Week 4: Demarcating Science - Karl Popper and Falsificationism

Course Context and Introduction to Demarcation

• Course Overview: This lecture is part of the "Philosophy and History of Cognitive Science" module at Aarhus University, led by Murillo Pagnotta.

• Theme: The central theme of Week 4 is "Demarcating science I: Popper," focusing on how to distinguish science from non-science and pseudo-science.

• Key Learning Goals:

  • Understand the "received view" of science within logical positivism.

  • Identify how logical positivists utilized the criterion of "Verifiability" to separate meaningful from meaningless statements.

  • Examine the problem of induction and the subsequent problem with verification.

  • Analyze how falsification overcomes the problem of induction.

  • Understand Karl Popper’s use of "Falsifiability" as a criterion to demarcate scientific from non-scientific statements.

• Quote of Note: "A subjective experience, or a feeling of conviction, can never justify a scientific statement" (Popper, 1935/2002, p. 24). This is contextualized by the Radiohead lyric from the song 'There, there': "Just ’cause you feel it doesn’t mean it’s there."

Logical Positivism and Verificationism

• Origin - The Vienna Circle: Logical Positivism (LP) is associated with the "Vienna Circle," a group of scientists and philosophers (led by Moritz Schlick) who met weekly in Vienna from 1924 to 1936. Many members eventually fled to the UK and USA due to rising anti-semitism.

• The Commonsense View of Science: The view (as described by Chalmers, 2013) suggests science starts with facts (states of affairs in the world), moves to factual/observation statements via observation, and eventually arrives at scientific knowledge (theories/laws) via induction.

• Foundational Schools of Thought:

  • Empiricism: The belief that all knowledge derives from experience.

  • Logical Positivism: The belief that meaningful statements are only those that can be verified through experience.

• The Criterion of Verifiability:

  • Definition: A statement is verifiable if one can specify the steps taken to determine if it is true or false.

  • Meaningful vs. Meaningless:

    • Meaningful: Statements for which verification is possible in principle. Note: A meaningful statement can be false (e.g., "$2 + 2 = 5$"); what matters is that it belongs to a system where its truth/falsity can be tested.

    • Meaningless: Statements that cannot be verified. These are considered "metaphysical nonsense." They are neither true nor false but simply not-understandable in a scientific sense.

• Classification of Statements:

  • Meaningful (Verifiable):

    • Empirical Statements: e.g., "The sun is shining."

    • Tautologies: e.g., "The sun is shining or not shining" (True by necessity).

    • Contradictions: e.g., "The sun is shining and not shining" (False by necessity).

    • Analytical Statements: e.g., "A square has four sides" (True by definition).

    • Mathematical Statements: e.g., "$2 + 2 = 4$" or "$2 + 2 = 5$".

  • Meaningless (Non-verifiable/Metaphysical):

    • Theological: e.g., "God exists," "God is omnipotent."

    • Metaphysical: e.g., "The Absolute is beyond space and time," "Everything that exists participates in Being itself."

    • Ethical/Aesthetic: e.g., "Murder is wrong," "Beethoven’s 9th is sublime." These are categorized as judgments rather than verifiable facts.

Observation and Theoretical Terms

• Direct Verification: Observation statements can be directly verified through perceptual experience (e.g., "The person in front of me is speaking").

• Indirect Verification and Theoretical Terms: Many scientific statements contain theoretical terms that are unobservable (e.g., force, atom, memory, motivation).

  • To verify these, the term must be linked to observations via an operational definition (defining abstract concepts by the procedures used to measure them).

  • Example 1: Height: The statement "My height is $1.56\,m$." Height itself is a theoretical term; we cannot directly observe it without an instrument. It is verified indirectly.

  • Example 2: MEG Data: "The on-scalp MEG data contained high-amplitude events." These are verified indirectly through specialized instruments and experimental procedures.

The Problem of Induction

• Inductive Process: Science often attempts to move from singular observation statements ($O_1, O_2... O_n$) to universal statements (Theories/Laws, $T$).

  • Example: "This swan is white" + "That swan is white" $∴$ "All swans everywhere are white."

  • Example: "Positive reinforcement increases behavior in animal $x$" $∴$ "Positive reinforcement increases behavior in all animals" (Behaviorist theory).

• David Hume and the Problem of Induction: Hume argued that no matter how many singular observations we have, we cannot logically justify a universal conclusion.

