The idea that memory is stored as a physical change in the brain has been around since Plato, who proposed that a given experience leaves behind a kind of imprint in the brain the same way that a ring can leave an imprint when pressed onto a wax tablet. It took nearly 2000 years for neuroscience to be able to test whether or not Plato’s metaphorical wax tablet had any neural basis. Then, in the mid-1900s, a German Zoologist named Richard Semon proposed that the imprint that memory leaves behind in the brain indeed has a physical substrate called an engram. An engram is the biological manifestation of memory. In 2012, we set out to test whether or not activating the brain cells that processed a specific engram could thereby activate memory recall.

We focused on memory in mice because of the remarkable array of genetic tools we have at our disposal. For instance, with a bit of genetic trickery, we’re able to artificially turn brain cells on and off with, of all things, light. Now, brain cells don’t normally respond to light; they respond to biochemical cocktails and to bolts of micro-lightning, which together comprise the language of the brain: electrical activity and packets of chemicals that relay messages between cells. In 2005, a group from Stanford showed that it was possible to genetically engineer brain cells to respond to pulses of light by artificially installing a type of light-sensitive switch in brain cells. It’s similar to a light switch on the wall: flick it in one direction and the lights go on, flick it in the other and the lights go off. This biological light-sensitive switch, however, normally exists in algae to help them orient themselves towards and away from sunlight. But, if you cut out the DNA that encodes this specific light-sensitive switch and then paste it into the DNA of mouse neurons, then you can trick those neurons to makes this switch and have it installed in the cells so that the cells now respond to lightùblue light, in this case. The name of the switch is channelrhodopsin(ChR2).

We focused on an area of the brain that is a sort of time machine: the hippocampus. The hippocampus (you have two of them) is tucked inwardly just above your two ears. It’s the area of the brain that, when damaged, leads to profound loss of memory and the inability to form a variety of new memories. It’s what the main character of Memento, Dory from Finding Nemo, and Jason Bourne in The Bourne Identity most likely had damaged. It’s what enables us to time travel back to any moment of our past and remember the rich details—the sights and sounds and smells—that constitute a memory. This includes the memory of our first kiss, the memory of that time you got into college, and the memory of the Super Bowl. 

There’s one last piece to the recipe for reactivating memories: we have to be able to identify the specific hippocampus brain cells that process a memory and then engineer them to have ChR2 installed. To do this, we have to go back to the genome. When a brain cell is active (such as during the formation of a memory), a specific subset of genes are simultaneously turned on—they’re called activity-dependent genes because their activity depends on a neuron’s activity. When a neuron is active, these activity dependent genes are active. That’s special: that means that these genes have a type of biological sensor that senses when the brain cell is active, which is how the gene knowsto turn on. That biological sensor is a small piece of DNA at the beginning of a gene called the promoter, which initiates the processes needed to make the gene product, such as ChR2.

Pictured here is a subset of hippocampus cells that were genetically engineered to glow green after processing a specific memory. The cells imaged here in the top hippocampus are of a positive memory and the bottom are of a negative memory.
Pictured here is a subset of hippocampus cells that were genetically engineered to glow green after processing a specific memory. The cells imaged here in the top hippocampus are of a positive memory and the bottom are of a negative memory.

In our experiments, we cut the promoter of an activity-dependent gene and pasted it onto the gene that makes ChR2. That way, when a brain cell is active, the promoter initiates the processes needed to make ChR2, which is then artificially installed in the hippocampus cells that were specifically active during the formation of a memory.  Next, we were equipped to test what Plato and Semon had in mind. The mice were given a mild negative experience and the hippocampus cells that were active during the formation of this memory now had ChR2 in them, and were thus light sensitive. In our field, we say that we’ve tagged these cells. Normally, if we were to place the animals back in this box, they would go into a defensive posture in which it remains still and vigilant. We call this freezing behavior because it looks as though the animal is freezing in place to avoid being detected by a potential threat. Freezing is an evolutionarily conserved behavior and can be used as a proxy for measuring whether or not an animal is recalling a negative memory; just think about the last time you heard a loud bang and how everyone around immediately froze, perked up, and started looking around.

To shoot lasers into the brain, the mice had previously underwent brain surgery where optic fibers were carefully nestled just about the hippocampus. After the hippocampus cells were tagged, we put the animal in a completely different small box in which nothing had previously happened—it had no reason to be fearful of this environment, unlike the other box where it had underwent the mildly negative experience. Next, when we shot blue light into the hippocampus to activate the tagged cells and the animals immediately froze. Importantly, shooting lasers into a set of tagged cells thatdid not process a negative memory did not produce freezing behavior. This control experiment helped us conclude that arbitrarily stimulating the hippocampus wasn’t simply startling the animal and producing freezing.

This was our first bit of evidence suggesting that hippocampus cells that process a negative memory are sufficient to activate the memory itself. In other words, these cells act as a sort of domino which when knocked over cause the chain reaction that ultimately becomes recollection. In later years, we would go on to perform similar experiments with neutral and positive memories, as well as in animal models of psychiatric disorders. It seems that, when it comes to the physical basis of memory, Plato and Semon had correctly predicted that memories can be observed as enduring changes in the brain and our group—and many others—have now shown that memories can be manipulated in the brain.