When someone encounters a new experience—say, a terrifying rabbit—groups of neurons in their brain fire together, the connections between them become stronger, and molecules accumulate at the places where neurons meet. Many scientists believe that these preserved patterns of strengthened connections are the literal stuff of memories—the physical representations of the things we remember. These connected neuron groups are called engrams.
When people bring up old memories, the engram neurons fire up again. They also enter a brief period of instability, when the molecules that preserved the connections between them disappear and must be remade. This process, known as reconsolidation, means that humans are partly reconstructing our memories every time they bring them to mind. And it means that the act of recollection creates a window of time in which memories can be updated, and fears can be unlearned.
“That was the theory,” says Daniella Schiller from the Icahn School of Medicine at Mount Sinai. “It’s been speculated, but [this new study] is one of the most direct demonstrations so far.”
Khalaf and his team, led by Johannes Gräff, worked with a special strain of engineered mice that are completely normal as long as they can eat a drug called doxycycline. If you remove the drug from their meals, a series of genes kicks into action, and drops a distinctive molecule into any active neuron. In this way, the rodents automatically label their own engrams. Whenever they learn something new, or recall an old memory, the buzzing networks of neurons in their heads get tagged.
The team made good use of this feature in a simple experiment. They trained the mice to fear a small box, by putting them inside and giving them some mild electric shocks. A month later, the team took the rodents off doxycycline and put them back in the same box. They froze—a clear sign that they were remembering their old distress. Meanwhile, they were labeling all the neurons that fired during this moment of recollection—the fear engram.
Later, Khalaf put the mice through exposure therapy, repeatedly returning them to the scary box without any accompanying shocks. As these sessions continued, their fear started to subside. But here’s the crucial bit: The more closely they reactivated the neurons from their original fear engram, the more thoroughly they shook off their fear.
Without the former, the latter doesn’t happen. Khalaf showed this by chemically silencing the rodents’ labeled neurons, and preventing them from reactivating their fear engrams. When he did this, the mice responded less well to their rounds of exposure therapy. But if Khalaf instead boosted the activity of the engram neurons during the rodents’ therapy sessions, they lost their fears faster than before.
“It’s an important advance, in that it suggests, for the first time, that extinction of fear involves the modification of the original fear-inducing memory,” says Jelena Radulovic from Northwestern University.