Diachronic Dutch book arguments, or Dutch strategy arguments, have been used to recommend the rule of conditionalization (C) Jeffrey’s rule for fallible learning (J), and the principle of reflection (R). The arguments have been criticized in a variety of ways—more ways than can be discussed in this paper. Here we will look at two of the criticisms. As we proceed, it will become apparent how some criticisms not explicitly raised can be met; others will go unexamined. The criticisms are these: First, Dutch strategy arguments are said to fail as defenses of rules J and C, and principle R, because rational agents may violate each of those constraints, yet avoid vulnerability to a Dutch strategy; their knowledge of Dutch strategy arguments enables them to do this. Second, it is said to be no wonder that the arguments fail, because there are counter-examples to each of these purportedly general constraints on rational belief.

Logicism as a philosophical enterprise died a sudden and unnatural death in the early 1930s. The logicist program was an attempt to secure our mathematical knowledge in the indubitable bedrock of our a priori logical intuitions. It was a program very much impressed by the remarkable achievements in formal logic and axiomatics in the early century. While that program is well dead and gone, a research program within artificial intelligence (AI) has come to be known by the same name, sharing with its homonymous predecessor at least that program’s esteem for logic.

Logicism in AI, briefly, is the view that prerequisite to developing an artificially intelligent system we must find means for representing and reasoning with a vast store of prepositional knowledge and that the only viable candidates for representing prepositional knowledge are formal logics (and their associated formal languages).

Nancy Cartwright’s new book Nature’s Capacities and their Measurement (1989)2 attempts to achieve something remarkable: refute Hume’s most crucial theses concerning causation and let in capacities largely through methodological arguments. The two central theses of Hume which Cartwright challenges are:

1. (1) Generic causal facts (e.g., causal laws) are reducible to regularities.

2. (2) Singular causal facts are true in virtue of generic causal facts.

Cartwright’s arguments against (1) and (2) are methodological in the sense that she arrives at her conclusions by considering our best methods for studying the relationship between probability and causation. Cartwright rejects (1) on the grounds that if we look at the way in which probabilities are related to causal relations as expressed in her principle CC, or if we carefully study the methodology of causal modelling employed in the social, behavioral, and life sciences, we see that to get certain causal information from probabilities we must presuppose the truth of other generic causal facts.

The priority of mathematics in the construction and deployment of the physical sciences has recently come under attack. Empiricism in the philosophy of mathematics, an offshoot of the trend towards naturalized epistemologies, subordinates formal disciplines to empirical facts (conceived variously as experience or experiment). (Kitcher 1983 and Maddy 1991) When allied with structuralism (Chihara 1973 and Resnik 1981,1982,1990), which denies to mathematics any specific subject matter of its own, it displaces intelligibility, heuristic force, and explanatory power away from mathematics into the realm of “fact”.

In this essay, I will argue that mathematics, and mechanics conceived as a formal science, do have their own proper subject matters, their own proper unities.

Recently, several authors have attempted to characterise causation by identifying causal connections with physical processes. While this approach may not live up to providing a complete analysis of causation in all its contexts, it has raised some significant points regarding the causal role played by various entities described in natural laws. To date, much of the discussion has involved the attempted identification of physical terms or concepts that necessarily or sufficiently characterise causal processes, and has centered around three main contentions:

1. Cl. Causal processes involve the presence of forces, and that forces are themselves a species of primitive causes (e.g., Bigelow, et al. 1988).

2. C2. Causal processes necessarily involve transfer of a physically measurable quantity, such as energy-momentum (e.g., Aronson 1982 and Fair 1979).

3. C3. Causal processes necessarily involve microscopic interactions between “fundamental” entities, such as is described by quantum field theory (e.g., Heathcote 1989).

The overall character of the ideas Thomas S. Kuhn has offered concerning the nature of scientific inquiry has been generally misunderstood, or, rather, misconstrued. (See Kuhn 1957,1970.) Kuhn’s ideas do not add up to a fully articulated analysis of the structure of the scientific process. Kuhn does not offer a theory of science which should be evaluated in the same way as, e.g., the hypothetico-deductive model of science or the inductivist one. What Kuhn does is best viewed as calling our attention to certain salient phenomena which a philosophical theorist of science must try to understand and to account for. We do injustice to Kuhn if we deal with his views as if they were finished products of philosophical theorizing. They are not. Rather, they are starting-points for such theorizing; they pose problems to be solved by a genuine theory of science.

