Humean metaphysics

Whereas the philosophy of science was dominated in the second half of the twentieth century by epistemological issues raised in the context of logical empiricism and its critics, the project of a metaphysics of Nature (metaphysics of science) has been rehabilitated recently. One can broadly distinguish three positions within that project: a Humean metaphysics that is close to empiricism (e.g. David Lewis, Barry Loewer and Helen Beebee), a metaphysics of universals (e.g. David Armstrong) and a metaphysics of powers (e.g. Sydney Shoemaker, Alexander Bird, Stephen Mumford as well as Charles Martin and John Heil). This chapter is about the opposition between the first and the third of these positions. I shall first outline the Humean view on properties, laws, causation and probabilities, and point out how on the one hand this view is parsimonious, whereas on the other hand it provokes the objection that it is deficient (this section). I shall explain how the metaphysics of powers seeks to remedy these deficiencies and consider its view of probabilities (Section 7.2). The chapter then recalls the standard argument against the conception of properties in Humean metaphysics, goes into arguments from physics and finally maintains that the metaphysics of powers, in contrast to Humean metaphysics, is able to do justice to both the ontological commitments of physics and of the special sciences (Section 7.3).

Introduction

It is part of our common sense conception of the world that what happens now can make a difference to the future but not to the past; events in the present, we believe, can causally influence the occurrence of future events but not of past events. What is the relation between this asymmetry and other physical asymmetries? Is the causal asymmetry fundamental or can our asymmetric notion of cause be shown to be reducible to some other physical asymmetry? There is a venerable tradition in the foundations of physics and the philosophy of science according to which the causal asymmetry is intimately related to the temporal asymmetry embodied in the second law of thermodynamics. This view has recently been forcefully defended by David Albert (2000) and by Barry Loewer (2007), who argue that the causal asymmetry can ultimately be grounded in the very same facts that give rise to the second law of thermodynamics, chiefly among them a low-entropy constraint on the initial state of the universe.

In this chapter I will critically examine aspects of their accounts and will argue that neither account is successful as developed so far. In Section 2.2 I will briefly summarize the Boltzmannian account of the thermodynamic asymmetry, from which Albert and Loewer aim to derive asymmetries of causal influence and control. Both accounts centrally involve the claim that it follows from the Boltzmannian account that possible macro-evolutions are much more restricted toward the past than toward the future.

Introduction

The (metatheoretical) structuralist program for the reconstruction of scientific theories (in the following referred to as ‘structuralism’) allows for the performance of several metatheoretical tasks, among them: (a) precisely explicating the inner structure of any given ‘textbook’ theory in all its complexity; (b) making a clear distinction between different theories that in standard expositions are usually not clearly distinguished from each other, and identifying their intertheoretical relationships; (c) reconstructing their evolution in historical time. The present essay applies the structuralist methodology to the analysis of two writings of the German physicist Rudolf J. Clausius on thermodynamic phenomena published in the middle of the nineteenth century. Conventional historiography interprets these writings as the presentation of a first well-founded theory of phenomenological thermodynamics. However, as our analysis will show, Clausius actually constructed in his papers three different theories – each of them with its own identity, though of course related to each other.

The essentials of the structuralist representation of theories

Structuralism (without this name) was initiated by the pioneering works of Joseph D. Sneed, The Logical Structure of Mathematical Physics (1971) and Wolfgang Stegmüller, Theorienstrukturen und Theoriendynamik (1973). However, the present chapter employs the more evolved conceptual tools and representation methods to be found in the joint work, An Architectonic for Science, by Wolfgang Balzer, C. Ulises Moulines, and Joseph D. Sneed (1987). For those readers who are not well-acquainted with structuralist ideas, a brief summary of them is laid out in what follows.

Statistical mechanics and philosophy

Statistical mechanics attempts to explain the behaviour of macroscopic physical systems (in particular their thermal behaviour) in terms of the mechanical properties of their constituents. In order to achieve this aim it relies essentially on probabilistic assumptions. Even though in general we do not know much about the detailed behaviour of each degree of freedom (each particle), statistical physics allows us to make very precise predictions about the behaviour of systems such as gases, crystals, metals, plasmas, magnets as wholes.

The introduction of probabilistic concepts into physics by Maxwell, Boltzmann and others was a significant step in various respects. First, it led to a completely new branch of theoretical physics. Second, as Jan von Plato pointed out, the very meaning of probabilistic concepts changed under the new applications. To give an example: whereas before the development of statistical physics variation could be conceived as the deviation from an ideal value this was no longer a tenable interpretation in the context of statistical physics. Genuine variation had to be accepted (von Plato, 2003: 621).

Furthermore, the introduction of probabilistic concepts triggered philosophical speculations, in particular with respect to the question whether the atomic world does indeed follow strict deterministic laws (cf. von Plato, 1994; Stöltzner, 1999). For instance, in 1873 Maxwell gave a lecture entitled ‘Does the Progress of Physical Science tend to give any advantage to the opinion of Necessity (or Determinism) over that of the Contingency of Events and the Freedom of the Will?

Arrows of time

Although most fundamental laws of physics do not distinguish between past and present, there exist classes of phenomena in Nature that exhibit a direction of time: their time-reversed versions, although in perfect accordance with the underlying laws, are under ordinary conditions never observed. Arthur Eddington called such classes ‘arrows of time’, see Zeh (2007) for a detailed overview. The main arrows of time are the following:

• radiation arrow (advanced versus retarded radiation);

• second law of thermodynamics (increase of entropy);

• quantum theory (measurement process and emergence of classical properties);

• gravitational phenomena (expansion of the universe and emergence of structure by gravitational condensation).

The expansion of the universe is distinguished because it does not refer to a class of phenomena; it is a single process. It has therefore been suggested that it is the common root for all other arrows of time – the ‘master arrow’. As will become clear below, this does indeed seem to be the case.

The radiation arrow is distinguished by the fact that fields interacting with local sources are usually described by retarded solutions, that is, solutions which in general lead to a damping of the source. Advanced solutions are excluded; they would describe the reversed process, during which the field propagates coherently towards its source, leading to its excitation instead of damping.