Single LMAN fragments gave 13% mean maximum HVC spine coverage within 20 µm paths in one Area X map

Local models of song learning use HVC and LMAN input on medium spiny neurons (MSNs). We tested how source identity limits the possible reach of a local shaft signal. We reused checked geometry for 186 MSNs in one Area X map, j0251 v8. The data came from one adult male zebra finch. We selected one, two, four, or all recorded LMAN axon fragments on each target. We found the largest share of HVC spine contacts within a stated path radius of each group. At 20 micrometres, mean largest coverage was 13.04%, 22.74%, 35.64%, and 47.32%, in that order. No single fragment reached half coverage. All fragments did so in 82 targets. Of these targets, 55 needed at least five fragments. A later calculation gave median radii for half coverage of 66.88 micrometres for one fragment and 21.30 for all fragments. Higher synapse and source scores and smaller projection offsets retained the source limit. An independent source rebuild matched 493,861 path entries exactly. A clean repeat matched all numerical outputs and both figures. The bound assumes that a signal uses only the stated recorded fragments. It also assumes that the signal cannot travel beyond the stated path radius. Several fragments can belong to one real neuron. Missing contacts can raise real coverage. The results constrain stated local signal models. The package provides a source-fragment distance atlas, code, provenance, results, figures, and repeat instructions.

connectome · Area X · HVC · LMAN · dendrites · credit assignment · source fragments · local signal models · maximum coverage