Language and thoughts
Hi all, I'm trying to understand how thoughts are being interpreted by BMI devices. More specifically I'm interested in the effect of language on our thoughts and what that means for direct brain communication via BMI's VS indirect brain communication via traditional interfaces such as a GUI.
To help me bridge the two interface types, I have 3 questions regarding this:
1. If I control a device with a BMI device by thinking "move forward", does it do so on a thought level independently or regardless of my language?
2. If this is the case, does this theoretically mean that a baby who has not learned a language yet, would be able to control this same device?
3. My 3rd and last question is a stretch question that might only dwell in the world of science fiction. Nevertheless, if we had to turn back time and somehow have inherited direct brain communication before we invented language, would our conversations have been richer, deeper and less ambiguous because of this?
Looking forward to hear your thoughts on this.
Cheers,
Steyn from sunny Cape Town
Comments
Steyn, hi. Great questions.
BCIs / BMIs generally do not "read thoughts" as how you might imagine. There are some common BCI paradigms. Here is a list on g.tec's page,
https://www.gtec.at/product/bcisystem/
The reason that the general concept of 'thoughts' are difficult to interpret, is that they activate so many different areas of the brain cortex surface. By using some of the simplifications shown on the page above, the EEG activation patterns are much more easily and reliably recognized.
Regards, William
Great, thanks for the quick response and link William! I'm busy working through it.
Just a follow up questions, is the challenge of interpreting thoughts, technology, i.e. resolution and bandwidth? If so, could we see a future where signals are being interpret on a broader brain level as the technology gets better?
Also, the website mentions the following: "The user is asked to imagine moving either the right or left hand. This produces specific patterns of brain activity in the EEG signal, which an artificial classifier can interpret to detect which hand the user imagined moving."
Does the word "imagine" refers to thoughts, but just a very focused thought level?
One issue with "non-invasive" BCI, is that the signals received (EEG, MEG, fNIRS, fMRI, etc.) -- have a great deal of noise and cross contamination with nearby signal generators. "Invasive" BCI, which opens the skull and places electrodes on the cortex surface, removes some of these limitations, but adds many other serious concerns. The non-invasive tech is definitely getting 'better', but has fundamental limits as to how far it can go; noise will always be present.
With the Motor Imagery, it's best activated when you 'imagine' moving the left or right hand. So 'thoughts' of left or right would not work as well. The reason for this is that the cortex "motor strip" has strong localization, which can be picked up by the EEG.
https://www.google.com/search?q=motor+cortex
Got it. ?
There is another whole area of BCI, called "affective BCI", which attempts to categorize the emotional valence ('affect') or cognitive states of the subject being measured. So this is more of a passive monitoring situation, where the BCI might alert you that your concentration or focus is less than optimal. So the user is not issuing 'commands', but is being kept informed of general brain states. And thus is related to neurofeedback applications. (Neurofeedback itself is another huge research area.)
https://www.google.com/search?as_q=affective&as_sitesearch=openbci.com
Unfortunately for those who want to read words off an EEG, the language areas of the brain are not organized by words, like a dictionary. The EEG mind reading problem is much easier when dealing with motor imagery or visual imagery, because the motor and vision parts of the brain do have an organization that reflects the body's or the visual image's spatial organization.
So if we tried to process the EEG based on just thinking of one phrase versus another, without a corresponding change in motor or visual imagery:
One possible avenue to pursue would be processing motor imagery involving the lips and tongue -- in effect, trying to lip read using EEG. This might be do-able with an implanted micro electrode array, but scalp EEG lacks the required resolution.
These are great Further avenues to explore. Thanks!
@Billh, I've been thinking about this and have a couple of other thoughts that might help place my first message in context:
On a broader level, I see language as compressed thoughts. This often creates a disconnect between people due to the lossy nature of it. Based on this, my current hypotheses is that if Brain Machine Interfaces can connect people on a visual thought level, we might be able to have a deeper connection with each other than what language can allow us. This however is based on the assumption that the visual would be accompanied with emotion.
I understand that the technology might be decades away from this or it might never even realise.
Either way, I find the technology incredibly interesting and has helped me understand the shortcomings of traditional UI's in so many ways.