[pymvpa] [mvpa-toolbox] Fast ER design (Dr. Aguirre please help ; -) )

[Yaroslav Halchenko]

Thank you!

The importance of "counter-balance" is what caused me to mention de
Bruijn sequence and your paper.  But what about the choice of ISI given
the contrast(s) and possibly pre-set TR:  am I correct that there is no
other way to figure out an optimal stimuli duration and ISI  given
the constraint of number of categories and run-time, besides exploring a
set of choices and selecting the one with best "efficiency"/"detection
power" ? 

On Tue, 12 Jul 2011, Geoffrey K Aguirre wrote:

> Folks --

> 	As I understand it, the question regards the best "rapid" event-related design for use with MVPA.

> 	For an MVPA study, the goal is to first measure the response of each of several stimuli or conditions versus a putative baseline condition. The activation strengths for these stimuli or conditions for each voxel are then entered into an MVPA analysis. An important design consideration is that it is desirable to have the stimuli "counter-balanced", meaning that each stimulus type precedes and follows each other type. This is desirable as it prevents carry-over effects (the influence of one stimulus upon another) from being mistaken for direct effects (the response to the stimulus itself).

> 	The ideal sequence for such an application is provided by a specific type of de Bruijn sequence, called a "type 1, index 1" sequence (REF 1). We have a web page on our wiki that describes the sequences, and provides a web-based tool for generating them:

> 		http://www.cfn.upenn.edu/aguirre/wiki/public:t1i1_sequences

> 	T1I1 sequences are ideal for this application as they maximize the "main effect" of stimuli versus the null-condition (REF 2), and provide nearly identical estimation power for all of the stimuli in the experiment.

> 	Hope this does the trick,

> Geoff

> REFs

> 1) GK Aguirre (2007) Continuous carry-over designs for fMRI. Neuroimage, 35, 1480-1494.
> 2) TABLE 1, GK Aguirre, MG Mattar, L Magis-Weinberg. (2011) de Bruijn cycles for neural decoding. NeuroImage 56: 1293-1300
-- 
=------------------------------------------------------------------=
Keep in touch                                     www.onerussian.com
Yaroslav Halchenko                 www.ohloh.net/accounts/yarikoptic