It was a work night in the Trump era, and I was in bed, reading a book about neoliberalism. Or, more accurately, trying to. The author’s stated ambition for the chapter at hand was to render certain late ideas of Foucault’s accessible to a general audience, but she seemed for long stretches to lose sight of this goal, forcing me to scan the same three pages over and over again hoping that something like sense might materialize. After having a brief panic attack re: all the things I’d need to read just to begin to grasp the contours of everything I don’t know, I settled into a thought along the lines of: wouldn’t it be nice if I could just get this stuff injected into my brain?
Thus the subject of this week’s Giz Asks: the possibility of uploading information straight to one’s skull, today or (more likely) in the future. There are people at work on this very question, and you can read some of their insights below.
Assistant Professor, Psychological & Brain Sciences, UC Santa Barbara and Assistant Professor, Computer Science, Psychological & Brain Sciences, UC Santa Barbara
We can already implant a sensory neuroprosthesis to, for example, restore some form of vision to those who are blind, so why wouldn’t we be able to upload the contents of an entire encyclopedia to our brain, or implant a device that instantly turns us into master pianists? Although the practical solution might not be as simple as plugging a USB drive into our brain, I think the prospect of augmenting our senses and our intellect with a brain device is certainly within our reach. However, before this is possible, we need to overcome several challenges.
One obvious hurdle is engineering. We need to be able to manufacture and implant biocompatible devices that can interface with the brain both safely and effectively. Although technology is rapidly advancing, current implantable technologies are often limited to a small number of electrodes that are orders of magnitude larger than the neurons they are trying to target. We may need smaller and more flexible devices that can be implanted deep in the brain and interact with specific subpopulations of neurons regulating different behaviors and memories.
However, the biggest challenge I see is that our understanding of the brain is simply not good enough to make brain uploads viable. We need to better understand how information is stored and accessed in the brain. For example, there is no one brain area that stores your piano or Spanish skills. As far as we understand, implicit memories (such as motor memories required to play the piano at master level) require a concerted effort of multiple brain areas such as the basal ganglia and cerebellum. Other information, such as semantic memories (e.g., general facts and information) rely on an interplay between the hippocampus, amygdala, and the neocortex at large. Hence there might simply not be a single place in your brain where you can plug in the USB stick.
Ultimately, the solution might have to be more creative than an implantable chip. An alternative might be to find ways in which we can aid the brain in learning or remembering stuff. Deep brain stimulation (DBS) is already making use of this principle to improve memory-related disorders such as anxiety and dementia. So maybe you’d still have to practice the piano a bit, but instead of putting in the 1,000 of hours it takes to become a concert pianist, you’d be done in a couple of hours.
Last but not least, there are ethical and philosophical considerations that come with such technology. Who would have access to it? How could it be abused? Even if we could, whether we should remains an open question.
Professor of Computer Science and Engineering and Electrical Engineering and co-Director of the Center for Neurotechnology at the University of Washington, Seattle
We are already in the “Kitty Hawk” days of sending information directly into human brains. For example, in collaboration with cognitive neuroscientists, my lab has shown that rudimentary information from one or more human brains can be transmitted directly to another brain using non-invasive brain-to-brain interfaces and “BrainNets.” We have also shown that information from a simple virtual reality video game can be sent directly to the human brain using noninvasive magnetic brain stimulation, allowing a player to successfully play the game without visual feedback. Finally, in collaboration with neurosurgeons, we have shown that electrical stimulation of the human brain surface over the somatosensory cortex can be used to send information about touch, allowing a human to adjust their hand posture according to this information.
These early results in my lab and many others in the field of neural engineering indicate that the human brain, with its remarkable ability to adapt, can make use of new forms of information delivered directly to different brain regions. Extrapolating into the future, uploading more complex information into a brain will require advances in at least three areas: (1) neuroscience, specifically, a deeper understanding of how abstract information is processed and stored in the brain, (2) neurotechnology, especially more precise, less invasive and larger scale brain recording and brain stimulation methods, and (3) artificial intelligence to compute the appropriate stimulation patterns that enable information to be uploaded.
To address this challenge, we have recently proposed the concept of a brain co-processor, a new type of brain-computer interface that processes information alongside the brain and acts as a “brain assistant.” Brain co-processors utilize artificial neural networks to adaptively translate external information, e.g., from sensors or the internet, in the context of ongoing brain activity to appropriate stimulation patterns, allowing the brain to parse and potentially store this information for future use. Such co-processors may eventually allow direct upload of complex information into the brain and, if designed and deployed within an ethically responsible framework, have the potential to transform a number of fields of human endeavor, from education and communication to rehabilitation and treatment of neuropsychiatric conditions.
