I had a bio professor put it this way: pretend you are reverse engineering alien technology. Throw away all your anthropocentric ways of thinking.
Biology was not designed by human minds and doesn’t work like something we would design. Everything influences everything else and its feedback control loops all the way down. It’s more like a vibrating causal cloud than a chain of causality.
That was in the early 2000s and while I’ve been out of the field mostly since then I still don’t get the sense that we have a solid conceptual vocabulary for things like this.
I think of causality as generally linked to our notion of invariants.
We can arrive conceptually at causes in physics because we have "laws of physics", which are invariant over all conceivable conditions (except in the most extreme cases).
Once you lay down your assumptions/invariants, the rest of the system can be "controlled" - intervene upon X and observe Y. If the system is truly invariant over repeated trials, like a computer algorithm or highly controlled experiment, then you can deduce some form of causality. As in a formal proof, causality necessitates a "starting point", an axiom, an invariant, from which all else follows.
In biology and many other systems, you simply don't have many invariants at all. The entire topology is variable, constantly "shifting" as a single node changes.
How can you ever control or reproduce such a system to deduce what causes what? It's possible, but as the author suggests, only with great difficulty, frequent missteps and irreducible imprecision.
There are no invariants because biology has an (effectively) infinite number of feedback loops that cross temporal and spatial scales.
As a computational biologist working on human health, I've started referring to my various approaches as "causal enrichment", because I don't expect to find causality in a clean way. Tweaking some feedback loops are more likely to result in a desired outcome than others, but only probabilistically.
Problem is that the models of how parts connect in biology are mostly descriptive.
However, try to use only words to describe how a circuit works that powers a radio, then use those words as a “model” to predict how a radio would work: you’ll end up with vague hand wavy text that poor quantitative and qualitative predictions.
One solution is to use the language of differential equations to explicitly describe in simple but precise language how parts are connected. Although not quantitative in their predictions, at least this “systems” approach makes qualitatively interesting predictions (eg limit cycles, fixed points, bistability).
And here I am, with this thought about how us humans, having an ingrained notion of science and discovery needing to be built on __analyzing__ things, are defied by such complex, interwoven systems that really can't be put into parts.
And while I'm totally an armchair scientist here, what comes to mind immediately is that the resources needed to describe such a system analytically would grow exponentially with a very high value of that exponent.
I wonder, what a horrendously complex system of differential equations would be needed to describe just one receptor, with its countless inputs and outputs -- and how brilliant in mathematics would a person reading it have to be.
Disclaimer: this comes from a university dropout with no background in biology and only a little in mathematics -- and moreso, whose native tongue is not English.
I think one thing to keep in mind is that coming up with these differential equations is an art. Classically it's about balancing simplicity (for human understanding) and being able to recapitulate some of the dynamics of the system.
Receptor dynamics is one subsystem that has been studied well theoretically. We know increasing number of binding sites can generate for example "cooperativity" and bistability dynamics.
> And while I'm totally an armchair scientist here, what comes to mind immediately is that the resources needed to describe such a system analytically would grow exponentially with a very high value of that exponent.
Yes, this relates precisely to Stephen Wolfram's notions on "computational irreducibility." For humans, it is an impossible project. For machines, it is perhaps possible, but we humans would not comprehend it.
There was quite a lot of writing like this in the aftermath of the initial expansion of chemistry into "bio-chemistry". Various thinkers, biologists, multi-disciplinarians all noting that it was foolhardy to imagine that biochemistry could "explain life".
As TFA notes at its opening, they all sort of went quiet as molecular biology became a real thing, and for a while it did seem to many folks that the systematists and cyberneticists were quite simply wrong.
> "The problem I had in mind was that of getting clear about the very nature of causation in biology. It differs from the problem of causation in the physical sciences. Organisms manifest a fluid, integral, harmonizing sort of causation that is more like a play of the multi-dimensional reasons for things than a set of one-dimensional mechanical interactions. It is more like the rich interplay of meaning in an unfolding poem than a rigid syntax or logic."
I don't think causality differs across different academic disciplines, and of course nature is indifferent to the political divisions found within academia. E.g., the life of a star is rather similar to the life of a dog, in many ways - birth, development, death - and so why this desire to draw this fundamental divide between 'living' and 'non-living' systems? Yes, there's the transfer of information from one generation to the next in living systems, but that's only possible because cells manage to maintain their physical integrity.
Some people might think this means everything is dead, mechanical, and boring - but one could just as easily adopt the philosophy that the entire universe is alive, dynamic and interesting.
Doesn't our intuitive notion of "causes" come, in the end, from human-scale objects and agency?
Something happened because some being willed it. That's how pre-scientific people thought of everything. The "something" might be the god of wind or the sea, but it's never just "probability."
Biology was not designed by human minds and doesn’t work like something we would design. Everything influences everything else and its feedback control loops all the way down. It’s more like a vibrating causal cloud than a chain of causality.
That was in the early 2000s and while I’ve been out of the field mostly since then I still don’t get the sense that we have a solid conceptual vocabulary for things like this.