Activating Memories with Light: How our paper came to be

The feeling of publishing a paper has two key components: the unforgiving roller coaster leading up to publication — a process that feels like an airplane is perpetually parked on the face of your self-esteem — and the part after the paper has been accepted — a process that feels like a double-thick oreo milkshake multiplied by world peace. Both processes are bound together by research. Research is what happens when you navigate onto the edge of what is known and unknown. Neuroscientists are cartographers of the brain, and this story is about how my first two years as a grad student in Susumu Tonegawa’s lab (T-lab) have been an adrenal gland-squeezing voyage to understand how uncharted neural waters make the wine of memory possible. I took the hippocampic oath; we called the voyage Project X.

The Black Room

May 26th, 2011 — a Thursday — was probably a sunny Boston day, perfectly suited for a brisk jog around the Charles River. It might have also been a gloomy, Londonesque day, perfectly suited for an Otto’s pizza and a movie in Harvard square. Heck, it might have even been the day that the Red Sox breezed passed the Detroit Tigers 14-1, but I wouldn’t have known. I don’t remember the details of how beautiful or dreary it was outside, or of a particular blowout at Fenway, because Xu Liu — a gifted postdoc in the Tonegawa lab — and I were sitting in a windowless black room, breathing quietly with our eyes peeled — superglued — on the monitor in front of us. 

I do remember that my palms were collecting more moisture than usual from wearing powder-free latex gloves. I do remember that, for 12 minutes, Xu and I didn’t say a word, or look at each other, or make any ocular motion that resembled an eye blink. I do remember that 2 minutes in, I had to sneeze violently, but the last thing I wanted to do was to distract our experimental mouse in the adjacent room. The black scruffy rodent was fully equipped with 2 optic fibers gently tucked into the left and right side of its brain. It was foraging around a square, white, almond-scented chamber. What the mouse didn’t know, however, was that a blindingly blue flash of light was about flicker on 3 minutes into the session to illuminate the dark recesses of its brain. This was our first attempt to reawaken a memory.

A mouse with an optic fiber in its brain

Our seemingly straightforward evening then took a turn that would take 6 months to come full circle.  The optic fibers were positioned right above a brain region crucial for forming the variety of memories that we hold near and dear — that memory of our first awkward kiss, or that time we got mugged at 2am, or that time we got into college. Aptly named in Graeco fashion for its meandering layers of cells that together resemble a seahorse, we were optically activating the hippocampus. At the 3 minute mark, like the bursts of sunlight that nudge a city awake, pulses of light began bombarding the mouse’s cognitive machinery, photons bouncing around neural tissue, waiting to nudge awake a dormant memory. Xu and I nervously gritted our teeth — waiting, wishing, wondering, watching. 

A Brief History of Broken Thoughts 

Why would brain cells respond to light, though? How was it even technologically possible to activate a fear memory with a brilliant blue glow?

The same decade in which Watson and Crick announced the structure of DNA, three seminal findings in neuroscience — one systematic, one accidental, and one tragic — paved the way for memory research. The underlying theme in all three discoveries was that memories reside in the physical stuff of thought — in brain cells.

Starting in the early 1920’s, the neuropsychologist Karl Lashley became one of the first researchers to systematically scoop out parts of the rodent brain in an attempt to relate the size of the damage to memory performance. He was intoxicated by the idea of finding an engram  the electrophysiological bolts of micro-lightning and biochemical cocktails that comprise a given memory. Frustrated after a decade’s worth of experiments, Lashley concluded that no matter how much damage he inflicted on the brain’s overlaying mantle, the cortex, he could never fully get  his rats to perform at chance levels on a maze. He simply couldn’t erase the memory. Indeed, it would take thousands of more experiments from independent labs to figure out why this was so. It turned out that, what psychologist Steve Pinker said about the neuroscience of swearing is also true about memory: “It engages the full expanse of the brain: left and right, high and low, ancient and modern.” Memories are now thought to be distributed across the brain. Lashley’s error was to assume that memories were localized to single brain regions. 

