[Eeglablist] EEG 'localizer'?

[Johan]

Dear Tarik and Makato, also best wishes for the new year, and hope you had
a wonderful christmas!

Thank you all for your suggestions! The main thing right now would be to
choose several things to be included that are 'most important'. For us,
this would be something visual and something auditory in any case, so we
could do basic ERP analyses - just capture N/P 1/2, etc. In addition, some
kind of frequency encoding (maybe different ones) would also be highly
desired.

Some more information: the reason is that we wish to use this 'localizer'
to compare different environmental conditions which are less than ideal
(i.e. a very loud and noisy room, a screen which can only be seen for 1/2
of it) - with the 'perfect' laboratory condition, in order to tell us
something about the quality of the EEG itself, using within-subject testing
(so we would measure it always twice, and quantify EEG signal 'decay' just
with paired T-test on several observables from the task such as N1 and so
forth). Also, the frequency encoding would be interesting in case some
(additional, location-specific) artifact corrections are required, such as
with EEG/fMRI - then you can check whether the frequency is still 'in the
signal' after all the corrections.

Kind regards,
Johan

PS. I think it would also be great if a localizer would be so standardized,
that they would allow for a reliable comparison between different
sites/studies - maybe even in past studies - but this is a somewhat harder
problem to tackle properly than just the scenario mentioned above: instead
of optimizing for just time (and not worry too much about the ERP shape as
such), you now also optimize for ERP shape in terms of its comparison with
the 'standard'. Regardless, I also think it would be of very high value to
have some kind of standardized EEG talk - one which could be distributed
free of charge (i.e. without spending too much on software like
presentation and such) to anyone with a wish to do anything with EEG. I
somehow doubt that a vendor-based task would be the way to go - these
require payment of some kind - which is a barrier for proceeding further.







---------- Forwarded message ----------
From: Tarik S Bel-Bahar <tarikbelbahar at gmail.com>
Date: Fri, Dec 26, 2014 at 10:22 PM
Subject: Re: [Eeglablist] EEG 'localizer'?
To: Johan <johanvandermeer at gmail.com>
Cc: "eeglablist at sccn.ucsd.edu" <eeglablist at sccn.ucsd.edu>



