Looking Like a Brain Cell Isn't the Same as Being One, Study Says

Here's the puzzle. Dendrites — the branching arms that brain cells use to receive signals — look strikingly similar to frost patterns on a window, mineral crystals, or even the wiry networks you get from simple computer simulations. A new paper by Nima Dehghani argues we've been fooled by that resemblance. These shapes look alike, he says, because they all grow the same way, not because they do the same job. Shape, in other words, is cheap.

So what actually separates a real neuron from an ice crystal that merely resembles one? The answer, the paper argues, is feedback. A biological neuron carries an internal electrical state and helps generate the electric field around itself — and that field then reaches back and changes the neuron's own behaviour. The author calls this 'field closure': a looping conversation between a cell's shape, its electrical activity and the field it creates. In his mathematical model, the crystal-like 'ice neuron' is simply the version of this system where that feedback loop is switched off.

When he turns the feedback on, something interesting happens. A dendrite grown with no feedback, then 'frozen' in place, suddenly conducts signals further down its branches — without its shape changing at all. The effect is small and concentrated in a few shortcut pathways, not the main cables. Below the firing threshold, this feedback is almost the entire signal. During an actual spike, it's just a small correction to an already-strong electrical event.

The takeaway: a branching tree of dendrites isn't automatically a brain. You need the internal state and the self-generated field. This matters because 'brain-inspired' AI, lab-grown brain tissue and fossil neurons are often judged by appearance. One honest caveat — this is a mathematical argument, not a wet-lab experiment, so the biology still needs to confirm it.