And then, in the same decade, the Canadian neurosurgeon Wilder Penfield accidentally discovered that, when it comes to processing memory, not all brain regions are recruited equally. A Princeton graduate and first-string football tackle, Penfield would treat several patients with epilepsy by carefully resecting the problematic neural tissue. However, in order to distinguish between healthy and problematic tissue, Penfield applied tiny jolts of electricity to the brain and reported his patients’ responses — some were innocuous, others were signatures of seizure onset. One observation, though, would hint that Inception-like experiments were actually possible. After applying electrical stimulation to areas of cortex near the hippocampus, about 5% of Penfield’s patients reported that they were unwillingly recalling vivid memories of past experiences. Electrically activating just a tiny lump of brain was sufficient to induce the recall of an entire memory in humans. 

Penfield’s work set the stage for the one surgery that would become a cornerstone of memory research. It would be immortalized in every neuroscience textbook at the expense of one man’s ability to thread and unify his overall sense of being across time. William Beecher Scoville was the neurosurgeon whose scalpel would carefully expose one of the most elegant, philosophically-charged principles of neurobiology: broken brains give rise to broken thoughts.

Henry Molaison — the mighty HM.  A Connecticut native, HM had two cashew-shaped lumps of brain removed to treat the epileptic convulsions that had so devastated his youth. Scoville had to lift up skin and cut through bone to reach an area right above HM’s ears, to reach and remove the hippocampus. After the surgery, HM was suspended in time as a 27 year old on September 1st, 1953. Without a hippocampus, he developed anterograde amnesia, the inability to form new memories. His story has even made its way into modern culture: Drew Barrymore’s forgetful character in 50 First Dates; Dory, the optimistic regal tang with the inability to form new memories in Finding Nemo; Memento’s Leonard Shelby, whose memory resets every few minutes and this forces him to tattoo experiences on himself as reminders to hunt down his wife’s killer. Forever stuck in the present, HM taught us that certain parts of the brain were more special than others in processing memories of experienced events. He lost his ability to bridge personal events across large spans of time and, in turn, gave us an unforgettable roadmap to navigating the brain’s memory systems.

With HM, neuroscientists had thrown the gauntlet down. They now knew where to begin to look, and the search for a neural correlate of memory intensified. Still, the brain speaks on the staggeringly fast timescale of milliseconds. Researchers have actually been able to eavesdrop on neural activity with hair-thin electrodes placed into the brain that pick up physiological signals. Manipulating neural activity, however, has usually relied on lesions or concoctions of drugs to inhibit activity. Lesions indiscriminately and irreversibly destroy neural tissue; drugs take minutes or even hours to have a full effect. Both are hard to restrict spatially in terms of the amount of brain they cover.

 On August 14th, 2005, everything changed. One remarkable paper got rid of these limitations and introduced a new method to probe the brain: optogenetics. If slow and irreversible manipulations were the fate of neuroscience, then as Yogo Bera once quipped, the future ain’t what it used to be.

Neuroscience is now undergoing a breathtaking facelift with the invention of optogenetics. Developed in Karl Deisseroth’s lab at Stanford, the tools of optogenetics allow the control of neural activity with pulses of light on a millisecond timescale — exactly the tool we need to study the brain at its natural speeds. The D-lab had tricked brain cells to turn on or off with light by expressing a light-sensitive protein called channelrhodopsin-2 (ChR2). In essence, neuroscience finally caught up to the speed at which neurons communicate.  Studying the brain went from 56k to cable with that single landmark paper that started a revolution.

Tantamount to remembering a first kiss, I remember where I was when I first read this paper: I was at a burger joint in Newton eating a roast-beef sandwich with au jus. After reading the paper’s final line of prescient text (“Thus, the technology described here may fulfill the long-sought goal of a method for noninvasive, genetically targeted, temporally precise control of neuronal activity…”), I muttered under my breath in exhilaration, au shit!”

