When something in the mind breaks, you see what it was for

Lex Fridman opens Karl Deisseroth's book *Projections: A Story of Human Emotions* at the weaving metaphor. Warp threads are structural, anchored at the origin, stretching from a deformed past through a ragged present into a still-featureless future. The tapestry of the human story, Deisseroth writes, has its own warp, rooted in the gorges of East Africa; each life is weft laid across that scaffold. In people who are ill, the fabric frays. The warp is exposed, raw, and revealing.

Asked what he has learned about human nature from that fraying, Deisseroth answers as a biologist: "When something breaks, you see what the original unbroken part was for." A mutated gene turned up or down lets you infer its true job. The same logic, he says, is the natural way into psychiatric illness, and it is the story he wanted to make communicable to people who care about their own minds and about those who suffer. The protagonist of the book is not a single patient. It is the human mind, still anchored in the moment of someone sitting in a clinic.

The line between function and dysfunction is fought over in psychiatry more than in any other medical specialty. Deisseroth still treats patients. In the emergency room the manifestation can be so powerful, the suffering so great, that there is no doubt the person cannot continue as they are. Elsewhere it is a spectrum: people who can still work, with a small dysfunction, and everyone in between. Psychiatry, he notes, refuses the words disease or disorder unless social or occupational functioning is disrupted. There are no blood draws and no imaging studies that make the diagnosis. The line is operational — how are things working, in this moment?

To cover the whole spectrum he prefers the word *disorder*: it names the thing when it is not working. Lifetime prevalence of a psychiatric disorder, estimated from structured interviews rather than self-report, approaches about 25 percent; some studies run higher, some lower. Shockingly high numbers in the news, he says, often come from self-report. The figure does not capture severity. Nearly every condition lives on a range from nearly functional to completely disabling and even fatal. Genetics are a large part of the biology. What psychiatry still lacks is the lab test.

Fridman asks the old cliché: are madness and genius two sides of one coin? Deisseroth will not ride the metaphor. He reaches for numbers. Autism, bipolar disorder, and anorexia nervosa are all heavily genetic. Bipolar and anorexia can be fatal and cause immense suffering. At the population level, each of the three is *positively* correlated with measures of intelligence, educational attainment, and even income. The wording matters: a statistical tilt in a group is not a claim that every patient is brilliant, or that every brilliant person is ill.

Fridman offers his own version of intelligence: optimism can look like naivete, as in Prince Myshkin in *The Idiot*, but feeling the world rather than only calculating its mechanics may be a deeper kind of smart. Deisseroth says medicine meets that sliding scale every day. Standing in the abyss of a patient's suffering, a physician has to decide how much to feel and how much to keep clinical. The ability to move along that range, he says, is part of how a doctor survives, and how they protect their own feelings.

Deisseroth quotes *Finnegans Wake*, and Fridman admits he never understood it — only that the words produced feelings the dictionary sense could not predict. For a psychiatrist, Deisseroth says, the book is loaded with the speech symptoms of schizophrenia: clang associations, where the next word is chosen for sound rather than sense; neologisms; loose and tangential thought. Joyce's daughter suffered from schizophrenia, and families of people with the illness often include relatives with schizotypal personality — magical thinking, unusual patterns, a view from a safer distance. Even this severe, genetic illness, he argues, is a range that human beings occupy at different spots.

His own book was written sober, slightly disinhibited, late at night or in the early morning, when a friend's advice proved right: the inner critic is still asleep. Each chapter is anchored in a real patient and a class of disorder, with details changed for privacy and symptoms kept true. The diction changes with the illness. Mania gets a torrent of complex, pressured words. Psychosis gets language that slowly fragments and separates. He could not snap those modes on and off. He had to sit still, inhabit the mind of someone he knew, and wait.

The title *Projections* is doing several jobs at once. In neuroscience a projection is a long-range axon that binds two brain areas — the path from auditory cortex to reward centers, for example, whose strength helps determine whether someone can enjoy music. Optogenetics works especially well along those paths. In mathematics, a projection is a drop into lower dimensions. In psychiatry, we project our inner states onto other people to understand them, and we introject what we see outside. And then there is the ordinary projector: light carrying information across space. English holds all of that. Translations often cannot. Some editions, including the British one, became *Connections*.

