[pymvpa] Help with analysis design

[J.A. Etzel]

Are all of your blocks structured like this, with the visual cue 
immediately preceding the somatosensory stimulus? If so, a cross-modal 
type analysis (e.g. training a classifier to distinguish categories of 
visual cue, then testing it with somatosensory stimuli) is probably not 
possible: your events are just too close together. If there's a lot of 
jittering in the within-block timing and enough examples you might be 
able to set up some sort of FIR model.

Or, can you analyze your dataset as full blocks? For example, block 
types could be visual-A-somatosensory-A, visual-B-somatosensory-A, 
visual-B-somatosensory-B, etc. Then, you could classify between these 
block-level types, like visual-A-somatosensory-A vs 
visual-B-somatosensory-A.

good luck,
Jo



On 3/10/2015 4:40 PM, Raúl Hernández wrote:
> Hi, I'm having problems thinking about how to analyze my data and I was
> wondering if any of you could give me some assistance.
>
> I want to determine if the activity generated from a visual cue is the
> same as the activity generated by a somatosensory stimulus. So I design
> a discrimination task in which the participant receives a visual cue,
> then touches the stimulus and presses a button to determine if the cue
> matched with the somatosensory stimulus.
>
> This is the design of a single block (in seconds):
>
> 0-12 baseline
> 12-16 visual cue
> 16-20 stimulus
> 20-22 response
>
> Now, I have 4 categories of stimulus. I want to analyze the activity to
> know whether there is a pattern that can be associated with each
> stimulus category. I have 4 different visual cues that should elicitate
> the same pattern as the actual stimulus. I need to categorize each event
> in order to proceed with the analysis but if I assign the same 4
> categories to the visual cues then I'm already biasing the data. If I
> divide the data to train the classifier with the stimulus and then test
> it with the visual cues I think that due to the closeness of the events,
> the activity related to the stimulus might be biased due to the previous
> cue.
>
> I would really appreciate any suggestions
>
>
>
> _______________________________________________
> Pkg-ExpPsy-PyMVPA mailing list
> Pkg-ExpPsy-PyMVPA at lists.alioth.debian.org
> http://lists.alioth.debian.org/cgi-bin/mailman/listinfo/pkg-exppsy-pymvpa
>

-- 
Joset A. Etzel, Ph.D.
Research Analyst
Cognitive Control & Psychopathology Lab
Washington University in St. Louis
http://mvpa.blogspot.com/