
/ In addition to a heavy emphasis on culture, relativism seems
to also require an extreme form of logical skepticism that challenges the
credibility (or even the possibility) of culturally-independent empirical
"reality checks" that might compete with cultural influence.
Or, instead of asking why scholars reach
relativism as a conclusion, perhaps it makes more sense to assume that --
due to cultural factors operating in society and in scholarly communities
-- a preference for relativism comes first, followed by the arguments (involving
logic and culture) that are enlisted as support.
In recent decades, radical relativism
has become surprisingly popular among scholars. A catalyst in the
rise of relativism was The Structure of Scientific Revolutions (Kuhn,
1962), which emphasized the role played by non-logical factors in the revolutionary
overthrow of one paradigmatic "way of thinking" by another.
This book helped inspire a wave of anti-rationalist intellectual activity
that pushed the boundaries of relativism far beyond the original claims
of Kuhn.
One group pushing the boundaries, the
"strong program" in the sociology of scientific knowledge, has
focused on the ways in which cultural-personal factors affect the content
of science. This is more controversial than claims about the process
of science, which is generally agreed to be influenced by social factors.
Scholars in the strong program usually adopt a radical relativism, claiming
that the content of scientific theories is influenced more by culture than
by nature. (or at least they claim that we should assume culture is stronger
than nature, when we are studying the process and content of science)
3B. Strong Criticisms of The Strong
Program
Many critics have described the
logical deficiencies of extreme perspectives such as the Strong Program
that is outlined in Knowledge and Social Imagery (Bloor, 1976, 1991)
and is manifested in Laboratory Life (Latour and Woolgar, 1986):
According to Slezak, radical relativism produces destructive effects on society that extend far beyond the realms of scholarly discourse where these ideas originate:
One of the most harmful features of radical sociology is that, in important ways, it can undermine the conventional view that "a central aim of education... is the fostering of rationality, or its educational cognate, critical thinking. (Siegel, 1989, p. 21)"
Slezak and Siegel are not alone in their distaste for extreme relativism. Their views are shared by many scholars, including myself and Laudan (1990, p. x) who declares that "The displacement of the idea that facts and evidence matter by the idea that everything boils down to subjective interests and perspectives is... the most prominent and pernicious manifestation of anti-intellectualism in our times." Therefore, it is disturbing to see large segments of the intellectual community either approving radical relativism, or not being active in arguing against it.
Briefly stated, my opinion, based on the principle that without balance "too much of a good thing" can be harmful, is that extreme relativism is the result of taking useful ideas -- such as critical thinking, logical skepticism, and an awareness of cultural-personal factors -- and stretching them to the point where they not only lose intellectual credibility, but they become dangerous for science and society.
3C. Two Analytical Tools for Critical
Thinking
When evaluating extremist interpretations
of science, it helps to have tools that encourage flexible critical thinking
and precise, accurate conclusions. I have developed two useful tools
for analysis: idealizations and range diagrams .
These tools can facilitate a critical examination of the ways that science
is influenced by cultural-personal factors, and will help avoid dichotomous
generalizations such as "no cultural influence" or "all cultural
influence."
The use of idealizations to study science is based on the
principle that an oversimplified model can be useful
for estimating the effects of a component that has been intentionally omitted
from the model. In this case, cultural-personal influence is studied
by trying to imagine what science (especially as it is exemplified in a
specific historical episode) would be like without this influence, and comparing
this idealization with the actual science.
The second type of analytical tool, range
diagrams, can be used to help determine how accurately a sample represents
a larger population, and in deciding what conclusions can be drawn about
a population based on a small sample of case studies. For example,
when studying the mutual influence between societal politics and science,
different conclusions will result from studying a sociobiologist (this field
can be very politicized) and a benzene chemist (very little societal politics
is happening here). Although each scientist is part of the total science
experience, drawing a general conclusion based on either sample by itself
would be misleading.
These tools are useful for recognizing
cultural influence without overemphasizing it. And they help clarify
my own views, to minimize misunderstandings. When I criticize extreme
relativism, I am not claiming that cultural influence is negligible.
My model of Integrated Scientific Method contains cultural-personal factors
(such as psychological motives and practical concerns, metaphysical worldviews,
ideological principles, and opinions of authorities, operating in complex
social and institutional contexts) because these factors play a role in
the process of science and (usually to a lesser extent but not always) in
the content of science.
