One question we often run into is "how long do my recordings need to be to run a full ICA"?
We understand that the length of the recording is not the sole determiner for the quality of the components, and that good quality data is more important than its length. Another potential, yet unexamined issue, is whether more exemplars of the component in question can help to improve component quality.
As a small experiment we could examine whether obtaining recordings of continuous eye blinks and eye movements, as well as closed eyes before each experiment, and including this section into our ICA's actually improves the quality of those components.
If it quality improves significantly (we can decide on some exact outcome measure later), then perhaps we could make it standard practice to record this sort of thing prior to each experiment in a systematic way, and make sure to include this in our ICA scripts.
One question we often run into is "how long do my recordings need to be to run a full ICA"?
We understand that the length of the recording is not the sole determiner for the quality of the components, and that good quality data is more important than its length. Another potential, yet unexamined issue, is whether more exemplars of the component in question can help to improve component quality.
As a small experiment we could examine whether obtaining recordings of continuous eye blinks and eye movements, as well as closed eyes before each experiment, and including this section into our ICA's actually improves the quality of those components.
If it quality improves significantly (we can decide on some exact outcome measure later), then perhaps we could make it standard practice to record this sort of thing prior to each experiment in a systematic way, and make sure to include this in our ICA scripts.
Who wants to run the experiment and analysis? :)