Scientific Explanation and Theories of Discovery

格莱莫教授正在研究科学解释的问题。在他的方法中，解释的概念具有结构属性，这些属性在不同的语境中都是不变的。他认为，这种方法首先是可取的，因为解释的传统目标之一是提供发现策略的一部分；也就是说，我们应该推测那些假设或理论，为任何相关现象提供最好的解释。为比较解释的好坏提供结构性标准的特征描述，提供了一种标准的前景，这种标准可以应用于在其实质不同的研究领域中帮助发现（或为已经宣称的发现提供理由）。如果这些程序可以实行，那是非常值得期待的。其次，他认为，通过推进结构性哲学理论并研究其局限性，而不是将其排除在考虑之外，可以更好地取得哲学进步。第三，在以前的NSF资助下，他在计算机程序中实施了解释良好性的结构标准，这些标准有助于在各种不同的实体领域中寻找因果解释。在有可能对这些程序进行测试的地方，它们被证明是惊人的可靠。因此，他假设这是一个合理的推测，即某些类型的结构标准，适当地使用，是可靠的指导发现，当然不是在每一个上下文中，但在具有某些一般性质的上下文中。人们不知道的是为什么会这样。为什么对最佳解释的推断，或者至少是在某些地方使用的某些形式的推断，就应该是通往真理的可靠指南呢？在这项资助中，Glymour教授将试图对科学解释做出更明确的特定解释，并调查为什么，如果有的话，它应该成为科学探究的可靠程序的一部分。这一事业部分是试图使似乎由许多例子揭示的直觉变得精确，部分是规范的事业，并关注为什么在寻求真理时，偏好更好的解释是一种理性的政策。对于解释的概念，还没有一个完全清晰、笼统的解释，方法论文献很少或根本没有讨论为什么用它做出的推论应该是可靠的。这个项目的目标就是“清晰、全面的说明”。

Professor Glymour is examining the issue of scientific explanation. In his approach, the notion of explanation has structural properties that are invariant across contexts. He argues that this approach is preferable first because one of the traditional aims of explanation has been to provide part of a strategy of discovery; that is, we should conjecture those hypotheses or theories that provide the best explanation of whatever the relevant phenomena may be. Characterizations that provide structural criteria for comparing the goodness of explanations offer the prospect of a canon that can be applied to aid in discovery (or to afford justification for discoveries already claimed) in areas of inquiry that differ in their substance. If such procedures can be had they are much to be desired. Second, he argues that philosophical progress is better made by advancing structural philosophical theories and investigating their limitations than by excluding them from consideration. Third, he has under previous NSF grants implemented structural criteria for goodness of explanation in computer programs that aid in the search for causal explanations in a variety of different substantive domains. Where it has been possible to test these programs they have turned out to be astonishingly reliable. He thus assumes that it is a reasonable conjecture that certain sorts of structural criteria, properly employed, are reliable guides to discovery, not in every context certainly, but in contexts that share certain general properties. What is not known is why that should be. Why should inference to the best explanation, or at least certain forms of it used in certain places, be a reliable guide to truth? In this grant, Professor Glymour will try to make more explicit a particular explanation of scientific explanation and to investigate why, if at all, it should form part of a reliable procedure for scientific inquiry. The enterprise is partly one of trying to make precise the intuitions that seem to be revealed by a number of examples, in part the enterprise is normative, and concerned with why preference for better explanations is a rational policy if one seeks the truth. An entirely clear, general account of the conception of explanation has yet to be given and the methodological literature contains little or no discussion of why inferences made with it should be reliable. It is this "clear, general account" which is the aim of this project.