High-pass filters

Post Reply
Rachel
Posts: 1
Joined: Thu Aug 14, 2008 4:08 pm
Location: Aberdeen

High-pass filters

Post by Rachel »

Hi

My recordings contain lots of low-frequency drift which I would like to remove (e.g. high-pass filter at 0.01 Hz). Lots of people recommend using such a filter (e.g. Luck's ERP text book) but I know that it causes artefacts.

In my study, I have lots of saccades during my epochs which differ in direction according to condition, so if I average left and right trials I see large deflections in the eye channels. After high-pass filtering the same data I see opposite deflections in the data during the baseline, i.e. before the event that triggered the saccades - this is clearly artefactual as a result of filtering. I understand why it happens, but what concerns me is: if this happens for eye-movement effects, surely something similar will happen for all ERPs, just smaller and maybe less obvious, but still with the potential to give significant effects which are not real.

Given this, why do people still use high-pass filters?

And if I decide not to ever use high-pass filters, how can I use my data?

Does anyone else have this problem?

I suppose the ideal is to prevent the drifts from being in the recording, but I understand that they are very common with DC amplifiers.

Thanks in advance.

terracop
Posts: 8
Joined: Tue Mar 03, 2009 9:57 pm
Location: Europe

Post by terracop »

Drifts are best to be avoided during recording, you are right. (GiGo)
But it might be possible that you are processing the data wrongly. Which program are you using and at which stage of processing are you filtering? (Wrong baseline? Filtering epoched data?Slope?)

Post Reply