Notes on Theories, Falsification, Induction, and the Scientific Method

The Nature of Theories and What Makes Science

• Theory definition: theories are usually made up of multiple interrelated claims that together provide an explanation of how the world works.

• A theory begins as a human claim about the world; nothing in it is inherently better or worse at first, but it becomes scientific if there are consistent tests that support it over and over.

• A theory remains our primary explanation until something better comes along.

• Distinction from opinion: avoid calling these untested ideas simply opinions; opinions are untested and personal, whereas scientific theories are subject to testing and revision.

• Some ideas are not false; rather, a better explanation may supersede them because it explains more or does so more effectively.

• The heart of science is a claim about how the world works, subjected to consistent testing that supports it repeatedly.

• In science, theories are not proven with final certainty; they are supported by abundant testing and improved upon when better explanations arise.

The Role of Observation, Creation, and Testing

• Observations can inspire theory, but a theory is not itself merely a copy of observation; it is a created, abstract explanation.

• The idea that theory is purely grounded in observation is a myth. Observation motivates invention, but a claim becomes theory only when it is supported by tests.

• Choosing to start with observation is fine, but a theory or explanation requires testing to become accepted as scientific.

• The process includes making abstract concepts tangible and testable, then evaluating them against data.

Reading and Analyzing Theories: Key Concepts to Identify

• Definitions: focus on what the terms mean (e.g., what is meant by abstract terms like "ownership papers" vs. class membership).

• Evidence: what data or observations support the claims?

• Rationalization: how do the authors justify the connections between concepts?

• Bureaucracy and social terms: terms like "division of labor" and "means of production" are concepts that explanations use; watch for how definitions relate to claims.

• Claims about relations: theories assert how concepts relate to one another (e.g., concepts connected by theories of society).

• Scope conditions: when a theory applies and when it does not; some claims have applicability limits.

• The narrative often embeds these claims in examples; identify the underlying concepts and how they relate.

Abstraction, Modeling, and Language

• Abstraction can be challenging; students often confuse abstract concepts with concrete facts.

• Modeling starts from culture and language to identify basic ideas; abstractions help explain observables by translating them into testable statements.

• Hedging language (e.g., "perhaps it works like this") is often used, but hedges are not testable outcomes; precise tests require definite claims.

• Algebraic and logical structure helps clarify reasoning: separating definitions, evidence, and conclusions improves testability.

Observables, Inference, and Hypotheses

• Observable and measurable factors → used to test hypotheses and connect to abstract concepts (e.g., welfare, education level).

• A hypothesis may link many observables; you can have many hypotheses for a single theoretical claim.

• The aim is to map abstract ideas to observables so testing can occur.

• Example: a hypothesis might connect years of college completed (observable) to welfare (abstract measure) and then test across data.

The Logic of Theory Testing: Deduction vs Induction

• Inductive reasoning (observations to general claims) is not a logically valid form of inference for establishing universal truths; it is an intuitive leap, a creative step, not rigorous logic.

• Deductive reasoning (general to specific) is the valid logical form for testing claims against particular cases:

  • General claim: $orall x\,(Taxi(x) \rightarrow Yellow(x))$

  • Specific case: observe a taxi t with $Taxi(t)$; deduce $Yellow(t)$

  • If you observe a counterexample, e.g., $Taxi(t) \land \neg Yellow(t)$, the general claim is falsified.

• The danger of induction: a single counterexample can falsify a generalization; many cases without counterexamples increase confidence but never guarantee truth.

• Concrete example: All taxis are yellow vs. a blue taxi observed → the universal claim is falsified.

• Another example: All people who wear glasses are smart. If you observe a person with glasses who is not smart, the universal claim is falsified.

Falsificationism vs Empiricism: Popper’s View

• Popper argued that science advances not by accumulating observations proving theories, but by attempting to falsify them.

• The starting point of science is an abstract claim, not raw observation.

• A claim is scientific only if it is testable and falsifiable; tests aim to show the claim false.

• Empiricism (relying solely on observation) is not sufficient for science; you must test abstract claims, not merely observe.

• The correct process: inspiration from observation → create an abstract theory → test it; if falsified, revise or reject; if not falsified after extensive testing, increase confidence but never reach absolute certainty.

