[FieldTrip] common filter with more than two conditions

[Lin Wang]

Hi Arjen,

Thanks for your help. :-)

Best,
Lin

On Tue, Mar 8, 2016 at 8:28 AM, Arjen Stolk <a.stolk8 at gmail.com> wrote:

> Hey Lin,
>
> Overall the more data (i.e. covariance matrices) the spatial filter is
> constructed from the better it'll be able to describe the spatiotemporal
> patterns present in the data (cf. ERPs, being smoother with more
> averaging). The latter indeed may be condition-specific, and since I do not
> know anything about your conditions, I'll refrain from having an opinion
> there. :)
>
> Arjen
>
> 2016-03-07 16:16 GMT-08:00 Lin Wang <wanglinsisi at gmail.com>:
>
>> Hi Arjen,
>>
>> Yes, it answers my question. Thank! :-)
>>
>> Then a further question is: is it better to include as many trials as
>> possible to have a good estimation of the covariance matrix (provided that
>> the signals are good for all trials)?
>>
>> For example, there are two experimental conditions, with 20 trials per
>> condition. There are also 40 filler trials with a similar structure as the
>> experimental conditions. In this case, can I combine all the conditions to
>> build the common filter and then only compare the two experimental
>> conditions later?
>>
>> The cognitive processes might be different between the experimental
>> conditions and the fillers, so I'm not sure whether combining them has any
>> influence on the spatial filter.
>>
>> Best,
>> Lin
>>
>> On Tue, Mar 8, 2016 at 12:19 AM, Arjen Stolk <a.stolk8 at gmail.com> wrote:
>>
>>> Hey Lin,
>>>
>>> Provided that there are no systematic confounds (e.g. head position)
>>> across conditions, you could construct a common filter based on data from
>>> all conditions. I would leave any statistical comparison to after
>>> source-reconstruction.
>>>
>>> Does that answer your question?
>>> Arjen
>>>
>>> 2016-03-07 1:51 GMT-08:00 Lin Wang <wanglinsisi at gmail.com>:
>>>
>>>> Dear community,
>>>>
>>>> I'm trying to do lcmv beamformer source analysis with a common filter
>>>> for more than two conditions. I have a 2A (A1, A2) * 2B (B1,B2) design, and
>>>> I am interested in both the main effect of A (A1 vs. A2) as well as the
>>>> simple effects (A1B1 vs. A2B1 and A1B2 vs. A2B2).
>>>>
>>>> My question is how to build the common filter. I could combine all the
>>>> four conditions to obtain a common filter for the contrast of A1 vs. A2.
>>>> Then can I also use this common filter to compare A1B1 vs. A2B1? Or do I
>>>> have to build a different common filter (to combine the A1B1 and A2B1
>>>> conditions) for the contrast of A1B1 vs. A2B1?
>>>>
>>>> Thanks for your help in advance!
>>>>
>>>> Best,
>>>> Lin
>>>>
>>>>
>>>>
>>>
>>> _______________________________________________
>>> fieldtrip mailing list
>>> fieldtrip at donders.ru.nl
>>> http://mailman.science.ru.nl/mailman/listinfo/fieldtrip
>>>
>>
>>
>> _______________________________________________
>> fieldtrip mailing list
>> fieldtrip at donders.ru.nl
>> http://mailman.science.ru.nl/mailman/listinfo/fieldtrip
>>
>
>
> _______________________________________________
> fieldtrip mailing list
> fieldtrip at donders.ru.nl
> http://mailman.science.ru.nl/mailman/listinfo/fieldtrip
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mailman.science.ru.nl/pipermail/fieldtrip/attachments/20160308/51608eec/attachment-0002.html>