Content Minimalism
Which parts are to be deemed “truthful” or even “reliable” in an empirically successful theory? SVV focus on the realist import of impressive predictions: If a prediction comes out right, it is because the theoretical claims invoked in its derivation have some significant truth-content, which they pass downstream to the prediction.
The thesis here is this: A theory that is false but otherwise empirically successful lodges theory-parts that, although of weaker content, are still strong enough to make the derivation of impressive predictions go through. Spotting those parts prospectively (as opposed to just retrospectively) requires purging the derivation steps of content superfluous to making the predictions at hand. On this matter, SVV variously draw inspiration from the structural realist line revived by John Worrall (1989). In examining a theory T, they look for content that plays a substantial role in the derivation of impressive prediction. They begin by taking a representative set of remarkable predictions from T and then for each prediction pick the most cogent derivation available. The inferential steps are then subjected to critical scrutiny from the bottom (prediction-statement) up, looking for assertions involving “causally active” posits that seemingly cannot be further abstracted away without blocking the prediction. Because of the emphasis on purging content, the theory-parts that gain selection will often be thinned down versions of richer and more detailed theory-parts available in the full theory, as Fresnel’s Core exemplifies.One key question is how aggressive a purge one should attempt here. To Votsis (2011, p. 1230), the only parts dependably involved in the logical deduction of impressive predictions are “mathematical parts,” chiefly equations and mathematically structured concepts. However, since these cannot generate predictions without interpretation, they must be given one, which opens the door to superfluous content that—SVV agree—realists must strive to reduce to just the minimum needed to generate the impressive predictions of the theory at hand.
Thinning down derivations can be problematic, however. Minimalism pushes realism towards antirealism. Where is one to stop? A major shortcoming of bare bones trimming is a lack of clarity regarding what content is “absolutely necessary.” Wary of the perils of minimalism, Vickers (2013) looks for a “natural” differentiation between the posits invoked in the derivation of predictions. A prediction typically resorts to premises that include claims drawn directly from the theory at hand (“internal”) as well as ideas that, although “external” to the theory-proper, have guided the beliefs and judgments of the scientists who developed the theory. Vickers assesses these two varieties regarding their realist import. Once an approach (e.g. Fresnel's) is articulated, he urges, many of its surrounding posits (e.g. the ether) become “optional” supplements no longer indispensable for deriving the prediction in question. Being optional, “external” posits do not directly gain epistemic strength from the successful predictions of the theory, Vickers argues. Nor does realist commitment extend to all the internal claims invoked in the derivation: Commitment, he stresses, goes only to the “working parts” of the internal claims involved.
So, how well does this strategy handle the old ether? The proposed distinction assumes that the conceptual relations between internal and external claims are “optional,” but this can be denied, as the already cited critics of Kitcher, Leplin and Psillos noted. Vickers presents the ether as a posit external to Fresnel's Theory, yet it seems implausible that Fresnel's ‘internal' tenets could be affirmed without claiming also every claim they clearly presupposed. This is especially serious in theories born with conceptual links to higher-level ideas (Cordero 2011). In the ether example, arguing in the 19th century that something (say, X) was a wave amounted to claiming that it was a traveling perturbation, which presupposed that X required a supporting medium (since, metaphysically, theorists then widely regarded waves as modes of being). If so, using the external/internal distinction to bracket the ether required an extra step.
The conceptual network associated with the predicate ‘being a wave' needed first to have the noted conceptual “necessities” broken up, i.e. turned into contingent relations. But attempting that seemed “nonsensical” in the 19th century. It is only in the early 1900s that theorists such as Einstein, thinking against the current, proposed the required fragmentation. Einstein did this in his Special Theory, in a move that proved far from popular in the early 1900 but which subsequently gained wide acceptance in the natural sciences. The point is that Vickers' proposal can work only if one places it within a “modern” naturalist framework in which all conceptual links rest on empirical generalizations (indefinitely open to the possibility of revision).Other hurdles stand in the way of the reformed selectivist approach. Vickers is aware of two of them: Pursuing interpretive minimalism pushes theory-part selection into enemy territory—at each step the “absolutely necessary” might seem no more than the step's constructive empiricist version, i.e. the step freed of commitment to non-empirical content in the original version. Secondly, no matter how one conducts the sought purge, part of what remains may still be idle. One question is thus whether one can make “reliable enough” selections along the proposed lines. Vickers pessimistically concludes “the realist is still some distance from prospectively identifying (even some of) the working posits of contemporary science.” Indeed, in his view, the realist should commit to just some unspecified parts of what remains once the posits acknowledged as idle are removed. Vickers exudes pessimism:
In all this, we find ourselves—even 30+ years after Laudan (1981)—unsure of the extent to which the divide et impera strategy can succeed. Even if the ‘working posits' of contemporary science cannot be prospectively identified, it remains possible that we might develop a recipe for identifying at least certain idle posits. That would be a significant achievement, even if not quite what the realist originally had in mind.
Vickers (2013: 209).The above gloom harks back to Vickers's way of seizing on content separation in terms of his internal/external distinction, which is problematic on yet further grounds. For example, although conceptual fragmentation makes the external/internal separation possible, it is not enough for realism. Abstraction keeps out of sight some of the original background of a claim, but for a claim to have realist import the parts left out must be left out for specific reasons—ones explicitly backed up by the extant confirmational arrow. Otherwise, the selection will be arbitrary.
The strategy just reviewed improves on earlier selectivist strategies for identifying theory-parts suitable for realist commitment, and is both explicitly focused on abstraction and “prospective” (as opposed to merely retrospective) identification. But now there is this additional hurdle: The gloomy conclusion in the above quote clashes with the realist expectation of warranting commitment to explanations that make the theory's observable domain more intelligible in terms of unobservable structures and mechanisms underlying the intended empirical domain. It also leaves underappreciated much of the array of theory-parts that the best current confir- mational practices in the natural sciences sanction as empirically successful and beyond reasonable doubt. Recognition.
To overcome the pessimistic trap, I suggest, selectivists need to do better at both discarding posits seriously off the mark (e.g. the ether) and identifying posits worthy of commitment. Otherwise, the resulting realist stance will be disappointingly bland, something like “In every successful theory, somewhere some abstract versions of some of its claims about unobservables get some things somewhat right about the intended domain.”
More is expected of selective realism—a criterion both aimed at identifying substantive specific theoretical content and found reliable against the history of modern science.
3