Illustration by Richard Wilkinson


Hive consciousness

New research puts us on the cusp of brain-to-brain communication. Could the next step spell the end of individual minds?

by Peter Watts + BIO

Illustration by Richard Wilkinson

You already know that we can run machines with our brainwaves. That’s been old news for almost a decade, ever since the first monkey fed himself using a robot arm and the power of positive thinking. Nowadays, even reports of human neuroprostheses barely raise an eyebrow. Brain-computer interfaces have become commonplace in everything from prosthetic vision to video games (a lot of video games; Emotiv and NeuroSky are perhaps the best-known purveyors of Mind Control to the gaming crowd) to novelty cat ears that perk up on your head when you get horny.

But we’ve moved beyond merely thinking orders at machinery. Now we’re using that machinery to wire living brains together. Last year, a team of European neuroscientists headed by Carles Grau of the University of Barcelona reported a kind of – let’s call it mail-order telepathy – in which the recorded brainwaves of someone thinking a salutation in India were emailed, decoded and implanted into the brains of recipients in Spain and France (where they were perceived as flashes of light).

You might also remember breathless reports of a hive mind emerging from the depths of Duke University in North Carolina during the winter of 2013. Miguel Pais-Vieira and his colleagues had wired together the brains of two rats. Present a stimulus to one, and the other would press a lever. The headlines evoked images of one mind reaching into another, commandeering its motor systems in a fit of Alien Paw Syndrome.

Of course, the press goes overboard sometimes. Once you look past those headlines you notice that Reaction Rat had been pre-trained to press his lever whenever he felt a particular itch in his motor cortex (in exactly the same way you’d train him to respond to a flashing light, for example). There was no fused consciousness. It was a step forward, but you don’t get to claim membership in the Borg Collective just because a stimulus happens to tickle you from the inside.

And yet, more recently, Rajesh Rao (of the University of Washington’s Center for Sensorimotor Neural Engineering) reported what appears to be a real Alien Hand Network – and going Pais-Vieira one better, he built it out of people. Someone thinks a command; downstream, someone else responds by pushing a button without conscious intent. Now we’re getting somewhere.

There’s a machine in a lab in Berkeley, California, that can read the voxels right off your visual cortex and figure out what you’re looking at based solely on brain activity. One of its creators, Kendrick Kay, suggested back in 2008 that we’d eventually be able to read dreams (also, that we might want to take a closer look at certain privacy issues before that happened). His best guess was that this might happen a few decades down the road – but it took only four years for a computer in a Japanese lab to predict the content of hypnagogic hallucinations (essentially, dreams without REM) at 60 per cent accuracy, based entirely on fMRI data.

When Moore’s Law shaves that much time off the predictions of experts, it’s not too early to start wondering about consequences. What are the implications of a technology that seems to be converging on the sharing of consciousness?

It would be a lot easier to answer that question if anyone knew what consciousness is. There’s no shortage of theories. The neuroscientist Giulio Tononi at the University of Wisconsin-Madison claims that consciousness reflects the integration of distributed brain functions. A model developed by Ezequiel Morsella, of San Francisco State University, describes it as a mediator between conflicting motor commands. The panpsychics regard it as a basic property of matter – like charge, or mass – and believe that our brains don’t generate the stuff so much as filter it from the ether like some kind of organic spirit-catchers. Neuroscience superstar V S Ramachandran (University of California in San Diego) blames everything on mirror neurons; Princeton’s Michael Graziano – right here in Aeon – describes it as an experiential map.

I think they’re all running a game on us. Their models – right or wrong – describe computation, not awareness. There’s no great mystery to intelligence; it’s easy to see how natural selection would promote flexible problem-solving, the triage of sensory input, the high-grading of relevant data (aka attention).

But why would any of that be self-aware?

If physics is right – if everything ultimately comes down to matter, energy and numbers – then any sufficiently accurate copy of a thing will manifest the characteristics of that thing. Sapience should therefore emerge from any physical structure that replicates the relevant properties of the brain.

We might be about to find out. SyNAPSE – a collaboration between the US Defense Advances Research Projects Agency (DARPA) and the IT industry – is even now working on a hardware reconstruction of a human brain. They’re hoping to have it running by 2019, although if physics is right, ‘awake’ might be a better term.