The theory-ladenness of perception argument is not an argument at all. It is two clusters of arguments. The first cluster is empirical. These arguments typically begin with a discussion of one or more of the following psychological phenomena: (a) the conceptual penetrability of the visual system, (b) voluntary perceptual reversal of ambiguous figures, (c) adaptation to distorting lenses, or (d) expectation effects. From this evidence, proponents of theory-ladenness typically conclude that perception is in some sense “laden” with theory. The second cluster attempts to extract deep epistemological lessons from this putative fact Some philosophers conclude that science is not (in any traditional sense) a rational activity (Feyerabend 1975); while others conclude that we must radically reconceptualize what scientific rationality involves (Kuhn 1970; Churchland 1979).

I argue here that for the past thirty years there has been a persistent misreading of Thomas Kuhn’s The Structure of Scientific Revolutions. Not everybody has participated in this misreading, but its persistence is remarkable. So one of my tasks is to diagnose the reasons for its continuance. It is also an important misreading, because it attributes to Kuhn a view that is from most people’s perspective highly untenable, that scientists in some strong sense construct the world by choosing a paradigm.

To avoid repeating work done by Gutting and many others (see Gutting 1980) I will focus on Kuhn’s constructivist metaphysical program, as seen by commentators.

Variation is the fuel of evolution, but as is well known, not all variation will drive Darwin’s engine along. Within population genetics, “additivity” is one concept used to characterize the salient type of variation. The characterization of variation (variance) in terms of additivity originates with R.A. Fisher (1918), and Fisher’s “fundamental theorem of natural selection” (1930) states that the rate of evolutionary change is proportional to the additive variance in fitness in the population at that time. Evolution proceeds until the additive variance is exhausted.

It would not be surprising if a concept so important in the mathematical structure of evolutionary theory had philosophical significance. In two important works, William Wimsatt (1981) and Elisabeth Lloyd (1988) have claimed that additivity can be used to give a criterion for deciding the “unit of selection” in an evolutionary process.

Ever since the publication of his biophilosophical ideas in the early seventies, Jacques Monod has always been criticized quite severely for his old-fashioned philosophical reductionism. These numerous critiques have created a picture of an eminent scientist whose philosophical considerations are founded on the mechanistic worldview of Cartesian metaphysics, a view which is completely unacceptable to contemporary philosophers. Today it is the usual practice to cite Monod as one of the strongest and most radical reductionists in modem biology.

Monod’s biophilosophical conception arose at a moment when vigorous discussions on reductionism followed the great successes of molecular biology. His ideas were criticized namely for their strongly reductionistic interpretation of biological knowledge, and for their mechanistic model of life which, as Monod believed, reflected the spirit of modem science.

The notion of a genetic program, as described by Mayr, is ripe for an analysis inspired, at least in part, by recent debates in the philosophy of science taken under the heading of realism versus anti-realism.

That there was some chance element to evolutionary change was clear even to Darwin (cf. Hodge 1987; also Hodge & Kohn 1986), but chance has assumed a more central role in evolutionary theory since the early decades of this century. It was certainly clear to theoreticians such as Dobzhansky, Wright and Fisher that an adequate evolutionary theory would have to take a stochastic form, though they disagreed as to the significance of chance in evolution (Gigerenzer et al., 1989, ch. 4; Turner 1987). As Beatty explains, the central difference is that in more modern treatments, not only is the origin of variations a matter of chance, their evolutionary fates are also matters of chance (Beatty 1984; cf. Beatty 1987). Probabilistic theories of evolutionary change, as we shall be concerned with them, emphasize this dependency of evolutionary fates on chance, and treat evolutionary change as irreducibly probabilistic.

What is relativistic quantum theory? How does it differ from non-relativistic quantum mechanics? Is there something more involved than the transition from the Galilean group to the Lorentz group?

These questions are evidently of great importance to the understanding of relativistic quantum theory (RQT), in particular to the distinction between the basic principles of quantum theory, and the characterization of a particular class of dynamical evolutions. Within non-relativistic quantum mechanics (NRQM) a reasonably precise response is possible (differences concern the number of degrees of freedom, the existence of internal symmetries, and the form of the Hamiltonian as a function of the canonical variables); what appears characteristic of RQT is the appearance of a dynamics of a qualitatively different character (involving the creation and annihilation of particles), involving new kinds of constraints (renormalization theory: anomalies). Nevertheless, it seems we must know how to pass from the inhomogeneous Galilean group (IHGG) to the Lorentz group (IHLG), if only because we have successfully constructed a RQT, surely, then, we know what it is that we hold invariant in this transition.

