Ah, I was under the assumption that 400 microns was implied to be at least within an order of magnitude of the threshold, so for example I assumed the signal transmission worked at at least 40 microns, which is still 1000x the synaptic cleft. If there's no information about where the cut-off is, only some upper limit, perhaps even due to the lack of precision in the technique itself, then this does seem pretty questionable, indeed.