Hi Michael,
If you want to be consistent with the majority of previous studies, then you should have four EVs (II, CI, CC, IC) where the three column regressor is:
Stimuli Onset .1 1
or if you use a single column regressor then set all TRs where a trial occurred to 1 and zero elsewhere.
To control for RT effects, most studies add a fifth regressor which, for each trial, is equal to the demeaned RT.
cheers,
jack
>Two Proposed methods:
>(1) 4 Evs with a Basic shape of Custom (3 column format), Convolution of double-gamma HRF. Each of the four Evs will be composed of trial types listed in the conflict adaptation above (II, CI, CC, IC).
>Example of the 1 line in the 3 column file:
>Stimuli Onset Reaction Time 1
>
>Comparisons:
> Stroop Effect:
>Congruence (II EV + CI EV) > (IC EV + CC EV)
> Note: that the part in parenthesis would capture all of each type of congruent or incongruent trial.
>
>Conflict Adaptation: II EV > CI EV
>
>(2) Three regressors/EVs:
>1) An all Trial regressor (used to absorb what is the same across all trials)
>
>2) A conflict regressor: best way I can describe it is to to describe the 3 column file:
> Column 1: Stimulus Onset Column 2: Reaction time Column 3: a number between 0-1 used to approximate the level of conflict. (Find avg Rt for each condition II, IC, CI, CC, then normalize to a scale of 0-1. CC would be 0 presumably and CI would be 1 presumably)
>
>3) A "control" regressor: same general idea as (2) except were looking for stroop effect rather than conflict adaptation
>
>Comparisons
>Compare 2 > 1 for conflict adaptation
>Compare 3 > 1 for stroop effect.
>
>
>I am just curious if I am on base here and what would be the preferably methodology at the first level.
>
>Thank You,
>Michael Cordell
|