Dear Muka:
> Per your discussion with Dr. Thomas Stephen on the issue of deciphering
> early and late response (mean and exponential decay), I hereby would
> like to inquire the same thing. We are interested in modeling the early
> and late responses within an epoch (block design). Questions as
> following:
>
> 1. To find out "early" region: "1" for decay term (presumably the 2nd
> basis function tailoring to expoential decay term) and "0" elsewhere ,
> right? comforming to your reply to Thomas. Can you be more specific
> about the "early"? Does the term "early response" coin the situation of
> "peaking" in the early half of the block?
Yes, essentially at the start of the epoch. Regions declared
signficant with a test of the decay term are those showing within-epoch
adaptation or, equivalently, a greater response at the beginning
relative to the end.
> Not certain whether we actually identify regions with biphasic response
> (early=increase, late=decrease) according to the expoential template
> intead of merely early positive and late flat behaviour.
You will identify both. The difference between these situations
depends on the contribution of the box-car or mean component (small in
the first instance and positive in the second). To disambiguate simply
plot 'all effects of interest' or mask the decay (inclusively or
exclusively) with the mean effect.
> 2. Following point 1, how should we locate regions of late response? Is
> it akin to extract regions of early deativation areas?
Yes.
> 3. If we make the contrast of the decay term with '-1' and input '0'
> elsewhere, what resopnse type would turn out? Furthermore, what kind of
> response will appear if we put "1" for both basis functions? would it
> become "monophasic early"?
A contrast of -1 0 will give you regions with an increasing response
over the epoch (or early deactivations which adapt). A contrast of 1 1
is simply the sum of both coeficients and would not be easily
interpreted.
> 4. Theoretically, there are various possibilities of cerebral reaction,
> i.e., monophasic early, monophasic late, biphasic early, biphasic late,
> sustained increased, sustained decrease, and so forth. My question is:
> can we utilize the "mean and expoential decay" functionality to model
> the aforemenitoned varieties of behaviour?
By performing conjunction analyses between the mean and decay in various
positive and negative combinations one can parse regions showing
different sorts of responses into different SPMs.
I hope this helps - Karl
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
|