If the brain uses extremely low voltage signals to communicate (from what I understand around 100 mV), what sort of breakthroughs would be necessary to intercept these signals and interpret them as exact thoughts? I know we interface with the brain's electrical field through already at a much higher level, but what is stopping us from being able to interpret it more precisely?
$\begingroup$
$\endgroup$
3
-
1$\begingroup$ No, it is not possible at all. There is no technology whatsoever at present that can "read" a thought. Only a brain can interpret electrical impulses as thought. I can imagine nothing that can even foreshadow this. $\endgroup$– anongoodnurseCommented Feb 24, 2015 at 3:51
-
$\begingroup$ I think it is not possible for one brain to interpret the thought pattern of another brain. $\endgroup$– One FaceCommented Feb 24, 2015 at 5:24
-
3$\begingroup$ Define "thought", please, and "exact". You can pick up EEG signals for creating brain-machine interface and move your PC cursor around, is that good enough? $\endgroup$– aaaaa says reinstate MonicaCommented Feb 24, 2015 at 15:58
Add a comment
|
2 Answers
$\begingroup$
$\endgroup$
2
The brain activity is electric and chemical. The male adult human brain contains about 86 billion neurons (Azevedo et al). There is about 100 trillion connections between them. Solving a puzzle like that is not easy..
what sort of breakthroughs would be necessary to intercept these signals and interpret them as exact thoughts?
What you are referring to would be called solving the neural code in neuroscience. With today's methods, it is probably not possible.
Consider electroencephalography (EEG). It records the electrical activity of the brain. Ag/AgCl sensors are placed along the scalp (typically 64-256 in research settings). When about 50,000 parallel neurons fire simultaneously, a change in a recorded signal can be seen. While the time resolution is about 1 ms for EEG, the spatial resolution is several centimeters (it is not easy to find which areas produced the recorded signals; it is an inverse problem). A newer technique, called magnetoencephalography (MEG), is becoming more and more used but is expensive. It records the magnetic fields related to the electrical activity and allows better localization.
Is there other complications? Well... lot's of them. Blink your eyes during the recording and there will be a major artefact masking the brain signals in most channels (10-100x greater in amplitude than the brain signals).
Anyway, you would probably be interested in the new research involving the use of machine learning techniques: it has been possible to decode the contents of dreams, reconstruct what a subject is seeing, etc. The best papers have been published by Science, Nature, and PNAS, so using their search engines with the keywords decoding and brain should allow you to explore the topic easily.
PS. I did not have time discuss fMRI; someone else can may be do that..
-
$\begingroup$ I disagree to some extent. If you read "thought" as "intention to perform a motor action", then BrainGate (braingate2.org) has been reasonably successful. $\endgroup$– kmmCommented Feb 24, 2015 at 14:04
-
$\begingroup$ Very thorough! That is precisely what I was referring to when I mentioned, "interpret as specific thoughts". In essence, understand the function the brain is performing as specifically as possible. This is part of what fascinates me about machine learning. I will take a look at those sources! I am curious as to how fMRI works and how accurate it is compared to other methods. I have read articles that employed this method. $\endgroup$ Commented Feb 24, 2015 at 20:44
$\begingroup$
$\endgroup$
Possibly but not for a very long time.
Professor Gallant is in the embryonic stages of researching this and is having some success. Right now the computer needs a pool of possible images to draw from to make the possibilities small enough to search and the image resolution is rough at best. But the fact that even something as simple as color of a remembered image can be read means it is possible. But that is space shuttle level technology while we are still at the stage of strapping cardboard wings to our arms.
In all likelihood even when this technology is mature the machine will have to learn your brain similar to how a speak to text program has to learn your speaking patterns, before it works. To me this is hugely interesting for the possibility of seeing what parts of our brain architecture is universal and thus likely genetic.
I am including two other sources for the research but I warn you they are behind pay wall. http://www.nature.com/scientificamericanmind/journal/v25/n6/full/scientificamericanmind1114-40.html
http://www.sciencedirect.com/science/article/pii/S1053811910009109
