No one seems to address a clear distortion of this technique, which really requires careful treatment, when considering the ramifications of research founded upon it.
The clear distortion is this: Instantiation of a "standard brain" produces a monoculture more pure and homogenous than a population of identical twins. As with any monoculture, all data is reduced very nearly to an N1 sample size, leaving populations prone to abrupt, global crashes when encountering environmental faults.
Even identical twins don't have identical fingerprints. Nor do they have identical memories. If identical twins have differentiable networks of blood vessels, then their "brainprint" will vary accordingly, for sure.
A connectome is more of a signature of averages, sure, but a computer simulation will probably only introduce artificial, pseudorandom variations of neural structure, if the implementation seeks to introduce subtle physical variations to the tissue structure at all.
Based on this intrinsic quality of practice via computer simulation, understand that running a model in a system like this could introduce unwarranted confidence in replicable results, because the model is a replica by default.
I’ve thought about twins and brain structure too, but I am a computer scientist not a neuro-scientist so I might be completely wrong.
Serious mental illness seems to be of higher incidence in the population than I would expect given evolutionary pressures due to its affect on survival. This suggests perhaps some unexplained benefit to mental illness for survival of an individual’s genes despite its seeming like a disadvantage. (As is the case for sickle cell anemia or nearsightedness)
Alternatively, I surmise that brain structures are not fully coded by genes, but develop through a system that is chaotic, in the sense of being very sensitive to initial conditions, and so despite the brain development system being controlled by genes the outcome in terms of exact neural connections is to some degree unpredictable due to the slightest differences in the environment between individuals (even identical twins).
Evolution implies that the development of a brain can not be straightforward given how rarely genius appears in the population.
The clear distortion is this: Instantiation of a "standard brain" produces a monoculture more pure and homogenous than a population of identical twins. As with any monoculture, all data is reduced very nearly to an N1 sample size, leaving populations prone to abrupt, global crashes when encountering environmental faults.
Even identical twins don't have identical fingerprints. Nor do they have identical memories. If identical twins have differentiable networks of blood vessels, then their "brainprint" will vary accordingly, for sure.
A connectome is more of a signature of averages, sure, but a computer simulation will probably only introduce artificial, pseudorandom variations of neural structure, if the implementation seeks to introduce subtle physical variations to the tissue structure at all.
Based on this intrinsic quality of practice via computer simulation, understand that running a model in a system like this could introduce unwarranted confidence in replicable results, because the model is a replica by default.
With the strongest emphasis possible: YMMV