Then again, if physics is right, we shouldn’t exist. You can watch ions hop across synapses, follow nerve impulses from nose to toes; nothing in any of those processes would lead you to expect the emergence of subjective awareness. Physics describes a world of intelligent zombies who do everything we do, except understand that they’re doing it. That’s what we should be, that’s all we should be: meat and computation. Somehow the meat woke up. How the hell does that even work?

they sat down over beers and took a show of hands on whether to admit bonobos to the Sapients Club

What we can get a handle on are the correlates of sapience, the neural signatures that accompany the conscious state. In humans at least, consciousness occurs when a bunch of subcortical structures – the brain stem, the thalamus and hypothalamus, the anterior cingulate cortex – talk to the frontal lobes. Integration is key. Neurons in all these far-flung regions have to be firing in sync, a co‑ordinated call-and-response with a signal lag of no more than 400 milliseconds. Tononi is using that insight to derive an integration metric he calls ɸ. It is designed not merely to detect consciousness but to quantify it: to hang a hard number on the level of self-awareness flickering in everything from roundworms to humans.

If it does all come down to neural integration – if self-awareness is a matter of degree, flickering at some rudimentary level even in the ganglia of nematodes – then the specific architecture of the conscious brain might be open to negotiation. This, at least, is the position of the so-called ‘Cambridge Declaration’ unveiled at the 2012 Francis Crick Memorial Conference on Consciousness. Its signatories – ‘cognitive neuroscientists, neuropharmacologists, neurophysiologists, neuroanatomists and computational neuroscientists’ – attribute self-awareness to a wide variety of non-human species.

I’m not sure how seriously to take this. Not that I find the claim implausible – I’ve always believed that we humans tend to underestimate the cognitive complexity of other creatures – but it’s not as though the declaration announced the results of some ground-breaking new experiment that settled the issue once and for all. Rather, its signatories basically sat down over beers and took a show of hands on whether to publicly admit bonobos to the Sapients Club. (Something else that seems a bit iffy is all the fuss raised over the signing of the declaration ‘in the presence of Stephen Hawking’, even though he is neither a neuroscientist nor a signatory.)

Still, we are talking about a cadre of renowned neuroscientists, the least of whom is far more qualified than I to make assertions on the subject. One of the things they assert is that self-awareness does not depend on specific brain structures. The declaration grants ‘near-human levels of consciousness’ to parrots (who lack a neocortex) and to octopuses (whose brains – basically a bagel of neurons encircling the esophagus – don’t have any anatomical resemblance to ours at all). It’s neurological complexity that’s essential to the conscious state, they tell us. The motherboard can take any shape so long as it’s got enough synapses on board.

This is all preamble, though, a set‑up to the question posed at the outset: what are the implications of a technology that wires brains together, that in theory at least permits the existence of hive minds? In fact, you know a lot more about that than you might think.

You already are a hive mind. You always have been.

This thing you think of as you: it spreads across two cerebral hemispheres connected by the corpus callosum, a fat meaty pipe more than 200 million axons thick. Suppose I took a cleaver to that pipe, split it down the middle. (That’s no mere thought experiment: severing the corpus callosum is a last-ditch measure against certain forms of epilepsy.) In the wake of such violent separation, each hemisphere would go its own way. It would develop its own tastes in clothes, music, even its own religious beliefs. Ramachandran tells of a split-brain patient with a Christian hemisphere and an atheist one. You’ve probably heard of Alien Hand Syndrome, or at least seen the movie Dr Strangelove: try to put on a certain shirt, your evil hand rips it off. Try to pick up a favourite pen, your evil hand knocks it away and picks up a Sharpie instead.

Except it’s not your hand at all any more, of course. It belongs to that other self living across the hall, the one that used to be part of you before the break‑up.

A shy introvert morphs into a flirtatious jokester. A pleasant woman turns sarcastic. But when the other half wakes up, the new entity vanishes

You’re still talking, at least. Still friends of a sort. Even when the corpus callosum is severed, the hemispheres can communicate via the brainstem. It’s a longer route, though, and a much thinner pipe: think dial-up versus broadband. The essential variables, once again, are latency and bandwidth. When the pipe is intact, signals pass back and forth across the whole brain fast enough for the system to act as an integrated whole, to think of itself as I. But when you force those signals to take the scenic route – worse, squeeze them through a straw – the halves fall out of sync, lose their coherence. I shatters into we.

You might expect that an established personality, built over a lifetime and then split down the middle, might take some time to develop into distinct entities. Yet hemispheric isolation can also be induced chemically, by anaesthetising half the brain – and the undrugged hemisphere, unshackled from its counterpart, sometimes manifests a whole new suite of personality traits right on the spot. A shy, whole-brained introvert morphs into a wise-cracking flirtatious jokester. A pleasant, well-adjusted woman turns sarcastic and hostile. When the other half wakes up the new entity vanishes as quickly as it appeared.

So while the thing that calls itself I typically runs on a dual-core engine, it’s perfectly capable of running on a single core. Take you, for example. Chances are you’re running on two cores right now. Does each contain its own distinct sub-personality? Are there two of you in there, each thinking: Hey, I’m part of something bigger?

