Dear TDT team,
In my experiment, I have 2 tasks with a similar structure. I assume both tasks rely on similar neural mechanisms. To test this, I would like to do the cross-task decoding (i.e., train a classifier on task 1 and test it on task 2, vice versa). My questions are:
Can I use a cross-classification design to achieve this? In your tutorial paper, it seems that cross-classification was used for the within-task decoding (all sets = 1). Do I have to change the second half of sets to 2 in this case?
I did 2 GLMs to get beta images for the 2 tasks, respectively. Should I copy these beta images to a single folder and create the design manually.
I will also do the within-task decoding with CV. Does it make sense to use cross-classification with CV (rather than a simple cross-classification) for the cross-task decoding to make things consistent?
Thank you very much for your time and help!