This is a short thing I wrote up for my philosophy of science class on Arthur Fine’s criticisms of van Fraassen’s position. I liked it, so I thought I’d share. Apparently in that seminar I didn’t make it pass the Scientific Image. The problem wasn’t solved, but I think I figured out what the problem is. Solving it will probably require figuring out why van Fraassen adopts the epistemological views he does. No mean task, that.

In his “Unnatural Attitudes,” Arthur Fine finds fault with anti-realism in the philosophy of science in its attribution of an aim to science, which, in the case of constructive empiricism, is that science aims at empirical adequacy.
Upon further examination, this objection breaks into two further, related criticisms, one that says constructive empiricism artificially splits the inferential practice of scientists and one that says the constructive empiricist policy about accepting unobservables is unjustified. I will argue that van Fraassen can adequately respond to the latter. The former turns on broader philosophical disputes between van Fraassen and Fine that cannot be resolved here, but I will attempt to give a response that does not rely on these. In the end Fine’s criticisms cast some doubt on van Fraassen’s view but they are not decisive against it.

Fine initially criticizes anti-realism for the attribution of an aim to science because an aim distorts our understanding of science. The distortion comes from a desire to “interpret science in accordance with a set of prior, extra-scientific commitments.” The extra-scientific commitment of constructive empiricism, namely to empiricist epistemological principles, is built into the demand of empirical adequacy, which van Fraassen says is the aim of science. Empirical adequacy comes bundled with a distinction between the observable and the unobservable.

Despite rhetoric to the contrary, the source of Fine’s problem does not seem to be that van Fraassen attributes an aim to science, but rather what the aim is. If van Fraassen did not claim that science aims to provide empirically adequate theories, then part of the motivation for the empiricist epistemology is removed. It need not drop out completely, though, as the epistemology is a consequence of, not equivalent to, the aim. Fine’s criticism does not essentially involve an aim being attributed to science. Rather, the heart of the criticism lies with the extra-scientific commitments of the empiricist epistemology.

According to Fine, constructive empiricism’s unnecessary need to interpret scientific practice comes from a commitment to empiricist epistemology; Fine says this has two primary principles, that belief requires justification and that only experience can justify belief, both of which are important below. The interpretation holds that scientific evidence builds for belief when the statements in question are about observables and for acceptance when they are not. Fine puts the point by saying that the constructive empiricist “can follow the usual lattice of inferences and reasons that issues in scientific beliefs only until it reaches the border of the observable, at which point the shift is made from belief to acceptance. But the inferential network that winds back and forth across this border is in no way different from that on the observable side alone.” There is a uniformity in the scientific inferences about microwaves, on the one hand, and about elephants, on the other. Despite the uniformity, the constructive empiricist divides the commitment and supporting evidence into two categories, one for acceptance and one for belief. The inferences drawn by scientists crisscross the epistemological division between propositions about observables and those about unobservables. Thus, the constructive empiricist imposes an unnatural interpretation on science where no natural boundary exists, which is the first criticism of van Fraassen. While constructive empiricism is, rightly in Fine’s eyes, set against “the needless multiplication of entities” it does not follow this policy with respect to “the significance of practices”. Even though the inferential practices are uniform, van Fraassen draws a line between some of the involved propositions, dividing them into beliefs and mere acceptances. Before getting to van Fraassen’s response to this, it will be helpful to lay out Fine’s other objection and respond to it.

Fine’s first objection leads naturally to his second which is that constructive empiricism’s policy of never going for belief in the unobservable is not justified. Fine doubts that van Fraassen’s epistemological considerations successfully underwrite merely accepting, and never believing in, unobservables in all cases. He asks, “Why must it be the case, say for electrons, that the complex history of evidence, successful use, and reasoning at the very best supports belief in the observational reliability of electrons and supports our commitment to behave just as though they exist but nevertheless fails to support the belief that they do exist?” Fine thinks that there is no a priori reason to think that the evidence supporting a theory with unobservables, such as electrons, should never support an inference to the existence of them. Constructive empiricism might be a good heuristic, but the details of the specific cases need to be examined to decide the question of belief. Fine thinks that specific cases will disagree the constructive empiricist stance, although some may agree with it. However this would be a conclusion based on a posteriori knowledge, not a priori commitment.

Against this second criticism, van Fraassen has a reply. Given the principle that only experience can justify belief, van Fraassen’s policy about belief is justified, but this would be question begging. Rather, a better reply is to focus on Fine’s example of the virtues supporting a theory and how this would support belief in the theory’s unobservables. Constructive empiricism says that we are not rationally compelled to believe in unobservables but are rationally compelled to believe in observables. Fine’s claim that the theoretical virtues and historical details about theories of electrons support belief in electrons is ambiguous. If he means that the extra virtues provide some motivation for believing, then van Fraassen can agree; they may motivate but not compel. Alternatively, Fine may mean that the extra virtues rationally compel us to believe in electrons, although Fine does not say much about rational compulsion. If the belief in unobservables that Fine recommends is justified, it is justified by something outside of experience. Constructive empiricism, according to Fine, includes the principle that only experience justifies belief. This objection is a consequence of Fine’s rejection of that principle.

