Tutorial alpha user feedback: Pyramidal-Interneuron Gamma network

[sbrincat]

General
I really like this tutorial! Love how it builds up step-by-step, and clearly explains what is happening/what should be learned at each
step. Also like the conversational tone.

• Not a high priority (and no offense), but the first figure is not great. Would be good to get Annabel to do a quick model schematic figure.
• Maybe this is a bit nitpicky, but a notable difference from the paper is that the E and I cells become progressively less synched over time. That seems not to be the case in Fig. 1A,C of the paper. Do we think this is important? Even if we know why that’s happening and are ok with it (maybe some simplification from the paper?), it’s worth at least mentioning the reason for the difference in a note after the raster plot.
• Would be nice to add links for Refs 2,3 so users can easily navigate to the papers

Cell 6

• Since all weights are the same, I wonder if we might want to set one of the sets of weights using create_adjacency_edges() (or maybe show it done both ways), in order to make explicit the two ways to create edges in NB?

[agchesebro]

Thanks @sbrincat - super helpful comments! My first thoughts for the comments are:

• No offense taken haha. The art neurons in my brain never fully synapsed ;) I'll ping Annabel to make sure this gets replaced ASAP.
• I'll double check this again. This is a bit simplified from the paper and it's likely due to the tuning of the strength of the inhibitory bath, which is something we plan on including in the IAP course as an exploratory parameter anyway. So it would be nice to mention what's driving this.
• Whoops yes will add the links.
• Good idea - will include a create_adjacency_edges() version in the PR for the other points above.

[sbrincat]

I can ping Annabel. I have some examples from the literature of nicely illustrated PING circuits that she can hopefully synthesize with yours, eg:

[agchesebro]

For the second point: this is indeed an issue with E/I balance essentially. Because I've switched to a completely driven input for excitatory neurons so that we avoid SDEs in a starter tutorial the excitatory drive is a bit higher than in the paper. If we drop the drive from 25% of the neurons and increase the inhibitory bath, we see more of the synchrony (esp. in the inhibitory population):

I'll make a note of this in the tutorial. Messing with the E/I balance is on the list of suggested exercises for the IAP, so I think this will be a good intro to that concept.