Biology has analogies, not paradigms
The philosopher of science Karl Popper argued that Arthur Eddington’s 1919 test of general relativity emblematized science in action. At that moment, in Popper’s view, Eddington’s observation of the shift of light during a solar eclipse falsified Newtonian physics, leaving Einsteinian physics the only theory left standing.
The later philosopher Thomas Kuhn provided a different view of this test. Kuhn argued that all scientists see facts through the lens of a paradigm. An effective paradigm can encompass any fact, even ones which seem to contradict the paradigm. However, if enough anomalies accumulate and there’s a convincing alternative paradigm, scientists will switch over to the alternative paradigm. So, in Kuhn’s view, Eddington’s observation was just another anomaly in a list of anomalies that prompted the scientific world to switch over to an Einsteinian paradigm.
Thanks for reading my newsletter! Subscribe for free to receive new posts, mostly about biology.
If you’re familiar at all with the philosophy of science, these two viewpoints are probably old hat to you. I used to teach them myself in an Introduction to Western Philosophy course that I taught for a Chinese MOOC. But I’ve been thinking about them recently in the context of the various biological mysteries I’ve been exploring in my blog posts. You see, I don’t think either of them really apply to our understanding of these biological mysteries.
In biology, there aren’t really paradigms or theories that attempt to explain everything. There’s no equivalent of the Newtonian paradigm for obesity or autoimmune diseases. Even for DNA, while there is a “central dogma”, there’s nothing resembling a paradigm. In other words, there’s something that you’re supposed to understand and believe in in order to make sense of it all, but inherent in that is the idea that there are some parts missing, and you’re just going to have to accept that for now until we come up with something better.
Most parts of biology don’t even get a dogma. The closest they get are analogies, which can be better fitting or worse fitting. So, for example, it’s common to say the heart is like a pump (or even just “the heart is a pump”), which is pretty accurate as it does pump blood throughout the body. This doesn’t capture all the complexities of the anatomy of the heart or the relationship of the heart to the lungs, but it does give a good idea of what the heart does.
It also gives a good idea of most problems that can go wrong with the heart. There can be a blockage in the pump itself, which can reduce the amount of blood that can go into the heart or even stop blood from circulating through the heart. There can also be weakness in the muscle, which makes the pump less effective.
This pump analogy doesn’t cover some of the other common problems in the heart, like defects that cause inappropriate mixing of oxygen-rich and oxygen-poor blood or disruptions in rhythm that can be caused by disruptions in the electrical signal that controls the heart’s beating. Those go beyond simple pump analogies. However, that really doesn’t matter.
We don’t rely on the analogy to predict or explain all heart problems, just most of them. There’s no situation in which a heart anomaly would cause us to switch to a different analogy. Gaps in this analogy have been known for at least 100 years, but the analogy is still useful for understanding most heart mechanics.
And that’s really the best of the analogies that we get. Other analogies are worse. For instance, it’s very common to hear mRNA (or DNA) referred to as code. This is a useful analogy in some ways. For instance, it lets us think about how we can upload a sequence of mRNA to Github, print mRNA sequences, measure the amount of information in a given amount of mRNA, or expect a certain predictable outcome from a certain sequence of nucleotides.
However, this analogy breaks down completely when we think about mRNA as a physical substance. There’s no code-based analogy for the importance of the parts of mRNA that physically link with the ribosomes, interact with the immune system, or are degraded with use. It would have been literally impossible to build the mRNA vaccines without an understanding of the chemistry and biology of mRNA, although, again, it was helpful in building the mRNA vaccines to think about the mRNA as code in certain concepts.
Even though all biological analogies are lacking in some way, they’re still important, though, which is why they stick around. There’s no better way to illustrate this than looking at a part of biology where analogies are lacking and seeing how confusing it is.
The one that immediately jumps to my mind is the immune system. Now, the basics of the adaptive immune system are easy enough to analogize, like talking about antibodies as police or soldiers. But the actual complexity of the immune system way outweighs the usefulness of any such analogy. Take a look at the complexity of this map of the innate immune system from Wikipedia.
Then, when it gets to actually understanding problems with the immune system, we’re at a loss. We still can’t explain autoimmune diseases any better than “it’s the immune system attacking your body” or explain why certain immunosuppressive medications work or don’t work. There’s no analogy that remotely captures the blizzard of immune responses that occur when a foreign body enters the bloodstream, so we’re largely at a loss.
It’s tempting to think, then, that this would be a good place for a paradigm, something that can reduce the immune system down to its component parts and predict each next step. But it’s hard to imagine what those component parts would be, and how such a predictive model would work.
So, from my point of view, there are two options here.
1. Assume that analogies will come as our understanding of the system grows. So, understanding will lead to analogies which will lead to greater understanding.
2. Assume that there are no appropriate analogies for this system, and try to develop a different way of understanding it, likely by some kind of advanced modeling or AI that’s not currently feasible.
It’s a pick your battle type of situation. Would you rather understand a system that seems impossible to understand, or develop a type of modeling/AI that doesn’t exist? It’s not easy to answer.