Science plays a fundamental role in modern society. But what exactly is science? In philosophy this question is known as the demarcation problem (Popper, Kuhn, Laudan and others).

1. What is Science?
Science, Pseudoscience, Non-science
Dennis Miller
07.09.2020
Presented to the Arbeitskreis Philosophie Kelkheim. Originally in German as Was ist Naturwissenschaft?

2. Introduction
Science plays a fundamental and increasing role in modern society. It is the basis of
widely used technologies, e.g. mobile phones, medical imaging, plastics. Many areas
of politics and legislation need a strong scientific input (e.g. mitigating climate
change, ensuring the safety of medicines).
But what exactly is science?
In philosophy this question is known as the demarcation problem: what is the
difference between science and other ways of knowing (or claiming to know) about
the world? Is scientific knowledge different (maybe better) than other types of
knowledge? These questions are part of epistemology (theory of knowledge).
Asking what science is often also has another purpose: an attack on anti-scientific
ideas and pseudoscience. These are a variety of teachings and practices that reject
scientific principles, though many of them claim to be scientific. They are at best
useless distractions, at worst positively harmful.

3. Terminology
Modern use (from about mid-19th century)
English German
Natural sciences Naturwissenschaft
Social sciences Sozialwissenschaft
Humanities Geisteswissenschaft
‘Science’ now generally refers to natural sciences (physics, chemistry, biology, ...). Before the modern use of the
term it referred to knowledge in general, corresponding to German Wissenschaft. Natural science, esp. physics, was
called ‘natural philosophy’ (not the same as the German Naturphilosophie movement).
The term ‘social sciences’ indicates that they aspire to the rigorous methods of natural sciences. However, their
subject matter, human behaviour, makes is much more difficult to avoid ideological and subjective influences.
The term ‘scientist’ was coined in 1834 by William Whewell. Before that, there was no general term to cover all
those working in fields we now refer to as science.

4. Science, Non-science, Pseudoscience
Science
Physics Astronomy
Chemistry Geology
Biology
Social Sciences
Psychology
Economics
Sociology
Humanities
Philosophy
Literature
......
Logic
Pure mathematics
Technology
Craftsmanship
Pseudoscience
Freudian Psychoanalysis
Marxist History
Anti-Vaxx
Creation Science
Climate Denial
Astrology
Homeopathy
*
**
Green: Science
Blue: In principle not science
but strong overlap in practice
Red: Pseudoscience (claims to be
science but isn't)
Grey: Non-science, doesn't claim to
be science
* Popper's examples
** McIntyre's examples
§ Majority of medical community considers it pseudoscience,
but a significant minority supports and practices homeopathy.
§
The social sciences are considered here
as a special category. They have become
scientific by taking on principles used in
natural sciences; this is an ongoing process.

5. What is Science? – Various Approaches
• Philosophical
 Logic, justification of arguments, analysis of method
 Science vs. other claims to obtain knowledge: demarcation problem
• Historical
 How has science developed?
 How and why have theories been discarded?
 Has the meaning of ‘science’ changed?
• Sociological
 Social framework: research institutes, universities, carriers, financing
• Education
 What is science for schools and universities? What is not science?

6. Demarcation Problem
General Description of Scientific Method
Observe
Hypothesize
Predict new observations
Test predictions (experiment)
Analyze results: revise hypothesis
Does a particular method define science?
Realistic description of science
in practice?
Applies to all branches of science?
What are the requirements for a
scientific hypothesis/theory?
Agreement between prediction and
result does not prove hypothesis is
correct (Induction Problem).
This is the description of scientific method often found in indroductory textbooks.

7. Induction Problem
Induction: deduce future observations from previous ones
general statement from particular one
‘All swans are white’
I see lots of swans, all of them white → It's true.
Perhaps, but that does't prove the statement is true.
One black swan is found → This single observation proves the statement is false.
But in practice, we accept this type of argument, even though it is not logically
correct and occasionally does give the wrong answer. Everyday life would be
impossible if we only reasoned according to formal logic.
In Europe the black swan became metaphor for something which could not exist. It was originally used by the Roman author Juvenal (AD 82). ‘All swans
are white’ was considered a standard example of a well-known truth. The first European to record seeing an Australian black swan was the Dutch
explorer Willem de Vlamingh (1697). Recently the term has come to be used for rare and unpredictable events.

