Lipton I B E Ch 8 Explanation As A Guide To Inference
—-
Summary
Overall strategy in argument for IBE: - 1) Describe what makes for likeliness (of inference) and loveliness (of explanation). - 2) Show how these virtues match up. - 3) Show how loveliness is the inquirers guide to likeliness.
Work done so far: - 1): — accounts of inference have brought out some inferential virtues — contrastive explanation has been used to highlight some explanatory virtues - 2): — A structural similarity has been observed between Mill’s method of difference and the conditions required for lovely contrastive explanation. — A match between the Bayesian calculation of P(E|H) and the explanatory quality of an hypothesis has been argued for.
Many other (potential) virtues seem both explanatory and inferential: — e.g. mechanism, precision, scope & simplicity
We need the concept of explanation to admit of degrees (i.e. we must be able to distinguish between two explanation on the basis of how good they are as explanations) in order for Lipton’s IBE to get off the ground. — Background beliefs play a role in what is considered to be explanatory and this interest relativity may give us the continuum we require.
Arguments against loveliness: - “Whatever is true explains” — Just obviously not true, there must be other criteria at work, criteria which may give us the explanatory continuum we require. - “Explanation is exhausted by giving a causal history” — This does give us a limited loveliness as potential causes may differ in their levels of explanatoriness. — Explanation level may also depend on how causes are described.
Correlation does not imply causation - Even if we are able to demonstrate the correlation of inferential and explanatory virtues we still need to provide an argument as to why this should be interpreted as explanation guiding inference. - However, arguments for matching (correlation) seem to provide ammunition for those who wish to deny guiding: — One can argue for matching by showing that other inferential practices also match up with explanation virtue-wise. — However, an opponent can then respond that it is actually this other inferential practice that is used to guide our usual one, not explanation.
Argument strategy for explanatory matching and guiding: - 1) Show that providing a role for explanation gives us a better description of inductive practices. - 2) Show that explanatory guiding is the best explanation for the large cognitive role played by explanation. - 3) Show that our inclination to infer in a causal way leads us to proceed explanatorily.
IBE as a better description: - broader scope than instantial model - broader and narrower scope than hyp-ded - Bayesian: — aids in calculations — gives account of how hypotheses are generated — gives account of how relevant evidence is determined - Mill’s method of difference (..M.o.D.): — gives account of hypothesis selection — gives account of how difference is determined (whereas M.o.D. just gives inference from difference to cause) — allows for inferred differences — and so allows for non-observable causes to be inferred — overcomes difficulties for M.o.D. : — requirement that only one difference exist — responds to this with method for choosing between competing differences
The prominent role of IBE in cognition: - why are we concerned with understanding as well as inference? — seems strange if understanding is something that takes place after inference (explanation reduced to epiphenomena) — involvement of explanation in inference accounts for it — and is supported by research of cognitive psychologists
It is easier to reason in physical rather than logical terms: - hence we create simulations as part of our inference procedures — and there is a role for explanation in these simulations
Is there a gap between the cause as why (lovely) and the cause as how (likely)? - deterministic view seems to imply that we don’t need explanation - hence need to see whether explan. considerations are causing inference or if our inferences and our explanations are both the result of our belief in the deterministic principle. — there are signs of explanatory involvement in what appears to be “pure” M.o.D. — explanation seems to play a role in deterministic reasoning — in the subjunctive case, it is much easier to deal with explanation rather than causation (as conditions are less stringent) - seems that it is more natural to talk about the joint explanatory/inferential virtues of mechanism and unification in explanatory terms - background beliefs play a large role in determining the nature of other virtues and are themselves the result of past belief and virtue considerations
—-
What do I think? - I’m not sure that appealing to background beliefs will actual yield the explanatory continuum that Lipton is after. Surely, in scientific circumstances one is looking to fix these sorts of beliefs and so they can’t be called upon to judge explanatory loveliness in this context. This would seem to rule out the role of loveliness in scientific inference. — It seems that one of the aims of scientific practice is too rule out this sort of explanatory subjectivity based on the background beliefs of individuals. — After discussion with Jason it looks like I’ve fallen into the Myth of the Given on this one! It appears that I’ve taken it as granted that when one holds fixed the relevant elements of the world, as it were, one normative stance towards the world should be determined too. - This leads to an interesting consideration connecting with some Mc Dowellian perception stuff: it appears that our inferences belong to the space of laws (i.e. scientific nature) whereas our explanations belong to the space of reasons. When I have a minute it’d be a good idea to have a think about how McDowell’s strategy for obtaining the peaceful coexistence of these two spaces may be brught to bear in this context. - Is step 2 of the guiding/matching argument question-begging? — {[green Will delay judgement on the circularity issue in general until I come to assess matters as a whole at the end of the book. ]}
—-