Lipton I B E Ch 5 Contrastive Inference
—-
Summary {[green Note: below summary below is only of the second part of this chapter thus far. My notes are in a different location at present. ]}
This chapter takes the form of a case study of methods employed to track down the cause of increased infant mortality in a particular hospital ward.
Explanation & Deduction - ~narrow inductivism: — can never gather all relevant data — relevance is often theory dependent - hyp.-ded. method can eliminate the second problem — relevant only if logically implied by hyp. - But it doesn’t give an account of the source of the hypothesis. - How are we to narrow the field of possible hypotheses? — contrastive evidence — explanatory inference - So we have the following cycle: — contrastive data leads to explanatory hypothesis leads to experiments to collect more data - The hyp.-ded. model then gives too much priority to the hyp. rather than its source(s). — And still have problem re: disjunctive inclusion. - Furthermore, the hyp.-ded. theorist needs to specify background conditions in order to obtain logical contradictions necessary for negative inference. — And hence needs to give an account of the principles by which this is to be done, which is problematic. - IBE on the other hand has the significantly lower failure standard of “failure to explain.” — i.e. failure to specify the difference in causal histories. — Of course still need to decide is failure => incomplete theory or failed theory. - Thus Popper is wrong in asserting that falsification is the only way in which theories are eliminated. - Additionally, IBE is still able to account for the (descriptively accurate) use of deduction in some cases. — e.g. mathematical calculations.
—-
What do I think? - The superiority of IBE over the hypothetico-deductive model (both in descriptive and normative terms) is pretty thoroughly demonstrated above. Although, the hypothetico-deductive model never really seemed that strong to begin with so one wonders how much of an achievement this really is.
—-