The term “experimental philosophy” has been used to refer to the solution of what were considered philosophical problems by means of laboratory experiments. A recent example of this was the experimental violation of the Bell inequality, which ruled out certain philosophically appealing “realist local” theories as alternatives to quantum mechanics (QM).

Following the spirit of such experimental philosophy, this paper proposes a feasible test between two different interpretations concerning the nature of “mixtures” in QM. The use of delayed coincidence techniques seems to show that the process of mixing beams of light in different polarizations is reversible, favoring a weak version of the so-called “ignorance interpretation” over the “instrumentalist” view.

Quantum field theory (QFT) poses the following well-known problem. Calculations of values for certain quantities begin with a first-order approximation to which are added higher order corrections. In many cases, the calculation of even the very first higher order correction term yields infinite values. These infinite values are due to divergences in the theory. These divergences can be traced to their source in the standard fundamental assumptions defining QFT. No matter which version of the formalism we adopt, we are trying to incorporate into the mathematics certain fundamental physical constraints on the nature of the vacuum state, on the states that can be occupied by systems, on the comeasurability of observables, and so on. The resulting formalism contains divergences.

Suppose we set about creating pairs of photons in the singlet state, allow them to separate to some considerable distance, and then send each photon into a polarizer oriented at some randomly chosen angle. Quantum theory predicts that in the long run we will observe the following behavior. On each wing of the experiment we will see perfectly random behavior. Roughly half of the photons will pass their polarizers and half will be absorbed. There will be no correlation between passage or absorption in one experiment and passage or absorption in the succeeding experiments. There will be no correlation between the passage or absorption and the angle at which the polarizer is set. But when we examine the correlations between the two wings of the experiment, something remarkable emerges.

Convergent realists typically maintain the following theses:

1. Tl: Some scientific theories, literally interpreted, are at least approximately true.

2. T2: Some scientific theories, literally interpreted, genuinely refer.

3. T3: The history of mature sciences is a progressive approximation to the truth.

4. T4: The world (reality) described by scientific theories is independent of our thoughts or theoretical commitments.

Convergent realists usually offer the infamous “success of science” argument as the strongest defense of their view. Scientific realism, according to convergent realists, is the only view that does not portray the success of science as miraculous (i.e., unexplainable). Various versions of convergent realism have been defended by writers such as Richard Boyd (1973,1980,1984), W.H. Newton-Smith (1978), Hilary Putnam (1975,1978), JJ.C. Smart (1963,1985,1989), and Russell Hardin and Jay Rosenberg (1982). Critics of convergent realism include Larry Laudan (1981), Arthur Fine (1984), and Bas van Fraassen (1980).

An effective way to challenge scientific realism these days is to ask whether evidence for a theory’s truth can ever outstrip evidence for its utility. Granted that we might have strong evidence of a theory’s explanatory, predictive, heuristic, and technological utility, might we have in addition evidence of its truth? A negative answer claims that all the evidence in support of a theory is, immediately, evidence of pragmatic virtues, and only indirectly, if at all, evidence of truth. Realism then reduces to a willingness to infer truth from utility. Evidently, none of the evidence by which theories are tested will pertain to this inference, so the only recourse for sustaining it is to a metalevel at which truth and utility are compared in the abstract. And there enormous problems loom.

The central theme of this paper is that historical contingency plays an essential and ineliminable role in the construction and selection of a successful scientific theory from among its observationally equivalent and unrefuted competitors. Let me make a few clarifying remarks at the outset. It is not my claim that in all cases of theory selection there is in practice this radical type of underdetermination. Nor am I concerned with the rather trivial and philosophically uninteresting historical contingency of who did what when. For example, in response to Napoléon’s fishing for a favorable comparison between himself and Newton, Lagrange is said to have lamented (Gillispie 1960, p. 117; Moritz 1914, p. 167): “Newton was the greatest genius that ever existed, and the most fortunate, for we cannot find more than once a system of the world established.“

If the appropriate cognitive model of the mind is a connectionist model, then what does the future of folk psychology (FP) look like? “Bleak” say the eliminative materialists. In several recent articles and books Paul and Patricia Churchland have been arguing by demonstration that there are models on which to explain all behavior—not merely cognitive behavior—models which are more fruitful than FP. These models, which are connectionist in nature, are supposed to describe a better theory of behavior than FP, and so will eventually replace it. In this paper I look at several arguments the Churchlands have advanced for the thesis that FP will be replaced in a mature cognitive science, i.e., that connectionism will supplant FP.