Not likely. Rather, the local personae are obliterated, absorbed into a greater whole; as the Finnish computer scientists Kaj Sotala (at the University of Helsinki) and Harri Valpola (Aalto University) recently declared in the International Journal of Machine Consciousness, ‘the biological brain cannot support multiple separate conscious attentional processes in the same brain medium’.

Remember that. It could end up biting us in the ass a few years down the road.

Krista and Tatiana Hogan of the city of Vernon in British Columbia are seven-year-old sisters fused at the head. Craniopagus twins are extremely rare in any event, but the Hogans appear to be utterly unique in that they aren’t just fused at the skull or the vascular system. They are fused at the brain – more specifically at the thalamus, which acts as (among other things) a sensory relay.

They share a common set of sensory inputs. Tickle one, the other laughs. Each sees through the other’s eyes; each tastes what’s on the other’s tongue. They smile and cry in sync. There’s anecdotal evidence that they share thoughts and, although they have distinct personalities, each uses the word ‘I’ when referring to the other. The Hogan twins are two souls with one sensorium. All because they’re fused at a sensory relay.

But the thalamus is lower-brain circuitry. Dial-up, not broadband. Suppose the twins were fused at the prefrontal cortex instead?

If two hemispheres can each run separate, standalone programs – yet fuse to form a single coherent entity – what about the fusion of complete brains, a single contiguous porridge of neurons spread across two heads? Given a slight developmental tweak to the left, would we still be talking about two souls, or a single conscious being with twice the neuronal mass of a normal human brain?

Whether you instil sights, sounds, political opinions or a craving for a certain brand of beer might come down to where you aim the beam

There are other ways to put our heads together. Neurosilicon interfaces, for example. We’ve had those for more than a decade now. In labs around the world, neuron cultures put robot bodies through their paces; puddles of brain tissue drive flight simulators. At Clemson University in South Carolina, Ganesh Venayagamoorthy is busy teaching tame neurons to run everything from power grids to stock markets. DARPA has thrown its weight behind the development of a ‘cortical modem’, a direct neural interface wired right into your gray matter (we’re already using implants to reprogram specific neurons in other primates). But DARPA may have already been scooped by Theodore Berger, down at the University of Southern California. Way back in 2011, he unveiled a kind of artificial, memory-forming hippocampus for rats. The memories encoded in that device can be accessed by the organic rat brain; they can also be ported to other rats. It won’t be long before such prostheses scale up to our own species (that is in fact the explicit goal of Berger’s research).

If the prospect of surgery squicks you out, Sony has registered blue-sky patents for technology that plants sensory input directly into the brain using radio waves and compressed ultrasound. They’re selling it as a great leap forward for everything from gaming to telesurgery. (For my part, I can’t help remembering that neurons fire pretty much the same way whether they’re processing sensory input or religious belief. The difference between instilling sights, sounds, political opinions – why not an irresistible craving for a certain brand of beer? – might come down to little more than where you aim the beam.)

None of these efforts are explicitly designed to connect one human mind to another. What they’re pioneering is an interface, the ability to translate thoughts from meat into mech and back again. What we are seeing, in other words, is the genesis of a new kind of corpus callosum that extends beyond the confines of a single skull.

We’re still in the Precambrian. Grau’s emailed brainwaves amount to a fancy kind of semaphore that happens to bypass the eyeballs. Pais-Vieira’s hive mind was a pair of distinct rat brains, pimped out so that a spark in one would trigger a poke in the other – a stimulus that would have been meaningless to the recipient if he hadn’t already been trained to respond in a certain way. That’s not integrated awareness, or even telepathy. It’s the difference between experiencing an orgasm and watching a signal light on a distant hill spell out oh-god-oh-god-yes in Morse Code.

So it’s early days yet. But it may be later than you think.

Cory Doctorow’s novel Down and Out in the Magic Kingdom (2003) describes a near future in which everyone is wired into the internet, 24/7, via cortical link. It’s not far-fetched, given recent developments. And the idea of hooking a bunch of brains into a common network has a certain appeal. Split-brain patients outperform normal folks on visual-search and pattern-recognition tasks, for one thing: two minds are better than one, even when they’re in the same head, even when limited to dial-up speeds. So if the future consists of myriad minds in high-speed contact with each other, you might say: Yay, bring it on.

I’m not sure that’s the way it’s going to happen, though.

