Dear Doug
Thanks for your suggestion.
For the task, I have no low-level visual or tactile condition. a blank
screen is the baseline for both visual and tactile runs. This is
different from your design.
The big problem I met is: When I compare the visual and tactile
condition, I can't reproduce visual and tactile activation themselves.
For your example, you can get visual or auditory activation by one
sample t test for each of them. Now the comparison between visual and
auditory conditions is wanted. Let's use full factorial design (one
factor with two level for visual and auditory), because it will give
the activation of each condition and the comparison between each
condition. In the full factorial design, it's supposed that, you will
reproduce the visual and auditory activation as those in one sample t
test, which make sure things are OK. Meanwhile, in the full factorial
design, you can get the comparison between visual and auditory
condition. My problem is that, in the full factorial design, I can't
reproduce the visual and tactile activation themselves, and the
comparison results are thus not believable. This issue may be more
close to the spm analysis but not the experiment design.
btw, in your study, when you made the simple visual and auditory
conditions as explicit conditions in glm, you got (visual > auditory),
I guess you mean (complex visual > complex auditory) or (simple visual
> simple auditory). Because complex visual minus simple visual or
complex auditory minus simple auditory are relative difference, which
is supposed not to be impact by whether you model simple visual and
auditory in glm. Am I correct?
best
Xiang
On Tue, Jun 9, 2009 at 10:40 AM, <[log in to unmask]> wrote:
> Dear Xiang,
>
> As I understand, you ran a visual task in one run and a tactile task in
> another. When you analyzed each task (e.g., visual > baseline), you got the
> expected activation, but you did not when you tried to create a contrast
> between tasks (e.g., visual > tactile).
>
> I have also found this to be the case when testing visual and auditory
> tasks. I believe this happens because session differences are removed when
> both tasks are included in the same model, which effectively eliminates task
> differences.
>
> In my case, I use a low-level visual stimulus as a control in the visual
> task and a low-level auditory stimulus as a control in the auditory task,
> each contrasted with a more complex sensory stimulus (complex > simple).
> Removing the continual visual activity during the visual session and the
> continual auditory activity during the auditory session effectively removes
> all significant sensory differences -- even i f I just try to compare
> activity between the more complex sensory stimuli (complex visual > complex
> auditory). If I instead create a task with a sensory stimulus presented in
> each modality (e.g., simple visual & simple auditory), I find the expected
> sensory cortex activation (visual > auditory).
>
> Hope this helps.
>
>
>
> Doug
>
>
>
> On Mon, 08 Jun 2009 3:31:54 pm CDT Xiang Wu wrote:
>
> I actually has a visual task and a tactile task. In situation 1
> (separate one-sample t-test), visual activation for visual task and
> tactile activation for tactile task were got. However, stronger visual
> activation for visual task or stronger tactile activation for tactile
> task was not observed when compare the two task in situation 2 (paired
> t-test), just because there are no visual activation for visual task
> or no tactile activation for tactile task in situation 2. Even though
> there is no problems with spm, how to explain th e above results?
>
> Xiang
>
>
>
> On Mon, Jun 8, 2009 at 1:10 PM, Dasa
> Zeithamova wrote:
>> Perhaps I misread your question, but If I understand it correctly, you are
>> doing two VERY different analyses in Situation 1 and Situation 2, so you
>> shouldn't be surprised that the results are different.
>> In 1, you look which areas are active in cond A, and which are active in
>> B.
>> You get areas where A> baseline, and areas where A>baseline. In 2, you are
>> essentially performing a t-test looking for areas where A>B (i.e. where
>> A-B>0). Neither of A or B has to be above baseline for A>B and both A & B
>> can be above baseline and equal.
>> Is that what you were after?
>>
>> Dasa
>>
>>
>> Xiang Wu wrote:
>>>
>>> Hi all
>>>
>>> Given two conditions A and B.
>>> In situation 1, I perform group analysis one sa mple t-test on A and B
>>> "separately". So I got activations for A and B.
>>>
>>> In situation 2, I put A and B in "one group analysis". I conduct two
>>> sample
>>> t-test, paired t-test, or full-factorial (one factor, two level), the
>>> results were similar. However, The activations for A and B in "one group
>>> analysis" are largely different from those by "separate" one sample
>>> t-test,
>>> not only the significance level, even the activation patterns look
>>> different.
>>> I really confused. Can somebody explain the reason for me? I think there
>>> is
>>> no problem to get activation of each condition with one sample t-test,
>>> but
>>> I
>>> have no idea what happened when put multiple conditions in one group
>>> analysis.
>>> Thanks Xiang
>>>
>>
>
> ===========End of original message text===========
>
> < br>
>
|