Consistency in nomenclature
vborghe opened this issue · 5 comments
[another nitpicking one] Small mismatches in naming convention make life harder when computing, for instance, behavioral performance: in social task, trial_types are Mental and Random (capital letter) vs. mental and random (all lower case) in response
thanks for catching this! I agree it's not the most glamorous revision, but it does increase the ease of re-use of the dataset, and this is the kind of things we want to catch while we're still in beta.
Not to be annoying but just trying to think head, I wonder if deciding that no capital letter should be used would help.
Right now it seems a bit arbitrary, for instance in task_relational trial_type has the main two capitalize [Relational/Control] while the other not [countdown, new_bloc]
Similarly, in some tasks the number of trials is stored under ntrial in other in trial
I like the idea of sticking to lower case only. That's what we've done for our dataset names (movie10 and not Movie10 for example).
I also like to start with n for number of things, and stick to singular. So ntrial and not trial or ntrials. If underscores are allowed, I would vote for n_trial
I'll do another pass of all events files (and relative json description file) but looks great!