• A key caution: some claims (e.g., divinity) may be non-falsifiable because there is no possible observational test that would contradict them; such claims lie outside the scope of science.

Confidence, Testing, and Practical Implications

• Confidence grows with repeated successful tests and lack of counterexamples; for example, vaccine science relies on extensive testing and repeated validation.

• There is no final, 100% guaranteed proof; a theory can be supported to a high degree, e.g., a 99% confidence level, but could in principle be overturned by new evidence.

• The statement: “It has been tested over and over; not found false” describes a high degree of confidence, but not absolute certainty.

• Real-world application: immunizations and public health rely on repeatedly tested explanations of how viruses transmit and how the body responds.

• The practical stance: keep using the best-supported theory while remaining open to better explanations in the future.

• We may retain theories that have been falsified if no better alternative exists; this is a pragmatic choice to maintain useful tools and explanations.

• The metaphor: a hand screwdriver is better than nothing; an electric screwdriver is better, but you keep the working tool until a better one fully replaces it.

Newton, Einstein, and the Limits of Theories

• Newton’s laws work in many everyday contexts but fail at extremes (e.g., strong gravitational fields or very small/large scales), requiring more advanced theories.

• The claim that gravity behaves differently on Mount Everest or in extreme trenches illustrates that some theories have limited scope.

• General relativity (Einstein) explains gravity in regimes where Newton’s laws fail; wave physics considerations become relevant for precise motion descriptions in some contexts.

• In teaching, simpler approximations (linear models) are used for accessibility; more complex models exist for accuracy.

• The takeaway: do not discard Newton or simpler theories who still have valid scope; recognize when a broader theory is needed.

Science, Tests, and Real-World Examples

• The COVID-19 vaccine discourse illustrates the scientific method in action: hypotheses about transmission and immune response were tested repeatedly; vaccines were approved based on extensive testing and ongoing verification.

• The process emphasizes falsifiability, replication, and updating beliefs as new data emerge.

• The role of evidence: strong, repeatable evidence supports claims; lack of falsifying evidence supports confidence but does not guarantee truth.

Practical Guidelines for Studying and Reading Theories

• Identify definitions: extract what the author defines as central terms.

• Look for evidence: what data or observations are used to support the claims?

• Distinguish between the claims about concepts and claims about how the world works.

• Note the scope: where does the theory apply? when does it fail?

• Recognize the structure: how do the concepts relate to each other? what are the causal or correlational claims?

• Watch for hedging: if a text says “perhaps,” it may not be testable; seek explicit, testable claims.

• Assess the logic: are conclusions derived by deduction from general claims, or are they asserted as generalizations from particular observations without falsifiability?

Key Formulas and Logical Forms (LaTeX)

• Universal claim about taxis:
  $\forall x\,(Taxi(x) \rightarrow Yellow(x))$

• Specific case leading to deduction:
  If $Taxi(t)$, then from the universal we get $Yellow(t)$.

• Counterexample falsification (indicating a falsified universal):
  Given $Taxi(t) \land \neg Yellow(t)$, the universal statement is falsified.

• Example of a generalization and its negation:
  $\forall x\,(Glasses(x) \rightarrow Smart(x))$
  If you observe an individual with glasses who is not smart, you obtain $\neg \forall x\,(Glasses(x) \rightarrow Smart(x))$.

• Confidence level in scientific claims (conceptual, not a precise Bayesian formula):
  High confidence = high consistency across repeated tests; 100% certainty is unattainable in science.

• Example of a non-falsifiable claim (divinity) context: there is no observational counterexample that would falsify it; thus not science.

Summary: Popper’s Core Idea

• Science progresses by forming bold conjectures and attempting to falsify them through rigorous testing.

• The value of a theory is measured by its testability, its explanatory power (scope), and its predictive precision.

• We prefer theories with broad scope and high predictive precision, but we accept that no theory is final or immune to falsification.

• Induction alone cannot establish universal claims; deduction and falsification provide a more robust route to scientific knowledge.

• The scientific enterprise relies on ongoing testing, replication, and openness to revise or abandon theories in light of new evidence.