1. TWO SCIENCES
The methods of physics and biology are different, and they produce scientific facts in different ways. This means that the question Do powerline EMFs affect human health? must be considered from two different perspectives.
Following the end of World War II, Herman Schwan, a German physicist, became a professor at the University of Pennsylvania, Department of Biomedical Engineering, and remained there until his retirement. Schwan's area of expertise was the biological effects of electromagnetic fields, and he played an important role in a 1960s government program aimed at determining a safe level of exposure to microwave radiation for servicemen. Schwan's approach was based on a series of calculations and assumptions, and in the legal dispute he applied them to powerlines and concluded that powerline EMFs would not affect human health.
Schwan was cross-examined for 2 days in April, 1976 regarding his opinion about powerlines, and he fared poorly. As I watched, I tried to put my finger on exactly why he was unable to sustain his opinion. On the surface it appeared that Schwan's mistake was to equate the absence of a known mechanism of interaction between EMFs and tissue with the idea of the absence of a health risk. But I knew that the problem must go far deeper. Somehow, it was related to his attitude toward science, which was so different than mine. I saw the possibility that EMFs could cause biological effects as exciting, a previously unanticipated and unexplored idea that might have profound implications. I therefore viewed the handful of reports that existed in 1976 which supported this idea as tiny flowers growing in the garden of science. Schwan, however, saw the reports as weeds.
In the succeeding years individual physicists and groups of physicists offered opinions regarding whether powerline EMFs affect human health. But their arguments were no different from those of Schwan. It dawned on me that Schwan and those who think like him were not offering poorly thought-out opinions. Rather, within the frame of reference of what science was to them, these physicists considered themselves to be correct and it was hard to imagine anything that could make them change their minds. Schwan, for example, reacted to his cross-examination not by conceding that he could not sustain his position, but rather by becoming angry at the cross-examiner. At one point he glared at the attorney and said that he was a "very poor physicist." Schwan really believed he was right and that he could convince a room full of good physicists that he was right because they would understand how he thinks.
Many professional physicists, including even Nobel Prize winners, believe that their approach to the study of the natural world is pertinent to and can be used to address the issue whether powerline EMFs affect human health. Somehow, I thought as I watched Schwan in April of 1976, this is not the case. He was being a good physicist on the witness stand. If all the physicists in the country were asked to vote, I think they would have backed him and simply equated being a good physicist with being a good scientist. Perhaps the problem was not Schwan's way of thinking, but the relevance of his way of thinking to the issue of powerline EMF health risks.
I begin mulling over how scientists think, and how they decide what is or is not a scientific fact. It's easy to see that specific questions like Do powerline EMFs affect human health? are meaningless unless one specifies how the scientific facts to be used in answering the question will be obtained. Why? Because if Dr. A requires that scientific facts be obtained in a particular way, and Dr. B requires that they be obtained in some other way, then Drs. A and B can never agree. The other guy's data is simply junk science.
If I am correct that in an important sense that physicist's opinions about whether powerline EMFs and human health don't matter because the way physicists think is inapplicable to the issue, then I should be able to prove this contention by an analysis not connected directly with the EMF issue. That is exactly my goal in this section and in the next section. First, I will show here that there are in use in science today two different reasoning processes for deciding what constitutes scientific knowledge - those of physics and biology. In the next section I will show why the physical approach has little to offer towards resolution of the powerline hazards question.
There have been many studies of the philosophy of science. Generally, the aim in these studies was to identify what the authors considered to be the basic features of scientific practice, and this was done by selectively choosing special cases for analysis. By choosing special cases, differing conceptions of scientific practice could be described. The purpose here, in contrast, is to establish how science is done today, without limitation to specially chosen cases, and in the absence of idiosyncratic ideas regarding how it ought to be done. Consequently, I employed representative sampling to facilitate identification of the rules and procedures of scientific reasoning that are used to establish a putative fact as scientific knowledge.
To characterize contemporary scientific thinking employed in experiments routinely performed in universities, government laboratories, and corporate facilities, and published in peer-reviewed journals, I randomly chose Issue No. 5248 of the journal Science (January 26, 1996). The Issue contained 12 reports that could be analyzed to ascertain the thinking that was employed by the investigators in arriving at a judgment that new knowledge had been found. The reports are summarized in Table 1. Four additional reports were not considered because they involved measurement or other activities (invention and discovery) that did not utilize formal reasoning.
A common feature of the reports summarized in Table 1 was the use of a model to facilitate reasoning. The model was either a physical system that was manipulated in the laboratory, or a conceptual simplification of a real system such as a particular arrangement of a small number of atoms. Use of a model was fundamental and absolutely essential in all cases of scientific reasoning.
Two kinds of studies could be distinguished. In one kind, the goal was to provide an explanation of a phenomenon in terms of mathematical equations (covering laws), which were regarded by the authors as governing the phenomenon of interest, and which were afforded a prominent role in accounting for specific changes in the model system. A force, explicitly or implicitly contained in the covering laws, was regarded as the necessary and sufficient cause of change in the model and, ultimately, of the phenomenon to be explained. No other factor or condition was needed to explain the changes. Thus, in the cover-law studies, a deductive form of reasoning was employed to rationalize particular observations, namely those for which the model used was deemed appropriate.