The tool of idealization
is useful for recognizing bias and coping with its effects, as recommended
by the first teacher of Section 1, who is expressing
my basic views. And range diagrams are
useful for avoiding generalizations that oversimplify and distort, for recognizing
that cultural-personal factors play different roles in different areas of
science and in different communities within each area, and exert different
influences on the process of science and on the content of science.
3D. Is there Scientific Progress? A
Million Dollar Wager
Although to most of us the answer
is obvious, skeptics can challenge a claim that scientific knowledge improves
over time. The progress in scientific utility is clear. But
progress in truth is impossible to verify, since none of us can be sure
we know the truth, so a skeptic asks "Can you prove it?"
My brief answer is "no, but it's a good way to bet."
For example, consider a million dollar
wager. Imagine that 1000 scientific theories from the year 2000, covering
a wide range of fields, are compared with 1000 corresponding theories from
500 years earlier, in 1500. You can choose one set of theories, either
1500 or 2000, and someone who knows the truth about nature -- such as an
omniscient being (God?) or an alien from a scientifically advanced civilization
-- decides which theory (in each of the 1000 areas) is closer to this truth.
If your theory is more true, you win $1000, but if the other theory is more
true you lose $1000. Should you care which set of theories you get?
According to those who claim that science does not improve with age, it
should not matter. If there is no scientific progress, the 1500-science
and 2000-science have an equal chance of being closer to the truth.
In my opinion, anyone who is not a fool (or who wants to give away a million
dollars) should have a rationally justified confidence, although no proof,
that the science of today is a better way to bet.
For a rough estimate of how superior
you think the theories of 2000 are, consider a wager with two options: you
can pay $600,000 and choose the theories of today, or decide not to play.
If you play, you break even with a 20-80 split between the theories of 1500
and 2000. I would eagerly pay the entry fee, with confident assurance
that I would win roughly $400,000. Would you take the bet if the fee
was changed to $800,000, so you need a 10-90 split to break even?
What do you think the majority of scientists would do? I think most
would take the bet, even if they had to pay $900,000 for the chance to win
$100,000. After all, 1-to-9 odds aren't too shabby when betting on
500 years of scientific progress.
Please notice that I'm not saying all
theories of 2000 are perfect, just that in general they are better than
the theories of 1500. If our theories continue to apparently improve
(as they have in the past), then in a million dollar wager comparing theories
of 2000 and 2100, I would bet against our current theories.
4. Instrumentalism and Realism
This section will be re-revised sometime soon, probably in October 2001.
One response to the impossibility of proof is an instrumentalist perspective, in which scientific theories are interpreted as making claims for usefulness, but not for probable truth. Instrumentalism and realism differ in their answer to the question, "Does science try to find the truth?" Realism says yes, but instrumentalism says no.
Section 4A describes a system of concepts that can help us increase the precision of our thinking and communication.
4A. Four Types of Status
As a reminder that the outcome
of theory evaluation is an educated estimate rather than a claim for certainty,
ISM uses a continuum of theory status,
ranging from very low to very high, to describe the degree of confidence
in a theory. To allow a more precise description of theory status,
seven additional distinctions are useful.
Each theory has six types of status
(in three pairs), and an interpretation, and a range of claims:
1a. Each theory has a relative status
(compared with alternative theories) and an intrinsic status.
1b. Each theory has a pursuit status
and an acceptance
status. As suggested by Laudan (1977), even if a theory is
not judged to be worthy of acceptance, scientists can rationally view this
theory as worthy of pursuit (for temporary application and continuing development)
if it needs to be tested more thoroughly, or it seems to have potential
for developing in ways that will improve its plausibility and utility, or
it is useful (even in its current form) for stimulating new experimental
or theoretical research.
1c. Each theory
has a truth status
and a utility
status. / In ISM, truth status
is an estimate of the similarity between the actual composition-and-operation
of systems and the composition-and-operation models (for these systems)
that are constructed by using the theory. { In doing this, I am using
a "correspondence" definition of truth,
that a theory is true if it corresponds to what actually exists. }
Truth status (which I usually call plausibility)
is a human estimate for the probability of truth, rather than a claim for
a certainty of knowledge about truth. / A theory's utility status is an estimate of the overall usefulness
of this theory, including scientific utility for cognition and research
and (if utility is defined more broadly) for cultural-personal usefulness.
