Historical research guide · 0342 · 2010

DIRD Technological Approaches to Controlling External Devices

DIA / AAWSAP contractor

In one minute

The Defense Intelligence Agency asked a narrow question with a very long reach: how do you operate a machine when no hand touches it? This 2010 reference document surveys every route then available from a nervous system into a circuit — scalp electrodes, magnetic sensors, blood-oxygen imaging, muscle sensors worn on the skin, and arrays implanted on or in the cortex — and grades each against one engineering target: five to ten bits per second of control, a command selected in under 300 milliseconds, and all of it while the operator walks around, not lying still inside a shielded room. The author’s verdict is unusually direct. Nothing then available cleared the bar, because scalp reading hears only the summed shout of tens of thousands of neurons at once. The near-term winner is muscle sensing, which borrows the body’s own amplifier. The far-term winner is a connection to single neurons made by light rather than by metal, or by growing living tissue onto the chip before it ever goes in.

Connection

Chapter 4 describes a craft whose pilot feels nothing and sees nothing of the outside; this is the Defense Intelligence Agency’s survey of what that pilot would actually hold, and it grades every candidate by a measured bit rate rather than a promise, which is exactly the discipline chapter 1 asks for. Its two long-term picks — optical links to single neurons, and electrode arrays sheathed in laboratory-grown neural tissue so the implant meets the brain as biology rather than as metal — are the kind of named, checkable forward marker chapter 13 is assembled from.

From the source

The report sets a hard engineering target and holds every technology to it: thought-based operation of remote machinery during normal human activities, with no shielded room, no requirement to stay perfectly motionless, and no hand-operated device — data transfer above 5 to 10 bits per second, and a 1-of-N command selected in 300 milliseconds or less.Introduction, p. v
The reason a direct brain-to-circuit connection has never been built is a difference in carriers. In an axon, sodium, potassium and chlorine ions move across the membrane, perpendicular to the wave travelling down it; in a copper wire the signal moves as a compression wave among drifting electrons. The underlying physics makes metallic and semimetallic circuits about a million times faster than the biological system, so every technology in the survey instead places sensors or transmitters close to the neurons and lets classical electrodynamics carry the signal across the gap.Direct Neural Signals, pp. 1-2
Neurons fire at a maximum of roughly 100 hertz to 1 kilohertz, so sampling at a few kilohertz is fast enough to catch any individual neuron firing, and there is no evidence that transduction frequencies above 100 kilohertz have any advantage in a brain-machine interface. Since even that ceiling is easily reached with current electronics, the whole difficulty of the field sits in the connection, not in the circuitry.Direct Neural Signals, p. 2
Scalp electrodes read the summed activity of roughly 50,000 local neurons, which holds EEG-based methods to 20 to 30 bits per minute — under half a bit per second, against a target of five. The report names the single observation that would change this: a study showing noninvasive spiking or event-related-potential detection from a coherent group of fewer than 10,000 neurons. Until localisation on that order is demonstrated, EEG-only devices remain low-bit-rate application technologies.MEMS, ECoG, and PSoC Circuitry, p. 14; Discussion, Noninvasive Electrical Devices, pp. 24-25
Muscle gives the near-term answer because it is a biological amplifier of the neural signal. Noninvasive muscle sensing reaches 1 to 2 bits per second with minimal training, about four times the rate of a comparison EEG forehead sensor; and in amputees whose unused nerves have been surgically rerouted into remaining muscles, twelve electrode pairs feed a classifier that issues a new prediction ten times a second and selects one of ten motions in less than 170 milliseconds.Noninvasive Technologies, EMG, p. 9; Trials Using Human Subjects, p. 23
The far-term architecture the report backs is a duplex network of 5 to 20 bits per second built on single neurons — either optical stimulation and gating, where a light link acts as a connected input dendrite on one neuron with spectroscopic read-out of the action potential and no physical contact at all, or electrode arrays first encased in laboratory-grown neural tissue so that the implant meets the brain as biology rather than as metal. Routed through the cerebellum, such a connection would make the controlled device part of the body as far as the brain is concerned.Discussion, Invasive Technologies - General, Optical, and Ex-Vivo Engineering, pp. 26-27; Conclusions, pp. 27-28

The way in

https://documents2.theblackvault.com/documents/dia/AAWSAP-DIRDs/DIRD_12-DIRD_Technological_Approaches_to_Controlling_External_Devices.pdf

Historical source-access record. This guide links to the original; it does not reproduce the paper text.

Defense Intelligence Reference Document, Acquisition Threat Support. DIA-08-1003-012, 23 March 2010 (IcOD: 1 December 2009). Full title on the cover: ‘Technological Approaches to Controlling External Devices in the Absence of Limb-Operated Interfaces’. Stated on its title page as one in a series of advanced technology reports produced in FY 2009 under the Defense Intelligence Agency Advanced Aerospace Weapon System Applications (AAWSA) Program. Released under FOIA and published by The Black Vault. AUTHOR. Withheld under FOIA exemption (b)(6); the preparing office is withheld under (b)(3):10 USC 424. The text moves fluently between magnetic-resonance physics, electrode metallurgy and neurophysiology and cites its own earlier survey work — internal evidence of a single physicist-author writing across both fields, recorded here as an inference and not as an attribution. TEXT. The report runs about 11,000 words across an introduction, four survey chapters, a discussion and conclusions, and is reproduced in full below. The document carries a copyright warning against further dissemination of its photographs, so the thirteen figures are not reproduced; the captions that carry technical content are kept. The sixty numbered endnote references and the classification banners and running heads are omitted; the complete text is at the source. Footnotes carrying substantive technical content are folded into the paragraph they annotate, in the report’s own words. Inequality signs and the Larmor relation are written out in words so the page renders.

How to cite it

DIA / AAWSAP contractor (2010) DIRD Technological Approaches to Controlling External Devices https://documents2.theblackvault.com/documents/dia/AAWSAP-DIRDs/DIRD_12-DIRD_Technological_Approaches_to_Controlling_External_Devices.pdf

Where it sits in the curriculum

The reference documents and the institutional record