The epigraph is from Borges's "Two English Poems." Fridman reads it as both a love poem and a scientist's hunger to get at a hidden life. Deisseroth wanted the offering-up: not only the yellow rose, but ancestors, dead men, two bullets through a grandfather's lungs. Writing the book meant putting hard things from his own life on the page. That vulnerability, he says, is what love and writing share. Fridman jokes that the yellow-rose line is a pickup line. Deisseroth says the form is universal — country songs do it too: here is what matters to me, I hope it matters to you — only Borges did it with unusual beauty and unusual hurt.

What is love, then? Deisseroth calls it the strongest and most stable connection one person can form with another, and that bond helped the human family survive against long odds. It is unreasonable on its face, then becomes reasonable by virtue of its own existence — the same trick, Fridman notes, that belief performs in other parts of life, though not in bioengineering, where imagining a sound bridge does not make it sound. The wild raw joy is not a small part of the story. Positive and negative have to be wrapped together for an honest description, which is why love takes a book, or at least a poem.

Fridman pulls up penguins huddling against Antarctic cold, then a clip titled "lonely deranged penguin" — a bird that left the pack for the mountains, declared mad by the narrator. He refuses the verdict: maybe it is going home, exploring, looking for a lost love. Deisseroth, as a psychiatrist, would like the prior therapist's notes. Then he gets serious. Individuals and species toggle between staying to reap what they have built and going out to forage. Even *C. elegans*, with 302 neurons, has rest and wandering, and individuals differ. Diversity is good for the species: some stay and keep a firm footing; some carry personal risk as explorers. Migration routes that later look clever almost certainly started as something like this.

Modern psychiatry began close to neurology and pathology. Early founders, Freud among them, looked at cells under microscopes and also sat with patients. They soon found that microscopy was too far from what a human being would actually say, and that what the human being said was more interesting and more mysterious. Many of them chose to inhabit the verbal domain, because words were the only real tool. One lasting inheritance of those decades, Deisseroth says, is the distinction between conscious and unconscious mind, and the claim that the unconscious is worth taking seriously. Psychoanalysis grew from that claim. It often aimed at insight more than cure, and that was useful only if one was honest about the aim.

Freud's break with contemporaries, including Carl Jung, came from tying the unconscious so tightly to libido that its operation could hardly be separated from sexual drive. Jung saw a much more complete parallel version of the conscious self, full of different motives, all of which had to be considered. Id, ego, and superego still make a sticky story: primal drives, later moral codes, and a mediator. Deisseroth's view from modern neuroscience is blunter. The idea of the unconscious is valuable. Those three boxes probably do not map onto any particular neurobiology. Moral drives and gratification drives are all drives, and even the moral ones are taught in ways that run back to survival. The real picture is closer to a town hall of competing desires, from different parts of the brain and a whole personal history. Jung, on that point, is nearer the modern view.

Psychoanalysis is mostly not considered a treatment today, but it persists as a source of insight and as a vast influence on literature, philosophy, and art. Fridman asks about other sticky maps — Kahneman's System 1 and System 2. Deisseroth does not referee them. He says that in modern neuroscience, whenever they seem to get closer to a question at the level of cells, the question recedes.

Deisseroth's lab, like many others, now listens in on individual neurons in mice, rats, fish, and monkeys — time series at sub-second resolution, from enormous numbers of cells. The old hope was to find where a choice begins: cortex first, then older striatum, then thalamus. What has emerged over the last few years is that these regions represent the action and the choice at about the same time. You cannot point to a spot and say the decision starts here. Nobody is close to a "free will neuron." He adds a tweet-sized caveat: all generalizations are wrong, including this one.

They finally have the data streams the field wanted for decades, and those streams have not revealed where actions initiate. The answer may be in the data, at the wrong scale or the wrong projection, or they may need a different kind of data. Passive observation — here is the activity, there is the choice — cannot settle cause. What has become valuable is reaching in, turning defined populations of cells up or down, and watching what the animal does. Optogenetics, using light to control cells, is still relatively untapped at brain-wide cellular resolution. That, he says, is how they hope to make headway.