8. Karl Popper - falsifiability
Scientific theories are falsifiable.
A statement is falsifiable if some observation might show it to be false.
Scientific theories make predictions which can be expressed as falsifiable statements (typically via an
experiment to test the theory).
Theories that can always offer an explanation for every imaginable result are not scientific.
Examples
General Relativity: the 1919 Eclipse test was a good example of scientific method. It showed that light was
deflected towards heavy bodies. “The theory is incompatible with certain possible results of observation – in
fact results which everyone before Einstein would have expected.” [1]
Psychoanalysis: nothing could, even in principle, falsify psychoanalytic theories. Their ability to explain every
possible form of human behaviour is, in fact, a weakness.
Marxist theory of history: though originally scientific (falsifiable) later versions were unscientific. Addition of
ad hoc hypotheses made it always compatible with the facts.

9. Karl Popper – falsifiability (cont.)
“I wished to distinguish between science and pseudoscience; knowing very well that science often errs, and that
pseudo-science may happen to stumble on the truth.”
Focus is on logical aspects rather than how scientists actually think science progresses.
- Science is not defined by a specific method.
- Naturalistic methodology, i.e. what scientists do in practice: gives interesting information but should not be the
basis of the philosophy of science.
Psychological theories of Adler and of Freud
- Both are systems that can explain every conceivable case: no set of symptoms or behaviour can be imagined that
would disprove them.
- Therefore these theories are not falsifiable, not science.
- However, their work contains important and interesting ideas and observations, which might well be useful in
developing scientific psychology.
Detailed account of falsifiablility in The Logic of Scientific Discovery [2]
- Theory (logical viewpoint): a single instance serves to falsify a theory.
- Practice (methodological): In practice a single conflicting or counter-instance is not sufficient to falsify a theory.
Reproducibility problems, experimental uncertainties and auxiliary hypotheses must be considered. Scientific
theories are often retained, at least for a time, in spite of conflicting evidence or anomalous results.
- Falsifiability is a criterion for demarcation, not for meaning.

10. Karl Popper - Quotes
“I wished to distinguish between science and pseudoscience; knowing
very well that science often errs, and that pseudo-science may happen
to stumble on the truth.”
Freud and Adler:
“The two psycho-analytic theories were ... simply non-testable,
irrefutable. There was no conceivable human behaviour which could
contradict them. This does not mean that Freud and Adler were not
seeing certain things correctly: I personally do not doubt that much of
what they say is of considerable importance, and may well play its
part one day in a psychological science which is testable.”

11. Falsifiability – criticism
Idealised picture of science
Focus on famous episodes such as 1919 eclipse experiment
→ not typical scientific research.
Non-scientific theories may be falsifiable:
→ Astrology can make specific predictions.
Astrology: The predictions to be found in newspapers and magazines may often appear to contain a surprising amount of truth.
They are written so that most readers will find something that can be interpreted as corresponding to a real event. But astrology
(and all sorts of esoteric systems) can also make specific predictions. If the prediction fails (or could have failed), it was falsifiable.
A theory can be scientific but wrong; indeed many scientific theories have been falsified by subsequent work.

12. Falsifiability – Duhem-Quine Thesis
A scientific hypothesis cannot be tested in isolation: auxiliary hypotheses are always required for an
empirical test.
Examples of auxiliary hypotheses
Measuring instruments. For complex instruments, e.g. electron microscope, there is great deal of
theory underlying their operation. This can make the interpretation more difficult.
Correction terms. To obtain accurate positions of astronomical bodies corrections must be made for
refraction by the Earth's atmosphere.
Previous scientific theory. The idea that the Earth is in motion was rejected by some people because
birds do not get thrown into the sky when they let go of a branch. Modern theories of mechanics
(Galileo, Newton) resolved this objection.
→ If an experiment appears to refute a theory or hypothesis we don't discard it at once. First
consider whether the auxiliary hypotheses are correct.
The Duhem-Quine thesis is based on ideas presented by Pierre Duhem (1861-1916) and Willard van Orman Quine (1908-2000).

13. Falsifiability/Demarcation: an alternative approach
Thomas Kuhn
Popper approaches demarcation from a logical point of view, whereas Kuhn concentrates on the history of
science.
The paradigm is a central concept in Kuhn's work. It refers to the accepted theories, methods, training, etc.
- Periods of normal science are punctuated by scientific revolutions involving a change of paradigm.
- The paradigm concept was criticised as being vague. Kuhn made tried to make it more precise and in later
work used other terms instead.
Falsification is not how science really works.
- When accepted theory disagrees with new results:
▪ Try to find explanations within the current theory.
▪ Patch up the theory as long as possible.
▪ When theory no longer tenable: paradigm shift to new ideas.
Kuhn's 1962 book The Structure of Scientific Revolutions [3] was considered revolutionary. His approach was different
from previous historians and philosophers of science. These had viewed science either from a logical point of view or,
from the vantage point of present, as an error to truth progression.

