I’ll confess to enlisting Occam’s razor to win an argument as often as the next person. But lately I’ve been seeing in it a fallacy, at least as it has increasingly been misused.
Occam’s razor, of course, is the principle that the simplest explanation is most likely to be the correct one. It is named after William of Occam (also spelled Ockham) back in the 14th Century, and it is considered, along with the work of Francis Bacon, to be instrumental in the development of modern science. Occam himself expressed the principle by employing the word “multiply” in a fashion that is not still used today. But the most common updating of how Occam technically stated his principle is that it advises us to “avoid postulating entities to account for what can be explained without them.” Or another paraphrase is, “Make no unnecessary assumptions.”
But notice that these more technical formulations do not use the word “simplest.” And indeed, the technical formulations have a slightly different point to them than the claim that the simplest explanation thereby has the ontological status of being Truth or even of being closer to the Truth.
The way that the word “simplest” has become incorporated into Occam’s original meaning is by stating the principle as a competition between two choices. It is usually stated something like: “If there are two explanations for the same observed event, give preference to the simpler.”
If the subject matter is said to be science, then the word “hypothesis” is often substituted for the word “explanation” to become the form of Occam’s razor that is so often heard today: “If there are two competing hypotheses, then the simpler one is more likely to be correct.”
The word “simpler” can be taken to mean the one with the fewest assumptions, postulates, or just plain verbiage. Then the notion of a “razor,” accordingly, means to do without the unnecessary additions.
Two of Occam’s students are credited with using this methodology to anticipate Copernicus and Newton. Nicholas of Oresme found it simpler to hold that the earth is in motion than that all the stars are in motion in the night sky. And John Buridan argued against Aristotle’s notion that the planets must be continuously moved by “intelligences” since it was simpler to hold that a planet, once launched, merely stayed in motion rather than needing to be always pushed. Buridan also surmised that the planets must be made of the same substance as the earth since there was no reason to assume that two kinds of matter were necessary instead of one.
Regarding the study of motion, Buridan argued that, when one body collided with another, the motion imparted from one to the other depended on the velocity of the first body and the quantity of matter in the second. He deduced as much by finding it simpler than alternative explanations of his day having to do with heat or with placement.
But did using Occam’s razor really settle the issues? Or did making a more lasting determination entail something else in the way of scientific practice?
I deliberately picked examples where Occam’s razor turned out to be right. But when it turns out to be wrong, then what is it that decides if it is really right or wrong?
My point is that, all by itself, being simple does not confer upon a premise the ontological status of being True. There is more to it than that. It does not even automatically give a notion the status of being scientific (although there are those today who argue that it does).
I think that most people will agree that the razor was never really intended in the first place to be a claim for establishing absolute Truth (although Occam himself, a monk, used it in support of miracles). The razor most typically has been used only for finding what is said to be “closer to the truth.” But what about using it for determining what is “more scientific”?
My argument will be that Occam’s razor can be used to bolster even pseudoscience, so it does not turn out to be a reliable indicator of being either true or scientific, contrary to some assertions made by others that I will discuss here.
I will start by giving an example of when Occam’s razor does not work.
first example of a problem
The problem is that sometimes a person will make a proposition, and to answer it we want to be able to say, “But that does not fit with the rest of what we know.” Yet to bring in the rest of what we know is to expand the discussion—it makes it be about other things—which is the opposite of keeping it simple and using the fewest number of parameters.
That situation arises frequently, for instance, regarding medicine where a proposal has to fit with the rest of what we know about the body, not just seem to make sense as an idea in its own right. Say that a person wants to tell us what food is good for us to eat, and we want to answer that the suggestion does not seem to go well with what else we know about physiology and how the body works. Should that determination be left to Occam’s razor? The proponent of the proposal could argue that it is simpler not to bring in the rest of physiology and so wins the disagreement.
Another way of putting it is that often, for many scenarios, context counts, and Occam’s razor seems to be telling us that we can successfully ignore what else is contributing to an outcome. It seems to assume that how a thing acts all by itself, as if it is floating in isolation in Outer Space, is all we need to know in making an explanation and that how it fits with others is not important. But often, how things fit together does contribute to an outcome, as the example of the body illustrates, which thereby casts doubt on Occam’s razor.
So I can give a still more specific instance of this problem. Gary Taubes, a science writer, has written a book (2017) in which he invokes Occam’s razor by name to try to prove the thesis that too much sugar in the diet causes a multitude of diseases, including cancer, heart disease, stroke, bowel syndromes, Alzheimer’s, and practically anything else you can think of. In a review of the book, Jerome Groopman, an internist who teaches at Harvard medical school, wants to make the argument that Taube’s thesis does not fit the rest of what we know about how the body works. But Taubes, who anticipates that this argument will be used against him, cites Occam’s razor as a reason to dismiss that complaint. Going even further, Taubes criticizes how modern medicine sees diseases as “multifactorial” and as “multidimensional,” citing that as a violation of Occam’s admonition that science should keep it simple.