Hello Johan,
> Hoping all is well and have a good new year!
> My strongest recommendation is that you do plenty of pilot testing from
> short & long versions of tasks, and show yourself that you can generate
> robust and intended metrics from the mini-tasks.
> ​Some other thoughts are listed
>  below that you might find useful to consider.
> Good luck with your efforts, it will be interesting to hear more
> ​about ​
> them someday.
> Best wishes, Tarik
>
>
> ************************
> EEG fingerprinting and localizer tasks are of interest to the general
> community for multiple reasons, but I don't think there're are yet
> candidates for batteries of valid "EEG mini-tasks". Your question is
> tricky, as ERP and EEG tasks/metrics are well-established, but EEG sources
> and "localizer" tasks are less well established. Overall, a lot of
> reliability, validity, and single-trial work still  needs to be done across
> the EEG/ERP/source-estimation field before we can have a good advanced
> mini-batteries of EEG tasks to deploy. This topic also connects with the
> search for quickly-generated neurocognitive markers for psychiatric,
> neurological, BCI, and operative monitoring endeavors. Exploring recent
> literature and reviews in that area would highlight some major candidate
> markers (ie, CENTRICS efforts). Some EEG companies sell standardized
> ERP/EEG tasks/metrics with their systems, and there is of course a broad
> market in non-EEG cognitive-testing batteries (e.g., NIH Toolbox, Cantab,
> etc.). Overall, responses to your question about whether it's safe to use a
> task are highly dependent on the evidence base considered appropriate by
> the respondent (e.g., # and quality of previous studies, consistency of
> patterns, constraints on metrics, psychometric properties, external
> validity, etc..).
>
> Of some help might be the several reviews of basic guidelines for
> ​major ERPs such as ​
> P300, MMN, ERN, FRN, N400
> ​, including
>  information about minimal recommended trial count and other constraints
> that are essential. You can also find articles that have tried to establish
> minimum-trial-numbers for the ERN, FRN, and MMN, often in pediatric
> populations.  All these can be found on Google Scholar, most within last ~7
> years. You may also want to search for other terms such as EEG +
> fingerprinting, cognitive state monitoring,  and localizer.
>
>
> There are some trade-offs that require consideration here in terms of
> fidelity (more trials, less tasks) vs. bandwidth (less trials, more tasks),
> in terms of the EEG metrics/constructs to be derived (established ERP/EEG
> metrics vs. simple "localizers" or "fingerprints"), and in terms of how you
> expect to use the resulting EEG-derived metrics (eg., diagnostics,
> exploratory analyses, tracking ICs from the min-battery in other
> conditions). If you want to connect with known ERPs, you will need some
> minimim number of valid trials for each ERP metrics to be able to compare
> to previous findings. The more you deviate from previous protocols, the
> more likely that the resultant metrics will be tentative/exploratory. See
> Luck's MONSTER method as an interesting approach in the ERP area. In your
> search for an EEG-localizer battery, a good rule of thumb cutoff is around
> 50 trials per each kind of tasks (50 face trials, 50 sound trials, 50
> finger-tap trials) - or 50 trials per condition within each task. You might
> cut things to 30 trials if you have few artifacts and can denoise the data
> accurately. If you're not looking for any particular well-established ERP,
> you could essentially just present stimuli in different modalities for ~50
> trials, mostly or all with passive presentation requiring no response, or a
> simple response to maintain attention to the stimuli. If you'd like
> involvement of executive processes, then you could vary the stimuli to
> include oddballs or responses. For a very brief task generating a
> ​n ERN or FRN
> ,
> ​ consider ​
> experimenting
> ​with ​
> ​protocols which require
>  a speeded-response-period
> ​ and/or false negative feedback
> , so as to maximize
> ​ the number of trials generating fronto-medial negativities.​
>
> If you want to get ICA maps of visual, auditory, or other networks, then
> you might be able to simply show a visual stimulus such as a check board or
> faces, for example, to get good visual Independent Components (though the
> latter stimuli would activate higher-order visual regions). Same for
> auditory and sensori-motor networks. There are a good number of tasks that
> give reliable activation within and/or across specific neuropsychological
> systems. Further, many research groups have decomposed brief periods of
> resting data, and derived Independent components associated with known
>  intrinsic brain networks. This suggests rest tasks can also work to get
> "localizers
> ​"​
> or
> ​"​
> fingerprints"
> ​.
>

> ********************
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
> On Tue, Dec 23, 2014 at 10:53 AM, Johan <johanvandermeer at gmail.com> wrote:
>
>> Dear all,
>>
>> I am searching for a single task, which can (well.. ideally..) be
>> performed in under about 5-10 minutes (no longer), and which can get me
>> (ideally!) an ERP of:
>>
>> visual
>> auditory
>> motor
>> go-nogo (response inhibition)
>>
>> ... and, if possible, also the possibility to impart specific frequencies
>> in the EEG signal that can be extracted later on - sort of like the case
>> when you have a flickering checkerboard with pattern reversals every 0.2
>> seconds for a 5 Hz 'base' signal in the EEG.
>>
>> The objective is to compress as much information as possible into a
>> single EEG task - so that you can compare the outcome (within subject
>> analysis) between 2 different situations... without (1) totally confusing
>> the subject, and (2) still have the time to do a larger experiment.
>>
>> So - it's sort of like an EEG 'localizer' task.
>>
>> Does anyone ever try this before, or has seen anywhere something like
>> this being attempted? Or rather you think this is just impossible and not
>> worth the effort? :-)
>>
>> Thank you very much for your time and kind regards,
>> Johan van der Meer
>>
>> _______________________________________________
>> Eeglablist page: http://sccn.ucsd.edu/eeglab/eeglabmail.html
>> To unsubscribe, send an empty email to
>> eeglablist-unsubscribe at sccn.ucsd.edu
>> For digest mode, send an email with the subject "set digest mime" to
>> eeglablist-request at sccn.ucsd.edu
>>
>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://sccn.ucsd.edu/pipermail/eeglablist/attachments/20141228/68c76d92/attachment.html>