2. Each theory can be viewed with
a realist interpretation
(with scientists thinking that this theory is intended to have two types
of function: to be useful and to describe what really occurs in nature)
or an instrumentalist
interpretation (that this theory is intended only to be useful, with
no claims to describe reality). This theory interpretation, which
can vary along a continuum from pure realist to instrumentalist) is independent
from estimates of status for truth and utility. For example, a scientist
might think that a particular theory is intended to portray reality (so
there is a realist interpretation) but does not do this very well (it has
a low truth status, a low plausibility). Also, a realist interpretation
is compatible with a strong emphasis on utility, because a theory can aim
to be both true and useful.
3. Another "flexibility
concept" helps us think more precisely about the specific applications
of a general theory. When a theory is applied to a particular experimental
system, to construct a theory-based model of this system, scientists can
use variable-strength hypotheses to make different
"similarity claims" for the same model and system. The truth
status that is an outcome of evaluation can vary with the strength of a
hypothetical claim. A strong claim (of an exact match between all
features of the theoretical model and the real system) may have lower truth
status than a weaker claim (of a similarity that is approximate rather than
exact, or a similarity between some features but not all). {
more details about the concept of variable-strength
hypotheses }
The definition of hypothesis
given above (borrowed from Giere, 1991) refers to claims about the similarity
between a model and system, so it is oriented toward truth status, which
is relevant only with a realist interpretation. But we can also think
of variable-strength hypotheses as making different claims about the expected
degrees of agreement between different types of predictions and observations
(for example, a "medium strength" hypothesis might claim that
there will be a close match for some predictions, but a less exact match
for other predictions) or as making different claims about the ways in which
the theory might be scientifically useful. This emphasis on utility
would be compatible with either realist or instrumentalist interpretations.
THE UTILITY OF THESE CONCEPTS.
Some terms used in this section (pursuit and acceptance, realism and instrumentalism)
are commonly used in philosophy, while others (intrinsic status and relative
status, truth status and utility status) have been defined by me, and one
(hypothesis) is used with a variety of different meanings. Of the
nine terms, only "acceptance" and "hypothesis" are common
in the language of scientists, but I think all of the concepts are
common in the thinking of scientists.
These concepts are useful because they
allow flexibility (and even encourage it) instead of forcing ideas into
narrow channels by rigid language. They do not limit a thinker to
dichotomous alternatives such as acceptance or rejection, verification or
falsification. If status rises above a certain level we can think
in terms of acceptance, and if it falls too low we can choose to reject,
so these binary categories are still available, but a yes-or-no choice is
not forced on us prematurely because our rigid concepts have limited the
options we are capable of imagining and thoughtfully considering.
In a community of critical thinkers,
these concepts -- especially when they are organized into a logical framework
that is coherent yet flexible -- will allow precision and accuracy in thinking
and communication. They allow a more accurate description of scientific
methods, when describing a specific situation or making a generalization.
If used in a classroom, this system of concepts will encourage students
to think and communicate more carefully, with increased precision.
4B. Critical Realism
In real life there is a range
of realist views, and a range of instrumentalist views. It is difficult
to define either of these views precisely, due to the wide range of positions
adopted by realists and instrumentalists. One useful thinking tool
is described by Leplin (1984) who, in order to portray the range of realist
views, describes ten claims that a realist may or may not believe.
By affirming or denying various claims, a variety of realist positions is
possible, ranging from modest to strong. And the short-list of claims
made by one modest realist might differ from the claims of another modest
realist. Instrumentalist positions are similarly variable. Therefore,
when discussing this topic it is important to avoid oversimplistic dichotomies
and strawman stereotypes. This section describes one type of realism
-- critical realism -- that seems to offer many practical benefits.
When thinking about critical realism,
two concepts are crucial.
First, a realist can place a high value
on both truth status and utility status. This is summarized in my
definition of "theory status" as an estimate of "a theory's
plausibility and/or utility." For a realist, the relative importance
of truth and utility can vary from one theory to another, or even from one
application of a theory to another. Compared with an instrumentalist,
who adopts a restrictive view that eliminates one of the two major criteria
by excluding a consideration of truth, a realist has a wider vision that
looks for both utility and truth.
Second, a critical realist (CR) distinguishes
between goals and claims. A CR is a realist about goals,
and a critic about claims. A CR combines realist
goals (wanting to find the truth) with critical
evaluation (willing to be skeptical about claims for the truth status
of a particular theory). As explained in Section 4A,
realism (for goals) is compatible with criticism (of plausibility).
