When I set the GLM, I use the dafault filter settings and get the following warning:
Warning: The low pass filter cutoff frequency is higher (or equal) than the nyquist frequency. The data
won't be low pass filtered!
I start off from a .bdf file, sampled at 1024 Hz.
Maybe I just missed it, but I couldn't find the details of the low-pass cutoff in the Bach 2013 Biol Psych paper, so I can't tell exactly what the problem might be.
The data generally look fine, not too noisy and without any obvious artifacts. I get consistently negative parameter estimated though, so I thought this might be a cause. Or can I ignore the warning if the data are not too noisy?
Thanks
Elisa
If you would like to refer to this comment somewhere else in this project, copy and paste the following link:
the low-pass cutoff is 5 Hz. The reason for this warning is an attempt to low-pass filter the design matrix which is already created at 10 Hz. The warning has no meaning, does not affect your results, and will disappear in future versions.
The consistently negative parameter estimates sound like an unrelated problem, usually to do with wrong model timing, too many events at very short intervals, or a few very large responses to non-modelled events (such as block start, or block end, or start of fixation).
Hope this helps
Dominik
If you would like to refer to this comment somewhere else in this project, copy and paste the following link:
Hi,
When I set the GLM, I use the dafault filter settings and get the following warning:
Warning: The low pass filter cutoff frequency is higher (or equal) than the nyquist frequency. The data
won't be low pass filtered!
I start off from a .bdf file, sampled at 1024 Hz.
Maybe I just missed it, but I couldn't find the details of the low-pass cutoff in the Bach 2013 Biol Psych paper, so I can't tell exactly what the problem might be.
The data generally look fine, not too noisy and without any obvious artifacts. I get consistently negative parameter estimated though, so I thought this might be a cause. Or can I ignore the warning if the data are not too noisy?
Thanks
Elisa
Hi Elisa
the low-pass cutoff is 5 Hz. The reason for this warning is an attempt to low-pass filter the design matrix which is already created at 10 Hz. The warning has no meaning, does not affect your results, and will disappear in future versions.
The consistently negative parameter estimates sound like an unrelated problem, usually to do with wrong model timing, too many events at very short intervals, or a few very large responses to non-modelled events (such as block start, or block end, or start of fixation).
Hope this helps
Dominik
It does help a lot, thanks!