This is a technology site, and we don’t normally write about medical tech. But since I’m interested in brain-computer interfaces , I was struck by science writer Ed Yong’s article on “ model human brains ” last month in National Geographic. They’re jelly-bean-sized neural lumps that include rudimentary retinas and cortices, layered similarly to normal human brains, recently grown in an Austrian lab. The trick was to take human skin cells, chemically rewind them into stem cells, then coax these to become neurons. I thought what any reasonable person would: This is it. This is how we end up making those brain plug human-machine interfaces in The Matrix.

As Neuroskeptic speculated on the Discover magazine blog, I too wondered about the ethics of growing human brains in laboratories. What if the thing that develops from human progenitor cells is eventually able to complete the program encoded in its genes? What if we end up with Terminators? This process is cut short in the current versions, because the brains have no circulatory system, no blood to bring oxygen and nutrients to the cells in the center of the blob, so they develop malformed with all the sophisticated parts on the outside, “like a car with its engine on the roof” as one author of the paper put it. But some individual brain parts develop much like a normal embryo’s, up to the ninth or tenth week.

All my heady sci-fi visions were scuttled by some grounded brain experts, but I learned some fascinating stuff. Here’s everything you ever wanted to know about “mini-brains.”

Matt Weber, University of Pennsylvania, Post-Doctoral Fellow, Psychology Department: One thing to keep in mind is that these little guys are developed to the equivalent of nine weeks old. That limits what you can do with them because they aren’t responsive to the outside world, so their development is insulated from the organizing influence of sensory input. And this influence kicks in quite early in development: Fetuses move their limbs, they taste what their mothers eat, they learn their parents’ voices. So the ability to generalize from anything we observe in these disembodied brains to even a 20-week-old fetus might be quite limited.

The upside is that we might be able to separate genetic contributions to development from experiential ones. But that assumes we can reconcile ourselves to the ethics of developing a disembodied brain to a stage where, if it were embodied, it could respond to the world. That’s a deep rabbit hole, and I won’t venture in except to note that “responding to the world” is a pretty low bar for meaningful mental life: Vegetative patients still have brainstem-mediated pupillary responses to light, and retinas detached from a brain not only respond to light, but actually can learn about regularities in visual displays. Which isn’t to say there might not be good reasons to resist developing a disembodied brain that far–it’s just that it’s a complicated question.

Ania Dabrowski, University of Michigan Medical School, MD-PhD Doctoral Candidate, Neuroscience:

This paper is a developmental neurobiologist’s wet dream. Deeming this a “mini-brain” is a bit of a stretch, however it’s important to put this into the context of how we’ve been studying human neural development. The most common model for studying these questions in humans are mice. Because they’re mammals, and they’re kind of similar to us. But a mouse is not a human, and we don’t know how so (kind of weird thing to say, right?). It seems obvious that there are differences between how human and mouse brains develop, but our resources to study this are limited!

In comes mini-brain. Suddenly, you can take human cells, and see how they form rudimentary brains. And you can manipulate human genes, and see how they are relevant to cortex development. Sure, it’s not a real human brain. But it’s also not a mouse brain.