To remember is to conjure the ghosts of places and people that no longer exist. Diving to touch the tiled bottom of the pool as a child and feeling the pressure in your ears. Walking through an orchard with your family and plucking apples from tree branches. Scenes like these float imperceptibly in our minds before a smell or sound fastens them into existence.
Although memories seem ethereal, scientists believe that they may be stored in the connections between neurons called synapses. In theory, a map of a person’s brain charting the location of each neuron and synapse could be a record of memories spanning a lifetime.
Having such a map, known as a connectome, would transform our understanding of the human brain and consciousness. By comparing the neural wiring of healthy and unhealthy brains, researchers could design new treatments for mental illness. Pushing the connectome to its most extreme, enthusiasts imagine a future in which people could achieve a form of immortality by uploading their memories into robots.
The promise of the connectome, though, is matched by the challenges inherent in mapping it.
The human brain, with its roughly 86 billion neurons and 100 trillion synapses, is nearly infinite in its complexity. And although scientists have already embarked on mapping specific dense tangles of neurons, it would likely take them thousands of years to scan an entire brain.
As Sebastian Seung, a prominent neurologist at Princeton University, has declared: “Finding an entire human connectome is one of the greatest technological challenges of all time. It will take the work of generations to succeed.”
Against this backdrop, Eugene Lee is toiling in a windowless room — part laboratory, part library — on the campus of the Massachusetts Institute of Technology. Lee, a Ph.D. candidate in the Department of Brain and Cognitive Sciences, has spent the past four years working with the connectome of worms. It is a project that, while lacking the grandeur of mapping the human connectome, could answer a fundamental question: How do animals learn?
Lee, on the verge of his 30th birthday, wears striped dress shirts and speaks in smooth tones, with hints of an English accent shaped by his time as an undergrad at Imperial College London. His desk is adorned with paraphernalia — colorful diagrams and a stuffed animal — related to the tiny roundworm known as C. elegans.
Lee has spent an enormous amount of time with C. elegans, analyzing their cognition by reprising the iconic study performed by Ivan Pavlov a century ago.
In Pavlov’s experiment, researchers taught dogs to associate the ringing of a bell with food. This type of behavior, termed classical conditioning, is a simple but powerful type of learning.
Lee is applying the same model to worms. The worm’s connectome — which was first mapped 30 years ago — tells him which nerve cells, amid the otherwise indecipherable tangles of neurons strung across the worm’s body, could be working together during the learning process. But only by teaching them to respond to stimuli can he understand how information actually flows through the nervous system.
If his research is a success, other scientists may be inspired to use Lee’s methods with more complex animals. Perhaps it will help convince researchers that mapping a mouse connectome or even the human connectome is worth the effort.
But Lee has discovered that there are many hills to climb before that point. There is, after all, no manual for training worms.
“You have to get into the mind of the worm,” he said on a recent day. “Why would it care to learn? What matters to the worm?”
To the naked eye, C. elegans seems unremarkable. In a Petri dish, the worms resemble pieces of lint scattered over a smooth layer of a beige, jelly-like substance; they spend most of their lives searching for bacteria to eat. But, under the microscope, they transform into supernatural creatures.
“You have to get into the mind of the worm. Why would it care to learn? What matters to the worm?”
The worm’s body is transparent and, as it undulates, it shimmers with different textures. Smooth round eggs are packed in a tight row. The intestine is dark and grainy. The worm’s long torso, pockmarked and gray, is dully luminous like the surface of the moon.
Lee is drawn to C. elegans because of its tantalizing combination of simplicity and complexity. In comparison to humans, C. elegans only has 302 neurons and 7,000 synapses. Although worms appear too simple to mimic human cognition, they actually have a striking ability to learn and form memories. This makes them the perfect test case for using the connectome to explain animal behavior.
“Studies of worms and of other simple and highly tractable animals, such as the fruit fly, have yielded enormous insights into how nervous systems work,” said Robert Horvitz, Lee’s thesis adviser and an MIT professor of biology who in 2002 was awarded the Nobel Prize in physiology and medicine along with two other scientists.
