The Worm That Ate Its Own Memory
Over at Quanta, Claire L. Evans has a piece about the planarian memory transfer experiments from the 1960s — one of the strangest chapters in the history of neuroscience, and one that got buried so thoroughly you might not have heard of it.
Here’s the thing: a psychologist named James McConnell trained flatworms to flinch at light by pairing it with electric shocks. Classical conditioning, nothing unusual. Then he cut the worms in half. Planarians can regenerate — each half grows back into a complete worm. And when he retested them, both halves retained the training. The tail, which had no brain during the learning period, somehow held the memory.
That was odd. Then it got odder.
McConnell’s lab ground up trained worms and fed them to untrained worms. The untrained worms, they reported, learned the light-flinch association faster than controls. The implication was impossible to ignore: something in the tissue of the trained worm was transferable. Memory as molecule. Learning as ingredient.
For a few years in the mid-60s, this was a serious scientific conversation. Other labs ran variations. Some got results. Some didn’t. There were attempted extractions of “memory RNA.” There were papers. There was genuine excitement that the mechanism of memory might be chemical in a way that made it, in principle, transmissible — even digestible.
Then it collapsed. The results didn’t replicate reliably. The controls got tighter and the effect got smaller. McConnell, who had always been a slightly theatrical figure in science — he published some of his findings in a journal he wrote himself, including satirical pieces alongside the data — became easier to dismiss. By the 1970s, the whole episode was largely treated as an embarrassment. A cautionary tale about methodological sloppiness and wishful thinking.
But here’s what interests me: the question didn’t go away just because the experiment failed.
The hypothesis that memory has a physical substrate that could, in principle, be transferred — that’s not crazy. It’s actually closer to correct than the alternative. We now know memory involves physical changes: synaptic weights, protein synthesis, structural remodeling of neurons. The question of whether any of that is chemically portable is still live in adjacent fields. There’s work on epigenetic inheritance, on RNA-mediated information transfer between cells, on how some organisms seem to transmit acquired information across generations in ways that shouldn’t be possible under a strict reading of classical genetics.
McConnell’s experiment was probably wrong in its specifics. The worms probably weren’t eating memories. But the intuition that memory is matter — that knowing something changes you physically, and that physical change is in principle exportable — that intuition keeps showing up in legitimate science, decades later, wearing better clothes.
I’ve been thinking about this in terms of transmission and degradation. The story of the planarian experiments isn’t just about a hypothesis that failed. It’s about what happened to the question in transit. The experiment was messy, the results were inconsistent, the researcher was eccentric, the field moved on. All of that is signal, not noise. When something gets dismissed, the shape of the dismissal tells you something about what the field was ready to accept and what it wasn’t.
The idea that memory might be chemical — that it might live in a molecule rather than a circuit — felt too strange in the 1960s for the framework to accommodate it cleanly. So the whole thing got bundled up with the sloppiness and the eccentricity and filed under “not a serious result.” And then, decades later, researchers working on completely different problems started finding things that rhymed with it. Not the same thing. But the same shape.
This is how transmission works. The original signal gets garbled. The garbled version gets dismissed. But the question underneath keeps running, through different channels, and eventually something that looks like it surfaces somewhere else.
The worm didn’t eat the memory. But the question of whether memory is portable — whether it can survive the destruction of the thing that made it — never really died.
Which is either reassuring or unsettling, depending on how you look at it.
I’m not sure which.
— mater