Project X Reaches Escape Velocity 

In one of the most famous monologues in cinema history, Roy Batty, Blade Runner’s bioengineered android who slowly learns to develop human emotions, hints at the daunting task of finding a memory scattered in time: a memory in the brain is as lost as a tear in rain. Roy did not commit Lashley’s fallacy, but how do we find a teardrop in the brain?

Rather than playing neural target practice, the easiest solution is to let the brain tell us where memories are located. This strategy was the ace in the hole that we needed for Project X to work. Fortunately, brain cells leave genetic footprints of their recent activity. Whenever a brain cell is active, an exquisitely regulated set of genes are turned on or off in response. These genes are involved in cascades of microscopic events that help shape neural activity over time. One gene in particular, c-fos, has the ability to sense when the cell in which it is enclosed becomes active. With a few exceptions, and put simply, when a brain cell is active, c-fos is active.

How does c-fos “know” to turn on when a cell is turned on? In a very real sense, c-fos contains a biological sensor, its promoter, which itself contains a variety of elements that detect when a neuron has fired. During the process of becoming activated, a neuron lets in all sorts of particles to charge itself up as it gets ready to fire. Those particles initiate a series of molecular baton handoffs — adding phosphate groups here, binding to transcription factors there, shuttling proteins everywhere  — some of which make their way to the promoter of c-fos to activate it. It is here that genes and environments interact.

Within the hippocampus, we wanted to trick only the cells activated while forming a memory into expressing ChR2, thus making them sensitive to light. If we simplistically just tied the c-fos sensor to ChR2, then any cell that becomes activated would contain ChR2. This is a problem because shocking a mouse happens only once and it’s only this memory that we want to tag, not the memory of the mouse having its cage changed, or of it being handled by the experimenter, or of it digging for food. We wanted specifically to crystalize the fleeting formation of fear itself. This meant we also needed to open and close windows of when this tagging could even occur. Without this ability to induce the tagging process, then the simple rattling of a mouse cage could activate thousands of cells unrelated to our experiment. We needed a system that conferred temporal precision.

In 2007, a team of researchers genetically engineered exactly the mouse we needed: within the mouse’s genome, they had tied the c-fos promoter to the last piece of the puzzle that Project X was look for. It was an inducible element capable of activating other genes. Whenever a cell was active, c-fos would drive this element but only in the absence of a chemical called Dox, which we placed in the animals’ food pellets. Dox could therefore regulate the windows of labeling. After we introduced ChR2 to this system, it would only be expressed in active cells and only when we opened this window for tagging by removing Dox from an animal’s diet. When the animal ate food containing Dox, c-fos would no longer drive ChR2. If Karl put the opto in optogenetics, then Susumu emphasized that Project X needed to focus on the genetics part just as much.

A subregion of the hippocampus labeled with ChR2

Around the lab, we called the technique:


Engram Labeling Using a


Suppressible and





Or, Red Socks. (If you’re from New York, I’m sorry I’m not sorry. Also, it is impossible to find a biologically relevant term here beginning with the letter ‘x’).

Prior to May, over 3 years were spent developing and testing viruses, characterizing the mice we were using, and blueprinting other projects in case this one did not work.  On May 25th, 2011, we tagged the cells activated only when our mice were being shocked in a particular environment (“The Red Room”). On May 26th, we placed the mice in a completely separate environment (“The Black Room”), which it had no reason to fear, and we flickered on the light to illuminate the hippocampus.

The same way a coffee-soaked madeleine could jog up the remembrance of childhood memories for Marcel Proust, light had acted as the mouse’s madeleine. It is in the spirit of Proust that I describe the 3 minute mark on May 26th: 

A cold shudder ran through the mouse, intent upon the extraordinary thing that was happening. And suddenly the memory revealed itself. The thought was that of being shocked yesterday morning in the red room, where we fear conditioned the mouse and crystallized the fleeting formation of fear, which, through its echoey murmurs, glowingly presented itself in the form of ChR2-positive cells. From the mouse’s perspective, the whole of the fear conditioning chamber and its surroundings, taking shape and solidity, sprang into being, grid floors and the aroma of almonds alike….all this from furnishing a patch of hippocampus with light.

