DOW-UAP-D130 — Brain-Machine Interfaces Without Manual Controls

The report defines its target as thought-based operation of remote machinery during normal activity, with more than 5–10 bits per second and command selection within 300 milliseconds.

Data transfer rates are sought to exceed 5-10 bits/second to be useful for operation of complex devices; this rate range and above is referred to as high-bandwidth BMis.

The BOLD response peaks several seconds after neuronal activity, which the report identifies as a major limitation for rapid control applications.

The BOLD response to any event peaks about 4-6 seconds after the event occurs, limiting the applications for which monitoring these signals and their associated delay may be useful

The report says ordinary manual interfaces already transmit roughly 5–14 bits per second, setting a practical benchmark that proposed neural interfaces must justify exceeding.

Baseline examples of performance include finger pointing, which can convey 14 bits/s of information, operation of a mouse about 8 bits/s, while stylus tapping a soft QWERTY keyboard on a PDA has a lower rate, around 5 bits/s.

EEG communication had been demonstrated for some locked-in patients, but the report says command response times were measured in seconds.

Successful communication has been established via EEG BMI in several studies based on both spiking activity and P300 signals (References 22-25). The response time to execute a command using these systems is measured in seconds.

A cited dry-electrode study reported 90 percent one-dimensional cursor accuracy with six electrodes and suggested about 5 percent error with twelve.

Popescu and colleagues have shown a system that uses 6 dry electrodes and is 90 percent accurate in operation of a 1-D cursor by untrained subjects.

The report treats field deployment of MEG as impractical because neural magnetic signals are extremely weak and require cryogenic sensors and heavy shielding.

the fact that 100 fT is about 100 million times smaller than the Earth's magnetic field will prove an insurmountable barrier to sifting signal from noise in anything but a heavily-shielded, metal-free environment.

The surveyed EMG interface achieved 1–2 bits per second with minimal training by using muscle activity as a biological amplifier of neural intent.

Recent work has shown that performance of 1-2 bits/s is possible with minimal training, about four times the current performance of a comparison EEG forehead sensor

The report warns that even a constrained two-dimensional primate motor task could not be fit with a linear model once realistic motion impedance was included.

no fit correlating the observed movements and neural activity could be obtained with a linear model when kinematic impedance was considered.

The report gives invasive ECoG a higher signal rate than scalp EEG but still characterizes its information bandwidth as low.

Currently EEG methods are limited to 20-30 bits/min ( < 0.5 bits/s). This drawback has prompted the use of a more accurate and speedy recording method based on invasive techniques such as the electrocorticogram (ECoG)

A lamprey brainstem experiment demonstrated a bidirectional closed loop between neural tissue and a mobile robot, while remaining an experimental animal model.

Signals generated by the two optical sensors of the robot were translated into electrical stimuli and applied to the vestibular pathways, and to two populations of reticular neurons.

Human ECoG studies cited by the report achieved one- and two-dimensional cursor control in roughly 1–2 seconds with accuracy up to 75 percent.

Movement times to target were on the order of 1-2 seconds with up to 75 percent accuracy with training time on the order of 30 minutes or less

Targeted muscle reinnervation allowed a classifier to select among ten intended arm motions in under 170 milliseconds, according to the cited work.

for controls, a motion selection from 1 of 10 possibilities could be achieved in less than 170 milliseconds.

A peripheral nerve-array self-experiment lasted three months before the physical nerve-to-array connection deteriorated beyond use.

This trial lasted 3 months before the physical connection between the nerve and the microarray deteriorated beyond use.

The report expects near-term real-world systems to rely mainly on noninvasive EMG, while saying EEG-only control for healthy users was unlikely within five years.

The current state-of-the-art BMI devices that have shown the most promise in the near term utilize muscle amplification of the neural signal.

The report concludes that practical high-bandwidth two-way invasive interfaces remained far from application because human trials showed short-lived function and no robust single-neuron duplex connection.

direct interface technologies based on proximal action potential stimulation and detection are still a long way off from practical application since the trials themselves in duplex operation show functional failure after only a matter of weeks.

For the long term, the report identifies optical interfaces and electrode arrays encased in engineered neural tissue as promising research paths, not demonstrated deployable systems.

Two research paths currently showing the most promise are optical interfaces and ex-vivo preparation of electrode arrays encased in engineered neural tissue.