Thus Groopman, in wanting to show how Taubes’ attack on sugar is wrong, has no choice but to first take on Occam’s razor, which he does. Concludes Groopman (the physician) in this review,
“Occam’s razor is hardly a fundamental law of the universe, however. No credible scientist would ever think of using it to prove or disprove anything.”
And indeed, Occam’s razor is usually found in philosophy books, not in science books.
Taubes actually employs a version of Occam’s razor invented by Thomas Aquinas a century before Occam, which states the doctrine in terms of causes instead of assumptions or postulates. When put this way, the razor reads, “The proposition that contains the fewest number of causes is most likely to be the truer one.”
And that brings into view Taube’s fuller argument. He is saying that all of these diseases have one cause, our diet, and more specifically, an overload of sugar causes all of these many effects. He argues that that is simpler than purporting that each disease has its own special cause. So by Occam’s razor, he must be correct compared to a multidimensional approach. And to disprove him, we have to disprove Occam’s razor (or so Taubes implies).
Taubes approach is to make a broad general argument—and a simple one—from which he draws conclusions about each disease, and then he looks for evidence that he is right in each case.
But Groopman wants to start with the details in every case and look at how they fit together and fit with what else we know. He does that in the review regarding cancer.
Regarding cancer, Taubes cites studies which show that cancer cells require additional glucose compared to normal cells. (Glucose, of course, is a kind of sugar and one-half of what makes up table sugar). Groopman answers that cancer cells, because they are growing much more rapidly than normal cells, require more of everything, including, for instance, vitamins. So that does not prove that sugar causes cancer any more than that vitamins cause cancer. In other words, using the same logic of Occam’s razor, Taubes must also conclude that vitamins cause cancer. But Groopman, by arguing from what else he knows about cancer cells, concludes that they do not. Also, other studies show that drugs that treat diabetes (and so lower blood glucose levels) do not reduce the incidence of cancer.
Taubes’ argument does not have the ontological status of being either true or scientific just by virtue of being the simplest one. Unless we want to believe that vitamins cause cancer, Taubes’ approach fails compared to Groopman’s multifactorial analysis.
My purpose here is not to resolve the issue of whether dietary sugar is good or bad for you but to show how a resolution will not be up to using Occam’s razor. To the contrary, a resolution will entail understanding how many aspects of the body are working together. Even if someone wants to argue that sugar is harmful, a scientific resolution with require a multifaceted analysis such as Groopman’s. (Anyone interested in pursuing the subject of dietary sugar can find more in Groopman’s article. Groopman discusses other schools of thought and other evidence on dietary sugar, besides Taubes’).
What I do want to discuss further is more about Occam’s razor. By considering why it fails (and when it fails), it becomes possible to make other observations about the structure of knowledge. I will address that in part two of this post.
But first, lest I appear here to be discussing mostly dietary sugar, I will offer another example of the failure of Occam’s razor. Since my hope in making these posts is eventually to point the way toward describing a philosophy of chemistry, I will make the second example be from chemistry.
second example of the problem
When molecules form new molecules in a chemical reaction by falling apart and then combining in new ways, they frequently first create intermediary structures which then twist or otherwise change until they fall apart in the ways that make the new products. Occam’s razor would seem to suggest that, in positing what these intermediary forms should consist of, the simplest should be what entails.
But often that is not the case. Indeed, frequently there is not even a single answer, but rather the intermediary will depend on the circumstances such as the temperature and air pressure. In itself, that is not the simplest way that things could be. (The simplest would be that, when the same two things react, they will always react in the same way every time). Studying the various kinds of intermediaries is an important aspect of chemistry since knowing them can help predict what the final products will be. But the intermediaries cannot be automatically assumed to be the simplest forms that are possible. Sometimes they are simple, but other times they are complex.
They do not follow a rule that the simplest will always be what happens.
And it is definitely not “more scientific” to assume that the simplest will always prevail. Instead, science studies the multifactorial circumstances which bring about the different answers.
But then why does Occam’s razor not always work? (I am not saying that it does not sometimes seem to work).
lessons from Occam
So far in this post we have seen two different answers for why Occam’s razor does not always work; we have seen two kinds of failure. One is that context often counts, and looking just at the simple can blind us to the contribution to an outcome of how things are fitting together. An example is the body. There are occasions when we want to answer a proposal by saying, “That does not fit with what else we know about how the body fits together.” Thus we require a multifaceted rather than a simplistic answer.
And the second kind of failure of Occam’s razor concerns when there might be intermediates to a process. If we look only at the starting and ending points of the process, to find a connection only between them, we might see what appears to be only a simple change. But in reality it might be full of complex intermediates that defy a simplistic explanation.
I will examine a third type of failure and how all of these failures can provide us with hints about the structure of knowledge in part two of this essay.
By looking closer at Occam’s razor and at how it fails, it is as if we are doing experiments to put many of our theories of knowledge to a test.