In the other kind of study, the goal was to prove that a particular factor was a but-for cause of a particular observation. In Table 1 Report No. 8, for example, the authors employed KD cells and demonstrated particular cause-effect relationships involving decreased cyclin-E/CDK2 activity and loss of anchorage. Similarly, in Report No. 11, A31.C1 cells were used to demonstrate that osteopontin activated CD44. In the cause-effect studies, no attempt was made to explain the results in the sense of showing that the relationship between the postulated cause and the observed effect was a necessary consequence of a general mathematical principle.
The authors of the cause-effect studies extended their results beyond the particular biological objects that they manipulated in their own laboratories by means of abduction, which is an inferential reasoning process distinct from induction and deduction. In these studies, it was either argued or assumed that the relationships observed were not specific to the respective laboratories, but rather would be found by others in appropriate replications of the studies. The term most frequently employed to describe the link between the study actually conducted and the larger conclusion advanced by the investigators was suggests, but many other euphemisms were used (Table 2). For example, if it were true that decreased cyclin-E/CDK2 activity generally led to loss of anchorage, then the results observed in the KD cells (the study actually conducted) could be viewed as a deductive consequence of that general principle. On the other hand, on the basis of the data, it would not be true to say (and the authors did not do so) that the results proved that loss of anchorage observed in KD cells was due to decreased cyclin-E/CDK2 activity, because the authors did not exclude all other possible explanations. The study only suggested that this is the case. Thus, no logical inconsistency would be entailed were it the case that investigators in a different laboratory failed to find the reported cause-effect relationship.
Moreover, it could be the case that the reported link between decreased enzyme activity and loss of anchorage occurs only for KD cells and not for other types of cells. It seems clear from the report that the authors viewed KD cells merely as a convenient model within which to study a model-independent phenomenon. I expect that the editors of Science regarded the observed cause-effect relationship as likely to be model-independent because KD cells have no particular significance in themselves, but served merely as a convenient tool for demonstrating a basic biological phenomenon. But nothing in the study precludes future investigators in other laboratories using non-KD cells from observing that decreased cyclin-E/CDK2 activity does not lead to loss of anchorage of the cells.
These considerations make it clear that whatever generality may appropriately be inferred using the KD model, the basis of the validity of the generalization is the following abductive argument: were it the case that it was generally true in nature that decreased cyclin-E/CDK2 activity causes loss of anchorage in cells, then the data and relationships observed in the present study could be explained deductively.
Each of the other cause-effect studies in Table 1 similarly relied on abductive reasoning as a means of generalizing the results beyond their individual laboratories.
The authors of the covering-law studies, in contrast, proved their point. For example, consider the report dealing with rupturing of adhesive bonds formed by short-chain molecules. A model was adopted that involved 2 walls containing 800 atoms each, coupled by stiff springs on a face-centered-cubic lattice; the space between the walls was occupied by 128 polymer chains that each contained 16 molecules of a given mass. Equations based on physical theory (electromagnetism and energy conservation), assumed forces (introduced in the guise of potentials), and numerical values of particular parameters in the equations were regarded as jointly controlling the process of rupturing of bonds between the polymers. In simulation, the walls were maintained at different temperatures and then separated from one another at different velocities, and it was shown that energy dissipation occurred by means of viscous forces at high temperature, but by particular structural rearrangements of the polymer chains at lower temperatures. The results obtained were absolutely certain, and would be obtained by any knowledgeable investigator who employed the same model and made the same assumptions. The molecular sequence of events in the model could be explained in the sense that it could be deduced from a covering law as the result of a particular cause (the force) via particular temperature-dependent mechanisms. Further, the results obtained necessarily apply to an important class of real systems, namely those systems for which the model was a true and accurate representation. The point is that, given the model and the assumptions, no conclusion other than that stated by the authors was possible.
On the basis of the evidence provided by the representative sample of Science reports described here, it can be seen that there are two fundamentally different approaches to doing science in the 1990s - two distinct scientific thought-styles. In the physical thought-style, the goal is to explain an observation by showing that it is compelled by basic physical laws or at least by phenomenological equations. In this thought-style, a scientific fact is a deduction from a relevant covering law made in the context of particular assumptions. The concept of causality does not occupy a central position in the physical thought-style because the necessary and sufficient cause of the observation to be explained - a force - is known in advance of the explanation.
In contrast, in the biological thought-style, the goal is to establish a scientific fact. In this thought-style, a scientific fact is a but-for cause of an observation established using orthodox measurement methods and appropriate statistical techniques. In the biological thought-style, covering laws are not employed and linkage with covering laws, even in principle, is not required as a precondition for accepting observations as valid. Scientific facts are generalizations that admit of exceptions.
The analysis of the reports in Issue 5249 of Science leading to the conclusion that two distinct thought-styles were utilized to produce scientific facts applies equally well to all subsequently published issues of Science that I have considered. That is, I can show that each report in any issue of Science that involves formal reasoning can be classified into one (or a combination) of the thought-styles described here. It can permissibly be concluded, therefore that there presently exist two distinct valid methods for producing scientific knowledge. Consequently, the scientific facts of the physicist and the biologist are fundamentally different objects. This analysis makes clear - I think for the first time - that there presently are two distinct pathways by which observations can rise to the level of scientific fact.
I will show how failure to distinguish between the thought-styles and to identify the applicable thought-style accounts, in part, for the present controversy regarding whether powerline EMFs affect human health.
Marino Home Page | Research Interests
| 46792 |