The Disappointment of Neuroscience
The forest is not in the trees
Complex is a word that gets used a lot in mental health. And yet perhaps not enough.
Since the enlightenment, people have believed nature follows certain rules and that the job of science is to figure out these rules. This is a form of reductionism, which is the idea that reducing something to its components enables you to figure out how the whole thing works and then use that to fix it or predict what happens next.
Then came chaos, and a realisation that sometimes even when you understand the components and rules you still cannot always reliably predict what happens over time. Weather, planetary motion and turbulence are all examples of things we can understand when we zoom into a few sequential snapshots, but struggle to foresee much beyond that.
And yet, these systems are not random. They follow patterns, establish equilibriums and have limits they don’t cross. Something is shaping and constraining their behaviour, which is why we don’t get snow at the equator (at sea level at least) or drops of random size and direction from a leaky tap.
But what if the individual components of a system can also change each other’s behaviour, so that they don’t always behave in exactly the same way? Atoms, markets and neurons are examples of adaptive components that when interconnected in sufficient numbers give rise to unexpected ‘big picture’ phenomena that we call chemistry, financial cycles and consciousness respectively.
These hidden rules that shape and structure the big picture cannot be directly seen anywhere within it. This is the domain of complexity science, which over the last few decades has grown as an independent field of study, perhaps most notably at the Sante Fe Institute in New Mexico, USA.
It turns out many (perhaps most) processes are complex, functioning somewhere between total predictability and randomness, where a new bigger picture somehow emerges from countless small-scale interactions. This can happen at multiple levels, with each level emerging - but not directly computable - from the previous one.
Typically, when we think we can make out a picture we put a frame around it and give it a name. While nature itself has no such boundaries, this tendency to frame things is how we get something called biology from the milieu of chemistry, which in turn emerges from physics, which in turn emerges from mathematics, which ultimately emerges from some property of the universe.
The ‘frames’ of these pictures represent the boundaries of our (human) understanding around a complex system. This means understanding what’s happening on one side of a boundary cannot be directly extrapolated across to the other side; each picture requires its own vocabulary, tools and theories. Each can spawn an entire field of study that spans a lifetime of learning. It is these boundaries - these transitions from one complex system to another - that cause us to see nature as discrete subject areas:
Mental health and illness are big picture equilibrium states that emerge from genes and neurons, in a similar way to how capitalism emerges from buying and selling. Trying to work out the overall state and behaviour of a country’s economy by delving into its billions of individual transactions would not work, not just because gathering and analysing that amount of data would be impossible but also because the level of understanding we are looking for cannot be seen in the data.
This is why “$20 billion [in research] has failed to move the needle on mental Illness" (Thomas Insel, former director of the National Institute of Mental Health in the United States). It turns out that understanding genes or neurons, which each form vast networks of adaptive interconnected units, cannot be directly extrapolated to improving our understanding of human behaviour and disorder. This is a limit of the reductionist approach; it cannot penetrate the transition between complex interactions at one level through to the emergent big picture phenomena at the next.
In physical health, studying the biology of health and disease is useful to the extent that the functional output of a tissue or organ is correlated with its anatomy and physiology. While this approach has some validity with the brain (e.g. the sensory and motor pathways), the brain’s grand unified output - i.e. us - is very far from what we see under the microscope. There is no straightforward chain of ‘because this, then that’ from biology to behaviour.
Through the lens of complexity then, it becomes clear why the reductionist approach has so far failed to significantly improve our understanding of mental illness and its treatment. There are complexity transitions between each of the levels of genes / neurons / networks / functions / behaviours that block the reductionist’s arrow of reason from shooting straight through them all.
"Everything should be made as simple as possible, but not simpler." Albert Einstein, 1933
The billions of dollars that have been invested in neuroscience are testament that the reductionist model is at best inefficient, and may ultimately be insufficient, for understanding mental illness. Complexity science tells us that we need to rebalance our research towards structures that operate closer to who we are at the big picture level, such as the social and environmental, because we will not find humanity in neurons any more than we will find forests in trees.
This is an impressive opening article! I enjoyed reading it and will now enjoy thinking about it for the rest of the year. Damn you Sir!
Hi Saj!
"It turns out that understanding genes or neurons, which each form vast networks of adaptive interconnected units, cannot be directly extrapolated to improving our understanding of human behaviour and disorder. This is a limit of the reductionist approach; it cannot penetrate the transition between complex interactions at one level through to the emergent big picture phenomena at the next."
Hmm. Isn't this a matter of the sheer complexity rather than the reductionist approach itself? We sure have been naive and presupposed that understanding individual or small groups of neurons well will give us a good understanding of the mind. But it seems to me that the problems, rather than reductionism itself, are more like:
1. we're still in a very early stage of understanding the larger structures, where the details really matter. But yes, most people in the 70's probably hoped that it would pretty much be sorted out by now.
2. Even if we had a complete description of the brain stored in a supercomputer or something, the complexity and transitions from low-level to high-level phenomena are probably not graspable for human minds in the same way as the standard model -> cell behaviour is (even though we only have simple internal models of that).
So, the main problem, as I see it, is that we do not have the right data, nor the right mental capacities to understand psychology in a reductionist fashion. And perhaps we never will.
Personally, I believe people like Dennett have figured out the core principles of the mind to a high degree, and he wouldn't have done so without neuroscience.