For example, it is difficult to deny that in the early 1950s, scientists
who studied the structure of DNA were aiming for a theory that would describe
the actual structure of DNA. They wanted to find the truth, so they
were realists. Before 1953, however, their claims were modest, because
all of their theories had a low truth status. They were evaluating
critically, in an effort to achieve their realist goals. But after
April 1953, the claims became bold, and those who were most knowledgeable
quickly decided that the "double helix" structure deserved to
have a high truth status.
4C. Pros and Cons of Instrumentalism
Laudan (1984) clearly expresses
the two most common arguments in favor of instrumentalism. One argument
is that the components of many abandoned theories were once considered real,
so why should we be confident that the components of current theories will
not meet this same fate? But this ignores the analogous counter-argument:
History also provides many examples of postulated components (for entities,
actions, or interactions) that are still considered valid. And sometimes
postulated components that initially could not be observed became observable
when improved observation technologies and techniques were developed.
Laudan's argument depends on inductive "boy who cried wolf" logic
that is not deductively valid. And it seems to imply that, in a "1500 vs 2000" wager, the fact that the theories of 2000
consistently win should be counted as evidence against the possibility that
these theories might be true, or at least approximately true.
In my opinion, the strongest argument
for the reality of many components of modern theories is that it seems extremely
unlikely that these theories could make accurate predictions if none of
their components (or very few of them) corresponds (not even approximately)
to what is actually happening in nature. In other words, a claim that
"this theory is approximately true" seems to be a plausible explanation
for why the theory can make accurate predictions. This is not a proof,
of course, but it does seem like a rational way to bet.
A second argument by Laudan is that a
goal is "utopian" if there can never be a way to know whether
it has been achieved. Since the truth of a theory can never be proved,
we can never know if we have achieved a realist goal, so this utopian goal
should not be held by rational scientists. Compared with the first
argument, I find this one more impressive. But I remain unconvinced,
for reasons similar to the "best way to bet" arguments against
logical skepticism. Even though there is no way to
prove a theory is true or false, scientists can have a rationally justified
confidence about it, and this is all that most modern scientists expect.
More important, we should remember that the defining characteristic of realism
is a goal (our search for truth), not a claim (for certainty).
To say that scientists do always
think instrumentally is inaccurate*, and to demand that scientists should
always think instrumentally (by never thinking of a theory in terms of its
possible truth) is too restrictive. { * In my experience, most scientists
have difficulty even understanding the concept that scientists don't
try to search for the truth, and they certainly don't agree with
it. } Compared with instrumentalism, the eclectic "best
of both" framework offered by critical realism seems to be a much better
way to describe the actual practice of science, because this framework flexibly
accommodates the fact that both types of thinking (in terms of utility and
truth) are used in science, with the relative proportions depending on the
scientist and the situation.
Attitudes toward utility and truth also
differ in science and design. An engineer whose main goal is to improve
a product will tend to be more satisfied with viewing a theory strictly
in terms of its usefulness in promoting progress toward the main goal, without
thinking too much about whether or not the theory is true. {This is
discussed in more detail in Introduction to Design
and [eventually] on the "Science and Design" page. }
Do scientists search for truth? Of course, searching for truth is not the only goal. Scientists are also motivated by the intellectual stimulation and satisfaction of solving problems, and by practical benefits such as obtaining grants, earning salaries, publishing papers, gaining respect from scientific colleagues and from nonscientists, and developing science-based technologies that will bring practical benefits like improved health care or new consumer products. Yes, all of these are motivations, but usually scientists also want to construct accurate theories, theories that match the reality of what is happening in nature.
Webster's New Collegiate Dictionary
defines instrumentalism as "a doctrine
that ideas are instruments of action and that their usefulness determines
their truth." Does this doctrine make sense? An idea may
be useful because it is true, and its usefulness may be an indication of
its truth. But an idea cannot be true
-- in the sense that it corresponds to reality
-- because it is useful. { And if we adopt any other
meaning of "true", this word loses its usefulness. }
As explained in Section 4B, "there
is a range of instrumentalist views." I can respect the instrumentalists
who disagree with the definition above (that "usefulness determines...
truth"), who are humble in their claims about the power of theories,
who say "an instrumentalist is not claiming (or denying) that a theory
corresponds with reality and is therefore true; the claims being made
are actually more modest than those of a realist, because an instrumentalist's
claims refer only to a theory's usefulness." { Arguments against
"truth as a goal" (and even against "truth" as a concept!)
are especially among scholars who get excited about postmodern theories
of radical relativism. But anti-realist perspectives are also held
by scholars who don't have postmodernist views. }
By contrast, the following section shows
the foolish self-delusion that occurs when people become arrogant about
the power of their own ideas.
