Scientific Explanation
Scientific explanation studies what it is for science to explain a phenomenon rather than merely describe or predict it, and which models best capture the difference.
Definition
A scientific explanation is an account that shows why or how a phenomenon occurs by relating it to laws, causes, mechanisms, statistical regularities, or a unifying theoretical framework, in a way that confers understanding beyond mere accurate description.
Scope
This area covers the major philosophical accounts of explanation in the natural and social sciences: the covering-law tradition, causal and mechanistic accounts, statistical and probabilistic explanation, and unificationist approaches. It addresses the relation between explanation and prediction, the role of laws, the asymmetry of explanation, and pragmatic and contextual dimensions of why-questions.
Sub-topics
Core questions
- What distinguishes an explanation from a mere description or successful prediction?
- Must explanations cite laws of nature, causes, or mechanisms?
- Why does explanation seem asymmetric when prediction is symmetric?
- Can there be genuinely statistical explanations of single events?
- Is explanatory power a matter of unification, causal information, or pragmatic relevance?
Key concepts
- explanandum and explanans
- covering law
- explanatory asymmetry
- causal relevance
- explanatory unification
- pragmatics of explanation
Key theories
- Covering-law (deductive-nomological) model
- Hempel and Oppenheim hold that to explain an event is to subsume it under general laws, deriving a statement of the explanandum from laws plus initial conditions.
- Causal-mechanical account
- Salmon argues that explanation consists in exhibiting the causal processes and interactions that produce the phenomenon, locating it within the causal structure of the world.
- Interventionist causal account
- Woodward analyses explanation in terms of relationships that would remain stable under hypothetical interventions, answering 'what-if-things-had-been-different' questions.
- Unificationist account
- Kitcher holds that explanation advances understanding by reducing the number of independent argument patterns needed to derive the phenomena of nature.
History
The systematic study of explanation begins with Hempel and Oppenheim's 1948 deductive-nomological model, which dominated mid-century philosophy of science. Counterexamples concerning explanatory asymmetry and irrelevance prompted causal accounts in the 1970s–80s (Salmon), pragmatic accounts (van Fraassen 1980), unificationist accounts (Kitcher 1989), and interventionist causal-modelling approaches (Woodward 2003).
Debates
- Are laws necessary for explanation?
- Covering-law theorists require subsumption under laws, while causal and mechanistic theorists argue that local causal information can explain without invoking exceptionless laws.
- Is explanation objective or pragmatic?
- van Fraassen treats explanation as a context-relative answer to a why-question, against accounts that take explanatory relations to be fully objective features of the world.
Key figures
- Carl Hempel
- Paul Oppenheim
- Wesley Salmon
- James Woodward
- Philip Kitcher
- Bas van Fraassen
Related topics
Seminal works
- hempeloppenheim1948
- hempel1965
- salmon1984
- kitcher1989
Frequently asked questions
- What is the difference between explanation and prediction?
- On the covering-law view they share the same logical structure, yet explanation seems asymmetric: a flagpole's height explains its shadow's length but not vice versa, even though either can be predicted from the other. Accounting for this asymmetry is a central problem motivating causal accounts.
- Can statistics alone explain anything?
- Probabilistic models such as Hempel's inductive-statistical account and Salmon's statistical-relevance approach hold that citing the right probabilistic relevance relations can explain, though critics dispute whether low-probability events are thereby explained.