22. Inductive Scientific Method

 

Summary

Building on the previous discussion of a posteriori (empirical) probability, this video introduces the basics of science and focuses on the inductive scientific method. Since empiricism is the foundation of science, judgments about probability here must be made after observation and data gathering.

What Is Science?

The word derives from the Latin scientia ("knowledge"). Restricting the definition to the hard sciences and some social sciences, science is described as a search for natural causes and an instrument for producing effects and making predictions about the natural world. It is a systematic pursuit of an organized body of knowledge based on reasoning from empirical observation and quantifiable data toward testable explanations.

Modern science has been so successful that we tend to treat it as the best and most natural route to knowledge. But it is a human-made system resting on a set of underlying assumptions that should be evaluated:

• Empiricism — sense experience is the source of knowledge and truth.
• Objectivity — we can study the physical world without bias.
• Materialism — everything in the universe is made of physical parts.
• Predictability — the universe consists of regular, interconnected causal relationships.
• Unity — the universe has an underlying unified structure.

Causal Arguments vs. Causal Explanations

Both deal with causal claims, but they differ. Causal arguments are persuasive: they offer support for the conclusion that one thing caused another, and (as with analogies) the key is to clearly define the two things and their relationship. Causal explanations instead reveal how or why one thing caused another; they are not arguments. An explanation is only adequate relative to what we are looking for (a leaking toilet explains the puddle well enough to call a plumber, but not to repair it yourself).

Conditions for an Adequate Explanation

A good causal explanation cannot be self-contradictory, vague or ambiguous, incompatible with established facts, or lead to false predictions. This points to testability: physical explanations generate expectations and predictions, which we test by checking whether the prediction comes true. When a prediction fails, this is falsification, though conclusions remain a matter of probability.

Some explanations fail because they are non-testable: a song giving off "good vibes" is meaningless (there is no "vibrometer"), and a heart condition blamed on past-life crimes cannot be tested because we cannot identify who counts as a past-life sinner. Other predictions are merely untestable for practical or technical reasons, and what is untestable today may become testable later. Circular explanations simply restate themselves ("the floor is wet because there is water on it") and tell us nothing new. Finally, unnecessary complexity is undesirable. By Occam's razor (William of Occam, 14th c.), a principle of parsimony, if two explanations work equally well, the simpler one is preferable.

Hypotheses and Two Kinds of Science

Before asserting a cause scientifically, we begin with a guess: a testable hypothesis offered as a possible solution. The scientific endeavor splits into experimental science (present events) and historical science (past events), but both rest on observation measured against the regular patterns observed in nature. There is no single universal method, only methods (plural).

The Inductive Empirical Method: Eight Steps

The inductive empirical method determines a conclusion's probability experimentally (a posteriori), generating a hypothesis to be confirmed or not confirmed:

1. The situation — a situation sparks a theoretical or practical problem or question.
2. Formulate the problem — be precise about what is studied and how (statistical, experimental, historical).
3. Observation — watch and record relevant phenomena. (A philosophical worry: theory and assumption often determine what counts as observable.)
4. Reflection — identify emerging patterns and a possible causal mechanism, drawing on background knowledge. A causal mechanism connects cause and effect; without one, events are merely coincidental.
5. Formulate a hypothesis — the central feature, an intelligent guess. This step is abductive (a leap beyond current evidence), using inference to the best explanation.
6. Make predictions — if the hypothesis is correct, certain things should follow (no oxygen, no fire).
7. Run experiments — repeat and observe. A predicted event gives possible confirmation, not proof; a failed prediction means the hypothesis is likely false.
8. Accept or reject the hypothesis — judged by how consistently results occur. Roughly: ~90% success means keep it, ~60% needs further study, under ~40% suggests abandoning it. A hypothesis confirmed with high probability becomes a theory; a universally confirmed theory becomes a scientific law.

Verification, Falsification, and a Logical Warning

Hypothesis testing has a logical structure. Concluding "the hypothesis is correct" from "if the hypothesis is correct, X follows; X follows" commits the fallacy of affirming the consequent. Following Karl Popper, falsifiability (not verifiability) is what distinguishes science from pseudoscience: hypotheses can be falsified but never truly verified. This connects to the larger problem of induction, to be explored later. The next video turns to the experimental procedures known as Mill's methods, developed by John Stuart Mill.