4D. Do Scientists Create Reality?
Do scientists study nature, or create nature? Somewhat amazingly,
In the words of Latour & Woolgar (1979, p. 64), "The bioassay is not merely a means of obtaining some independently given entity; the bioassay constitutes the construction of the substance." Matthews (1994, p. 152) comments on this type of wild claim -- that "objects do not lie around ready made in the world but are mental constructs (Wheatley, 1991, p. 10)" -- by explaining a crucial distinction: "Where he [Wheatley] goes wrong is in failing to distinguish the theoretical objects of science, which do not lie around, from the real objects of science, which do lie around and fall on people's heads."
A description of the way scientists typically think about the observation of real objects (no, it is not necessary to "create the reality" of the objects) is provided by a cell biologist:
Another excellent description of "truth" and "reality" is provided by a prominent philosopher:
Next, Sober illustrates what he considers to be a valid meaning for "thoughts becoming reality" by describing a situation in which a person's thoughts (he thinks he won't hit a baseball) affect his actions (he swings too high), thus causing a result (he doesn't hit the baseball). By contrast,
These quotations, from a scientist and a philosopher, summarize the most important concepts in "Reality 101" so I'll just close this section with an example from science: Anyone who really thinks that "beliefs create reality" should be eager to explain how the real motions of all planets in the solar system changed from earth-centered orbits in 1500 (when this was believed by almost everyone) to sun-centered orbits in 1700 (when this was believed by most people, at least in the scientific community). Did the change in beliefs (from theories of 1500 to theories of 1700) cause a change in reality (with planets beginning to orbit the sun at some time between 1500 and 1700)?
5. Positivism (science and unobservables)
One way to avoid some limitations
of hypothetico-deductive (HD) logic is to avoid speculating about anything
that is not observable. Positivism is
the claim that scientific theories should not postulate the existence of
entities, actions or interactions that cannot be directly observed.
By contrast, empirically based hypothetico-deductive logic allows "unobservable"
components in a theory, if this theory makes predictions (or retroductions)
about observable outcomes.
The motivations for positivist constraints can be due to beliefs about utility (what is useful) and/or ontology (what exists). utility: One motivation for positivist philosophy is to build science on the firm foundation of empirical observations, thereby making scientific knowledge more certain. ontology: Or positivists may want to purge science of metaphysical proposals for unobservables. But most philosophers have concluded that positivism does not necessarily make science more certain; and instead of making science non-metaphysical, it simply replaces one type of metaphysics with another type.
In the early 1900s, very few scientists
decided to abandon atomic theory (even though this abandonment was being
urged by Ernst Mach, based on positivist principles) or to stop thinking
in terms of "forces" (which have been considered unobservable
by many positivists). But positivist perspectives did dominate two
major fields, psychology and philosophy of science, for several decades
in the first half of the twentieth century. Behaviorist
psychology, based on positivist limitations, enjoyed a quarter-century
reign of dominance, but since the 1950s it has been surpassed by a less
restrictive cognitive psychology (whose focal point is unobservable cognitive
activities within the brain) that has provided a liberating perspective
for most psychologists. And the influence of logical
positivism, which was the dominant philosophy of science for several
decades, has declined dramatically.
Although positivism is considered a legitimate
perspective in philosophy, it is rare among scientists, who welcome a wide
variety of ways to describe and explain. Contrary to the restrictions
of positivists, scientists practice science the way they feel is most effective,
and most modern theories include unobservable entities (photons, electrons,...)
and interactions (electrical fields and forces,...) among their essential
components. Although scientists sometimes generate and utilize a theory
that is limited to a description of empirical patterns, usually this type
of theory is seen as a temporary stage along the path to a more complete
theory that probably will include unobservable components. This feeling,
that "we're not there yet" when we have only a descriptive theory
for an empirical pattern, contrasts with the positivist view that this should
be the logical ending point for science.
For example, according to a prominent
contemporary defender of positivism (empiricism), a theory should be only
a way to conveniently summarize a large amount of data, to make generalizations
about observable quantities, and to make predictions:
A "positivist account" is a philosophical theory about what
scientists should do and, more important, what they should not do,
rather than a description of what scientists actually do. It
is prescriptive, not descriptive. When van Fraassen states that in
a positivist account "the demand for an explanation... plays
no role in the scientific enterprise," the meaning would be
more clear if he claimed that "according to positivist views of science,
the demand for an explanation... should play no role
in the scientific enterprise."
