FacesAomic selection of main contrasts

[ferponcem]

I would like to pick different contrast as main_contrasts for the FacesAomic task, see list.
For instances, instead of using anger-neutral, I would prefer to use anger alone, and so on for competent, joy and pride.
I think the subtraction of the neutral condition takes away a lot of information about the faces effect.
Let me know what you think @bthirion

[bthirion]

I don't like the idea: looking at an angry face vs baseline is not specific enough. You need to contrast it against a more relevant baseline. Maybe we could check individual data a bit more in detail.

[ferponcem]

I have launched a quick test to compare the maps. Since these are very sparse maps I am using glass brain visualization, please see here. In that folder you can browse through different subjects.
I am just thinking that the joy-neutral contrasts look to me like noise, while for most of subjects the joy and pride maps look a bit more like signal.
For now I have just generated for joy and pride, let me know if it's informative so I can also push contempt and anger.

[bthirion]

Maybe joy does ot work well. Do we have contrasts that group together positive/negative emotions ? They may be more powerful.

[ferponcem]

We don't 😞 we only have female-male and mediterranean-european (and their reverse). And actually those are very bad, there's something fishy because those two contrasts look exactly the same, take a look here. I checked and this is consistent with the snapshots we have stored with the data, see:

• /ibc/smooth_derivatives/sub-04/ses-44/res_task-FacesAomic_space-MNI152_dir-ffx/stat_maps/female-male.png
• /ibc/smooth_derivatives/sub-04/ses-44/res_task-FacesAomic_space-MNI152_dir-ffx/stat_maps/mediterranean-european.png

Maybe we should have a look again at the contrast definition and create something like positive-negative instead?

[ferponcem]

Commits 8a4302e and 60d35d5 added three contrasts: positive-negative, positive-neutral and negative-neutral 🚀
I still see that there might be a mistake on the definition of the european contrast, see here.

[bthirion]

Yes, can you fix it ?

[ferponcem]

Fixed in #85