Occam was wrong

Parsimony has become very popular with scientists and philosophers. The idea is that the simplest explanation for any phenomenon – the one that includes the fewest possible number of causal factors – is the best explanation. This principle was used by William of Occam in logical arguments and has been popularly called Occam’s Razor. It is often applied to situations where one must choose between alternate theories, one of which includes a fewer number of causative factors or a simpler mechanism than the other.

“Entities must not be multiplied beyond necessity.”

  • William of Occam

“We are to admit no more causes of natural things than such as are both true and sufficient to explain their appearances.”

  • Newton

“…the simplest hypothesis proposed as an explanation of phenomena is more likely to be the true one than is any other available hypothesis, that its predictions are more likely to be true than those of any other available hypothesis, and that it is an ultimate a priori epistemic principle that simplicity is evidence for truth.”

  • Swinburne

The idea certainly seems reasonable, but is it correct? It might be for most physical theories, but it’s quite wrong when applied to biology. There are in fact no simple phenomena in biology, and simple hypotheses simply fail on a regular basis. Biology works according to its own principles and laws, which are not the same as the logical constructions of the human mind. Nor are these laws easily reduced to the laws of physics. Simplicity might be evidence for truth in physics; it almost never is in biology. Here are a couple of examples:

Which hypothesis is more likely to be true?

  1. In a biochemical reaction where A is converted to B, an excess concentration of B inhibits the reaction.
  2. The same as 1, AND an excess concentration of A stimulates the reaction.
  3. The same as 2, AND an excess concentration of A stimulates another reaction C to D, and D also stimulates the reaction.
  4. The same as 3, and a couple more redundant stimulatory and inhibitory processes.

Note that 1 is sufficient to regulate the amount of A and B. It is the simplest solution. But it is almost never what actually happens. The correct answer is usually 4. One example is the control of the synthesis of deoxyribose trinucleotides (used in DNA replication).

Here is another one. We know that coloration in plants and animals is usually of some evolutionary benefit or purpose. Which hypothesis is likely to be true for fish living in the deepest ocean where no light exists?

  1. They should be devoid of color, since there is no possibility of there being any selective advantage for color where there is no light. Simple and logical
  1. Such fish should exhibit an enormous variety of brilliant and beautiful colors. Illogical, crazy.

Yes, the second one is correct. Fish living in the deepest part of the ocean have amazing colors – brilliant yellow, blue and orange patches on their scales that are never seen since all the fish and other creatures in this environment are blind. The fish have such amazing colors because in that environment, colors are irrelevant and cannot be harmful (by attracting predators, for example). They are the result of random mutations in pigment genes (carried over from their ancestors) that produce a phenotype with no detrimental effect on the survival of the individual, and are therefore never selected against.

Welcome to Biology.

There is a popular belief that phenomena have a single, simple explanation, and that once the best cause of anything is identified, all other potential causes can be ruled out. Like the principle of parsimony that underlies Occam’s razor and the other quotes above, this certainly does not work in biology. Whether at the organismal level or at the level of biochemistry, many different causes can produce the same effect. And one cause can produce a myriad of effects. This is why biological reality is so complex and so hard to capture in mathematical terms. It is true for metabolic pathways, for gene regulation, for animal behavior and for evolutionary biology.

In the next post I will discuss the implications of the failure of parsimony and simplicity in biology for theology. Comments from readers who might have a view on this issue are especially welcome here.


This entry was posted in Uncategorized. Bookmark the permalink.

10 Responses to Occam was wrong

  1. SheilaDeeth says:

    I love the fish example – simple, elegant, and so well-known we never even think about it.

  2. gregwelty says:

    I was doing a web search checking a student paper for plagiarism, and I came across your post (https://thebookofworks.com/2016/05/02/occam-was-wrong/), since the Swinburne quote I was checking is also contained on your site. So, I got here randomly. But your bold headline (“Occam was wrong”) led me to read further!

    While I agree with you that quite often nature is *not* simple – and so I dispute none of your examples – unfortunately this is not enough to overturn the principle of parsimony as a relevant criterion for theory-selection. And that’s because of an important phrase contained in the Swinburne passage that you elided via ellipsis: “I seek in this essay to show that — other things being equal — the simplest hypothesis proposed as…” etc. “Other things being equal” is a crucial qualification to the principle of parsimony, and as far as I know, always has been. What we are most interested in when it comes to scientific explanatory hypotheses are theories that have *predictive power*, that lead us to expect the evidence that we see. The principle of parsimony comes in at a later stage of intellectual inquiry, when we have two or more theories with *equal* predictive power, when ‘things are equal’ in that respect. Here the evidence itself won’t tell us which theory is more likely to be true, since each theory on offer has equal predictive power. (For example, the theory that a single criminal caused the bloodstains, broken window, and exploded safe, versus the theory that three criminals caused these phenomena.) The principle of parsimony says that the rational thing to do in *that* situation – the ‘other things being equal’ situation – is to prefer the simplest hypothesis as more likely to be true than its rivals.

    I presume that in each of the biological examples you’ve given in this post – biochemical reactions and coloration in plants and animals – the false hypotheses just are those hypotheses that contradict the data. One doesn’t need the principle of parsimony to help us choose here, for there is no need for tiebreaking among hypotheses that don’t equally lead us to expect the data. The tie has already been broken, because only one hypotheses leads us to expect the data.

