'Scientific Explanation: Core Problems'

Core Idea

Scientific explanation differs from mere prediction: explaining why an event occurred appeals to underlying causes, principles, or laws, not just recognizing patterns in data. Understanding what constitutes a genuine scientific explanation—and whether all explanations share a common structure—is a central concern of philosophy of science.

Explainer

You already know from your prerequisite on causation that there is a difference between correlation and causation, between prediction and explanation. Saying "the barometer falls before storms" lets you predict rain; it does not explain why it rains. Explanation demands more — it demands the right kind of connection to what actually produced the event. The question this topic asks is: what *kind* of connection is that, exactly?

The most influential answer was Hempel and Oppenheim's deductive-nomological (DN) model, proposed in 1948. On this view, to explain an event is to show that it *had to happen*, given the laws of nature and the initial conditions. You deduce the explanandum (what's to be explained) from the explananda — a set of premises that include at least one law of nature and statements of initial conditions. Why did the metal rod expand? Because: (1) all metals expand when heated (a law), (2) this rod is metal and was heated (initial conditions), therefore (3) this rod expanded. The explanation is a valid deductive argument from lawful generalizations. The DN model honors your prerequisite on deductive reasoning: explanation and prediction have the same logical structure — the only difference is temporal, whether you derive the event before or after it occurs.

But the DN model faces sharp counterexamples that reveal something important. Consider this "explanation": the length of a flagpole and the angle of the sun *logically entail* the length of its shadow. The same facts, taken in reverse, also entail the height of the flagpole from the shadow length. By DN standards, both are valid explanations. But we feel that the shadow doesn't explain the flagpole height — the flagpole height (together with the sun angle) *causes* the shadow length, not the reverse. This is the asymmetry problem: the DN model cannot distinguish explanatorily relevant from irrelevant factors.

The asymmetry problem points toward causal theories of explanation — the idea that genuine explanation must track genuine causal structure. To explain why the flagpole casts a 15-meter shadow, you describe the causal process from the sun illuminating the flagpole to the shadow being cast. But causation faces its own philosophical difficulties (as your prerequisite established), which is why philosophers of science continue to debate whether causal, mechanistic, unificationist, or pragmatic accounts best capture what scientific explanation achieves. The upshot is practical: scientists don't just seek true predictions; they seek accounts of *why* — accounts that reveal the mechanisms, causes, or deep structure that generate the phenomena. What exactly that requires is this topic's central question.