For Lee, the experiments are meditative tests of endurance. Training one worm takes 45 minutes, and a typical session, when he handles multiple animals one after another, lasts up to eight hours.
To conduct his “lessons,” Lee exposes each worm to a violet laser light, which it hates, and a fruity-smelling alcohol, which normally elicits no reaction. After 10 to 20 sessions, C. elegans learns to associate the two, reacting to smell alone by gagging or spitting up any bacteria it has eaten and sliding backward.
With each iteration, sensory information — in the form of electricity and chemicals — passes through groups of neurons called neural circuits. In the worms, the circuit begins with neurons that detect light or smell, then loops through a chain of other nerve cells, before ending in the muscle cells responsible for gagging or moving backward. By examining the connectome, Lee expects to identify a particular nerve cell that connects to both a smell-sensing and a light-sensing neuron.
“Perhaps,” “he said, “that is where information from both senses is first collected and learning occurs.”
To test this hypothesis, Lee, using a laser, will kill this nerve of interest. If he’s severed a link in the neural circuit, the worm will no longer learn properly. Systematically repeating this process — killing neighboring cells and looking for changes in behavior — will help him trace the entire neural circuit responsible for associating laser light and smell.
As he works, sitting before a microscope focused on a single worm, Lee has the syncopated and coordinated movements of a drummer. His right hand, holding a pipette filled with alcohol, wafts the smell above the worm, always taking care to hover over the animal as it twists and winds its way across the Petri dish. After a few seconds, his foot presses a pedal on the floor to activate the laser and pair the two stimuli. Later, when testing whether the worms react to smell alone, the index finger of Lee’s remaining hand taps the rhythm of the animals’ digestive muscles as they clench and relax, a measurement of the gag reflex, into his computer.
Lee, who prefers seclusion, generally works from 12 p.m. to 3 a.m., surrounded by shelves packed with faded academic journals and stacks of Petri dishes. The building at rest seems both heavy with silence and buoyant in its emptiness.
He has little distraction, and doesn’t need to worry about sharing equipment with other researchers. There’s also the chance of having a private moment of triumph amid the grind of experiments.
“I like it at night,” he said. “You can scream and shout if you see something exciting. You can linger on the feeling of being the first to know something. When people are around, the magic is gone.”
Lee first become interested in science while conducting a middle school experiment that involved extracting DNA from tissue samples using household materials. After reading Francis Crick’s “What Mad Pursuit” and other accounts of discovering the structure of DNA, he decided to work in a lab while in high school.
“I was helping another scientist who was using quantum dots to detect cancer,” he recalled. “Whenever something worked the postdoc and I would skip dinner and work past midnight. You get that rush in science and you need to get another and you find yourself pushing your curiosity.”
Every scientist strives to make a great discovery, but breakthroughs often require tiny, repetitive steps of progress. Research can be particularly perverse — how hard scientists work may have little bearing on how much they accomplish. Oftentimes scientists are the most busy when they are having problems with their experiments.
Lee has spent months troubleshooting how best to train worms by adjusting the strength of his laser, deciding which smell to use, and tinkering with other variables.
To maintain his sanity, he goes horseback riding in New Hampshire twice a week. Although he chose the sport on a whim — he was looking for an activity that was “English” and vastly different from his childhood hobbies in Singapore — there are parallels to the steady grind of research.
As a beginner, Lee spent two years learning to control a trotting horse before moving onto jumping. Even today, he and his trainer alternate trotting and jumping lessons to keep his fundamentals sharp. “I always jump better after flat lessons,” he said. “It taught me that basics matter and that incremental steps will eventually lead to a big jump.”
As with science, there are also unavoidable rough patches.
“Sometime there are bad days even if you have learned to do your stuff right,” Lee said. “You’re distracted … things go wrong. It is just what it is.”
For the past 200 years, researchers have approached cognition from two contrasting directions. One takes a large-scale view of the brain, assigning specific functions such as controlling body movements or emotion to particular areas of the organ. The second takes place on a microscopic scale, describing the properties of individual nerve cells and how they communicate using electrical and chemical signals.