Associate Professor and Co-Director of the Brain Initiative at UC Santa Barbara
With today’s technology, it should be possible to upload very small amounts of information already. By small, I mean 1 bit or so. (Depending on how inclusive we are with the question, maybe we already have done so.) However, anything complex is much more difficult. Uploading expertise in a new language or a detailed memory—that won’t be possible anytime soon, for two reasons: (1) our technologies for manipulating neural circuitry are too crude, and (2) our understanding of what to manipulate and how is too primitive.
However, it could be possible in the future. Provided a million years of technological development (it probably wouldn’t take that long, but for the sake of argument), it should be doable. To say that it is flatly impossible implies that there is something that cannot be measured and manipulated in the brain that is essential for learning a language or storing a memory, and that is not a defensible position. There are limits from fundamental physics, but I doubt they would prevent us from realizing success in this task. The brain can learn, and it uses physically possible processes. We can manipulate many of the processes involved in changing brain circuitry. It’s just challenging to do this precisely with the crude tools we have today. Plus, even if we could, we don’t know exactly how to manipulate them to get the desired outcome. But those are solvable problems.
My lab is focused on understanding neural circuitry and developing better tools for measuring and manipulating it. There is tremendous potential for further advances, but we won’t be uploading complex information to the brain very soon. That’s fine. It’s job security for both sides of my job: research and teaching.
Let me take this opportunity to focus attention on something that is closer than the general public might think. I bet that within our lifetimes, neurotechnology will provide radically improved strategies for treating conditions like depression, anxiety, and compulsive behaviors. Not new pills, but new neural interfaces. Neuroscientists have made great strides in understanding that circuitry, and our latest tools are technically capable of specifically manipulating these circuits. The tools are too crude for complex information, but they can do the job with this circuitry and enhance people’s lives. Brain-machine interface technology for prosthetic devices is also progressing. However, I do not anticipate uploading useful, complex information to my brain anytime soon. I’ll have to keep learning the old fashioned way.
Associate Dean and Professor at the School for the Future of Innovation in Society at Arizona State University
This is a classic sci-fi trope: injecting stuff directly into the brain, so that we suddenly know a vast range of material or possess a vast range of skills. It’s a very attractive concept, and of course it’s what’s driving companies like Elon Musk’s Neuralink, which are dealing with very advanced brain-machine interfaces. But there are very big problems here.
To start with: yes, we’re going to be able to do this, at a very, very crude level. We already have the ability to change the electrical signals in clusters of neurons—which means that we can do very crude things, like inject images into somebody’s head. But that’s nothing approaching a new skill-set or knowledge-base.
As far as the sci-fi stuff’s concerned—i.e., instantly uploading a 15-week course—that’s nearly impossible, for a number of reasons. First of all, the brain is not like a computer, and we can’t program it like one. Imagine this: you take the chip in your smartphone and ask a chip designer whether they could reprogram it without using any of the standard materials—in other words, instead of writing apps, they’d take a super-fine drill and drill holes in it and insert electrodes and reprogram it that way. The chip designer would laugh in your face, and tell you there’s no way on Earth—it’s too complex. And yet the chip in your phone is infinitely simpler than your brain. If it’s laughable for the chip in your phone, how on Earth are we going to be able to do it to our brains?
The brain is organic—it hasn’t evolved to have stuff easily written into it, other than through our senses. We would have to somehow build a secondary system into it, and every single neuron and every single synapse would require its own connection. The only way to do that would be to actually grow this secondary system from birth. Which, yes, we could maybe do in the distant future. But even if we could, that only deals with the electrical stuff—it doesn’t factor in the biochemical systems that also determine how our memories work.
And even if we could do this, we almost certainly shouldn’t. Our brains have evolved to be incredibly efficient—we use every ounce of grey matter to do stuff, which means if we mess around with some part of our brain, we’re likely going to interfere with our identities. Because our brains are so complex, uploading material to them would likely displace something that defines who you are—you’d end up a different person. So despite the potential benefits—it would be great, for instance, to know your surgeon has had the latest techniques uploaded to their brain—I think the risks are much more significant.
Emeritus Professor of Engineering at Coventry and Reading University, described elsewhere as ‘The World’s First Human-Robot Hybrid’
We already do upload information into the brain—through the normal sensory route. Already, in research studies, we can directly transmit signals into the brain either for communication purposes or for motor control—through a port such as Braingate. As for downloading things like memories (which you haven’t actually had) into the brain, I can’t see any reason why this will not be possible in the future, but to do that we need to learn a lot more about how memories are stored and the process of recall.
Research Associate Professor of Biomedical Engineering and Director of the Neural Modeling and Interface Lab at the University of Southern California
If we define “upload information into my brain” as using brain-machine interface technology to directly write in information to the brain with high speed and convenience, making traditional forms of learning unnecessary—as in The Matrix—my answer would be: it’s complicated.
