Characterise the scientific method, contrasting inductivist and hypothetico-deductive accounts, and assess whether a single method defines science

What this dot point is asking

SEAB wants you to characterise the scientific method and to ask the harder question of whether any single method defines what makes inquiry scientific. Science is our most successful knowledge-producing enterprise, so understanding how it works, and whether it works by one method or many, is central to the philosophy-of-science strand. Your task is to contrast the main models of method and to weigh the claim that there is one method that demarcates science.

The answer

The inductivist picture

The traditional picture, often called naive inductivism, holds that science proceeds in stages: scientists observe the world without prejudice, collect a large body of data, and then generalise inductively to laws and theories that the data support. On this view, theory is built up from neutral observation, and the more confirming instances accumulate, the more secure the theory. It is an appealing image of science as careful, bottom-up fact-gathering.

Why naive inductivism fails

Two problems undermine it. First, observation is never theory-free: what a scientist notices and records depends on the concepts, instruments and expectations they bring, so there is no stage of pure, prejudice-free data collection. Second, induction itself cannot be justified without circularity, which is the problem of induction: no number of confirming instances logically guarantees the next case, and arguing that induction works because it has worked before is itself an inductive argument. So science cannot rest on a foundation of pure data plus induction.

The hypothetico-deductive model

A more accurate model is hypothetico-deductive. Science begins not with data but with a hypothesis, however it is conceived (guess, analogy, inspiration). From the hypothesis, together with auxiliary assumptions, the scientist deduces observable predictions. These predictions are then tested by experiment or observation. If they hold, the hypothesis is corroborated and retained; if they fail, it is revised or rejected. The genius of the model is that the origin of a hypothesis is irrelevant to its scientific standing; only its testable consequences matter.

The asymmetry of confirmation and refutation

The hypothetico-deductive model carries a logical asymmetry. A successful prediction only supports a hypothesis, it does not prove it, because the prediction could be true while the hypothesis is false (this is the affirming-the-consequent point from the reasoning area). A failed prediction, by contrast, can in principle refute a hypothesis by modus tollens. So tests can refute more decisively than they confirm, an asymmetry that Popper will build a whole philosophy upon. But refutation is not perfectly clean either, because a failed prediction may be blamed on an auxiliary assumption rather than the core hypothesis (the Duhem problem).

Is there a single method?

Whether one method demarcates science is contested. Popper proposes falsifiability as the criterion: a theory is scientific if it forbids something observable and so could be refuted. Kuhn replies that real science is governed by paradigms and the puzzle-solving of normal science, not by a timeless method, and that scientists do not abandon a paradigm at the first refutation. A defensible position is that science is unified not by one rigid algorithm but by a cluster of methodological virtues, testability, openness to evidence, control of variables, intersubjective checking, so "the scientific method" names a family rather than a single rule.

Examples in context

Example 1. A drug trial. To test whether a drug works, researchers do not simply gather cases of recovery; they hypothesise an effect, predict that a treated group will outperform a placebo control, and run a randomised trial. The design controls variables so that any difference can be attributed to the drug. This is the hypothetico-deductive method made concrete, and it shows why a passed trial corroborates without proving, since the result is consistent with other explanations.

Example 2. The discovery of Neptune. Astronomers found that Uranus deviated from its predicted orbit. Rather than abandon Newtonian mechanics at once, they posited an unseen planet (an auxiliary hypothesis) whose gravity would explain the deviation, and Neptune was duly found. The episode shows the Duhem problem in action: an anomaly can be met by adjusting an auxiliary assumption rather than rejecting the core theory, which complicates any simple story of refutation.

Try this

Q1. State two reasons why naive inductivism is an inadequate account of scientific method. [6 marks]

• Cue. Observation is theory-laden (no prejudice-free data stage), and induction cannot be justified without circularity (the problem of induction), so science cannot be built up from pure data.

Q2. Describe the stages of the hypothetico-deductive method. [6 marks]

• Cue. Propose a hypothesis (any origin); deduce observable predictions with auxiliary assumptions; test them; corroborate and retain the hypothesis if predictions hold, or revise or reject it if they fail.

Q3. Explain the asymmetry between confirming and refuting a scientific hypothesis. [8 marks]

• Cue. A passed prediction only supports the hypothesis (affirming the consequent), so it cannot prove it; a failed prediction can refute it by modus tollens, though the Duhem problem means the fault may lie in an auxiliary assumption.