... I'm too looking for studies which could help me making a Full-dive device... I hope (believe) that one day we will invent that device.
A lot of science and engineering literature can be found online by searching for “brain-computer interfaces”
. Much of it is pretty technical, and all takes some effort to digest, but it’s a lot of fun to just surf/wade through. As somebody who’s been interested in the subject since he was a teenager (in the 1970s, when BCIs were a distant dream, almost wholly in the realm of speculative fiction), I make a habit of following its literature, though I’m doing a lot of catching up, and revising many of my assumption, since this thread started.
The best general summary of the subject I’ve found in a professional journal are the introduction and “review of state-or-the-art systems” in IEEE Signal Processing Jan 2008, which can be read here
. My July 2014 post
in this thread is pretty good for an amateur, I think, and rare in that it directly compares real and fictional examples.
Here’s my partial synopsis of the current state-of-the-art in brain-computer interfaces, and speculation about its likely near future.
BCIs can be categorized on a few main axes:
- Invasive (fictional example: The Matrix) vs. non-invasive (Sword Art Online);
- Brain->computer (“read only”) vs. computer->brain (“write only”) vs. brain<->computer (bidirectional, “read/write”) (most fictional example are of fully bidirectional r/w interfaces);
- Realtime vs(most fictional example are realtime) vs. non-realtime.
The major difference between invasive and non-invasive “reading” approaches is resolution. Invasive systems consisting of arrays of surgically implanted electrodes are able to narrow down the source of a given signal to a smaller collection of neurons, in the ideal case individual ones, while non-invasive systems such as EEG and fMRI can narrow it down only to collection larger than about 1000000 neurons. This isn’t because electrodes actually touch individual neurons, which are many times smaller than present-day electrodes. Whether with implanted or surface attached EEG, sophisticated algorithms taking signals from many electrodes and often run much slower than realtime, on recorded data, are needed to achieve maximum resolution.
The difference in resolution between invasive and non-invasive “writing” is even greater, so much so that at present, only implanted electrodes can truly be said to write to neurons. Worse, the resolution improvements made by signal-processing algorithms is essentially possible only in the “reading” case – you can in principle improve he signal-to-noise ratio of brain output signals by using better algorithms and faster computers, but can’t take similar approaches to make the brain do the same with noisy input. The brain’s “hardware and software” sophisticated as they already are can’t be “upgraded”.
So 150 days after my July 2014 post, I still think the only way we’ll be able to create a real SAO-style BCI is invasively. I’m optimistic that improvements in signal processing technology may be able to much improve non-intrusive read-only BCIs, but don’t think such interfaces will be able to do much more than trigger simulated-world actions in the same way the button controls on a present day video game system trigger game actions, such as in a typical fighting game like a SAO PS Vita game
from Bandai Namco. Given that our brains are pretty good at interfacing via fingers and buttons well enough for this kind of game mechanics, I’m skeptical if a BCI would seem to gamers to be an improvement or even as good, much as we’ve found with motion-sensing interfaces like the Wii remote
and the PS Move
.This doesn’t mean I’m abandoning the dream of a deep dive BCI. I just think such a system will have to be an invasive one, and a nanotech one.
Invasive BCIs are creepily unattractive to most people now, I think, because of the current state of the art of invasive electrodes, which requires a surgical team to cut open your head and implant them. My hope rests on the vision that electrodes can be made much smaller (about 100 nm, 10-7
m), approaching or even smaller than individual neurons. At such scale, they could pass though tissue, including bone, without the subject noticing, and, while technically still invasive, practically be less noticeable than the magnetic and radio-frequency signals the makers of SOA imagined for the fictional NerveGear.
I elaborated on the idea, in the context of its use in medicine, rather than for CBIs, and inserted though blood vessels rather than directly though the scalp, in this 2005 post
. From it:
Now, imagine that rather than being just a featureless fibre mechanically pushed, pulled, and jiggled with the aid of external radiological imaging, each fibre ends in a “nanobot” with 10 nanometer (10-8) features including manipulators, propellers, touch sensors, light emitters and sensors – all the features of a free-swimming machine, but without the ultra-miniature powerplant, brain/memory, communication system, and waste heat radiator/exchanger a free swimming design requires, and, rather that being at the whims of Brownian motion, being anchored to the end of what, at that scale, amounts to a massive, near stationary cylinder. The cylinder provides course, push/pull movement, but the manipulator and propeller features handle fine guidance. In addition to emitting light for its own sensors, this “nano-head” can send minute blips of high-energy (1e15 – 1e18 hz) radiation to provide high-resolution position data to external sensors, include an identifying data tag. The fibre supplies power, removes heat, and streams data both ways. On the end opposite its nano-head sits all the information processing power the macroscopic scale world can provide. In between, movement of the fibre can be facilitated by specialized clusters of features similar to the head.
The technology for long, thin wires like has existed in prototype for 10+ years. A proof of concept of electrodes this thin was done in 2005 (see NFS press release Wiring the Brain at the Nanoscale
).In the fictional imagery of Sword Art Online, what I’m imagining would be a helmet like the NerveGear, but when you turn it on, it doesn’t shoot EM radiation into your brain, but many invisibly thin wires.