Advertisement

What did you do today? Your brain ‘remembers’ it cell by cell

Share

No one will forget the amnesiac Henry Molaison.

An epileptic who had his hippocampus removed in 1953, Molaison lost the ability to record episodic memory. He had no idea if he’d eaten lunch. He greeted acquaintances each time as if it were the first.

Largely because of Molaison, who died in 2008, researchers discovered that humans can’t convert the experience of rapidly changing events into memory without this small, sea-horse-shaped brain structure. Decades of experiments confirmed this.

There was one big problem. No one seemed to be able to identify the neurons associated with coding these “episodic” memories.

Advertisement

Brain imaging studies showed signs of activity on a broad scale, but on the rare occasions when scientists got to put electronic probes into the human hippocampus, the coding pattern proved elusive. The hippocampus remained a kind of black box between experience and memory.

“It’s kind of paradoxical,” said John Wixted, an experimental psychologist at UC San Diego. “We know that the hippocampus is involved in episodic memory, yet when you go looking for the neurons that do the job of episodic memory, they seem not to be there. … And there turned out to be a very good reason.”

In a study published online Monday in the journal Proceedings of the National Academy of Sciences, Wixted and fellow researchers suggest that memory processing takes place in an exceptionally small fraction of neurons scattered widely in the hippocampus. The team also included researchers from the Barrow Neurological Institute in Phoenix, Veterans Affairs Medical Center of San Diego, the University of Montana, Louisiana State University and Arizona State University.

Computer modelers had long ago suggested that more localized or pattern-based coding schemes would lead to overload, causing our brains to jettison one episode to process another. Neuroscience was essentially casting lines randomly in a large lake, looking to hook a very finicky and solitary fish, in other words.

“We’re only measuring a few neurons in the hippocampus,” Wixted said. “But if each memory is coded by a small fraction of neurons, the odds that your micro-wire landed on a neuron associated with a memory is really low.”

Epileptics no longer have their hippocampuses summarily removed, but some still require surgery. To map that surgery, doctors implant electrodes in their brains. Researchers leap at the chance to test these patients during their hospital stay, and that’s what neuro-engineer Peter Steinmetz at Barrow offered up to Wixted: nine volunteers.

Advertisement

Researchers briefly presented the volunteers with a set of words, then mixed those words with others. All they had to do was rate, on a scale of 1 to 8, which ones seemed novel and which ones familiar. Over multiple trials, each electrode measured activity of a single neuron or a small group of them.

The researchers calculated the difference in responses to familiar versus novel words, and detected episodic memory coding.

But researchers were soon bitterly disappointed when they looked for areas where signal spikes were stronger in response to all the familiar words. Only about 5.5% showed a stronger average response, and that was about what would be expected by chance.

“That is at first puzzling,” Wixted said. “But we should’ve realized from the beginning that it would be that way. They’re not responding generally to the targets, they’re responding to one of them.”

Researchers took a closer look at the data to separate the outliers, using a statistical “quantile” analysis not often used in their field. It showed the data fell in a pattern that the computational experts had predicted: Episodic memory neurons essentially were few and far between.

“I literally almost fell out of my chair,” Wixted recalled of the results. “The way I tell the story now is we’re confirming long-standing ideas, but to be honest with you, I didn’t think it would come out that way.”

Advertisement

The result won’t necessarily change the way we treat diseases that affect memory, such as Alzeheimer’s. But it could make the research path more clear, Wixted said.

“The degeneration is happening at the level of individual neurons,” Wixted said. “It’s important to know what those neurons are doing in order to make that memory happen.”

Advertisement