MVPA temporal generalisation matrix vs. decoding time

Hi,

As I understand, for the outputs of the SVM classifier, the values of the diagonal from the temporal generalization matrix should be similar to the decoding time course values.

I have been doing some pilot analysis, and they are similar but not identical... (see attached figure). Are the used measures different between the two kinds of outputs?

Perhaps I am conceptually confused here (it is the first time I do this analysis).

Thank you very much in advance.
Best

Elena

plot_diag_vs_pairwise.pdf (272.6 KB)

@pantazis @Sylvain @MartinC
Can you please help with this?