The Hypothesis Dump

Opinion,

Within neuroscience, I have taken part in the practice of reductionism during my work studying the hippocampus, trying to link circuit mechanisms to behaviour. Now that’s all well and good, but here is the question: How deep can we dig until we get lost in natural variation no longer relating to function? 

Let’s say you target a receptor in the brain (hypothetical scenario), like the GABA-R, and divide it into subunits, which can be done. Then investigate the effects that one particular subunit has on receptor functionality. For example, let’s say that subunit increases the affinity for the ligand. Now let’s check the concentration of this subunit on GABA receptors across cells. If there is a particular concentration distributed topographically, now it’s time to look at the other subunits. You find different distribution of the subunit concentrations across an area of the brain. Let’s do some intervention: ablate, knock out, or modulate the cells expressing the different subunits and we find differential effects on behaviours related to the brain area. 

Now let’s go deeper: Are there any mutations in the subunits increasing the variations within the subunits?

They seem heterogeneous. Now let’s investigate how the heterogeneity within one subunit affects the functionality of the receptor. Let’s say the different mutations affect the affinity in different directions. If we plot the different variants of the subunits against the affinity level of the receptor we find a nice and interesting pattern. Now we investigate the distribution of the mutant variants across that topographic map we made earlier for the brain region and we find that the distribution again is topographically organised. The obvious question now is: what are the amino acid compositions in the different mutants of the GABA receptor subunits? 

Furthermore, are the different compositions distributed topographically consistent with our previous findings? 

We find that there are clusters of distinct neuronal subtypes expressing GABA receptors comprised of specific subunits, which in turn harbours a specific mutant variant, with a specific composition of amino acids. Now how do these clusters of mutant GABA receptor subtypes affect the affinity? 

Remarkably, they do affect affinity differently! And lets test those effects on behaviour. There is a trend towards some behavioural changes. After all our investigations, we find a negligible effect on the overall output. Have I made the point? 

At some point, we will find variations in biology that are merely due to individual differences that do not have a function, except a “readiness” to environmental selection that might occur at a later stage. It’s not a novel idea that organisms harbour a reservoir of variations in response to future selective pressures. Very similar to individual variations across a species but at a substrate level. These variations could at some point just be noise of diversity. 

Another issue is the failure to explain the emergent properties of substrates. We sometimes lose track and correlate properties of the individual components with a complex functional and behavioural outcome, which arises instead from emergent properties of the whole system, perhaps the whole organism. 

But how else would you investigate a complex biological system without reductionism? 

Perhaps there is no other way, however, this just means that the scientific community needs to be more careful about applying functions on described attributes until such is found. Even if it is found, we need to be careful about concluding the importance of the described mechanism on the investigated function. 

Nervous systems are a great example of emergent properties. The whole system acts together to produce behaviour. Now, of course, there is distribution of functionality, but that does not mean it is not dependent on other systems, not even mentioning the dynamic shifting of functionality across networks. 

This leaves us with the initial question: when does reductionism lead to non-functional substrate variations (by non-functional, I mean that there are variations that in the end do not make a meaningful change in the system’s overall output)? 

This is very difficult to answer, but the take-home message would be that at some point, this should occur if we talk about biological systems, especially if we assume normal distributions. That means that we need to be diligent when drawing conclusions between behavioural outcomes and neuronal mechanisms, especially the deeper we go. We should be careful not to attribute causation of emergent properties as the properties of individual components. 

Declaration of AI use: The author used Lumo AI for spelling and grammar-checking during the preparation of this article.