First, I think this is definitely something theoretically possible. The common understanding in the scientific community is that information is stored in the brain in the form of synaptic weights and/or neural activities, and that these can be altered externally in many different ways, including via brain-machine interface. If they are altered in the right way, information will then be uploaded into the brain.
That said, it will still be extremely difficult. Much better basic neuroscience, interface technology, and computational models are required to make it happen. We are still very far away from being able to do this. Most of the current brain-machine interfaces or cortical prostheses aim to restore impaired memory functions, as opposed to enhancing them. A few cases of memory enhancement, including my own study, were conducted with very specific experimental paradigms and very specific forms of information/memory. The Matrix-like information uploading won’t happen in the near future.
And even when the technology is mature enough to allow this to happen, I am not sure whether it will be necessary, since information storage is no longer the bottleneck of human intelligence. With the internet and search engines, people now can easily access almost unlimited amounts of information. Having more information uploaded to the brain is unlikely to be a priority for improving human intelligence. Instead, the ability to integrate information and make better judgments seems to be much more important. However, how the human brain can achieve this ability is still unknown.
Assistant Professor, Electrical Engineering and Computer Sciences, UC Berkeley
It depends on the complexity of the information. We can (today) have people hallucinate sounds, involuntarily perform motor movements, disrupt seizures or alter emotional experience through direct stimulation in specific areas of the cortex. It is not inconceivable that one day, we could “upload” more complex information like a new skill or delete a traumatic episode from memory. This could very clearly be used for the wrong reasons (much like gene editing), and there is a clear need for emphasizing neuroethics, to the same degree that we’re focusing on advancing the science and engineering of the human brain.
Theory Fellow, Janelia Research Campus, Howard Hughes Medical Institute, whose research has included stabilizing brain-computer interfaces, among other things
The short answer is yes. In fact, if you are willing to accept a somewhat limited definition of “information,” it has already been demonstrated in humans. Recent studies have used intracortical microstimulation (ICMS) to stimulate the brain directly. Participants in these studies perceive the stimulation as sensations of pressure, vibration, tingling, or touch from different parts of their body. Of course, there is nothing being done to their physical bodies—these sensations are entirely “written” into the brain by ICMS. A team out of the University of Pittsburgh recently used this technology to transmit information from sensors in a prosthetic arm directly to the brain of someone controlling it.
While this is exciting, the more complete answer is that we have a long way to go before we might be able to upload what we more typically think of as “information,” knowledge of different subject areas or facts, directly to a person’s memory. To use an analogy from computers, to upload information to a computer you need to at least know the binary code computers use to represent information and have a means of setting the individual bits on a hard drive used to store that information. In neuroscience, we have a sense of which broad areas of the brain process certain types of information, but our knowledge about the code for representing information and how that code is persistently stored in the brain—things that will come down to the level of individual neurons and how they are connected—is very limited. And while ICMS can activate neurons in the brain, it is a crude method of doing so—stimulating large numbers of neurons at once. We will need methods of activating individual neurons in a finely controlled way to write in detailed information to the brain.
In addition, before we begin to upload information into our brains we will need to consider the ethical questions that come with this ability. Under what circumstances should we do this, and how will this interface with free will? If information can be uploaded, we will likely be able to download it as well. Who will have access to this information, and since these technologies will likely be expensive, how do we ensure the rich do not gain an unfair cognitive advantage over those with fewer resources?
With all of that said, the field of neuroscience is progressing rapidly. A recent technology called optogenetics uses light to activate neurons, and researchers are developing methods to direct the light to target individual neurons. With much additional work, technologies like this may one day provide tools for transmitting more detailed information directly to the human brain. Our knowledge of how information is represented and stored in the brain is sure to advance as well, and recognition of the ethical issues that come with technologies for interfacing with the brain has already begun.
So while we are surely many years, likely decades, away from systems that could be routinely used to upload information to our brain, it seems likely that one day this will be possible.
Professor, Computer Science & Engineering, University of Washington, whose research has applications in the domains of human-computer interfaces, among other things
Experiments are already going on now to provide sensory input via direct electrical stimulation of the surface of the brain. In addition to direct electrical stimulation by implanted electrodes (wires inserted into the brain!), transcranial magnetic stimulation (TMS) uses a strong magnetic field to stimulate the brain from outside—no surgery needed. My bet is that it will be possible to provide certain forms of sensory input directly to the brain. For example, a prosthetic hand or foot might be able to generate some sensation of touch, or the perception of an audio alert sound could be generated. Perhaps interpretable speech could be generated directly in the brain. That said, I find it much harder to imagine that one could ever successfully generate in the brain higher level cognitive input in the brain, such as words or thoughts, or even sophisticated visual information at the level of readable text.
Do you have a burning question for Giz Asks? Email us at email@example.com.