This article is more than 1 year old

Building bits of brain in the lab will change our minds

Wake up and smell the cortex

Opinion In Hollywood, brains in jars mean three things. Mad scientists at work, we are tampering with forces we cannot control, and something monstrous is coming. We are now making brains in jars for real: the mad scientists are at Johns Hopkins University and the forces they are tampering with include video gaming. Yes, there will be new monsters, but some old ones will be slain along the way.

No need to rush to Amazon for next-day delivery on pitchforks, though. The scientists are no madder than anyone else who chooses to work in academia these days, and are carrying on work that's been brewing worldwide or nearly two decades. The brains are tiny lumps of nerve cells called organoids, the jar is a Petri dish, and their sole superpower is playing Pong, that game with two white rectangles and one whole square. Last week's hype, that these could bring the astonishing computational capabilities of the brain to AI, is true and almost entirely irrelevant. It'll be science fiction for a very long time. Organoids are far more important than that.

This all began with the observation that some adult animals could regrow limbs, replicating the way embryos turned from bundles of identical cells into vastly complex systems. All cells that have DNA have the information to do this, but only some – called stem cells – can.

Experiments with stem cells taken from embryos produced the first organoids, or miniature tissue bundles replicating organ structure and function, but it was the ability to turn ordinary cells from adult humans back into stem cells and repeat the experiments that kicked things into high gear. Human embryo research is highly regulated, but nobody much cared about the ethics of a skin cell. Human organoids were on the menu. This all happened a while back: The Scientist named organoid research one of the most important developments of 2013, the same year that formal protocols were introduced for cerebral organoids – ones that replicated structures within the human brain. Off to the races.

Ten years on, and we have Johns Hopkins serving up a dish of 100 million human neurons capable of interacting with stimuli, learning recognition and reaching decisions. These are not simulating or modelling the neurological processes that build up to consciousness in our heads, they are those processes. That doesn't mean they are conscious, nor that they will be soon. We now know a great deal about how brains are structured into modular columns handling standard tasks, combining into larger functional units through a hugely complex structured network; the cerebral organoids in the Johns Hopkins dishes are by neuron count barely one three-millionth of all that. They're as far from the full function of what's between your ears as a Neolithic settlement is from California. The developmental process to get from one to the other is not going to yield its secrets easily.

Make no mistake, the unique research potential for learning about brain development, structure and function is hugely exciting and important. If you care about what some called the hard question of AI, and others cognitive science, then this is the field to watch. But it won't be commercially interesting for many decades, if at all, at least not in the model suggested of super low-energy, massively efficient data analytic add-ons for more conventional computers. Even keeping the blobs fed and alive at scale is a roadmap of mysteries. Talking about "organoid intelligence" is a fine and honorable way to fund further research, in these days where AI/ML is hyper-fashionable, but it's not the real meat.

There are three areas among many where organoid research is most important and will see the most immediate impact on society. By getting organ development out of the organism and into the dish, it immediately makes basic research on healthy and diseased organs much more accessible. It opens up genetic modifications that are simply unavailable otherwise, and makes them accessible. For cerebral organoids, this means laboratory access to degenerative and disruptive processes that cause huge suffering.

Another important area is innovation. Performing innovation on living complex animals, not just humans, is much more ethically fraught than investigating illness, which is itself fraught enough. A glance at the allegations festering around Elon Musk's Neuralink project shows how badly that could go: nobody worries about Petri dishes. So here, brains in jars actually guard against mad scientists.

Which leads to the third and, for cerebral organoids in particular, most significant aspect of this work: ethics. More precisely, by opening up the way to Blade Runner levels of artificial systems that don't emulate but recapitulate biology, we are finally giving ourselves the tools not only to investigate what being a conscious entity means, but demand that we do that investigation. Biologists now see consciousness not as a uniquely human attribute, but a spectrum of awareness that stretches deep into other species with boundaries we neither see nor understand. By what right do we give ourselves special status? And for those who hold, in the only mildly hyperbolic words of Monty Python, that every sperm is sacred, what when that's true of every flake of dandruff? In such matters, many researchers and clinical staff are already called monsters: doubtless this research will create more such.

It's easy to boil AI/ML ethics down to "Will it hurt me and how do I stop it?" It's much harder to be so simplistic when empathy for the machine is based on common ancestry. The world that will evolve from the organoid Petri dish will change the way we see ourselves, and we can make monsters or angels from that as we wish. But we will have to make that choice. The days of Pong won't last for long. ®

More about


Send us news

Other stories you might like