27/7/2018

Carl Hempel and ‘Covering-Law’ Models of Explanation

Filed under: — Bravus @ 11:27 am

As part of my on-going reading in the philosophy of explanation I’ve been focusing on the work of Carl Hempel, who talks about what Dray has described as ‘covering-law’ approaches to explicating explanation in science.

Together with Paul Oppenheim, in 1948 Hempel described ‘deductive-nomological (D-N)’ explanation in science: explanation in terms of scientific laws combined with initial conditions.

His 1965 work, which I’m reading now, expands this understanding to include ‘inductive-statistical (I-S)’ explanations, noting that some scientific laws are inherently statistical in character rather than deterministic. While Dray originally included only D-N explanations when coining the term ‘covering-law’, Hempel expands the term to include I-S explanations.

Hempel talks only about I-S explanations which make the probability of the outcome ‘practically certain’, or very close to 1, however I already know from reading David-Hillel Ruben that there also exist I-S explanations that explain outcomes with low probability, and even explanations that decrease the probability of the thing they explain.

The relevant chapter is about 130 pages long and includes a lot of defenses of this approach against a number of challenges, as well as expanding the discussion to include historical and other explanations as well as scientific ones.

Relevant to my interests, he also considers the ‘pragmatic’ features of an explanation given to an individual person, as well as the general explanations given in science. What is required to explain to an individual depends on characteristics such as the person’s existing knowledge and interests, whereas a general explanation does not depend on these things.

After finishing Hempel’s account, next step is to move on to Wesley Salmon… and then Peter Achinstein.

Leave a Reply

Your email address will not be published.

This site uses Akismet to reduce spam. Learn how your comment data is processed.