The Standard Arguments Against Empirical Equivalence

Laudan and Leplin introduce the themes of empirical equivalence and underdeter­mination in the following way Laudan and Leplin (1991, p. 449): “By the 1920s, it was widely supposed that a perfectly general proof was available for the thesis that there are always empirically equivalent rivals to any successful theory.

The idea that theories can be empirically equivalent, that in fact there are indefinitely many equivalent alternatives to any theory, has wreaked havoc throughout twentieth century phi­losophy. It motivates many forms of relativism, both ontological and epistemological, by supplying apparently irremediable pluralisms of belief and practice. It animates epistemic skepticism by apparently underwriting the thesis of underdetermination. In general, the sup­posed ability to supply an empirically equivalent rival to any theory, however well supported or tested, has been assumed sufficient to undermine our confidence in that theory and to reduce our preference for it to a status epistemically weaker than warranted assent.

Laudan and Leplin identify two pillars on which the doctrine of empirical equiv­alence and theoretical underdetermination rests. First, the doctrine relies on the sup­posed fact that for any theory T and any body of observational evidence E, there is another theory T' such that T and T' are both in agreement with E. The support for this, they say, is twofold. Some authors claim that there exist general algorithms that are able to generate an empirically equivalent theory T' given any theory T. In addi­tion, there is the Duhem-Quine thesis that a hypothesis can only entail observable consequences with the help of auxiliary assumptions.

E, there exists another hypothesis H' that also entails E by using an alternative class of assumptions A'.

Second, the doctrine of empirical equivalence and underdetermination of theory choice is said to depend crucially on the premise that only observational statements that are directly logically linked to a theory can legitimately count as empirical evi­dence for it: an observational report provides evidence for a hypothesis only if its content can be derived, explained or predicted from that hypothesis.

From the first premise it follows that there exist empirically equivalent rivals to any theory; and from the second premise that all these rivals are equally supported. It then appears that the objectivity and even rationality of scientific theory choice come under severe threat (see for a more extensive discussion, also of the other material in this section, Acuna and Dieks (2014)).

A straightforward way out of this problem would be to have recourse to non- empirical features of empirically equivalent theories. For example, if one of the the­ories is easier to understand, leads to a simpler world picture, or is easier to use than its rivals, one certainly has a reason to prefer this theory. However, theoretical virtues of this kind are controversial—they will often have a pragmatic flavor and depend on subjective preferences. Although the consideration of such non-empirical virtues can certainly provide reasons to prefer one of the theories, and in this sense dissolves the problem of the rationality of theory choice, the objectivity of theory choice still seems endangered—something to which we shall return.

Laudan and Leplin therefore choose a more robust way of countering the threat of relativism, by directly attacking the two premises of the above argument.

A natural response to this manoeuvre is the observation that any new theory may again have countless empirically equivalent rivals so that the underdetermination problem has not been solved but only shifted. Kukla (1993, 1996) in fact argues that given any input theory, one can always formulate an alternative and empirically equivalent theory by applying simple transformations to the form of the theory. But this criticism is rejected by Laudan and Leplin, on the grounds that such artificially formed rival theories T* are parasitic on T and fail to provide genuine alternatives; they do not satisfy “theoreticity criteria” that would make them serious contenders.

From Laudan’s and Leplins rebuttal of Kukla’s criticism Leplin and Laudan (1993) three such criteria can be extracted: non-superfluity, plausibility and testability.

A hypothesis is superfluous if it can be dispensed with without any loss of empir­ical content, i.e., if it does not contribute to the derivation of observational conse­quences.

Plausibility comes in because we should not consider theoretical schemes as rivals if they contain far-fetched and artificial characteristics Leplin and Laudan (1993, p.14):

Provisions that fly in the face of what we have good empirical reason to assume must claim some offsetting rationale if they are to be admitted as part of a theory. It would be different if the course of nature were known to exhibit such vast and mysterious ruptures or bifurcations as T' envisions, if natural law did not exhibit isometry, at least. One might then be willing to entertain wild, unexplained and unconfirmable scenarios as genuine possibilities. But the world is not known to be like that.

So in order to be considered as genuinely scientific, a hypothesis must possess a minimum degree of plausibility—which can be judged on the basis of a background of already well-confirmed empirical knowledge.

Finally, if rival theories contain superfluous additional hypotheses, in the sense that these hypotheses do not play a role in the entailment of observational conse­quences, these hypotheses will be untestable Leplin and Laudan (1993, p. 13):

Because the purpose of theorizing is, at least in part, to gain predictive control over the subject matter under investigation, a theory must, at least in principle, be open to test. A ‘propositional structure’ that is not even in principle confirmable, that could not logically be an object of epistemic evaluation, is not a theory; for it could not in principle impart understanding nor advance practical interests.

The second part of Laudan and Leplins argument is directed against the premise that only the confirmation of observational statements that are derivable from a the­ory can count as empirical support. As they argue Laudan and Leplin (1991, p. 460), “significant evidential support may be provided a theory by results that are not empir­ical consequences of the theory”. Accordingly, “the relative degree of evidential sup­port for theories is not fixed by their empirical equivalence”.

When we take stock at this point, we see that Laudan and Leplin have pointed out that empirical equivalence, if it occurs in practice, does not need to be of a strong kind—it need not be permanent. Moreover, they have argued that not any arbitrary formal concoction should be regarded as a genuine rival to a scientific theory; there are certain requirements of scientific respectability that have to be satisfied. The three requirements that they list all relate to the way a theory connects to empirical evi­dence, and in this sense are meant to be objective. Finally, Laudan and Leplin have called attention to the fact that even if there were serious cases of persistent empirical equivalence, this would not necessarily mean that no evidence-based choice between them could be made.

However, the critical remark can be made that Laudan and Leplin have not actu­ally disproved the possibility of respectable rival theories that are empirically equiv­alent, nor demonstrated that the progress of science is bound to break any such equiv­alence.

3