3 Scientific Explanation

What explanations aim to do

A scientific explanation aims to show why something happened or how it follows from wider patterns. Three influential approaches emphasize laws, mechanisms, and unification. They focus on different explanatory virtues and can complement one another.

Explanation through laws

The covering-law account treats an explanation as an argument that derives the phenomenon to be explained from general laws and relevant conditions. The phenomenon is the explanandum; the laws and conditions that do the explaining are the explanans.

In a deductive-nomological (DN) explanation, the explanandum follows deductively from the explanans, which must include at least one law of nature. For example, laws of motion and gravitation, together with information about a planet’s position and velocity, can be used to derive features of its orbit.

An inductive-statistical account uses statistical laws and relevant circumstances to make an event likely, rather than guaranteeing it. This approach is useful for genuinely probabilistic phenomena, but showing that an event was likely may not explain why it occurred rather than a particular alternative.

Law-based explanations make logical structure and generality clear, and connect explanation with prediction. However, a valid derivation may still seem unexplanatory. A derivation relating the length of a flagpole’s shadow to its height, for example, does not by itself show whether the pole’s height explains the shadow or the shadow explains the pole. This illustrates the concern that deriving a result from laws may not be enough if the explanation fails to identify a relevant causal direction.

Explanation through mechanisms

A mechanistic explanation describes how a phenomenon is produced by organized entities and activities. It identifies relevant parts, what they do, and how their interactions generate the outcome, rather than relying only on a general rule.

For example, an explanation of a neuron’s electrical signaling can describe ion channels and ions and how their coordinated activity changes the cell’s electrical state. Mechanistic explanations are especially prominent in biology and neuroscience, where understanding a process often involves identifying its working parts and organization.

Mechanistic accounts can connect explanation to experimental investigation and control: they can show what would happen if a component were blocked or altered. A mechanism need not replace laws; generalizations can describe how its components behave, while the mechanism shows how those components produce the phenomenon.

Mechanistic descriptions can be incomplete, and which components matter depends partly on the phenomenon being explained. Mechanistic detail also does not automatically explain broader patterns that apply across many different systems.

Explanation through unification

Unificationist accounts treat explanations as powerful when they show how many different phenomena can be understood through a small set of general patterns. Philip Kitcher’s influential version emphasizes reusable argument patterns: a theory gains explanatory power by deriving many accepted claims through a relatively limited collection of such patterns.

Newtonian gravitation illustrates this approach. The same framework helps explain both the fall of objects near Earth and the motion of planets. Its unifying appeal is not merely that each event can be derived from a law; one connected account brings together phenomena that might otherwise seem unrelated.

Unification helps explain why scientists value theories that reduce apparently separate facts to a common pattern. But simplicity or breadth alone cannot guarantee a good explanation: a theory might unify phenomena through an inaccurate or overly flexible account. Unification must therefore be assessed alongside empirical support and explanatory relevance.

How the approaches fit together

The three approaches ask different questions about an outcome:

• Laws: What general principles, together with the circumstances, entail the outcome? This approach highlights generality and how explanation can support prediction, but derivation alone may miss causal direction or relevance.

• Mechanisms: What process produces the outcome? This approach reveals how a process works and what interventions may change it, but may not capture broad patterns, and the relevant level of detail can vary.

• Unification: How does the explanation connect this case with others? This approach shows how diverse facts fit into a coherent account, but a unified account is not necessarily true or causally informative.

In scientific practice, an explanation may combine all three approaches. A law can describe a mechanism’s behavior; the mechanism can show how a general theory applies to a particular system; and the theory can unify that system with other cases. No single approach captures every explanatory aim equally well.

Assessing an explanation

Judging an explanation involves considering whether it is well-supported, relevant to the question, and adequate to the phenomenon’s causal and broader context. Law-based accounts clarify deduction and prediction, mechanistic accounts emphasize organized causal processes, and unificationist accounts emphasize explanatory reach and coherence. Their different strengths and limits make the appropriate approach depend on the explanatory aim.