Dear SPMers
I've haven't received a response re. this issue, apologies if seems v simple but would
greatly appreciate input.
I want to model response accuracy in an efMRI design (I'm assuming a canonical HRF). I
have conditions A & B, which, in some but not all cases, can be subdivided into correct &
incorrect responses. I want to compare correct-A to correct-B. My question is, is it
MORE valid to –
(a) Specify a model with four canonical-HRF-convolved regressors (four conditions), or
(b) Two canonical-HRF-convolved regressors (Conditions A, B) and one binary accuracy
regressor for each condition?
I think a is valid but b is not. Is this right? Reasoning follows.
------
I can see modelling as four conditions makes sense if I take it from first principles. That
is –
-I am interested in brain regions that respond to stimuli when they are both (1) A given
condition and (2) accurate.
-By corollary, these theoretical brain regions should not necessarily respond to inaccurate
stimuli – i.e. I can ignore inaccurate scans.
-If I include a binary regressor, I do not think I'll have a regressor that models predicted
blood flow for accurate regions. I have one that models brain regions that respond to all
stimuli, and one which models an additive effect of 'accuracy'.
I don't think I can make the assumption that accuracy will effect signal additively. E.g., I
have attached a figure overlaying the predicted blood flow when all responses are
accurate V most are accurate, (red all accurate, blue some inaccurate; i think the
difference would be bigger with a short TR and more inaccurate events). I think while the
influence of incorrect responses may alter the signal in a unitary way, a series of missed
trials early on may effect the value of the signal at later time points in a different
manner.
-Thus I think I should have a blood-flow convolved regressor that models blood flow ONLY
in accurate scans.
I've concluded that having a binary regressor may be valid, but that a condition-accuracy
specific convolved regressor is definitely valid.
Thanks
Christopher
------ Forwarded Message
From: Christopher Benjamin <[log in to unmask]>
Date: Sun, 10 Aug 2008 11:16:16 +0100
To: <[log in to unmask]>
Cc: Christopher Benjamin <[log in to unmask]>
Subject: Modeling response accuracy
Dear SPMers
I have a question re. modelling response accuracy in an efMRI design (I‚m assuming a
canonical HRF) which differs slightly from a number of related previous posts I can find.
I have conditions A & B, which, in some but not all cases, can be subdivided into correct &
incorrect responses. My question is, it more valid to specify a model with ˆ
(1) Four canonical HRF-convolved regressors for accurate and inaccurate trials in each
condition; or
(2) Two canonically-convolved conditions with two non-convolved Œaccuracy‚ regressors?
The former makes more sense to me theoretically but I‚d greatly appreciate the
confirmation.
With thanks
Chris
------ End of Forwarded Message
|