In the 3rd year of my PhD studies, I still feel like a newcomer to neuroscience. Reorienting myself around the history and religions in this field has been challenging.
Wait, history and religions? That doesn’t sound very scientific! The more I learn, the more I realize that we (or at least, I) have been bamboozled by the narrative that peer-reviewed science is truth, and that experimental evidence is objective. At least in systems neuroscience, where there are big egos and little consensus, the prevailing driver seems to be belief. The brain is just too unknown!
What major ideas can systems neuroscientists all agree on? I’ll pick a canonical example from visual neuroscience: In the early 1960s, Hubel and Wiesel discovered cells that selectively responded when shown stimuli such as an edge with a specific orientation. They found roughly two classes of cells, and called them simple and complex. This concept is relatively uncontroversial, and forms the basis for current models of vision in brains and in machine learning. For the most part, we understand the anatomical connections in the brain. But take one more step, and we are suddenly lost: what happens after complex cells, and how does this lead to visual perception? We agree on shockingly little about the visual processing stream.
One major controversy is whether a deep neural network is a good model of the brain. There are well-respected experts on both sides of this argument, and they are indeed supported by seemingly-robust, peer-reviewed studies (e.g. for, against). But here lies a contradiction: if all evidence were objective, and our conclusions totally dependent on evidence, there would be no disagreement. We certainly can (and will) get into the weeds of whether certain methods are sloppy or unprincipled1. But fundamentally, what differentiates one scientist’s opinion from another’s? I argue that in the presence of all our evidence, it must be religion2.
To be clear, I am not saying we should not hold different beliefs; in fact, it is necessary that we do. What has always troubled me is the paradox of the hypothesis, which is ultimately driven by a belief: you must pick a side before you start. Is it a convincing research proposal if I refuse to pick a side, and propose studying all possibilities? Correct me if I’m wrong, but it seems that scientists necessarily cannot be agnostic.
This brings me to my existential struggle as a young researcher. How do I develop the confidence and knowledge to assert my position, when there is so much I know I/we don’t know? I can form a well-supported hypothesis, but a peer will point out evidence that directly refutes it, or a professor will disagree based on a perspective upon which they have built a career. I really am not criticizing neuroscience as religion, rather, I am curious about how people got there! As much as it frustrates me, I think sorting through what we know and understanding what I believe is the most urgent of my current pursuits. One way I know must be true is to read more, i.e. to understand history better. But beyond that, I don’t know.
A mentor once said to me: “Most of the fundamental problems in physics have been solved by very smart people. For the unsolved ones, it’s unlikely you are the smart person. And that’s why we study neuroscience.” For a field with little truth, there is plenty of room for our beliefs to make progress.
This is just the beginning of reflections on my scientific identity and experience. And I know these are not groundbreaking revelations (I just need to read more, probably). But by writing this down, I hope to figure some stuff out!
My favorite example of weird things from seemingly-sound data analysis is the dead salmon experiment.
There is certainly nuance here between belief and religion that I’m not able to give full treatment. I choose to use religion because I’ve observed people discussing schools of thought in ways that invoke religion: steadfast beliefs even in the face of opposing evidence; rituals in analysis and argument; key historical ideas and players within communities.
Hehe - I remember the dead salmon experiment! More like this please.
Nice essay. On your Appendix #2, I think what you referred to is belief, not religion. Religion is immutable; belief is malleable. Philosophically, belief has always been a part of science.