Telepathy. Mind Control. Possession. These and related phenomena are subjects of science fiction, movies, and comic books. Professor X from Marvel’s X-Men can read and control thoughts. Unholy demons and devils from The Exorcist can take the driver’s seat as it were in a poor possession victim at will. And now it seems the leading edge of brain-machine interfaces has arrived at telepathy, mind control, and possession from brain-brain interfaces.
There is one name that comes immediately to mind upon hearing brain-brain surgical interfaces. And that is Dr. Miguel Nicolelis. Having had the pleasure of meeting with Dr. Nicolelis years ago at a neural conference, I was already briefed on his work and his research directions. His latest feat appears to consolidate and proceed with additional brain-brain interfaces and propose a quite literal natural neural network (Pais-Vieira, et al, 2015).
To provide context, an artificial neural network follows the microscale brain neurons as below:
The artificial neural network is a simulation of the cellular neurons and their interactions. The inputs (1&2) are simulations of sensory neuron stimuli, say a feather stroking the chin whiskers of a rodent. The inters (1-3) are representations of some internal processing inter-neurons in the brain. These might be memory cells whose sole purpose is to detect a pattern and make sense of the tickling sensation on the chin whiskers. The output (here, only 1 for simplicity) might represent an action cell, say the impulse to suddenly jump or simply scratch the chin. In an artificial neural network, each input connects to all internals, all internals connect to the output, and all connections are handled via mathematical transfer functions. The abstraction works flexibly enough to borrow from regression models such that the artificial neural network as a whole can be translated as a construct of multiple, hierarchically stacked regression models. (e.g. Each of the three inters is the end result of three individual regression models each using two inputs AND the last output is another regression model using the three inters as inputs. Any regression fitting function/network transfer function would suffice.
What Nicolelis’s lab did was to essentially replace the computational simulation of the internal neurons with actual, literal neurons from living, active rodents.
Literally, a brain-machine interface. More poignantly, since the insertion-of-the-living-animal-brain portion encompasses more than one rodent, it can be more accurately termed a brain-brain interface. For a light summary of the research and discussion, read here. Essentially, the latest research report has multiple rodents working together as a group to accomplish a variety of tasks, including synchronizing activity, recalling a pattern, and classification. Results showed that teams of rodents outperformed individuals of the same.
From a first pass, layman perspective, one can be forgiven for wondering, “What is the world coming to? Is this a Matrix style ‘animal farm’ where our privacy, our very identities are blurred and mashed into a livid dream?” These are literally cyborg rats, with microelectrode implants surgically inserted into target neurons. These implants record a set of single unit recordings from individual neurons. They work like a wiretap into the neuron so if that neuron generates an action potential, the electrode detects the ionic spike.
(Above) A sample microelectrode array. Each microscopic needle inserts into a neuron and “wiretaps” into the cellular action potential activity. Each rodent received an array of electrodes, resulting in up to about a dozen viable unit cell recordings. The great advantage of a microelectrode array is that it records directly from the isolated neurons. There is very little noise to worry about filtering out as would be the case using an EEG recorder that listens through the skull, muscles, and skin to hear millions of action potentials from millions of cells at once. EEG recordings tend to be extremely noisy, leading to highly imprecise results. So while a microelectrode array may appear scary and invasive, it is merely a refined, pinprick version of a sensitive EEG cap. Of course, microelectrode arrays are not available for human use. Nor are they suitable for long-term use since the neurons so punctured do not live for extended periods. So one need not fear living “Borg” hybrids from Star Trek fiction.
In addition, the concept of listening to neurons is not so avant-garde strange after all. Since neurons connect eventually to executive decision centers and motor execution units, monitoring and listening to neurons has essentially been monitoring and observing behaviors. Having four rats think together to generate whisking and licking movements with their synchronized, microelectrode-studded neurons to get the reward is not much if at all different from having four rats watching each other whisking and licking together to get the same reward. According to Bill Bryson in his book, At Home (2011), rodents have been observed to work collaboratively to steal raw eggs without breaking them. One rat would wrap itself around the egg and extend a tail whereupon a partner rat would drag the tail and the connected rat into a safe place. The pair of rats would then consume the egg at their leisure in safety. No connecting microelectrode arrays were ever noted.
In similar fashion, the concern of scanning and recording neural action potentials as the ultimate invasion of privacy is at first glance a horrifying, shades-of-Orwellian-1984 nightmare. However, anyone living in any occupied apartment building in any city may be more than mildly surprised or horrified at realizing how much the neighbors really figured out. In fact, the matter may be even more serious in suburban homes since everyone is far more exposed entering and exiting the house. Plus, all incoming packages and outgoing garbage are single destination/sourced and isolated. Again, there are no microelectrode arrays, wiretaps, or even EEG caps involved. We normally have far less privacy than we may have been led to believe.
Far more disturbing would be research experiments not only connecting networks of collaborating rodents in parallel but connecting the motor outputs of one rodent to the muscles of another, anesthetized one. This can result in one mouse effectively controlling for example two tails, its own plus the tail of its sleeping partner. Now that is truly mind control. Then it would only be a matter of making the sleeping, mind-controlled rat spin its head around 360 degrees with another rat chanting, “The power of Christ compels you!”