Have you seen this study? It made headlines all over the world – first direct brain-to-brain interface! Brain activity of one person is recorded, decoded and sent over the internet to a computer which controls a brain stimulator, which in turn stimulates another person’s brain! Sounds quite sci-fi, right?!
Well, I hate to spoil the fun, but apart from a pretty cool demonstration of two well-established techniques in cognitive neuroscience, this study literally tells us nothing about brain function we did not already know, nor leads to any practical application in the (near) future. Besides – it’s not even new. Andrea Stocco and Ranesh Rao did almost the exact same thing over a year ago. Both these ‘brain-to-brain’ interfaces are hardly spectacular, however, not even as a proof-of-concept. At best, they are the neuroscientist’s equivalent of a Rube Goldberg machine. Fun art projects, not science.
So, what’s going on? Brain-computer interfacing, or BCI, refers to the decoding of brain activity in realtime and transform this into a control signal. BCI has been around for a while, and there are several reliable ways to use brain activity to generate control signals. The P300 is an example – ‘P300’ refers to a positive peak in the electroencephalogram that occurs roughly 300 ms after an observer spots something that captures her/his attention. The P300 is widely used as a dependent measure in cognitive neuroscience and psychophysiology, and can be detected in single events with quite a good accuracy. So, if you are monitoring a participant’s brain responses to particular events, finding a P300 tells you that the this event captured the participant’s attention. You can use this, for example, to guess someone’s PIN code: let a participant hold his PIN code in mind, and flash the numbers 0 – 9. The numbers that evoke that largest P300s are likely to be in his/her PIN code. How to get to the full PIN code, I leave up to your own imagination and evil genius…
However, there are more benevolent applications. P300s are used widely to drive spelling programmes for patients with different kinds of disabilities. However, the P300 is a passive measure: it’s something your brain does in response to an external stimulus. Ideally, for BCI applications, you want a measure you as a user can have some control over. Motor imagery is increasingly used to generate control signals. If you think of moving your arm, this generates activity in a specific part of the motor cortex. Think of moving your foot, and you activate another part of your motor cortex. With EEG you can pick this up and discriminate between different imagined movements with high accuracy. My colleague Ritske de Jong has developed algorithms that pretty much work out-of-the box and are up to 99% accurate. Once you know what movement a participant is imagining, converting that to a control signal is trivial. Both Rao and Stocco, and Grau et al. use this type of BCI for their Rube Goldberg machines.
But what about the brain stimulation part? In the brain-to-brain interfacing studies, researchers use transcranial magnetic stimulation (TMS) to induce brain activity. TMS is a very useful tool to study brain functioning because it allows us to interfere with brain processes. You cannot just disrupt brain processing though – stimulating the motor cortex can lead to (rather jerky) movements, and stimulating the occipital cortex results in seeing a flash of light, a phosphene. In order to stimulate a given area, you simply position your coil over that area. However, that is as sophisticated as it gets. You can induce a jerky movement, or you can induce a very brief lasting phosphene with TMS. You cannot influence someone’s thoughts, or fine-tune someone’s actions with TMS, apart from annoying your participants.
Both this ‘first’ demonstration of a brain-to-brain interface and the actual first demonstration of a brain-to-brain interface use TMS to stimulate the brain of the receiving participant. And it’s exactly this what makes me characterize it as Rube Goldberg machines. The sophistication of the computer-to-brain interface is about that of hitting someone with a hammer. There is simply nothing practical that you can do with TMS, as opposed to implanted electrodes. Even worse, there is reason to believe that TMS has a theoretical limitation in terms of what kind of brain activity and brain areas can be stimulated. It is very unlikely that TMS or any other kind of non-invasive brain stimulation will ever have the level of sophistication to induce brain activity at the fine-grained level required to control thoughts and actions.
So, what do both brain-to-brain studies, that are supposedly ‘proofs-of-concept’, show? Well, they show that:
- you can use BCI to generate a control signal. Great, we already know that for quite a while.
- you can transmit this control signal via the internet. Well, that’s obvious too, you wouldn’t reading this blog if we could not signals via the internet
- you can use a control signal to trigger a TMS device. Since TMS devices are around, they can be triggered by a TTL-pulse… so this is not really new, too…
All in all, there is really nothing new or surprising in these studies. There are no fundamental issues that are resolved, or new technological insights that truly show a new way of getting information from one brain into another. The setups presented in these studies are really nothing more than Rube Goldberg machines – hilariously complicated setups to perform a task that can be solved much, much simpler.
But they do make cool demonstrations. It’s just not science.
Oh, and as for brain-to-brain interfacing… what about this?