Picking a particular channel is currently relatively complex for new users.

It either involves pythonic tricks:

picks = [ii for ii, ch in enumerate(epochs.ch_names) if ch == 'mychan']
epochs.plot(picks=picks)

Or master of MNE's functions:

picks = mne.pick_channels(evoked.info. [`my_chan`])
epochs.plot(picks=picks)

Proposal 1

👍 0
Since picks is always referring to a list of int and ch_names is always a list of str, I propose that we accept picks ot be a list of str:

evoked.plot(picks=list(range(10)))
epochs.plot(picks=['my_chan'])

Proposal 2

👍 0

Accepting the parameter picks to be a list of int (for indexing), a list of str (to find channel names) or a dict (to quickly select channel types).

evoked.plot(picks=list(range(10)))
evoked.plot(picks=['MEG 1132', 'MEG 1022'])
evoked.plot(picks=dict(grad=True, eeg=False))

• advantage: no deprecation, simplification of channel picking
• disadvantage: keep clunky pick_type API

Proposal 3

👍 0

Same as proposal 2, but change pick_types to accept a types key that would accept list of str. This would allow

evoked.plot(picks=list(range(10)))
evoked.plot(picks=['MEG 1132', 'MEG 1022'])
evoked.plot(picks=dict(grad=True, eeg=False))
evoked.plot(picks=dict(types=('eeg', 'mag')))
mne.pick_types(info, types=('eeg', 'mag'))

• advantage: easier to loop across channel types, no deprecation
• disadvantage: more than one way to do pick_types

Proposal 4:

👍 4 (@larsoner, @kingjr, @nbara, @choldgraf)

Same as proposal 1, but automatically detect if str refer to ch_names of ch_types.

evoked.plot(picks=list(range(10)))
evoked.plot(picks=['MEG 1132', 'MEG 1022'])
evoked.plot(picks=('eeg', 'meg'))

• inconvenient: risk of conflict between channel names and channel types
• advantage: easier to loop across channel types