[Mne_analysis] psychocomputational analysis toolkit in Python?

Phillip Alday phillip.alday at mpi.nl
Fri Mar 9 06:33:25 EST 2018
Search archives:

On 08/03/18 19:20, Etienne Robillard wrote:
> OK I got it, thank you! :)
> Essentially, I want to develop a psychocomputational analysis toolkit in 
> Python for studying real-time neuroimaging data (of cortical activity) 
> from low-intensity, focused ultrasounds (LIFU) and possibly emulate a 
> functional brain-to-brain coupling system on my 32-bit computer as a 
> proof-of-concept to demonstrate the vulnerability of the human cochlear 
> pathway to become a target of sophisticated ultrasonic side-channel 
> attacks from mobile devices.

MNE-Python does provide a lot of tools that are useful for the usual
signal-processing techniques used in processing electrophysiological
data, take a look at the API documentation and example gallery ....

That said,

-  "brain-to-brain coupling" sounds horribly vague and more of a fashion
statement. See e.g.
for some commentary. The suggestion there to "just use language" has
recently shown up in a more serious Current Biology paper on oscillatory
patterns in classroom settings (Dikker et al., 2017).

- I have serious doubts about the biological feasibility of your
proposal ranging from "dog whistles don't summon humans" to "most adults
in developed countries have mild hearing loss and can't hear very high
pitched sounds because their cochlea no longer respond to them". It
sounds like you were inspired by something like *Snowcrash* and the idea
of perceptual input that can crash the human "computer". While every
sufficiently complex computational device has inputs (see *Gödel,
Escher, Bach* for a longer, philosophical discussion of this) that will
"break" it in some sense, it is not clear that such inputs can even
reach the "core" computations of the brain (whether you define that as
consciousness or something else) or if problematic input is implicitly
dumped along the perceptual pathways. Moreover, "break" can also mean
something as trivial "not be properly processed", which in a brain
setting could mean "not (consciously) perceived". The halting problem is
solvable in the real world: *every* program realized on a physical
computer will terminate at some point, even if that point is just the
heat death of the universe.

- If you want to simply use sound as non-lethal attack ... well, there
is a lot of research into that as part of broader research into
"non-lethal weapons".

- I'm not sure why you want to emphasize that you have a 32-bit
computer. Are you worried about real-time constraints? Even then, many
other aspects of your computing architecture will make a much bigger
difference. My RPi3 has a 64-bit processor but for many tasks it is
decidedly less powerful than the 32-bit computer I used in high school.

- Finally, there are serious ethical issues in your proposed project.
You definitely need to have an ethics board involved. You goal is quite
literally an attack on the lowest levels of the human brain.


> Etienne
> Le 2018-03-08 à 10:31, Christopher Mullins a écrit :
>>> Would it be possible to emulate a minimally functional 
>> brain-to-brain coupling system entirely in Python?
>> I don't know what that would entail, but the links I shared have a 
>> mailing list and a very responsive gitter, both of which would be 
>> great places to ask about that!  (You're welcome to ask here of 
>> course, but this being the general python mailing list, it's a shot in 
>> the dark.)  Good luck!
>> Chris

More information about the Mne_analysis mailing list