Presentation on theme: "The rationality of science"— Presentation transcript:
1The rationality of science Imre Lakatos ( )The rationality of scienceZoltán Dienes, Philosophy of Psychology
2Popper (1934) said scientists must commit in advance as to what results would falsify a theory – then stick with that decisionKuhn (1962) said science is best understood in terms of large-scale “paradigms” relatively resistant to falsification but are kept or discarded according to power struggles
3Lakatos (e.g. 1970):Falsifiability is very important“Blind commitment to a theory is not an intellectual virtue it is an intellectual crime”BUT we need not stick with our original decisions2) Large scale relatively unfalsifiable research programmes are important in characterising scienceBUT there is a rational basis to their rejection
4Popper’s falsificationism (as construed by Lakatos) Popper was a fallibilist: we might be wrong about any piece of knowledge. Theories must be made to stick their neck out. But we can only criticise one thing (e.g. a theory), if we take for true other things. To be critical about anything, we must decide to accept other things.
5Methodological falsificationism Popper was a fallibilist: we might be wrong about any piece of knowledge. Theories must be made to stick their neck out. But we can only criticise one thing (e.g. a theory), if we take for true other things. To be critical about anything, we must decide to accept other things.Decision 1Decide what can be legitimate independent and dependent variablesScores on a certain extroversion scaleReaction times
6Decision 2Decide which measurements you will accept.Take a measurement. But do you trust that reading/score? Safety control: repeat the measurement/experiment (matter of convention how many times)In taking these measurements fallible theories are involved; but for the sake of testing another theory we take them as unproblematic background knowledge.
7How can we reject probabilistic theories? Decision 3: Accept a certain level of significance e.g. p < .05
8Decision 4:Check assumptions of tests, data well behaved, background knowledge safe.Decide to accept all is in order so can be guided by the test outcome.Then results against a theory falsify it.We must “reject the theory and not work on it again on pain of being irrational.” (Lakatos’ description of Popper)
9We set out to test the theory that extraverts are evening people and introverts morning people. We are happy with all four decisions and get results contrary to expected:No relation between extraversion and most productive time of dayMUST we reject the theory?
10We set out to test the theory that extraverts are evening people and introverts morning people. We are happy with all four decisions and get results contrary to expected:No relation between extraversion and most productive time of dayMUST we reject the theory?Some one proposes a possibly better measure of extraversion.Might we test which measures is bestFind the new measure is better at measuring extraversionthen use the new improved measure in our experiment?
11NBPopper (1960) onwards softened his position:One can work on a falsified theory if it has passed more tests than its competitorsOne theory can be closer to the truth than anotherAlso:Popper (1970)Any aspect of background knowledge provisionally assumed as safe can be opened up for critical discussion at a later stage
12According to Lakatos (1970), the original 1934 Popperian account differs from the history of science in several ways.Lakatos:Historically all theories have been ‘born falsified’ into an ‘ocean of anomalies’.Scientists can rationally work on a theory already ‘falsified’ and also change their mind about the conditions that would falsify a theory.A scientific theory can become a hard core of a research programme, treated (for some time) as immune to falsifications.
13The methodology of scientific research programmes A research programme has a hard core and a protective beltHard core: the central beliefs of the programme –e.g.In connectionism: psychological states consist of activation flowing between units through adjustable weightsDopamine theory of schizophrenia: Schizophrenia results from imbalances in dopamineWe are ‘forbidden’ from falsifying the hard core
14Protective belt:Specific theories based on the hard core.Must invent auxiliary hypotheses that form a protective belt around the core, and direct falsifying conclusions to them – they get adjusted, re-adjusted or replaced to defend the thus hardened core.For example, a specific connectionist network of children learning to readPostulating of specific dopamine receptors or pathways as involved in specific symptoms of schizophreniaIf evidence goes against any of these theories, the specific theories are falsified, not the hard core generally
15The notion of a research programme emphasizes the continuity and unity of theories over time: “All swans are white” is falsifiable but not in itself science(there is hard core generating a protective belt of falsifiable theories)
16A scientifically relevant unit of evaluation is the whole research programme. If an adjustment predicts some hitherto unexpected fact the change is theoretically progressive;if some of these predictions are corroborated, it is empirically progressive.Otherwise the adjustment is degenerating.
17Thinking in terms of progressive or degenerating changes rather than Popper’s falsification: Less arbitrary.Any of our decisions can be appealed. We don’t have to stick with any of them.The decisions can be overturned if that leads to a progressive change.
18(Contrary to Popper 1934) It is rational to work in a research programme that is already “refuted” – anomalies can be pushed aside with the hope they will turn into corroborations in the fullness of time.(Contrary to Kuhn) research programmes have achieved monopoly only rarely; the history of science is and should be the history of competing research programmes. The sooner competition starts the better.
19Like Popper:we must retain the determination to eliminate, under certain objectively specifiable conditions, certain research programmes.
20Like Popper:we must retain the determination to eliminate, under certain objectively specifiable conditions, certain research programmes.A research programme is given up if it is degenerating while another is progressive.It is only when a progressive programme explains the failures of a degenerating one that the degenerating one is treated as falsified.
21Evolutionary psychology Parental investment theory and mate selectivityCheater detection moduleAttractiveness of symmetryThe mind evolved by natural selection;Most fundamental human characteristics evolved in the PleistoceneTiming of sex with lovers and long term partnersHuman sperm competitionHard coreProtective beltPositive heuristic: Think more carefully how Pleistocene conditions map onto modern day life; could the apparent adaptation really be a consequence of another adaptation; was the experiment ecologically valid
22One approach to learning When learning complex grammars, people learn small chunks of wordsConditioning results from conscious hypothesis testingAll learning is consciousPeople often base decisions on memorized exemplarsHard coreProtective beltPositive heuristic: Is something else simple people may have learned? How could the test of conscious knowledge be made more sensitive?
23Note on novelty:Progressive means making novel predictions that are confirmed.But what exactly counts as a novel prediction and why is novelty important?Early Lakatos: A novel prediction must be predicting a finding never discovered before (“temporal novelty”).Late Lakatos: A prediction is novel for a theory if it was not used in constructing the theory (”use novelty”).Contrast Bayesian theory