[Eeglablist] Makoto's pessimism explained: Talk at Santa Fe Institute on July 11, 2019

[Makoto Miyakoshi]

Dear electroencephalographiles,

For those who are interested in basic problems of EEG, my one-hour talk at
Santa Fe Institute on the basic issues of scalp-recorded EEG is available
on Youtube.
https://youtu.be/lRyZxt2WeKk?t=222

NOTE: this is not so much about preprocessing. It is rather an attempt to
determine the upper bound of possibility of scalp-recorded EEG. July is my
month to answer questions to the EEGLAB mailing list. I want to take
advantage of it to discuss the issue with those who are interested in this
issue.

Here is a glossary and synopsis of my talk.

*Makoto's pessimism*: It refers to a doubt whether scalp-recorded EEG can
reflect brain activity in full detail. It was proposed for the first time
in Neuroadaptive Technology 2017 at Berlin.
https://www.neuroadaptive.org/conference/2017/keynotespeakers This time, I
added biophysical explanations for the reason why scalp-recorded EEG's
effective degrees of freedom (eDOF) is very low.

*Poorman's X-ray fallacy*: A wrong belief that scalp-recorded EEG should
work as X-ray does. This X-ray analogy refers to the description in
https://www.fil.ion.ucl.ac.uk/spm/doc/history.html that goes "*The ensuing
SPM can be thought of as an X-ray of the effects significance*."
Unfortunately, this is not the case for the scalp-recorded EEG due to low
eDOF. See this slide https://youtu.be/lRyZxt2WeKk?t=1875

*Monadologistic fallacy*: A wrong belief that even one-channel
scalp-recorded EEG signal should contain profound brain information, and
depending on our choice of analysis method the signal can answer any
question we have. Also, remember that anything below EEG recorder's
sensitivity (typically 1 microVolt) at the scalp sensor cannot be picked up.

*Easy and hard problems*: Easy problems refers to forward model of a human
head. There are uncertainties in 1) skull conductivity, 2) presence of
foramen in each hemisphere, and 3) blood vessels under the skull. These
problems could be addressed by advance of technology in future. Meanwhile,
the Hard problems refers to the fact that 1) compared with ECoG, scalp
measurement has broad point spread function, and 2) to be
scalp-measureable, the active cortical patch should have size of > 1
inch^2. The hard problem of the scalp-recorded EEG cannot be addressed by
technology. To go beyond the Hard problems of scalp-recorded EEG, we would
need to develop chronically-implanted, ultra-high-density, wireless
(CI-UHD-WL) ECoG system.

Thank you for reading. I look forward to your comments and criticisms.

Makoto

-- 
Makoto Miyakoshi
Assistant Project Scientist, Swartz Center for Computational Neuroscience
Institute for Neural Computation, University of California San Diego