I don’t necessarily buy into the hokey old trope of an internet that ‘wakes up’. Then again, I don’t reject it out of hand, either. Google’s ‘DeepMind’, a general-purpose AI explicitly designed to mimic the brain, is a bit too close to SyNAPSE for comfort (and a lot more imminent: its first incarnations are already poised to enter the market). The bandwidth of your cell phone is already comparable to that of your corpus callosum, once noise and synaptic redundancy are taken into account. We’re still a few theoretical advances away from an honest-to-God mind meld – still waiting for the ultrasonic ‘Neural Dust’ interface proposed by Berkley’s Dongjin Seo, or for researchers at Rice University to perfect their carbon-nanotube electrodes – but the pipes are already fat enough to handle that load when it arrives.

And those advances may come easier than you’d expect. Brains do a lot of their own heavy lifting when it comes to plugging unfamiliar parts together. A blind rat, wired into a geomagnetic sensor via a simple pair of electrodes, can use magnetic fields to navigate a maze just as well as her sighted siblings. If a rat can teach herself to use a completely new sensory modality – something the species has never experienced throughout the course of its evolutionary history – is there any cause to believe our own brains will prove any less capable of integrating novel forms of input?

Not even skeptics necessarily deny the likelihood of ‘thought-stealing technology’. They only protest that it won’t be here for decades (which, given the number of us who expect to be alive and kicking 30 years from now, is not an especially strong objection). If we do stop short of a hive mind, it’s unlikely to be because we lack the tech; it’ll only be because we lack the nerve.

So I don’t think it unreasonable to wonder if one day, not too far from now, Netflix might change its name to Mindflix and offer streaming first-person experience directly into the sensory cortex. I suspect people would sign up in droves for such a service. Moore’s Law will work its magic.

What might that mean to us as individuals?

Ask one-half of the supersized self that the Hogan twins might have been, if their brains had fused just a little further up. Ask the poor bastard who awakened into a single hemisphere and had a few minutes to live some fraction of a life before the drugs wore off and his other half swallowed him whole. Oh, but you can’t ask him. He doesn’t exist any more. Right now he has as much individuality as your parietal lobe.

Consciousness remains mysterious. But there’s no reason to regard it as magical, no evidence of spectral bonds that hold a soul in one head and keep it from leaking into another. And one of the things we do know is that consciousness spreads to fill the space available. Smaller selves disappear into larger; two hemispheres integrate into one. The architectural specifics aren’t even all that important if Tononi is right, if the Cambridge Declaration is anything to go on. You don’t need a neocortex or a hypothalamus. All you need is complexity and a sufficiently fat pipe.

Does a thought know to turn back at the edge of one skull when the paths lead into another? Does an electron know the difference between a corpus callosum and a brain-computer interface? Titles in the popular press – ‘Google Search Will Be Your Next Brain’ – might not be so much ominous as childishly naive; they assume, after all, that ‘you’ will continue to exist as a distinct entity. They assume that brains can support multiple separate conscious attentional processes in the same medium.

Keep the bandwidth too low and you lose the experience; edge it too high and you lose yourself

Throughout history we’ve communicated via the equivalent of dial‑up, through speech and writing and images on screens. A fat enough neural interface could turn everything broadband, act as a next-gen corpus callosum that fuses we into some new kind of I that’s never existed before.

Of course they’ll put safeguards in place, take every measure to ensure that nothing goes wrong. Maybe nothing will. Keep your baud rate dialled back far enough and you’ll be fine. But there are always those who push the envelope, who might actively embrace the prospect of union with another mind. They’re not all that uncommon in transhumanist circles. Some regard it as an inevitable step in abandoning the flesh, uploading consciousness into a gleaming new chassis with a longer warranty. To others it’s a way to commune with the souls of other species, to share consciousness with cats and octopuses. It’s a fine line, though. Keep the bandwidth too low and you lose the experience; edge it too high and you lose yourself.

Even if you’re not into that kind of thing, you use the internet – which neuroscientists and game developers, even now, are reshaping into an explicit embodiment of neural intelligence. The web’s ɸ score isn’t going anywhere but up. And servers hiccup sometimes. Floodgates fail. Shit happens, and – as Batman’s butler once pointed out – some men just want to watch the world burn. Given the option, those folks might get tired of distributed denial-of-service attacks and leaked celebrity emails, they might try hacking Mindflix for Allah or the lulz. God help anyone who’s streaming the latest Marvel Total-Immersion Extravaganza when that happens.

These are some of the things we might want to start thinking about now – because they won’t matter that much to you after some failsafe has failed, or you’ve been talked into trying the whole mind-meld thing by someone who figured out how to disable the bottleneck. You might not care about the potential of an emergent consciousness built from silicon or a network of 1,000 brains, or whether logging out of a freshly integrated hive mind should be defined as murder or mere lobotomy.

Immersed in that pool – reduced from standalone soul down to neural subroutine – there might not be enough of you left to even want to get out again.