Both of these approaches struggle to explain how neurons, which by themselves are relatively simple cells, organize themselves into circuits to share information, make complex decisions, and create consciousness.
By pairing the connectome with functional tests — like killing neurons or turning them on and off — researchers are uncovering how neural circuits work. In C. elegans, scientists have explained a variety of behaviors, including how worms use smell to avoid dangerous bacteria, how the neurotransmitter serotonin stimulates egg laying, and how males choose to prioritize sex over food.
Perhaps the strangest demonstration of the connectome’s potential occurred in 2015, when a group of biologists and computer scientists downloaded the C. elegans connectome into a Lego robot.
In the blocky machine, composed of gray, beige, and red pieces, mechanical equivalents stood in for parts of the worm’s anatomy. Instead of nose neurons, the robot had a forward-facing sonar sensor. In place of neurons controlling muscles, motors controlled two wheels on the sides of the robot.
Guided by just an abridged version of worm consciousness, the robot behaved like its biological counterpart. When researchers pressed the anterior and posterior touch sensors, the robot moved forward and backward. When the worm occasionally hit a wall, it paused, reversed its wheels with a raspy mechanical whir, and tried a different direction.
Researchers concluded that even a crude version of the connectome is enough to generate simple behaviors.
Can the worm connectome help explain the behavior of more complicated organisms? Scientists believe that lessons learned in C. elegans are applicable to mice and humans.
To survive, all animals — from insects to vertebrates — rely on basic behaviors like detecting and responding to motion and associative learning. According to Marta Zlatic, a group leader at Janelia Research Campus, a Howard Hughes Medical Institute research center, the neural circuits underlying these behaviors in different animals are likely variations on a common pattern that has been standardized by evolution.
Similarities in circuit organization may allow researchers to study larger brains in a targeted way. In other words, rather than mapping the whole mouse brain, scientists could look for specific neural patterns that have already been studied in worms.
“What we’ve learned in simpler organisms already gives us very testable hypothesis for bigger organisms,” Zlatic said. “ One can just go in and directly test that hypothesis as opposed to be comprehensive which is still very, very difficult in a bigger brain.”
Studies in C. elegans and other simple animals also suggest that scientists can make discoveries using connectomes with less resolution — maps of complete brains that describe general structures instead of the precise location of every single neuron.
“What do I really get by looking at everything compared to 10 percent of everything?” asked Cori Bargmann, a professor at the Rockefeller University in New York and head of science at the Chan Zuckerberg Initiative. “What have I learned from looking at incredibly high detail as opposed to stepping back a little bit and getting the overall picture?”
For scientists, these questions are an area of “energetic discussion,” as Bargmann says.
The answers are unknowable because the connectome, for all its utility, is still just an entry point into the intricate workings of the brain. Overlying the connectome are layers of fluctuating complexity that can’t be captured by a static map. The connectome cannot indicate whether neurons are working with or against each other. It does not record changes in synaptic strength, a measure of how much influence neurons have over their neighbors. And on top of all this, there is the rush of neuromodulators — chemical signals, also invisible to the connectome, that seep through the brain and alter its activity.
“It’s been 30 years since the [C. elegans] connectome was published and we still haven’t really solved the worm brain,” said Bargmann, who completed her postdoctoral training in the Horvitz lab. “The results and outcomes [of our experiments] are much more complex and dynamic than we expected them to be.”
Still, scientists agree that the connectome is a critical component of modern neuroscience and perhaps the best tool we have for organizing the near infinite complexity of the brain into something that the human mind can comprehend.
Although they are not sure exactly what they will learn, researchers, inspired by success in worms, are mapping the brains of other animals including sea squirt tadpoles, fly larvae and adults, zebrafish, and the mouse retina.
These projects, with uncertain timelines and payoffs, are an act of faith. Like Lee meticulously training his worms in the deep of night, neuroscientists believe that the incremental process of mapping the brain neuron by neuron will result in a breakthrough.
“With science,” Lee said, “you might not know exactly where the research will take you, but you trust that when you arrive all the effort will have been worth it.”