Interesting work, and good energy efficiency, and it is a very long way from mimicking a human synapse.
We have identified about 60 biochemical modifiers of synaptic response thus far, and I suspect the final tally will reach well into the hundreds.
Biological neuronal function is so much more than simple transmission, it is the way in which such transmission is modified that is crucial.
Secondary and tertiary modulation of some of those actions has been observed.
Biological function is hugely complex.
It is not neatly coded and silo’ed systems.
Evolution works on everything at once.
Really weird stuff results from that process.
Some of the early evolutionary algorithm work done with FPGAs (field programmable gate arrays) gave some amazing results, which worked, but in one case involved capacitive couplings between circuits that weren’t even supposed to exist – but did actually.
So yes – interesting power requirements, and still a couple of orders of magnitude away from actually simulating a biological neuron.
I suspect we are still quite a way from understanding all the subtleties of human cognition, and we are making great progress with many of the major principles. I suspect Ray is closer than most others to the key principles, and there is a lot more that will need to be done before real, human like, cognition is possible.