"A neuroscientist who finds a correlation between a neuron or brain region and a specific but in principle arbitrary physical parameter, such as pain, will be tempted to draw the conclusion that this neuron or this part of the brain controls pain. This is ironic because, even in the neuroscientist, the brain’s inherent function is to find correlations – in whatever task it performs.
But what if we instead considered the possibility that all brain functions are distributed across the brain and that all parts of the brain contribute to all functions? If that is the case, correlations found so far may be a perfect trap of the intellect. We then have to solve the problem of how the region or the neuron type with the specific function interacts with other parts of the brain to generate meaningful, integrated behaviour. So far, there is no general solution to this problem – just hypotheses in specific cases, such as for recognising people...
In particular, neuroscience needs to start investigating how network configurations arise from the brain’s lifelong attempts to make sense of the world. We also need to get a clear picture of how the cortex, brainstem and cerebellum interact together with the muscles and the tens of thousands of optical and mechanical sensors of our bodies to create one, integrated picture."
This feels really accurate. I think it's what sometimes annoys me about systems neuroscience: it can seem reductive, like we're still focused on constructing a comforting map instead of actually primarily focused on the nature of these behaviors and how they are generated by the substrate of our brains. But then again, in order to do an actual study with a clear methodology and to perform a productive analysis of the data, you need to be concrete. You need to test a specific part of the brain, using specific stimulations.
But good news: the inadequacies of the physical mapping approach means that we'll never be able to make standardized mind-reading devices. If you put someone in an fMRI and train an algorithm to their brain for a really long time, with that person's full cooperation, you can get some basic things (ex. "they are visualizing a house; they are visualizing a cat; they are feeling a kind of sadness that is very distinct from other emotional states"). Individuals' neurological networks are so different - and individuals change across their lifetimes and evidently when they are on LSD - that we won't ever have mass biotech mind control. That's a dystopia we can cross off our lists.
(What we can do is create devices to help people with, say, PTSD to be aware of what is going on in their brains and better understand their triggers and what is healing.)
(also this essay says SSRIs don't work, and that's wrong; they just don't work for everyone and they probably don't work for the reasons we theorized in the 1960s but lots of scientists are working on it...)
Related: The amygdala; optogenetics might not be accurate;
FB: why standardized "mind reading" devices will never be a thing --> "Some researchers now believe the brain and its diseases in general can only be understood as an interplay between tremendous numbers of neurons distributed across the central nervous system. The function of any one neuron is dependent on the functions of all the thousands of neurons it is connected to."
No comments:
Post a Comment