Explanatory Unification and the Problem of Asymmetry

Citation: Barnes, E (1992) Explanatory Unification and the Problem of Asymmetry. Philosophy of Science 59 (RSS)
Internet Archive Scholar (search for fulltext): Explanatory Unification and the Problem of Asymmetry
Tagged:

Summary

In this paper, Barnes discusses the problem of unificationist theory of explanation proposed by Kitcher (1981). First, Barnes summaries the argument patterns and some terms used in Kitcher’s article, and thinks Kitcher’s theory of explanation can establish the very possibility of understanding the causal structure of the world which causal explanation cannot do. However, Barnes pointes out the problem of asymmetry in unification explanation by using example of height of tower and length of its shadow. In this example, tower height can explain shadow length, but shadow length cannot explain tower height. Despite Kitcher widens the range of explanation by using additional pattern with general patterns, Barnes thinks this way will reduce unifying power of original general explanation. Moreover, genuine explanation should be the most highly unifying based on Kitcher’s argument. However, he cannot explain putative nonexplanatory but unifying argument pattern.

For the unifying but nonexplanatory argument patterns, Barnes gave two examples to explain the failure of unification explanation. One is temporally symmetric closed system (retrodictive pattern), which is different from predictive pattern. Predictive pattern means that the fact it explains is based on the system which connects former explanation and latter subsequent occurrence by laws. However, the retrodictive pattern did not have the system which links the former explanation and latter occurrence. Therefore, the explanation is based on temporally symmetric at that time.

In addition, Barnes points out the difference between explanatory argument and evidentiary argument, and thinks Kitcher lacks the resources to respect the nonexplanatory charactor of such evidentiary argument. The conclusion for evidentiary argument should be an answer to “Why should we believe that conclusion”, while the conclusion for explanatory argument is that “Why such conclusion”. Then Barnes proposed causal process to distinguish the difference between retrodictive argument and predictive argument. A prior causal process of open systems explanatory is easier influence by retrodictive pattern rather than predictive pattern. In other words, retrodictive pattern cannot explain their conclusions, and it is a nonexplanatory and merely evidentiary argument.