Can New High Tech Machines Read Your Mind?

The rock that moved all on its own

I had mounted a decorative rock onto a radio-controlled toy car’s chassis and connected the car’s motor to a wireless interface, that was in turn under the control of my brain.

The experimental toy I was developing, called Rocky: The Rock That Rolls, did move forward and stop under mind control via electrodes on my scalp—more or less reliably—but I could never get it to turn in the right direction consistently, or to go in reverse. Oh, and the AI that interpreted my brain waves to control Rocky had to retrain itself with each new user, which explains why you won’t see Rocky in the toy aisle anytime soon.

Rocky’s shortcomings are typical of many of today’s brain-computer interfaces (BCI) in that these interfaces work, sort of, but not well enough to enter wide mainstream use.

But that may be about to change as advances in sensors and AI bring us closer to machines that can read our minds in limited but useful ways. Here are examples of some of those promising advances.

A growing list of “mind-reading” machines

• Reconstructing mental images: Researchers in Russia and Japan, using scalp electrodes and functional Magnetic Resonance Imaging (fMRI) respectively, have reproduced crude, but serviceable images that reproduce, solely from brain activity, the images that experimental subjects are viewing.
• Controlling exoskeletons: Neuroscientists at Duke University, Korea University, and TU Berlin employing both scalp electrodes and brain implants, have enabled spinal cord injury patients to control motorized exoskeletons.
• Mentally type and produce synthetic speech: Scientists in the US and Switzerland, using brain implants in the motor cortex of injured patients have enabled these patients to “type with their minds” at 90 words per minute and to generate synthetic speech merely by imaging they are vocalizing speech sounds.
• Detect addiction: Brookhaven National Labs has shown, through fMRI responses of brains of both “normal” and addicted subjects, that it is possible to spot brain activity associated with addiction.
• Reproduce speech from throat electrodes: Investigators in Wisconsin showed that surface electrodes capturing Electromyographic (EMG) signals from muscles used in speech in the neck can capture “subvocal” (words that are imagined but not spoken) information.
• Monitoring drowsiness and fatigue: Oregon Health and Science University neuroscientists showed that scalp electrodes can reliably detect whether operators are experiencing drowsiness or fatigue.
• Controlling drones with the mind: Engineers at the University of Minnesota controlled a quadcopter, even flying it through an obstacle course through brain commands picked up by surface scalp electrodes.
• Brain-to-brain communication: A global team in Spain, the US, India, and France, used Magnetoencephalography (MEG) to pick up “yes-no” type brain commands from one person, then communicate those commands, via Transcranial Magnetic Stimulation (TMS) to the brain of another person.

Where machine mind-reading is going

All of the techniques just described hold far more promise than mind-controlled rocks, especially in offering hope to patients with brain or spinal cord injuries and aiding both neurologists and psychiatrists in diagnoses and treatment of disease.

As new sensor technologies emerge (my favorite is one that, non-invasively, beams infrared energy through the skull and analyzes changes in the energy coming back out of the skull) and computers and AI algorithms get more powerful, it’s likely that, in the not-to-distant-future, mute patients will speak and paralyzed people will be able to walk or move their arms through robotic exoskeletons.

In other words, the age of mind over matter is fast approaching.