Lakatos describes three types of falsification: dogmatic, methodological and sophisticated. Popper held to a methodological falsificationism, but sometimes it seems Lakatos attributes sophisticated falsificationism to Popper (but this is probably mistaken).
- Science cannot prove a theory, but it can disprove one with the application of the empirical basis. The empirical basis is the set of observational propositions that are the potential falsifiers of a theory; the empirical basis is taken to be infallible.
- Scientific honesty consists in specifying, in advance, an empirical basis, and the unconditional rejection of the theory once falsified.
- Science progresses by repeated overthrow of theories that are falsified by observation (e.g., Descartes vortex theory, Newton’s theory, Einstein’s theory).
However , dogmatic falsification has problems.
- Dogmatic falsification does not take into account the theory-ladenness of observation: it assumes that observational propositions are true.
- Dogmatic falsification assumes the observational proposition can be proven true, but this is false. If the observational proposition isn’t true, then it can’t be used to falsify a theory. This means an observational proposition doesn’t falsify the theory, but is only an inconsistency.
- Even if observational propositions could be proven true, well-established theories still tend to resist falsifying observations: scientific theories are typically protected by ceteris paribus clauses. Probabilistic theories also resist falsification based on finite observations.
- Dogmatic falsificationists require scientific theories to be disprovable by a finite number of observations. Consequently, our strongest theories—Newton’s, Maxwell’s, Einstein’s–are demoted to metaphysics, while “all swans are white” would be scientific (since a single black swan would disprove it).
- Methodological falsificationists deem unfalsifiable, by decision or convention, some observational (or basic) propositions, which are to serve as the empirical basis. By contrast to dogmatic falsificationists, methodological falsificationists recognize this decision as fallible. Successful theories can also be tentatively accepted as unproblematic background knowledge, widening the range of theories open to testing that was restricted by the dogmatic falsificationists.
- In contrast to dogmatic falsificationism, Methodological falsificationists accept that theory may be falsified, even though it could be true (due to the fallibility of the empirical basis).
- Dogmatic falsificationism and methodological falsificationism shares two features: (1) a test is a two-cornered fight between a theory and the empirical basis; and (2) there can only be falsification and not confirmation.
- A theory is scientific iff it predicts novel facts over it’s competitors.
- A theory T1 is falsified iff certain criteria are met: (1) there is another theory T2 that predicts novel facts that are improbable or forbidden by T1; (2) T2 explains all the unrefuted content of T1; and (3) T2’s predictions are successful.
- Sophisticated falsificationism distinguishes between auxiliary hypotheses that represent progress and those that represent degeneration. Theories that have their novel facts confirmed by experiment are call empirically progressive. If a theory is not empirically progressive, it represents degeneration.
- In contrast to naïve methodological falsificationism, sophisticated falsification is not only a relation between the theory and the empirical base, but also includes a third variable—novel facts from competing theories.
Lakatos builds upon lessons learned from these the types of falsification into his methodology of scientific research programmes, where science doesn’t consist of a series of theories, but a bigger research programme. Lakatos can be seen as a compromise between incorporating the Popperian virtue of being open to rejection and Kuhnian concerns of describing how science is historically done.
Lakatos, Imre. The Methodology of Scientific Research Programmes