Dear Martin,
It seems that fMRI MVPA studies may be divided into two categories based on model specification: trial-wise and run-wise decoding. I was wondering if any studies directly compared these two approaches.
I also found that some trial-wise decoding studies used a design called Least Squares Single (LSS), which is very different from the typical 1st-level model we used for univariate and run-wise multivariate decoding analysis. Does it mean that I have to do n (number of trials) separate 1st-level models? I was wondering if there is an easy way to perform this LSS analysis and subsequent decoding using SPM/Matlab.
Best,
Frodo