Dear all,
I am using a block design to explore the neural basis for grammatical
processing-- blocks of noun or verb stimuli, alternating with fixation blocks (fix
sampling, i.e., one TR corresponds to one stimuli presentation).
However, there is an irrelevant variableˇXimageability that can not be matched
between nouns and verbs in advance, so I have to partial it out in statistical
analysis. But since imageability value is a property of stimuli, I could not use
the ˇ§covariate functionˇ¨ in 2nd-level anaylsis.
What I do instead, is to break down the block, and treat each trial individual,
as in ER analysis (set up onset for each trial, with duration of 0). In this way,
I can add the imageability value for each trial in ˇ§parametric modulationsˇ¨
function.
I wonder whether this method above sounds reasonable for you? Any input is
greatly appreciated!
Thanks in advance,
Xi
|