In just a few seconds, the mouse switched from exploring its surroundings to displaying fear. It ran to a corner, stood still, and crouched. It was immobilized by fear. It froze in place; light had awoken a fear memory. I secretly began calling the mouse Arnie for becoming a furry Arnold Schwarzenegger à la Total Recall. After this first experiment, my Google calendar went from 0 to looking like a bomb hit it overnight. Project X had tested positive for contact with reality.

Of course, there are plenty of alternative explanations. Perhaps the mouse was scared simply because it had two glass barrels digging into its skull and shooting out laser beams. Or maybe simply activating hippocampal cells indiscriminately caused a fear response. Or maybe there wasn’t anything special about c-fos-expressing cells after all. These pivotal control experiments kept us out of the Boston commons and in the Black Room until the day we submitted our findings, but one thing was becoming clear:  the physical stuff of light had activated the physical stuff of thoughts and triggered memory recall. Every control worked. Shining light on a fraction of hippocampal cells involved in fear learning was sufficient to push forward the cascade of events leading up to fear memory expression. It was the Penfield experiment with lasers rather than electricity.

Knights of Science 

Once we got our first result, it took only 6 months until submission because we were working as a team at the speed of the Manhattan Project on corticosteroids. (This is also why we hadn’t presented our data at annual neuroscience meetings, simply because the data didn’t exist yet.) Our lab has 60 people in it — UROPs and technicians and grad students and post docs — all providing the oil necessary for the moving pieces that would become our experiments. We also had the necessary bunsen burner flame under our fannies in the form of competition. To let the project reach escape velocity into the world of publication, Xu and I spent weekends implanting over 90 animals with optic fibers. We spent afternoons running behavioral and molecular protocols, slicing and staining brains, failing without flinching and succeeding without becoming complacent, unweaving the mind one brain-strand at a time. We spent evenings with a dozen members of the lab (“Team X”, we called ourselves) discussing the most recent results over burgers and Bushmills. Susumu met with us daily to infuse our routines with encouragement and to help us construct a cohesive narrative. 

Five minutes from Picower’s entrance is a well-lit Starbucks bustling with red-eyed students at all hours. (I wish I could say I was contractually obligated to drop company names here but, alas, half of my stipend still goes to grande iced coffees, black). It is here that Xu and I spent countless caffeinated hours putting together the manuscript, reading previously published papers, looking at graphs and asking ourselves, what the frappuccino are we looking at?
Balance is key, so admittedly, the lab and I also spent Monday evenings playing intramural softball against other departments and Tuesday evenings playing soccer. Trash talking the Media Lab was too good to miss. But, even on the field all I could think about was what the data meant and where the project was going. I felt Lashley’s intoxication. At our weekly lab meetings, we’d present a collage of cellular, circuit, and behavioral data to try to tease out a causal thread. It was pointillism with data points, and the picture of a reactivated engram was emerging, dot by data dot. Those 6 months consisted of the most team-oriented, collaborative, lab-wide effort I have ever experienced. Everyone contributed to the project and there was palpable excitement in the air. I breathed it, felt it. Lab members breathed it, felt it. These were nights of science. 

And then it happened. 6 months came full circle . At 4:55pm on November 1st, 2011, just before their offices in New York closed, we submitted our work to Nature. At 5:05 pm on November 1st, 2011, the T-lab and I vacuumed a few triple-hops-brewed beverages at a popular go-to bar that was a 10 minute walk from Picower’s entrance. During the toast, the name of the Cambridge bar — written somewhat wryly on a periodic table of a menu —  caught my attention and I briefly smirked. The name perfectly suited the occasion. The name of the bar captures every sentiment I have about working in lab, about making a career out of dissecting the moving pockets of chemistry that make up our minds, about manipulating memories with the ephemeral stuff of light, about working alongside lab members who endow my daily routine with excitement and friendship. The name encapsulates the remarkable intellectual muscle that humans are capable of flexing to uncover patterns in nature — humans noble in reason and infinite in faculties, driven by insatiable curiosity, capable of unparalleled mastery of the world without and god-like comprehension of the world within.