Even without a defense of the epistemological principle, van Fraassen can respond by denying that anything is gained by believing in unobservables. Believing in electrons does not make the theory more vulnerable to refutation or more empirically contentful. As van Fraassen said, “since the extra opinion is not additionally vulnerable, the risk is – in human terms – illusory, and therefore so [are the gains].” This is not a problem that Fine tackles. He could respond that what is gained by the belief is simplicity of our story about science, but this is not justification for a belief to van Fraassen. The van Fraassen ian point is that what is gained is the claim that there are things that do not and cannot figure in our experience of the world. van Fraassen’s response and Fine’s rejection of the claim that only experience can justify belief bring this exchange to a standstill. A more promising route is for Fine to point to the first problem, the uniformity of inferential practice, as an example of what is gained by believing, maintaining the uniformity, although this is not justification for a constructive empiricist. This leads us back to Fine’s first problem for van Fraassen.

The first problem, the uniformity of inferential practice, seems to be the bigger stumbling block for van Fraassen. The epistemological division does not cut at the joints of the inferential practice of scientists. The challenge is to justify the epistemology. Since settling the issue of whether epistemology of science is needed is beyond the scope of this paper, I will focus on the question of what the relation between the inferential practice and justification by evidence is.

van Fraassen only talks about evidence for a theory as a whole. This evidence will be based on the observable, although it will lend support to the statements about unobservables through supporting the theory. Statements about unobservables serve to simplify connections among observables. The inferential connections are among propositions or statements. Whether these propositions are believed or accepted seems to be a separate issue from what inferential connections they have. The inferential connections are determined, at least for the most part, by the theory which receives evidential support from the statements about the observables. The distinction among the types of commitment to propositions that Fine focuses on does not depend on these connections. The types of commitment and the inferential connections are orthogonal. If they are orthogonal, then it is not surprising that they do not naturally line up. The question now is why this should count against van Fraassen’s view.

Fine is oddly silent about what the uniformity among inferences is, even though it is, on this reading, central to his criticism of constructive empiricism. The inferences are not epistemologically uniform, since the evidence for them comes in different forms, depending on their subject matter, although all such evidence will be from observables. Inferences about electrons will be based on different things than inferences about elephants, e.g. the former will may be based on voltmeter readings while the latter may be based on tissue samples. They aren’t uniform in their subject matter. Fine’s point may be that they are uniform in the scientists’ commitment to them, in the form of belief in the premises and conclusions. van Fraassen could agree that they are committed but take issue with the further commitment in the form of belief. van Fraassen does not see it as a problem that most scientists take themselves to believe in all the entities they study, because the question is a philosophical one. No matter the actual views of scientists, all that is needed is the commitment to test the theory’s empirical adequacy, and this requires no more than acceptance of unobservables and belief in the observable. van Fraassen’s view can then maintain a uniformity among the inferences, commitment to using them to test the empirical adequacy of a theory, even though this is likely not the uniformity that Fine points to.

The heart of Fine’s criticism is that the epistemological division is artificially imposed on the inferential practices, which are not themselves problematic. A view that adds something artificial to an otherwise unproblematic phenomenon is not thereby refuted by the artificiality. van Fraassen’s reasons for drawing the line where it is relies on a criterion, observability, that while artificial with the respect to the inferential practices is not itself ad hoc. In fact, van Fraassen is able to maintain an important, natural uniformity among the inferences, as outlined above. However, given that van Fraassen thinks that being able to make sense of scientific practice without the inflationary metaphysics of realism is reason in itself to choose constructive empiricism, by parity of reasoning being able to make sense of scientific practice without van Fraassen’s extra epistemology is reason to choose Fine’s view over constructive empiricism. It looks like Fine’s criticism should count against constructive empiricism by van Fraassen’s own lights.

In conclusion, Fine’s objections push one’s intuitions against constructive empiricism but they do not decisively refute it. They cast doubt on the necessity of the epistemological distinctions for our understanding of science, undermining the impulse to enrich our epistemology, but they do not provide us reason to think that the epistemology gets things wrong. van Fraassen is able to respond to Fine’s criticisms, but the response to the second objection results in both sides digging in their heels. More completely responding to Fine would require a more detailed defense of the empiricist epistemological principles van Fraassen uses. Without that, the responses give on behalf of van Fraassen get him a draw.