But if scientists "do what they
should" (according to positivism) they will operate at a disadvantage
compared with scientists who misbehave, because most of the best modern
theories are non-positivist. Faced with this choice, to "behave
as they should" or to be effective, most scientists will choose freedom
and effectiveness.
{ note: Since modern versions of positivism can be called empiricism, in an effort to avoid confusion I'll call attention to an important difference between two similar terms: empirical science, which uses empirical observations and empirical evaluations, includes all science, both empiricist science (with only observable components in theories) and non-empiricist science (that allows both observable and unobservable components). Since HD logic allows scientists to empirically evaluate the plausibility of components that cannot be observed (but that produce effects which can be observed), non-empiricist science (which can include theories containing non-observable components) requires HD logic. But empiricist science (i.e., positivist science) can be done with or without HD; or, viewed from another perspective, HD logic can be done for theories with observable and/or unobservable components. In fact, a desire to accomodate non-positivist theories in science was a major motivation in developing the current importance of HD logic. }
Here is part of the section on positivism from my "Details of Scientific Method" page:
CONSTRAINTS ON UNOBSERVABLE COMPONENTS. A positivist
believes that scientific theories should not postulate the existence of
unobservable entities, actions, or interactions. For example, behaviorist
psychology avoids the concept of "thinking" because it cannot
be directly observed. A strict positivist will applaud Newton's theory
of gravitation, despite its lack of a causal explanatory mechanism, because
it is an empirical generalization that is reliable and approximately accurate,
and it does not postulate (as do more recent theories of gravity) unobservable
entities such as fields, curved space, or gravitons. But most scientists,
although they appreciate Newton's descriptive theory for what it is, consider
the absence of explanation to be a weakness.
some comments about terminology:
Positivism was proposed in the 1830s by Auguste Comte, who was motivated
partly by anti-religious ideology. In the early 20th century a philosophy
of logical positivism was developed to combine
positivism with other ideas. In current use, "positivism"
can be used in a narrow sense (as Comte did, and as I do here) or it
can refer to anything connected with logical positivism, including the "other
ideas" and more. Logical positivism can also be called logical empiricism. { Notice that empiricism
(i.e., positivism) is not the same as empirical.
A theory that is non-empiricist (because
it some components, such as atoms or molecules, that are unobservable) can
make predictions about empirical data
that can be used in empirical evaluation.
}
Although positivism (or empiricism, the
name typically given to the modern versions of positivism currently being
proposed) is considered a legitimate perspective in philosophy, it is rare
among scientists, who welcome a wide variety of ways to describe and explain.
Many modern theories include unobservable entities and actions, such as
electrons and electromagnetic force, among their essential components.
Although most scientists welcome a descriptive theory that only describes
empirical patterns, at this point they think "we're not there yet"
because their limited theory is seen as just a temporary stage along the
path to a more complete theory. This attitude contrasts with the positivist
view that a descriptive theory should be the ending point for science.
The ISM framework includes two types of theories (and corresponding models) -- descriptive
and explanatory -- so it is compatible with any type of scientific theory,
whether it is descriptive, explanatory, or has some characteristics of each.
My own anti-positivist opinions, which are not part of the ISM framework,
are summarized in the preceding paragraph, and are discussed in more depth
on the X-RATED page. [i.e., in the section you've been reading]
To check any of the references in this page, CLICK HERE and a page with references for the whole website (but mainly for this page and for "Details of Scientific Method") will open in a new window.
IS THERE A SCIENTIFIC METHOD?
( re: alternative elaborations of ISM )
IDEALIZATIONS AND RANGE DIAGRAMS
INTRODUCTION TO DESIGN ( re: science and design )
DETAILS
OF SCIENTIFIC METHOD:
When you click the link above,
the large "Details of Scientific Method" page
will open in a separate new window.
Then both pages (X-Rated and Details) will be open
so you can quickly move back and forth between them.
You can move to a specific location inside the "details" page
by
adding the appropriate suffix (such as #contrast) to the end of the URL.
( re: predictive contrast and "so what", add #contrast )
( re: theories used to interpret observations, add #supp )
( re: statistical conclusions, add #agree )
( re: the utility of simplified models, add #oversimp )
( re: variable-strength hypotheses, add #hyp )
( re: unobservable entities in theories, add #emp )
( re: descriptive and explanatory theories, add #theory )
copyright 2000 by Craig Rusbult
http://www.sit.wisc.edu/~crusbult/methods/xrated.htm