In her 2000 paper “Inductive Risks and Values in Science,” philosopher of science Heather Douglas argues that science must take non-epistemic values into consideration because its outputs concern non-epistemic values. Epistemic values are those that exist for the attainment of knowledge, such as the value of precision in aeronautic measurements that allows the invention of complex aircrafts, or the impartiality that is expected in the selection of test and control subjects for a psychological study.
In the 20th century, Carl Hempel described the goal of science as the attainment of an increasingly reliable and systematized body of knowledge. Epistemic values are those concerned with this goal. Conversely, non-epistemic values constitute a myriad of humanmade values that exist for anything else. Moral values is an example. They do not exist to create a systematized body of knowledge that lies elsewhere from us, but are aimed at doing what is “right” to ourselves and others.
For example, the value of justice serves to prevent conflicts and the mistreatment of one by another, not to gain a better idea of a system that exists outside of us, as physics may do. A system of justice, such as a country’s judicial system, may systematize the value of justice into a set of rules, but these rules too are up to change by humans through the legislature; we have the agency to decide what’s just, although not always at the individual level.
Conversely, science is believed to be free of these human values. Factors that may bring any kind of human bias are suppressed by techniques such as a double-blind study and using mathematics to demonstrate a result’s significance. Of course, this phenomenon occurs to prevent the errors of past sciences whose accuracy was undermined by the non-epistemic factors. Galileo Galilei’s theory was criticized by the Catholic church, who found him a “vehement[ly] suspect of heresy” due to his refusal to teach the Copernican model.
However, our ability to refine our scientific methodology is limited, and non-epistemic values inevitably affect what we consider “objective” science to this date. As I discussed in the article on March 12th, titled “From theory to observation,” science proceeds by inductive reasoning from local observations to universal laws. Douglas argues that this “inductive risk” is inevitable, since non-epistemic considerations cannot be completely removed.
Douglas introduces an experiment that tested the toxicity of dioxin — an industrial pollutant — to exemplify her argument. The first long-term study of the carcinogenic effects of dioxin was performed by Richard Kociba in 1978. Kociba and his team gave regular doses of dioxin to rodents throughout their natural lifespan of two years. An autopsy was performed afterwards to study the effect of dioxin in each of the rodent’s organs. The tissue and organ slides from the autopsy were preserved and analyzed twice again by researchers other than Kociba.
The United States Environmental Protection Agency gave their interpretation in 1980, and a consulting firm — PATHCO, Inc. — was commissioned by a paper company to analyze the slides again in 1990. These three groups disagreed on how to classify various abnormalities in Kociba’s slides, whether they showed signs of benign or malignant tumor, or neither. As a result, the three groups gave varying interpretations of the toxicity of dioxin on rodents, especially on the dose of ten ng/kg/day, which was the second highest dose considered in the experiment.
While the EPA found 27 out of the 50 slides of this dose to be tumorous, PATHCO found only nine of the slides to show actual signs of tumor, and zero that showed a malignant tumor. Douglas argued that these inconsistencies result from non-epistemic considerations. The organizational affiliation of researchers may act as an explicit non-epistemic consideration, because of the EPA’s role as a regulatory body, while the consulting firm was hired by a paper industry. Furthermore, the disagreements may point toward the subtlety of actual differences between cancerous and non-cancerous slides, which Douglas believes to involve a higher inductive risk.
Non-epistemic considerations in scientific research are not only inevitable, but also necessary to account for its non-epistemic outcomes. If the dioxin interpretation from 1990 is to be believed, it can be used as a justification for industries to pollute the ecosystem with more carcinogenics. It would be naive to assume that the commissioned researchers did not have this goal in mind, but they did not approach this non-epistemic value the same way they did epistemic considerations, with caution and objectivity. It is necessary that the non-epistemic consequences of scientific research are considered not only by the researching group themselves, but by those in the scientific community at large.
~ Benjamin Ha `27
Be First to Comment