You offer a confusing reading of the policy. Nowhere does it say that primary determinants should be anything other than relevancy and truth, and it very explicitly suggests that authors simply think a bit harder about distributing citation credit to avoid unthinking rich-get-richer-ism.

Also, it turns out that if you want to "down", you might also need to curve over. The descending projections of upper layer cells and deep layer cells in V1 coherently curve below layer 5. This matters a lot for measuring distance-dependent connectivity.

There were a few other interesting details that I couldn't emphasize but I really like. For example, it turns out there are cells that specifically target the apical dendrites (Iight pink) of deep layer neurons:

Some excitatory subtypes like layer 5 ET cells had particularly strong specificity, with their inhibitory neurons like this basket cell doing virtually nothing but connecting to ET neurons.

The third surprise was that this specificity was pervasive. Individual inhibitory neurons (columns here) not only put most of their synapses onto one or two excitatory clusters, but diverse "motif groups" of different inhibitory neurons had similar output budgets across target clusters.

But did this matter for the circuit at all? It turns out it did! For example, the inhibitory neurons that target layer 2 cells tend to talk to other layer 2 cells and not so much layer 3 and vice versa. Moreover, all layer 5 clustered seemed had quite different sources of inhibition as well.

But that's exactly what we found! And they look very different, with the SST-targeting population being largely bipolar cells, as expected, but the basket-targeting population being wispy bipolar cells. Now we have a new, highly specialized knob controlling cortical computations!

Reconstructing the axons of inhibitory neurons, we could reconstruct a columnar map of inhibition of inhibition. This revealed the first surprise! While it broadly agreed with the literature, in the upper right of this matrix, you see inhibitory neurons that only target basket cells (PeriTC here)

For the first time ever, we could look at many nearly complete neuronal arbors where we didn't just know their shape, but the size and location of all 4 million synaptic inputs. Individual cells had 1000-25000 inputs.

The idea was to take a deep dive into the morphology and connectivity of cells sampled along a cortical column. There were ~1300 cells, 150 inhibitory and 1150 excitatory.
(Render from @quorumetrix.bsky.social)

In my work, I use morphology and synaptic connectivity in datasets like MICrONs to understand what kinds of cells exist in cortex, what rules govern how they connect, and what this network architecture might suggest about how the brain works. (Panel from www.biorxiv.org/content/10.1...)

You can see its thick apical dendrite is totally covered in spines, each of which gets a synapse from an excitatory neuron.

Here's a close-up of its cell body, which is pockmarked with inhibitory inputs from basket cells that inhibit it and help control its activity, perhaps because of the huge amount of input it gets.

Look at this absolute unit of a neuron! This is a layer 5 ET neuron in mouse visual cortex, which gets a crazy amount of synaptic input (15,326 synapses in the dendrites here) and sends its outputs not just locally, but to subcortical areas that can more directly affect behavior.

Neuroglancer, if you don't know, is a web app developed by Google (with a lot more community involvement recently) for exploring very large, cloud hosted 3d image volumes and segmentations.