    I believe that reading further along in Swinburne’s little book that you cited, *Simplicity As Evidence of Truth*, as well as his presentation of this principle in *The Existence of God* and *Epistemic Justification*, would further explain the point. Ripped out of context, the principle of parsimony is obviously absurd. But in context, it has been appealed to by scientists for hundreds of years, and continues to be applied today across a wide variety of disciplines (including biology).

    Note that the same point is made in the Newton quote you gave. According to Newton, we admit no more causes than those that “are both true and sufficient to explain their appearances.” If the proposed causes aren’t even sufficient to explain the appearances – because they contradict the data or do not lead us to expect the data – then theories that propose such causes are not ‘in the running’ as possibly true. They have been disqualified from consideration, and so there is no opportunity for parsimony to enter in as a kind of tiebreaker. But if the theories (which specify “the causes of natural things”) *are* sufficient to explain the appearances of natural things, and there are two or more theories like this, then the principle of parsimony says to choose the theory which posits “no more” than what is sufficient to explain. It has application at that stage, not in the earlier one.

    To illustrate, you explain the fish colors this way: “They are the result of random mutations in pigment genes (carried over from their ancestors) that produce a phenotype with no detrimental effect on the survival of the individual, and are therefore never selected against.” But what if someone said, “They are the result of seemingly random mutations (etc.), each of which were produced by the intention of a distinct god, none of whom are any longer around.” This alternative theory would equally lead us to expect what we see (the variety of fish colors). But you would regard the first hypothesis as more likely to be true – since it doesn’t multiply explanatory entities beyond necessity. Ruling out *unnecessarily* complex hypotheses that predict the data is the job of the principle of parsimony, and that is quite different from ruling out complex hypotheses *simpliciter* (which it is never the job of parsimony to do). Complexify your hypotheses as much as you need, to explain the actual data (which can often be quite surprising and particular). But go no further than this in the process of complexification. That is what Ockham’s Razor is all about, as I understand it.

    Full disclosure: I got my MPhil and DPhil under Swinburne at Oxford, so I was forced to work through these issues for a chapter in my dissertation, and in my tutorial work. I am very interested (like you!) in faith/science issues, and I am quite happy to find someone so accomplished in the scientific arena working on these issues!

  3. gregwelty says:

    My comment was overly complex, so here’s a parsimonious summary 🙂

    Simplicity is a secondary criterion, *after* predictive power. That is, it is only applied after we have two or more theories with equal predictive power, that equally lead us to expect the data. At that point, simplicity enters in as a tiebreaker. It’s OK for explanatory theories to be pretty complex, since reality is often complex. What simplicity says is don’t make your explanatory theory *more* complex than it needs to be, in order to explain the data we have. That’s different from saying that all theories should be simple in nature.

    • Professor Welty

      Although I (as a former academic) am appalled at the idea of plagiarism, I am somewhat grateful for this case, since it brought you to this site, and led to you to posting those comments. I entirely agree with your comments, and I will say that for many years, I used to counter comments on the order of “Ockham’s razor says follow the simpler explanation” with “yes, if ALL ELSE IS EQUAL, which it almost never is”. So, there is no dispute there at all. For the biochemical examples, in most cases, what needs to be taken into account are some fairly subtle details of the effects of evolutionary change, some of which actually involve parsimony at a functional level. For example some pathways that appear to be unnecessary by any possible criterion, remain in place simply because the cost of removing them is not “worth the benefit” or in stricter evolutionary terms, supplies no selective advantage to the organism. This is why vestigial organs and molecular pathways exist.

      Having said that, I will say that I do believe that the principle of parsimony is often misused and abused, not just in the context of ignoring the “all things being” equal clause, but also in some atheist arguments against theism. For example in counter to a statement that God is not disproven by the fact that science can explain all natural phenomena, I have heard it claimed that if God is not necessary to understand how the world works, it only adds useless complexity to posit a God. I think that is false, for a number of reasons, chief of which is that an understanding of how the world works is only a small part of everything there is to understand. But that is a fairly large subject to be well covered here.

      I am assuming that you are the Greg Welty at the Philosophy Dept. of Southeastern University, and if so, I feel honored to have your comments here, and would welcome any further thoughts you might have on this or any other posts. (or tweets). Peace.

      • gregwelty says:

        I entirely agree that the principle is often abused. It certainly is not obvious that atheism is a simpler hypothesis than theism. Indeed, you might find it interesting that appeal to the principle of parsimony is the centerpiece of Swinburne’s program of natural theology in *The Existence of God*. Swinburne (a theistic evolutionist, BTW) argues that the atheist ‘stopping point’ of explanation is not nearly as simple as the theist stopping point. Here’s an example of how that argument would go:

        http://users.ox.ac.uk/~orie0087/pdf_files/Papers from Philosophical Journals/God as the Simplest Expanation of the Universe.pdf

        So the idea that simplicity automatically favors atheism over theism seems to me quite dubious.

  4. gregwelty says:

    The link I offered in the previous comment seems broken. Perhaps this will work instead.

  5. Pingback: Whatever Works | The Book of Works

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s