• Imre Lakatos (1922-1974)
  • studied at LSE (a disciple of Popper)

Falsificationism

• presents 3 versions of Popper
  • Popper0: Dogmatic falsificationism
  • Popper1: Naive methodological falsificationism
  • Popper2: Sophisticated methodological falsificationism
    • Lakatos thought Popper was heading from 0 -> 1 (but was unable to reach 2)

Popper0: Dogmatic Falsificationism

• Simple version of Popper's falsificationism

  • theories really can be falsified

    

    • cf. modus tollens
• Lakatos assumes that Popper thought this way when he was young but it didn't appear in his published works

Popper1: Naive Methodological Falsificationism

• We need an observation statement for testing

  • the scientists' community needs to have a consensus on the observation statement
    • ex. is black swan a swan or a shwan?

• We also need an auxiliary statement from the theory to make a prediction ⇒ Falsificationism is based on decisions

  • ex. decide to accept the observation statement
  • ex. decide to accept auxiliary hypotheses
  • ex. decide to accept there are no confounding variables

  then, proceed with falsificationist logic

• Lakatos purports that this is the real Popper (what he has published)

  • What if those decisions are wrong?
    • Popper allows "appeal procedure" but doesn't explain the details
  • Lakatos’ criticisms:
    • there are risks involved in each decision
    • is this really rational? (real-rationalism or pseudo-rationalism?)
    • doesn't fit in with the history of science
  • you only give up on a theory when there's a better alternative

Popper2: Sophisticated Methodological Falsificationism

• Lakatos completing Popper's theory (improvement on Popper)
• A theory T is falsified iff another theory T' has been proposed such that:
  1. T' explains the previous success of T
  2. T' has excess empirical content over T (i.e. makes novel predictions)
  3. Some of the excess of T' is corroborated
• A series of theories T, T', T'', ... is progressive if each new theory has corroborated excess empirical content (= leads to the discovery of a new fact)
  • A series is degenerating if that is not the case
  • T’ → T’’ is a progressive problem shift if T’’ has corroborated excess empirical content over T’
  • T’ → T’’ is a regressive problem shift if T’’ does NOT have corroborated excess empirical content over T’
• Demarcation criterion
  • a theory is "scientific" only if it has corroborated excess empirical content over its predecessor
• The tacking problem
  • a problem with specifying the continuity of the series of theories.
  • In principle one can simply tack on independent hypotheses to obtain a progressive problemshift:
    • let T’’ = T’ + H (where H has nothing to do with T’ but explains a new fact)
    • then T’ → T’’ is a progressive problemshift

Research Programmes

• a research programme consists of a hard core and a protective belt
  • hard core ← unfalsifiable
  • protective belt ← a set of supplementary assumptions that can be modified
  • progressive research programme ← one that has developed through progressive problem shifts and retained its coherence
  • degenerating research programme ← one that has failed to develop through progressive problem shifts or has lost its coherence in doing so
    • lose coherence (hard core weakens)
    • novel predictions were falsified
    • failed to make novel predictions
    • NB: empirical progress is often only seen with hindsight: what were once seen as refutations can come to be seen as confirmations from the perspective of a subsequent theory in the sequence. Thus Lakatos advocates methodological toleration.
    • methodological toleration (cf. adding the assumption that "the universe is infinite" to Galileo's theory ← ad hoc but not degenerative)
    • Wanted to establish a "coherent theory"
    • Scientific revolution: degenerating research programme gets replaced by a progressive one (thru a progressive problem shift)
      • Lakatos wanted to show how this is rational (↔ Kuhn: it's not rational)
• heuristics
  • negative heuristics ← tells you NOT to touch the hard core
  • positive heuristics ← consists of suggestions and hints on how to develop the refutable protective belt
• novel prediction: prediction that wasn't made by previous theories