Brain-Computer Interface: Difference between revisions
Caseorganic (talk | contribs) Created page with '=== Subvocal Recognition (SVR) === The Ambient Corporation is marketing a device called the Audeo that helps people speak [http://www.theaudeo.com/]. On February 26, 2008 the A…' |
Caseorganic (talk | contribs) No edit summary |
||
| (20 intermediate revisions by the same user not shown) | |||
| Line 1: | Line 1: | ||
=== | ===Definition=== | ||
A brain-computer interface (BCI) is a system that establishes a direct communication pathway between the brain and an external device, bypassing the ordinary output channels of nerve and muscle. A system that only sends information the other way — stimulating the brain without reading it — is more precisely a neural interface or neuromodulation device. | |||
The | The term was coined by Jacques Vidal at UCLA in 1973.<ref>Vidal, Jacques J. "Toward Direct Brain-Computer Communication." ''Annual Review of Biophysics and Bioengineering'' 2 (1973): 157–180.</ref> | ||
===History=== | |||
Hans Berger recorded the first human electroencephalogram in 1924 and identified the alpha rhythm, establishing that electrical activity of the brain could be measured from the scalp. Vidal's 1973 paper proposed using such signals as a control channel and demonstrated a system in the late 1970s. Clinical work accelerated from the 1990s: the Utah array, a silicon microelectrode bed implanted in cortex, became the standard research device, and the BrainGate trials from 2004 onward demonstrated cursor control, robotic arm control and, later, rapid communication by attempted handwriting or speech in people with tetraplegia and locked-in syndrome. | |||
== | ===Approaches=== | ||
* '''Non-invasive.''' EEG from the scalp; also fNIRS and MEG. Safe, inexpensive and portable, but the skull blurs and attenuates the signal, so bandwidth is low and setup is fussy. Most consumer "mind-reading" headsets are EEG devices with a small number of dry electrodes, and their claims should be read with the signal-to-noise problem in mind. | |||
* '''Partially invasive.''' Electrocorticography, with electrodes on the cortical surface beneath the skull but outside the brain tissue. A substantial improvement in signal quality with less tissue damage than penetrating arrays. | |||
* '''Invasive.''' Penetrating microelectrode arrays recording from individual neurons or small populations. The highest bandwidth, and the only approach that has produced fine motor control — at the cost of surgery and of gradual signal degradation. | |||
The | ===Limiting Problems=== | ||
The obstacles are biological rather than computational. | |||
'''Foreign-body response.''' Implanted electrodes provoke inflammation and glial scarring, which insulates them from the neurons they are meant to record. Signal quality declines over months to years, and long-term stability remains the central unsolved problem. | |||
'''Bandwidth.''' Even the best current systems read from a tiny fraction of the cortex. The gap between what is recorded and what a thought consists of is enormous, and popular expectations run far ahead of it. | |||
'''Non-stationarity.''' The mapping between neural activity and intention shifts as tissue changes and as the user adapts, so systems require recalibration — and the user is learning at the same time as the decoder, which makes the two a coupled system rather than a device with a user. | |||
===Applications and Ethics=== | |||
Established clinical uses are restorative: communication and movement for people with severe paralysis, cochlear implants (a neural interface, though not a BCI in the strict sense), and deep brain stimulation for Parkinson's disease, which is now routine and entirely one-directional. Speculative and commercial ambitions — consumer control, augmentation, memory enhancement — are considerably less mature than their publicity. | |||
Neural data is not like other personal data: it may contain information the subject has not consciously accessed. Closed-loop stimulation systems raise questions about agency, since a device that modulates mood or movement in response to detected states is participating in decisions. And a person whose speech or movement depends on a proprietary decoder is dependent on a company's continued existence — see [[Implant|implant]] on the several documented cases of abandoned neural implants. | |||
===Relevance to Cyborg Anthropology=== | |||
The BCI is the limiting case of the [[Cyborg|cyborg]] premise: integration at the level of the nervous system itself, where the boundary between the person's intention and the machine's action becomes genuinely hard to locate. | |||
===Related Reading=== | |||
* [[Neurological Computing]] | |||
* [[Implant]] | |||
* [[Cyborg]] | |||
* [[James Fung]] | |||
* [[Chris Gray]] | |||
===References=== | |||
<references/> | |||
[[Category:Cyborg Anthropology]] | |||
[[Category:Wearable Computing]] | |||
Latest revision as of 09:38, 26 August 2026
Definition
A brain-computer interface (BCI) is a system that establishes a direct communication pathway between the brain and an external device, bypassing the ordinary output channels of nerve and muscle. A system that only sends information the other way — stimulating the brain without reading it — is more precisely a neural interface or neuromodulation device.
The term was coined by Jacques Vidal at UCLA in 1973.[1]
History
Hans Berger recorded the first human electroencephalogram in 1924 and identified the alpha rhythm, establishing that electrical activity of the brain could be measured from the scalp. Vidal's 1973 paper proposed using such signals as a control channel and demonstrated a system in the late 1970s. Clinical work accelerated from the 1990s: the Utah array, a silicon microelectrode bed implanted in cortex, became the standard research device, and the BrainGate trials from 2004 onward demonstrated cursor control, robotic arm control and, later, rapid communication by attempted handwriting or speech in people with tetraplegia and locked-in syndrome.
Approaches
- Non-invasive. EEG from the scalp; also fNIRS and MEG. Safe, inexpensive and portable, but the skull blurs and attenuates the signal, so bandwidth is low and setup is fussy. Most consumer "mind-reading" headsets are EEG devices with a small number of dry electrodes, and their claims should be read with the signal-to-noise problem in mind.
- Partially invasive. Electrocorticography, with electrodes on the cortical surface beneath the skull but outside the brain tissue. A substantial improvement in signal quality with less tissue damage than penetrating arrays.
- Invasive. Penetrating microelectrode arrays recording from individual neurons or small populations. The highest bandwidth, and the only approach that has produced fine motor control — at the cost of surgery and of gradual signal degradation.
Limiting Problems
The obstacles are biological rather than computational.
Foreign-body response. Implanted electrodes provoke inflammation and glial scarring, which insulates them from the neurons they are meant to record. Signal quality declines over months to years, and long-term stability remains the central unsolved problem.
Bandwidth. Even the best current systems read from a tiny fraction of the cortex. The gap between what is recorded and what a thought consists of is enormous, and popular expectations run far ahead of it.
Non-stationarity. The mapping between neural activity and intention shifts as tissue changes and as the user adapts, so systems require recalibration — and the user is learning at the same time as the decoder, which makes the two a coupled system rather than a device with a user.
Applications and Ethics
Established clinical uses are restorative: communication and movement for people with severe paralysis, cochlear implants (a neural interface, though not a BCI in the strict sense), and deep brain stimulation for Parkinson's disease, which is now routine and entirely one-directional. Speculative and commercial ambitions — consumer control, augmentation, memory enhancement — are considerably less mature than their publicity.
Neural data is not like other personal data: it may contain information the subject has not consciously accessed. Closed-loop stimulation systems raise questions about agency, since a device that modulates mood or movement in response to detected states is participating in decisions. And a person whose speech or movement depends on a proprietary decoder is dependent on a company's continued existence — see implant on the several documented cases of abandoned neural implants.
Relevance to Cyborg Anthropology
The BCI is the limiting case of the cyborg premise: integration at the level of the nervous system itself, where the boundary between the person's intention and the machine's action becomes genuinely hard to locate.
Related Reading
References
- ↑ Vidal, Jacques J. "Toward Direct Brain-Computer Communication." Annual Review of Biophysics and Bioengineering 2 (1973): 157–180.
