Today’s prompt is about an unsolved scientific mystery in our field. In neuroscience, the entire brain is basically an unsolved mystery. We simply don’t understand it. But today we’d like to talk to you about a mystery within this very sentence.
We consider that the discussion surrounding our understanding of the brain is not limited just to how advanced this understanding is, but also includes the expectations we have when we wish to be able to comprehend this organ and its functions. There is often an implicit assumption that “understanding” naturally means “being able to put it in simple terms” and “abstracting away the unnecessary”. Borrowed mostly from physics, where concepts can be explained through somewhat simple equations, it is unfortunately unclear whether the same will hold true for neuroscience.
Up until relatively recently, one way of trying to easily explain neurological functions was to try and describe smaller units of the brain (and assuming, to a certain degree, that the findings can be extrapolated). Empirical neuroscience was mostly focused on the study of the different components of the brain in isolation, be it neurons or cortical regions. Computational modelling followed a similar trend: researchers attempted to reduce these components to their bare minimum, thus assuming that things such as neuron shape or type were not important for brain processes. For example, one popular approach towards simulating individual neurons was to assume they were points, thus discarding both dendrites and axons. But a paper published this year argues that ignoring dendrites is a mistake. Another example, this time at the regional level, showed that constructing an anatomically realistic model of the hippocampus (a region of the brain important for memory consolidation) is superior to one that assumes that neurons are uniformly distributed and connected in this network.
Of course, abstraction has its role. One subfield of computational neuroscience has managed to build low-level simulations of the brain making use of this. Basically, instead of simulating every single neuron in the brain, mean-field modelling replaces that by one single variable describing the activity of an entire population. To understand it a bit better, imagine you’d like to draw on a map the migration path of a flock of birds. You could painstakingly draw 1000 slightly different paths, corresponding to each of the birds in the flock. Or you could assume the flock moves as a unit and draw a single line. This approach has already been successful in advancing our understanding of brain dynamics. Yet, even though the complexity is considerably reduced, these models remain incredibly complicated to understand, as they usually have many parameters that interact in a nonlinear, and hence tricky, fashion. Plus, any given mean-field model can theoretically be extended to include even more details that make it more realistic, but also less tractable.
To sum up, whether the brain can be explained in simple terms is currently unclear and only time will tell how “graspable” this knowledge will be. But from the current state, we’re leaning more towards the complex view.
What did you think about this post? Let us know in the comments below.
You might also like:
Aussel, A., Buhry, L., Tyvaert, L., & Ranta, R. (2018). A detailed anatomical and mathematical model of the hippocampal formation for the generation of sharp-wave ripples and theta-nested gamma oscillations. Journal of computational neuroscience, 45(3), 207-221.
Byrne, Á., O’Dea, R. D., Forrester, M., Ross, J., & Coombes, S. (2020). Next-generation neural mass and field modeling. Journal of neurophysiology, 123(2), 726-742.
Larkum, M. E. (2022). Are dendrites conceptually useful?. Neuroscience, 489, 4-14.