| If the world is physical why are there so many phenomena we explain in non-physical terms---phenomena that we instead explain in sociological, or economic, or biological terms? For example, we typically give an economic explanation of a decline in the value of the dollar and we give an evolutionary explanation of the change in the color of a certain population of moths.;My dissertation is motivated by this puzzle. Some philosophers react to this puzzle by, either explicitly or implicitly, dropping physicalism. Others take a pragmatic approach, claiming that ideally, we would explain everything using the tools and vocabulary of physics, it's just that our intellectual limitations prevent us from forming or grasping physical level explanations of, for example, economic phenomena.;My account works differently. I develop an account on which, often, the correct level of explanation is higher than that of physics; we are not just forced towards higher-level explanations by our limitations.;Chapter 1 surveys the literature on levels of explanation, and finds the existing approaches flawed. Many accounts are flawed because they end up denying physicalism. Notably, informational approaches---which say that higher-level explanations contain some information that lower-level explanations do not---end up only being tenable if we accept that there are some (contingent) facts that are not determined by the physical level facts.;Other popular approaches end up failing to recommend good high-level explanations. In particular, the difference-making approach---which says that explanations should cite the things that make a difference to the phenomenon---fails to recommend explanations at a high enough level because physical facts always make a difference to special science phenomena. A couple of other accounts end up favoring less specific explanations in a way that seems unjustified.;In chapter 2 I give an account of levels of explanation that avoids these problems. I work with a very simple causal account---to explain an event is just to provide information about the physical causes of the event. The right level of explanation is picked out by adding an account of explanatory goodness. Explanations at the right level score better on explanatory goodness than those at the wrong level.;My account of explanatory goodness involves two dimensions---PRECISION and ROBUSTNESS. (In indeterministic cases a third dimension, CHANCE, is also relevant.).;PRECISION says A explains B better if A explains B in more physically possible worlds where A holds than if A explained B in fewer such worlds. In effect, it says that explanations should render the phenomenon, in a certain sense, expectable. ROBUSTNESS says A explains B better if A explains B in more physically possible worlds where B holds than if A explained B in fewer such worlds. In effect, it says that the things cited in the explanation should be required, or close to required, for the phenomenon. Together, they imply that the ideal situation is one where A and B hold in exactly the same range of physically possible worlds.;More generally, PRECISION tends to favor lower-level explanations and ROBUSTNESS tends to favor higher-level explanations. I argue that the combination of these dimensions picks out the right level of explanation for each phenomenon. Sometimes there is more than one acceptable level of explanation for a phenomenon---my account can capture this by allowing multiple acceptable trade-offs of PRECISION and ROBUSTNESS.;In chapter 3 I give an account of a distinction between natural and unnatural properties in the special sciences. Without such a distinction there would be explanations involving 'gerrymandered' properties that score maximally on PRECISION and ROBUSTNESS but are clearly bad explanations. There are further theoretical needs for such a natural/unnatural distinction.;The account is based on the idea that gerrymandered properties are 'cooked up' for one specific purpose while natural properties are explanatorily useful in a wide range of situations and have rich explanatory connections to other properties. I cash this thought out by developing the notion of explanatory clusters. Sets of properties are explanatory clusters when most of the facts about those properties can be explained well by other facts about those properties. The members of explanatory clusters are natural properties.;I end in chapter 4 by discussing how my view explains certain important features of the institutional structure of special scientific practice. In particular, I use the accounts developed earlier in the thesis to make sense of (i) Why various special sciences are methodologically autonomous; that is, why different sciences seem to have such different methodologies with little interaction between them. And (ii) why certain sciences, like economics, tolerate exceptions to their theories while other, like fundamental physics, do not. (Abstract shortened by ProQuest.). |
