Confirmation and Induction
Confirmation and induction concern how, and whether, observational evidence can rationally support general scientific hypotheses.
Definition
Confirmation is the relation in which evidence raises the credibility of a hypothesis; induction is inference from observed cases to unobserved cases or to general laws. The central philosophical task is to articulate and justify the principles governing such inference.
Scope
This area covers Hume's problem of induction, qualitative and probabilistic theories of confirmation, the major paradoxes (the ravens and the new riddle of induction), Bayesian approaches to evidential support, and the underdetermination of theory by data. It addresses the logic of evidence, the meaning of 'confirmation', and the rationality of inductive inference.
Sub-topics
Core questions
- Can inductive inference be rationally justified without circularity?
- What does it mean for evidence to confirm a hypothesis?
- Why do the ravens paradox and Goodman's 'grue' challenge syntactic theories of confirmation?
- Does Bayesian probability provide an adequate theory of evidential support?
- Is theory choice underdetermined by all possible evidence?
Key concepts
- induction
- confirmation
- degree of belief
- conditionalization
- grue
- raven paradox
- underdetermination
Key theories
- Hypothetico-deductive and instance confirmation
- Hempel develops a logic of confirmation on which a hypothesis is confirmed by its positive instances and by evidence deducible from it.
- Logical (inductive) probability
- Carnap seeks to ground confirmation in a logical measure of probability quantifying the degree to which evidence supports a hypothesis.
- Bayesian confirmation
- On the Bayesian account, evidence confirms a hypothesis when it raises its conditional probability, with belief revised by conditionalization.
- Falsificationism
- Popper denies that theories are ever confirmed, holding instead that science proceeds by attempting to falsify bold conjectures.
History
Hume's eighteenth-century critique of induction set the agenda. Twentieth-century logical empiricists (Hempel, Carnap) sought a formal logic of confirmation; Goodman's 1955 'new riddle' and Hempel's ravens paradox exposed its limits, while Popper offered falsificationism as an alternative. From the 1960s, Bayesian confirmation theory became the dominant probabilistic framework.
Debates
- Is induction rationally justified?
- Hume argues any justification of induction is either circular or question-begging; responses range from Popper's rejection of induction to Bayesian and pragmatic vindications.
- Logical versus subjective probability
- Carnap's programme treats confirmation as objective logical probability, whereas Bayesians typically interpret probabilities as rational degrees of belief.
Key figures
- David Hume
- Carl Hempel
- Nelson Goodman
- Rudolf Carnap
- Karl Popper
- Thomas Bayes
Related topics
Seminal works
- hume1748
- hempel1945
- goodman1955
- carnap1950
Frequently asked questions
- What is the problem of induction in one sentence?
- It is the question of how we can be justified in inferring that unobserved cases will resemble observed ones, given that any such inference seems to presuppose the very uniformity of nature it tries to establish.