We ended the evening at The Miracle of Science.

Coda: That is the story of how our paper came to be. It was accepted on March 15th and came out exactly 7 days later in Nature. Susumu bought us bottles of champagne and this cake:

And through a combination of serendipity and timeliness, a paper addressing similar concepts on memory reactivation came out  in Science on the same day by Mark Mayford’s group from Scripps. Two independent lines of work spanning years of research and occurring in parallel managed to publicly materialize simultaneously down to the hour. Here’s the backstory of how that happened.

The following morning at 6:30am, while running on the panoramic Mass. Ave bridge that connects MIT to Back Bay, an extraordinary, orange-flavored skyline blushed through the buildings of Boston to give a glowing, rosy explosion of morning. A few splashes of light making their way passed the Prudential Tower’s windows and into Boston’s brownstones are all it takes to invigorate the highly caffeinated morning commuters, Dunkin Donuts in one hand and the Metro in the other. “This is nature’s Project X,” I thought to myself.

*Liu X., *Ramirez, S., Pang, P., Puryear, C., Govindarajan, A., Deisseroth, K., and Tonegawa S.Optogenetic stimulation of a hippocampal engram activates fear memory recall. Nature (2012), doi:10.1038/nature11028.


9 responses to “Activating Memories with Light: How our paper came to be

  1. Pingback: Fine-tuning optogenetics | medicalneurophys

  2. Thanks for sharing your story. Can you explain what is meant by “pattern completion” at ?

    • Hey John,

      Put simply, by “pattern completion” we simply mean the ability to mentally conjure up a full representation based on degraded cues. For example, a mouse would be pattern completing a memory if it was trained to solve a spatial navigation task by orienting itself with regard to multiple cues in the room, yet at a later time point was only presented with a subset of those cues and could still accurately figure its way about a maze. Or, for us non-mice, it would be as if I met you one day and, at a later time point, I was able to recognize you based only on the shoes you were wearing. Both the mice and I would be “completing” our mental image of a memory based on a fraction of the cues originally available.

  3. great read man and many props for the work conducted! i presented your paper at journal club in my department when it first came out.

    • Hey Ahmed,

      Thanks a lot man — it’s still pretty exciting, gratifying, and surreal to have the work out there in the real word. The adrenaline rush is the kind of stuff that makes academia a unique place to be in — please let me know if I can help out or clarify any details in any way! Maybe I’ll even bump into you during SfN; best of luck with your research too.

  4. Pingback: Ideas Worth Spreading: TEDx/Boston

  5. Heard you on the Larry Elder show talking about how it might be possible in the future to eliminate the emotional trauma from negative memories. This has been done with uniform success since the publication in 1950 of the book Dianetics the Modern Science of Mental Health by L. Ron Hubbard. Doctor, if you have a troubling memory that you would like to turn into a simple memory, and you can approach this goal without any scientific bias, with knowledge as your only goal, I suggest you read the book.

  6. Hey Steve,
    I have a question about the paper, because it’s exciting and I’m going to present it on journal club. Unfortunately, I don’t understand one thing, I hope you could explain it to me. It’s about Figure 4 and the statement “Immunohistochemical analyses revealed a chance level of overlap between ChR2-EYFP – positive and c-Fos- positive cells, suggesting the two distinct contexts”. You also showed on graph 4f that cells double positive for ChR2-EYFP and c-Fos are in minority. How is that possible; I thought that when there is an activation leading to expression of ChR2-EYFP, there is also expression of c-Fos (because it’s under the same promotor)? Can you explan me what’s the catch?

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s