It's pretty much what Maxwell was said to have done.
Trying to create taxonomies is often worthwhile. You find patterns, holes and other interesting things when you succeed and learn more about what you don't know when you fail.
Last year's news, but still has to sink in. Read it, but beware exaggerations like "electronics had no rigorous mathematical foundation", which would set quite a few physicists and engineers spinning in their graves.
What Leon Chua rightly noted, and described, was a missing fourth passive electrical (not electronic) two-terminal component. The difference is that electronic is usually associated with so-called active devices. Those have three or more terminals (e.g. vacuum valves or semiconductor transistors), and the third terminal controls the electron flow between the others.
Given enough active devices, you can mock up practically anything. For example, with capacitors and transistors (and a power supply), you can make an imitation inductor - with a limited range, but good enough for an integrated circuit when the real thing was hard to build.
The great novelty with memristors is finding a real physical basis to provide the desired behavior in a simple, manufacturable two-terminal device.
P.S. The A.I./neural-network angle seems to originate with the device's creator (Stan Williams). With all due respect, I doubt that it will have all that much impact from that side.
Improvements in hardware are a necessary condition for AI. They are not sufficient. The other necessary conditions are algorithms (correct and efficient). The latter is much much more difficult. That is the biggest bottleneck in AI not hardware.
It never ceases to amaze me how we can be so certain that we have the best of something (Neural networks) for a particular job (simulating neural activity) and then something completely different comes along and blows it away (Memristors).
These are the sorts of things that give me hope for our future...
I think I missed something. What is the mechanism that causes the slime/memristor to keep changing states after the external stimulus is removed? I thought the memristor just remembers a voltage. After you remove the external voltage then wouldn't the voltage across the memristor just remember the state it was in when the last external voltage was applied? Likewise for the slime and air.
The research paper that the newscientist article is referring to ( http://arxiv.org/pdf/0810.4179v2 ) points to some resonance behavior of memristors' "LC contour". I don't understand memristors very well, but it seems that the only connection between memristors and the amoeba's "predictive" behavior is resonance, if there is a connection at all.
While memristors are cool and may in fact help us understand / mimic the learning ability of animals, the amoeba's ability to "predict" external forces is a weak, weak link.
Ah. I misread the NewScientist article. It wasn't clear to me that in the experiment to mimic the slime they, in fact, modeled the slime with a circuit that _included_ a memristor and not with just a memristor. The paper the article cites also made things much clearer: http://fr.arxiv.org/abs/0810.4179
