Agent Causation

What is an organism and how does it differ from a machine? We have this vague idea that it has something to do with self-agency and made out of “soft” (organic) matter rather than “hard” inorganic matter. Is there downward causation? I’ve read a fair amount of philosophy and science on this question, much of which I’ve found challenging to understand. Thus, it was refreshing to read Potter and Mitchell’s lucid article “Naturalising Agent Causation” (Entropy 2022, 24, 472. DOI:10.3390/e24040472). Maybe it was having this groundwork, or maybe the authors were just clearer in their exposition, but I found it easier to read.

 

Their opening questions: “When an organism acts in the world, is it right to say that the organism caused the effect? Or is that simply a useful metaphor or a convenient level of description? Perhaps the more accurate statement is that some biochemical pathway or neural activity within the organism caused the effect… is it right to think of organisms as agents capable of action?” On the first day of my biochemistry class, we talked about definitions of living systems and I joked that to study of biochemistry is to take “life” out of biology via reductionist methodology. Probably many biologists do the same thing unless they operate on the organism or systems-level, even though there is a shift to more holistic approaches.

 

The authors begin with definitions. I’ve chosen this statement to represent their thesis: “To justify agent causation, systems need to exhibit a causal power that irreducibly inheres at the level of the whole system while still maintaining that the causal power is instantiated in, or realized by, the system’s physical constituents.” Essentially, non-reductive and obeys physicalism. But to distinguish their conceptual framework from its detractors, the authors discuss vertical reductionism (every macroscopic description is “fixed” by a microscropic description), horizontal reductionism (attributing causation to a subset of the system), and external determinism (“the environment wholly determines the system’s actions”).

 

Their counter-argument is to propose eight criteria to exist as a causative agent. The first three (thermodynamic autonomy, persistence, endogenous activity) are familiar. I was reminded of the importance of having physical boundaries between system and environment. While I drum into students the importance of the conceptual separation when introducing thermodynamic model, sometimes we get lost in the math and forget that there are physico-chemical boundaries. Persistence ties into thermodynamic autonomy because an organism needs to constantly extract energy from the environment, even while this persistence is dynamic rather than static. But the organism has to actively do so.

 

The other five criteria are a little trickier. Holistic integration argues against reductionism, and reminds us scientists not to be overly enamored with our methodology. When we isolate a small part of a system to probe its workings in detail, that act of isolation is also a blinder. And the more we learn, the more complex things get as we scope outwards. As students in biochemistry class complain: why is it so complicated? Low-level indeterminacy relates to the quantum level, but only to set up the next criterion, multiple realisability: many microstates might lead to a single microstate and there seems to be all sorts of coarse-graining taking place as intervening layers complexify the system (as a means of regulation and control). The historicity criterion reminds scientists that we can move beyond asking how to asking why; and the agent-level normativity criterion argues for the importance of meaning.

 

But meaning, like beauty, is in the eye of the beholder. Thus, receiving and processing information must be integral to an organism. The authors do a nice job connecting signal reception from the external environment to a discussion on chemotaxis in bacteria and neural signals in “higher” organisms, to the historic evolution to the abstract processing of information, to the response (or non-response) of the organism. The middle part where physical signals are coarse-grained and manifested abstractly in some sort of network, which informs a response, is not only critical but cannot be cleanly separated from its beginning (the stimulus) to the end (the response). Hence, the organism coheres as an individual agent in its own right.

 

I end by quoting parts of their addendum (“The Reductive Instinct”) on why this is so challenging both for me as a scientist and my students, the budding scientists, to grasp. In science, there is “a slippage from methodological into theoretical reductionism… [we] employ experimental techniques that powerfully manipulate individual components and measure so specific outcomes while attempting to hold as much of the background activity of the system constant… This approach naturally lends itself to thinking that an organism’s components truly act in isolation from each other… [but] just because it is possible and often useful [as an approach]… does not mean that these elements actually work separately or have truly isolatable causal efficacy [normatively]”. While I haven’t fully figured out if their argument works philosophically (there might be some sleight-of-hand), I found their framework helpful, and more importantly their article was lucid and one that I can introduce to students. That’s a major plus in my book.