Occam's Razor: Valid tool of reason or evil mind control plot?

eris:

[Lurch voice]

You rang, sir?

[/Lurch voice]

Regarding Occam’s Razor as I understand it: as a guideline for theoretical speculation, it’s very useful. But it depends at what level one places it within the theoretical structure. That is to say, working within a given paradigm, it can be very useful to formulate a hypothesis or prediction as simply as possible. But paradigms as a whole cannot be compared to each other on the basis of simplicity, generally speaking.

As a metaphysical principle – i.e., that the truth content of theory is determined by its simplicity – it’s useless.

As a description of the way “science” works, or the manner in which scientists decide between competing theories, it’s a misleading simplification. The example posited by Truth Seeker is a case in point. We are asked to consider two possibilities: that SDMB consists of a collection of human beings, or a collection of aliens. Since assuming that the SDMB consists of a collection aliens involves a more subsidiary theoretical concepts, Truth Seeker judiciously applies Occam’s Razor (OR), and comes to the conclusion that in all probability the posters here are human. But my point is that we do not really need OR to come to that conclusion in this case, and the example doesn’t really have much to do with science.

Let us instead look at actual historical examples, and see how useless OR really is in such a cases. The turn-of-the-century debate within immunology, between Ehrlich and Bordet, is a case in point. I’m no expert in the field, but will try to summarize it as best I can. One of the fundamental questions within immunology between approximately 1890 and, say, 1910, concerned the exact nature of antibodies: could the immunological properties of the blood best be understood as a “function of the sera,” or should these properties be conceptualized as a function of discrete biochemical entities in the bloodstream? Scientists at the time were divided on the issue, and many resisted the assumption that Behring and Kitasato’s* findings indicated the existence of antibodies as a discrete substance. This group included Behring and Kitasato themselves, by the way; as late as 1896, if not later, they were still of the opinion that “antitoxins should be conceptualized as forces rather than chemical substances” (Cambrosio et. al., 1993). They believed, along with many of their colleagues, that assuming the existence of antitoxic agents was a “simplistic” explanation of immune phenomena. Here we have an example, by the way, of perfectly rational scientists resisting the demands supposedly imposed by OR.

Ehrlich, on the other hand, was an early proponent of the antibody theory, as was Bordet. But the debate between Ehrlich and Bordet is even more telling, because in this case, neither of the theories they put forward was necessarily “simpler” than the other. On the contrary: they were both equally “complex.” In fact, although theoretically ingenious, Ehrlich’s specific theories on antibodies were contradicted by a good deal the experimental evidence, as Bordet often pointed out. But Ehrlich won the day, and turned out to be the more influential of the two. Why? Well, certainly one of the most important reasons was due to Ehrlich’s artistic skill. Ehrlich employed pictorial representations of antibodies and their functions as a heuristic device, and this in turn had a powerful impact on his readership and lecture audiences. Bordet objected to this technique because he felt it led to an oversimplified conceptualization of the chemical relations of antibodies. In this case, OR worked in a strange way: Ehrlich’s heuristic representations, though simplifications, were more influential than Bordet’s experimental evidence. Since Bordet felt that such representations were misleading, he refused to employ them, and thus, in the end, he lost the debate (both Ehrlich’s and Bordet’s views on the way in which antibodies function have been abandoned since then).

Rather than thinking of OR as a rule, I would like to suggest that we consider it to be a scientific value. As such, it would take it’s place among many other such values, such as accuracy, scope, consistency, and fruitfulness (to name a few at random). Kuhn suggests this line of reasoning in his essay Objectivity, Value Judgement, and Theory Choice. All things being equal, the simplest of two theories is to be preferred; but the number of cases in which “all things are equal” is vanishingly small. Generally, when choosing between theories, scientists also must make choices among opposing values: the simplest theory may not be the most fruitful, or the most accurate, or the most consistent. Faced with making such choices, a scientist who values accuracy over simplicity will naturally select the theory that is most accurate, even if it is the most complicated theory as well. The history of science if rife with such conflicts: I could probably come up with another half dozen examples off the top of my head. Check out the discussion of various methods for differentiating between species at talkorigins for a good example.

By the way, underdetermination is a significant factor here as well. In the debate over how antibodies should be conceptualized, and how they work, scientists were forced to chose between a number of competing theories, all of which were equally well supported, and equally contradicted, by the experimental evidence. What is of historical interest here is the fact that scientists often achieve consensus on a theory well before the evidence supporting it becomes unequivocal. Kuhn suggests that this occurs not because of further evidential support, but rather due to the fact that scientists gradually come to share the values that underlie a specific paradigm. That is to say, the evidence doesn’t become less equivocal: instead, the majority of scientists in a given field come to agree that in this particular field, simplicity is more important than accuracy (or vice versa). The evidence supporting the theory then comes to be viewed as unequivocal, but, Kuhn argues, that is because scientists have, in effect, redefined what it is that they are studying, what constitutes valid questions within that field, and what constitutes evidence that supports their view.

  • Behring and Kitasato are credited with discovering “antitoxins” in 1890.