From the article:
"Then, the monkeys' brains were wired together [...]"
So that doesn't tell us shit. On to the paper:
"Electrophysiological recordings
A Multineuronal Acquisition Processor (64 channels, Plexon Inc, Dallas, TX) was used to record neuronal spikes, as previously described15. Briefly, differentiated neural signals were amplified (20000–32,000×) and digitized at 40kHz. Up to four single neurons per recording channel were sorted online (Sort client 2002, Plexon inc, Dallas, TX).
Intracortical electrical microstimulation
Intracortical electrical microstimulation cues were generated by an electrical microstimulator (Master 8 , AMPI, Jerusalem, Israel) controlled by custom Matlab script (Nattick, USA) receiving information from a Plexon system over the internet. Patterns of 8–20 (bipolar, biphasic, charge balanced; 200sec) pulses at 20–120Hz were delivered to S1. Current intensity varied from 10–100A."
So, we're talking about roughly a maximum of 64 * 4 = 256 neurons (at 40KHz) participating per brain. It's not that many, but also not few for an artificial neural network. Because that's what happened. The researcher trained the mice (via reinforcement learning) on specific problems after interconnection. He didn't interconnect them and immediately let them perform some random complex task:
"In one test, for instance, different rats brains were given different barometric pressure and temperature information, and then the computational power of the Brainet itself was used to calculate the probability that it would rain (given those inputs) at a rate higher than chance.
Nicolelis said that, essentially, he created a "classic artificial neural network using brains." In that sense, it's not artificial at all."