The Engram

The Engram

Karl Lashley spent thirty years looking for the memory trace. Beginning in the 1920s, he trained rats to navigate mazes, then systematically destroyed portions of their cerebral cortex with a soldering iron, then tested whether they could still find the way. He expected to find the lesion that erased the maze — the specific cortical region where that memory was stored. He never found it. No matter which portion of the cortex he destroyed, the rats retained something. The severity of impairment correlated with the total amount of cortex removed, not with which part was removed.

Lashley called this mass action: memory was not stored in a location but distributed across the whole cortex. He called the corollary equipotentiality: any part of the cortex could support the memory, at least partially. After three decades of meticulous lesion studies, he wrote: "I sometimes feel, in reviewing the evidence on the localization of the memory trace, that the necessary conclusion is that learning just is not possible."

The joke conceals the real problem. Lashley had proved that memory was not localized. He had not asked whether it was neural.

In 1955, James McConnell trained planarian flatworms to associate light with electric shock. The worms learned to contract at light alone — classical conditioning, unremarkable except for the organism. Then McConnell cut them in half. Planarians regenerate: the head half grows a new tail, and the tail half grows a new head, complete with a new brain. Two weeks later, both halves retained the conditioned response. The half that grew a new head — with a brain that had never experienced the training — remembered.

McConnell pushed further. In 1962, he ground up trained planaria and fed them to untrained ones. The cannibal worms learned the conditioned response significantly faster than controls that had eaten untrained worms. McConnell proposed that memories were encoded in RNA, transferable by ingestion. The claim was ridiculed. McConnell was marginalized. The work was dismissed for decades.

In 2013, Tal Shomrat and Michael Levin revisited the question with automated training systems that eliminated the handling confounds McConnell's critics had raised. They trained planarians to navigate across a rough-textured surface to reach food. They decapitated the worms. After two weeks of head regeneration, the worms retained their learned preference. The result held across controlled conditions McConnell never had access to. The memory survived the destruction and complete replacement of the organ that was supposed to contain it.

Beatrice Gelber had already gone further. In 1952, she trained paramecia — single-celled organisms with no neurons, no synapses, no brain, no body plan — to approach a wire that had been paired with bacterial food. Trained paramecia approached the wire when it was presented without food. Untrained paramecia did not. She demonstrated retention over three hours with spaced training. She showed the response depended on light — the paramecia used photosensing to detect the wire, meaning they were discriminating stimuli, not just reacting to chemical gradients.

Most remarkably, Gelber demonstrated the spacing effect: spaced training produced lasting memory, massed training did not. The spacing effect — one of the most robust findings in learning science, replicated in bees, birds, rats, and humans since Ebbinghaus described it in 1885 — appeared in an organism with zero neurons. The phenomenon we have studied for 140 years as a property of neural systems appears to be a property of living cells.

Gelber's absence from the historical record, as a recent review noted, testifies to the prevailing orthodoxy that single cells cannot learn. The finding was not refuted. It was ignored. The premise that memory requires neural architecture was too strong to be challenged by a paramecium.

Physarum polycephalum is a slime mold — a single-celled organism that can grow to cover a square meter. It has no neurons. It navigates spatial environments by extending tube-like pseudopods, retreating from unfavorable conditions, and optimizing its network geometry for nutrient transport. Researchers at Hokkaido University demonstrated that Physarum could encode spatial memory in the diameter of its tubes: thick tubes mark paths the organism has found productive, thin tubes mark paths it has abandoned. The tube architecture IS the memory. There is no separate storage system. The organism's shape is its record of what it has encountered.

When Physarum was subjected to periodic environmental stresses at regular intervals and then the stresses were removed, the organism continued to anticipate the stress rhythm — contracting at the intervals when it had previously been stressed. It remembered a temporal pattern and it encoded that pattern in the physical oscillations of its body.

Lashley looked for the engram in the cortex. McConnell found it in the body. Gelber found it in a single cell. Physarum carries it in the geometry of its own shape.

The engram was never missing. It was never exclusively neural. Memory is what happens when a system retains enough constraints that past states continue to influence present behavior. Synaptic weights are one implementation. Long-lived plasma cells in bone marrow, maintaining antibody production against a pathogen encountered decades ago, are another. Tube diameter in a slime mold is another. The spacing effect in a paramecium — whatever cellular machinery produces it — is another. Each medium stores differently. Each stores.

The question was never where in the brain the memory is kept. The question is why we assumed the brain was the only place to look.