One of the most-discussed memristor characteristic is its synaptic biomimesis. “State-of-the-art computers have difficulty mimicking the operation of the brain,” [NCSU Professor] Dickey notes. “Memristors, on the other hand, are effective at mimicking synapses. If you were interested in only mimicking brain function, then solid-state memristors would be more practical because they contain many more memory elements and are much more optimized at this point. One of the things distinguishing our work is that the device behaves like a memristor and has other properties similar to the brain. Conventional electronics tend to be rigid, 2-D, moisture-intolerant, and operate using electrons; the brain, in contrast, is soft, 3-D, wet, and operates using ions and in addition to adopting many of these properties, our device is composed of biocompatible hydrogels.” _Physorg
Prof. Orin Velev, Prof. Michael Dickey, and graduate students Hyung-Jun Koo and Ju-Hee So, have devised a new class of easily fabricated memristors based entirely on so-called soft matter – hydrogels doped with polyelectrolytes sandwiched with liquid metal electrodes – that operate using ionic conductance in aqueous systems rather than conventional electron transport.They are still in the very early stages, but the possibility of an implantable soft, biocompatible brain augment is too important to overlook.
...In essence, this suggests that in addition to having the potential to realize memristor-based neuromorphic structures, the polysaccharide hydrogel core of these devices is biocompatible, could possibly be interfaced with live neural and other tissue, and could lead to three-dimensional soft circuits and their in vivo operations.
...Going forward, Dickey continues, “We hope to take advantage of the fact the water-based gels in the device are biocompatible, and could in principle be integrated with biological species, such as cells, enzymes, proteins, and tissues. We also made no attempt to optimize the memory capacity in our prototypes, which is an area for improvement. Finally, we’re working to understand the subtle aspects of the operating mechanism.” _PO
Quite a few different interfacing techniques could be used, but the optical approach would seem to be the least intrusive for tissues such as the brain, which are sensitive to electromaqnetic forces. Optical materials have high bandwidth and may be less likely to be bio-rejected than electrically conductive materials. Some people have discussed optical brain control in the context of optogenetics.
Another fascinating type of bio-to-machine interface is the piezoelectric interface being developed at Georgia Tech. The piezoelectric interface can be operated by exquisitely subtle mechanical movements, such as a muscle fibre twitch. In other words, a thought -- even a subconscious though -- could cause a pattern of muscle twitches which would activate a particular machine command or subroutine via the piezoelectric interface.
The human brain was not evolved for the ultra-long lifetime of a next level human. Cell debris accumulates, DNA repair mechanisms begin to fail, immune systems weaken, hormonal support falls off, etc. Scientists are learning a lot about how normal aging leads to memory loss in even the sharpest minded senior citizens. The intricate network of cellular connections in the brain slowly loses definition and resolving power.
Well-designed brain implants could sense this process occurring and engineer work-arounds to compensate for the changes. Long term solutions would require a rejuvenation treatment to restore -- or improve -- the resolving power of brain networks, but sometimes work-arounds are the best one can do at the time.
Where would you place your soft bio-compatible brain implant? There isn't a lot of room inside the skull itself, but implants could be placed under the scalp in a relatively unobtrusive manner as long as they were not too large. Alternatively, some women might choose to place their augments in the breast area, and some men might choose augments shaped to serve as muscle implants. If the connections to the interface are via optical fibre, the distance from anywhere on the human body to the brain is negligible, in terms of the speed of light. The interface itself would need to be placed close to the brain.
Depending upon its sophistication, an implanted brain augment could come to know how an individual's brain works quite well, over a period of time. Such augments could even learn how to simulate their hosts in a rudimentary way. The possibilities arising from such pseudo-emulation are worth considering, but perhaps not here and now. (See Old Man's War by John Scalzi)
It is important to stress that these NCSU memristors are not at all close to anything that could be used as a brain augment. But it seems to be the goal of the researchers there to develop biocompatible sensors and intelligent interfaces using these materials. It is not a long stretch from there to an implantable computer augmentation for the brain.
Although memristors are often referred to as neuromimetic or synaptomimetic, in the aggregate, memristor computing devices will function nothing like the brain. But they will not need to. They will only need to function like competent and clever computers that provide reliable memory and I/O capability for mental computations, speculations, and interfacing with the outside world -- including the ability to control machines mentally and to communicate remotely with machines and other individuals who have similar augments.
No comments:
Post a Comment