14. Demarcation Problem: are we asking the wrong question?
Larry Laudan
Since antiquity many attempts have been made to define science, but “philosophy has largely failed to deliver
the relevant goods”.
There is no satisfactory demarcation (i.e. well defined necessary and sufficient conditions that define science).
→ We should discard the demarcation problem.
Alternative approach: distinguish between reliable and unreliable knowledge.
Lee McIntyre
Science cannot be defined by its method.
The scientific attitude is important: protects us from pseudoscience, denialism and fraud.

15. Scientific attitude
Good
View of empirical Evidence
Evidence is the primary consideration for theories
- Other considerations (scope, fruitfullnes, simplicity)
less important
Earnest attempts to find the relevant evidence *
Willingness to revise theories in the light of evidence
Personal qualities
Earnest
Open-minded
Intellectually honest
Curious
Self-critical
Bad
View of empirical Evidence
Disregarding evidence (Denialism)
Uncritical acceptance of favoured theories
Personal qualities
Narrow-minded view (often idealogically motivated)
Group-think
Lack of objectivity (e.g. career or economic motivations)
Acceptance of poor experimental methods
Tendency to believe implausible conspiracy theories
* There are, however, different views on evidence: Baysian vs. frequentist statistics,
dealing with outliers, excluding artefacts, deciding what is relevant.

16. Scientific Errors
Statistical and Methodological
Cherry picking: biased data selection to support favoured theory (or discredit one that is disliked)
Curve fitting with too many variables
p-Hacking
Data set too small
Keeping experiment open until desired result found
p-Hacking: This refers to trying various correlations until one is found that statistically significant according to the p < 0.05 test. If many correlations are
tried, a more stringent criterion should be used. Tukey described such a test as which he called the Honestly Significant Difference Test, implying a lack of
honesty elsewhere.
Psychological
Cognitive bias: lack of objectivity because one prefers one's own theory

17. Scientific attitude – the scientific community
Personal competence, integrity and professional ethics are important for good practice, but we cannot rely only on these.
Science is a group activity: the scientific community has various principles and instruments to promote high quality.
Peer review of scientific publications
Guidelines for good scientific practice: documentation, statistical methods, standards for
publication of results
Criticism by others → science as self-correcting system
Replication (often this is closely related work rather than exact replication)
Retraction of papers for major errors or scientific misconduct
Peer review: typically, the editor of a journal sends the paper to two scientists working in the same field. They recommend acceptance (often with revision)
or rejection.
Retraction: retractions, though uncommon, are an important correcting mechanism. In 2018 the retraction rate was reported as 4 per 10,000 papers [7].

18. Anti-Scientific Positions
Denialism: rejection of well-established theory in spite of overwhelming evidence
Typical denialists exaggerate uncertainties:
‘the science is not settled’
‘some other theory might be true’
This is a misunderstanding about how science works:
- Scientific theories can never claim complete certainty: there is always the possibility that they may be revised.
- It is unreasonable to reject a theory for which there is good evidence and instead favour a hypothetical future one.
- Complete agreement is seldom – there are usually a few scientists who disagree with generally accepted theories.
Pseudoscience: claims to be scientific but is not
Bad methods (experimental and/or theoretical)
Lack of scientific attitude

19. Denialism and Pseudoscience - examples
•Anti-vaxx movement
•Climate denial
•Creation science

20. Conclusion
Science
Pseudoscience
Non-Science
Difficult to define:
many attempts and controversies
Upholding quality of scientific work
Protection against pseudoscience, denialism, fraud
Improve the social sciences
Help protect society from harmful anti-scientific trends
Demarcation
Problem
A task for Philosophy
of Science
Scientific
Attitude

21. Literature
1 Karl Popper, Conjectures and Refutations, Routledge & Kegan Paul (1963) p. 36
2 Karl Popper, The Logic of Scientific Discovery, Hutchinson & Co. (1959). [Reprinted Routledge Classics 2002]
3 Thomas Kuhn, The Structure of Scientific Revolutions, Univ. Chicago Press (2nd ed. 1970) [50th anniv. ed. 2012]
4 Larry Laudan, The Demise of the Demarcation Problem, in R. S. Cohen and L. Laudan (eds.), Physics,
Philosophy and Psychoanalysis, 111-127, D. Reidel Publ. Co. (1983)
5 Lee McIntyre, The Scientific Attitude, MIT Press (2019)
6 Eric Loken, The replication crisis is good for science, The Conversation (Apr. 8 2019)
https://theconversation.com/the-replication-crisis-is-good-for-science-103736
7 Julia Belluz, A huge database of scientific retractions is live. That's great for science. (2018)
www.vox.com/2018/10/29/18022148/retractions-science-database