• Bertrand Russell’s Examples (1912):

  • The Chicken: A chicken is fed every day and expects to be fed again today. One day, the farmer wrings its neck. Past regularity does not guarantee future outcomes.

  • Uniformity of Nature: We expect the sun to rise because it always has, but there is no logical reason (only habit/inclination) to believe in the absolute uniformity of nature.

• Logical Failure of Induction:

  • Premises: Singular statements ($O_n$).

  • Conclusion: Universal statement ($T$).

  • The conclusion can be false even if all premises are true.

• Principle of Induction: Some positivists proposed a principle stating that if a large number of $A$s are observed under variety and all have property $B$, then all $A$s have $B$. However, this leads to a skeptical regress: justifying the principle of induction requires another inductive argument (it worked on occasion $x_1$, worked on $x_2$… therefore it always works), which is circular.

The Problem with Verification Logic

• Affirming the Consequent: The logic of verifying a theory via its predictions takes the form:

  • $If\, T, then\, O$

  • $O$

  • $∴ T$

• This is a DEDUCTIVELY INVALID argument. Example: "If it rains, the ground is wet. The ground is wet. Therefore, it is raining." (False: the ground could be wet because of a sprinkler).

Karl Popper’s Falsificationism

• Background: Born in Vienna (1902-1994). He interacted with the Vienna Circle but was never a member. He argued that the difficulties of inductive logic are insurmountable.

• The Solution: We do not need induction to progress in science. Instead, we can use deductive reasoning to test and potentially falsify theories.

• The Logic of Falsification (Modus Tollens):

  • $If\, T, then\, O$

  • $not\, O$

  • $∴ not\, T$

• This is a DEDUCTIVELY VALID argument. While singular observations cannot prove a universal theory true, they can conclusively prove it false.

• The Process of Science (The Critical Method):

  1. Propose a new idea/conjecture/theory ($T$).

  2. Deduce specific predictions ($O$) from the theory given certain initial conditions.

  3. Test the predictions:

     • If predictions are verified, the theory is corroborated (not proven, just supported temporarily).

     • If predictions are refuted, the theory is falsified.

• Key Distinction: Science does not start with observation; it starts with theories.

Falsifiability as Demarcation

• Demarcation Criterion: A statement is scientific only if it is falsifiable.

• Two Meanings of Falsifiability:

  1. Syntactic Property: A theory must allow for the deduction of statements that can be empirically tested.

  2. Critical Attitude: The scientific community must actively try to improve theories by testing them and being willing to abandon them if they fail.

• Fallibilism vs. Falsifiability:

  • Fallibilism: The philosophical attitude of accepting that we might be wrong; science is part of a "critical tradition."

  • Falsifiability: The technical criterion of demarcation for science vs. non-science.

• Examples of Falsifiability:

  • Falsifiable: "All metals expand when heated," "$F = G \frac{m_1 m_2}{r^2}$."

  • Not Falsifiable:

    • Tautologies: "It will rain or not rain."

    • Metaphysics: "Any event can be causally explained."

    • Psychoanalysis: Popper criticized theories like "Human behavior is driven by unconscious desires" because they could explain any possible behavior and thus were not falsifiable.

Case Study: Power Posing in Psychology

• Original Study (Carney, Cuddy, & Yap, 2010): Claimed high-power poses cause elevations in testosterone, decreases in cortisol, and increased risk tolerance.

• Outcome: Initially corroborated the theory ($posture \rightarrow hormones \rightarrow behavior$).

• Replication Crisis: Many attempts to replicate failed.

• The Falsificationist Attitude in Action:

  • Dana Carney (2016): Stated, "I do not believe that ‘power pose’ effects are real," based on mounting evidence and p-curve analyses showing no effect. She explicitly updated her view based on evidence.

  • Amy Cuddy (2018): Contested the p-curve analyses, arguing that a more comprehensive set of 55 studies still showed evidential value for postural feedback.

• Scientific Practice: This illustrates that falsification is not always a simple, immediate matter but involves complex debates over evidence and methodology.

Practical and Ethical Implications

• Negative Results: Research by Fanelli (2012) shows that negative results (failed hypotheses) are disappearing from many disciplines, which contradicts the falsificationist ideal that we should value findng where theories fail.

• The Critical Question: To test if someone (or yourself) has a falsificationist attitude, ask: "What evidence would convince you that your theory is false?"

• The Problem of Regress: Theories depend on observation statements, which in turn depend on other theories. Popper acknowledges that in principle, we could test implications ad infinitum, mirroring the skeptical regress.