Roles for Values in Science: The History of DES
Heather Douglas

In the philosophy of science, the value-free ideal has been widely held as the proper ideal for scientific reasoning for the past forty years. The ideal holds that when evaluating evidence and deciding which theories to ultimately accept, scientists should only consider cognitive values. Philosophers as diverse as Isaac Levi (1960, 1962), Thomas Kuhn (1962, 1977), Ernan McMullin (1983), and Hugh Lacey (1999) have articulated and/or defended the ideal, which has become a mainstay of philosophy of science. However, concerns have been raised over the viability of the ideal, particularly concerning the porousness of the distinction between cognitive values and all other kinds of values. (Rooney 1992, Longino 1996, Machamer and Douglas 1999) If this distinction fails, that is if non-cognitive values bleed into cognitive ones, the normative force of the ideal is undermined and perhaps lost. In addition, it is not clear that this ideal is the correct ideal for values in scientific reasoning. Early arguments to support it (especially Levi and Kuhn) presupposed that science was an activity best pursued in isolation from the surrounding society. Such a model of scientific isolationism no longer seems plausible (if it ever did), and arguments against it have gone unanswered. (e.g. Gaa 1977)

In order to delve deeper into the place of values in scientific reasoning, a closer examination of the how values play a role in the decision to accept or reject theories is needed. The view to date has been largely a “gap-filling” notion, that is that values fill the gap between evidence and theory. (e.g. Mitchell 2004) However, this is a rather unsatisfying account of how values influence reasoning. Based on insights from Hempel 1965 and Heil 1983, I propose an upgrade for our understanding of the function values can serve. That is, there are two (at least) distinct roles that values can play when one is deciding whether to accept a theory given a body of evidence: 1) the values can provide warrant to a particular choice, i.e. they can serve as a reason in themselves to make a particular choice, or 2) the values can shift what would count as sufficient warrant, where the warrant itself derives from other sources. The question is whether this philosophical account can provide a better ideal for actual scientific practice.

In order to answer this question, careful examination of historical episodes is needed. In this paper, I will take a closer look at a thorny episode in the history of science, the decision to keep prescribing DES to pregnant women after 1953. DES was approved for medical use in 1947 by the FDA, but in 1953, a large case control study was performed to see if DES reduced miscarriages in pregnant women. (Dieckmann et. al. 1953) The study found no effect from the use of the drug. There was already available evidence that DES could cause birth defects in animals. Yet scientists of the day decided to ignore this body of evidence in favor of the prevailing view that DES was beneficial for all pregnant women. (Dutton 1988) Why was the evidence neglected in this case? Why did the scientists of the day continue to support the claim that DES would reduce the likelihood of miscarriages, contrary to the best available evidence, until the early 1970s?

There were complex sets of values at play in this case, which make it rich ground for exploring the roles for values in scientific reasoning. The idea that DES was good for pregnant women was reinforced by prevailing gender essentialism of the time, a gender essentialism that believed hormones were the chemicals that made humans into two completely distinct sexes. The theory of hormone physiology dominant in the day had strong explanatory power, scope, and an elegant simplicity, thus exhibiting the traditional cognitive values. (Oudshoorn 1990, Borell 1985) And there were ethical concerns over the health of women and their children.

In this paper, I will examine in close detail the arguments scientists made for various interpretations of the available evidence. At issue will be how the different ideals for values in science illuminate, or fail to illuminate, the episode. The continued prescription of DES after 1953 seems to be a failing of some kind, and pinning down exactly what kind of failing will be the purpose of this paper.