For more than half a century, neurologists have thought of long-term memories as pieces of short-term memory simply shuffled into cold storage.
New research on the circuits responsible for memory suggest this model might have it all wrong, with both forms of recollection created simultaneously with the long term version kept silent.
The MIT study on consolidation of memory took previous research involving the labelling and activation of specific 'memory' cells to the next level, forcing mice to react to certain past experiences at the flick of a switch.
If you've ever watched the 2000 psychological thriller Memento, you'd be familiar with the concept of short-term memory loss. While the movie is fiction, the lead role played by Guy Pierce was loosely based around a man by the name of Henry Molaison.
In 1953, the 27 year old Molaison underwent brain surgery in an attempt to cure the epilepsy that had plagued him for most of his life. While the treatment was a success as far as the seizures went, Molaison lost his ability to remember anything for more than about half a minute or so.
Memories formed prior to the surgery were left intact, and he was still able to learn new procedures and broad concepts – Molaison simply couldn't recollect events in the form of mental images or stories.
As tragic as this was for Molaison, neurologists learned from the damage done during his surgery that long-term memory formation had something to do with the hippocampus, while their actual storage occurred somewhere else in the brain, such as the outer layers called the neocortex.
This shuffling of memories led researchers to believe short term memories are formed and stored in the hippocampus before being transferred as a whole package into long-term storage.
A second model called the multiple trace theory of human memory was proposed later in the 1960s; based on the observation of people recalling a sense of an event without the details, it was suggested the hippocampus might hold onto traces of a past episode while the neocortex stores a general sense of familiarity in long-term memory.
Since specific memories aren't stored in localised bunches of cells that are the same in all people, studying memory has been rather challenging, mostly relying on chance encounters with people who have had strokes, trauma, or lesions.
So providing evidence for one model over another has been slow, until recently.
In 2012, researchers at MIT found a way to label neurons called engrams that are involved in specific memories, allowing them to trace the circuits involved in recalling an event.
More excitingly, they found they could stimulate these cells using light channelled down optic fibres, literally switching them on and off to make them instantly remember an anxious event.
Now researchers have used these tools to discover long-term memories develop at the same time as short-term memories, only stay quiet while they mature.
"This is contrary to the standard theory of memory consolidation, which says that you gradually transfer the memories. The memory is already there," said researcher Takashi Kitamura.
To study how memories are made in mice, the team labelled memory cells in the hippocampus, the prefrontal cortex, and an area known to play a role in fear responses called the basolateral amygdala.
The mice were then subjected to mild electric shocks in a situation intended to create a fearful event; just one day later they were found to be forming memory networks in both the hippocampus and the prefrontal cortex.
Tellingly, while both could be 'switched on' to force the mouse to remember its ordeal, only the hippocampus cells activated in moments of natural recall.
Two weeks later, the situation was reversed; while those cells in the hippocampus could be forced to work when hit with a pulse of light, the mice no longer used them to remember, activating only the prefrontal cortex 'long-term memory'.
"They're formed in parallel but then they go different ways from there. The prefrontal cortex becomes stronger and the hippocampus becomes weaker," said researcher Mark Morrissey.
More long term studies are needed to see if the hippocampus completely loses all trace of memory after a while, yet researchers will need better techniques to trace engrams for a longer period.
There is also now the question of how the long-term memory cells 'mature' – while it's clear cutting off communication with the hippocampus impedes the building of new long term memories, the exact mechanisms are unclear.
It's exciting to imagine ways we might be able to rapidly reinforce the memories we want while erasing those which might be painful.
Until then tattooing names, addresses, and numbers on your body might be the only sure-fire way to make certain you don't forget.
This research was published in Science.