[DOC] Add Frites section for information theoretic connectivity

[adam2392]

PR Description

First step for: #57

cc: @EtienneCmb

this docs PR should go hand-in-hand presumably with a minor docs PR I'll make in Frites.

After looking at the API of Frites more, it seems like the focus of Frites is primarily "information-theoretic connectivity analysis", so for now, perhaps we make mne-connectivity "standard/basic connectivity analysis"?

Merge checklist

Maintainer, please confirm the following before merging:

• All comments resolved
• This is not your own PR
• All CIs are happy
• PR title starts with [MRG]
• whats_new.rst is updated
• PR description includes phrase "closes <#issue-number>"

[EtienneCmb]

Hi @adam2392,

After looking at the API of Frites more, it seems like the focus of Frites is primarily "information-theoretic connectivity analysis", so for now, perhaps we make mne-connectivity "standard/basic connectivity analysis"?

Partly yes. The big picture of Frites is to be able to put in relation brain data at different scales like local activity (e.g. mne.Epochs) or global scale (functional connectivity) with an external variable (stimulus types / behavioral models etc.). To achieve this goal with connectivity metrics, we have two types of approaches :

1. A two-levels analyses : first we estimate the FC across time points and at the single-trial level (e.g. frites.conn.conn_dfc, frites.conn.conn_spec). In a second level, we estimate the information shared between the FC and the external variable (using the frites.workflow.WfMi). See this example
2. We are also testing methods to directly asses feature specific comodulations or feature specific information transfer (not in Frites for the moment)