Ellis starting the day off by reading out Carl Sandburg's poem "Chicago" in its entirety.

I'm really curious what factors drive var(nTc). Both from basic perspective, but also from practical -- of someone that wants to put these dynamics under tighter experimental control / to have more certainty about what's happening when :)

It is so interesting that the nTc is so variable across trials.

But this is also sobering: clearly, time-aligned trial averaging has *deep* flaws -- yet so much of human cog neuro relies on it, due to unavailability of high-SNR recordings...

You sound like an all-American to me !

post-script: "all-american upbringing" seems to preclude this usage

i think it could have a variety of connotations (see other comments), but one dominant usage is in reference to being competitive in athletics or academics on national scale, a la en.wikipedia.org/wiki/All-Ame...

…Btw I do not mean to suggest there is no worth to mixed-selective perspectives, or that they are wrong! Personally I have just struggled when thinking about how exactly to provide strong positive evidence for them. And curious to hear your perspective.

You can’t be, until you find one :) Its existence is not guaranteed. The same I do not think is true for the mixed basis case. IMO, this is what often makes it compelling when an interpretable factorized basis is identified.

What appears mixed and non-linear from one basis could be sparse & factorized from another. When you see mixed selectivity, how can you be confident that your basis isn't just wrong?
doi.org/10.1038/s415...

What is this, a fitness test for Presidents?

Fully agree. Our story feels incomplete to me without those things, but we simply had to get it out the door. But, I think Todd & co. are working on those follow-ups now!

of course, don't mean to imply that one method is better or more sensitive in general than the other (that depends on exp design & theoretical questions). just that they can be sensitive to different neural features.

:) Borne mostly from reflecting on reactions from family/friends ("you're STILL in school/a student?!") --- yes, aren't we all?

(...but correct me if i'm misremembering)
So the decoders we used didn't require the neural signatures to be aligned across subjects, but only across sessions w/in subject. Seems likely that at finer spatial scales, multi-subj decoders would lose sensitivity more quickly than subject-specific ones.

hey thanks for mentioning David! Interesting ideas in this thread. Yes we were able to identify relatively reliable signatures at the region level. One potentially relevant difference btw our studies: our decoders were subject-specific, whereas yours were fitted on data from multiple subjs, IIRC.