If you want to know what actually causes sensation, cognition, and action, you need to play activity into some cells and not others, at the brain's own rates. Electrodes cannot do that. Every neuron is electrical, so a wire stimulates everything nearby — including cells that oppose each other. Deisseroth's image: trying to control fish by spraying them with water. Francis Crick, late in life, wrote in 1999 that neuroscience needed a way to turn specific types of neurons on and off in a behaving animal, and that light would be the ideal signal. He had no idea how.

Deep brain cells do not respond to light, which is a gift: any sensitivity you add is signal above noise. Optogenetics borrows genes from microbes. Algae and bacteria make proteins that catch a photon, open a pore, and let ions flow. That ion flow is exactly the electricity neurons use. Put the gene into some cells and not others, shine light, and only the cells expressing the protein are directly activated. The proteins are microbial opsins. Deisseroth's group and others then used X-ray crystallography, the same method that revealed DNA's double helix, to see how they work, and re-engineered them for red light, different speeds, and different ions.

He likes to start the story in 1866, with Andrei Famintsyn, a botanist in St. Petersburg who published — in German, in a French journal — that single-celled green algae swim toward light. Famintsyn could not have imagined those genes spliced into a mouse or a human retina. The modern timeline is compressed: 2005, cultured neurons; 2007, behaving mice; 2009, a general method for essentially any cell; 2012, single-cell resolution in a living mouse; 2019, twenty to fifty specified cells, enough to make a mouse act as if it had seen a stimulus that was not there; 2022, hundreds of specified cells across mouse visual cortex. The principles, Deisseroth says, are now clear. What remains is more experiments.

He treats optogenetics as a discovery tool, not a therapy he is driving through trials. The clinical proof of principle still landed. Botond Roska in Switzerland, who had used one of Deisseroth's opsins on a cadaveric retina years earlier, spent about a decade on regulators and primates, then injected an opsin into one eye of a man blind in both eyes from retinal degeneration. With specialized goggles projecting pulsed light, the man could reach for objects on a table. The paper in *Nature Medicine* was titled "Partial recovery of visual function in a blind patient after optogenetic therapy." Andrew Huberman thinks Deisseroth deserves a Nobel. Deisseroth's reply is almost shrugging: prizes will be forgotten, all of us will be forgotten, a cool idea is a cool idea.

Deisseroth's outpatient specialty is treatment-resistant depression, plus autism spectrum; he also works the emergency room. He stays with depression because help can feel tantalizingly close. People may have every objective marker of a fine life and still be hit with unstoppable hopelessness, a future that looks worthless, and anhedonia: no reward in food, company, movies, or books. In severe cases there is active psychic pain that feels inescapable. Twin studies and chromosomal regions show a strong genetic contribution. Stress and trauma worsen symptoms, but many people fall ill without an obvious external cause.

There is still nothing to measure. No known chemical, structure, or EEG pattern diagnoses major depression. Doctors talk, elicit symptoms, and rule out other causes. Medications help some people. If they fail, brain stimulation is next, and electroconvulsive therapy is effective as a last resort. What is missing is a physical interpretation. Congestive heart failure can present as shortness of breath and swollen feet — symptoms that look unrelated to each other and to the heart — because the heart is a pump that is not pumping. Psychiatry does not yet have that unifying level. The mystery is not the separate symptoms. In mice, Deisseroth's group can already turn up or down the motivation to overcome a challenge, the apparent experience of reward, energy, and social interaction, each tied to specific projections and cells. What they cannot explain is why those symptoms arrive together as a human syndrome.

Fridman asks whether understanding depression therefore requires a theory of everything for the mind. Deisseroth would like unifying principles. He does not think you need them before you treat. If someone comes in with profound anhedonia, and animal work says a particular intervention hits anhedonia, helping that one symptom in that one person is already a good thing. Zebrafish give up in a hard situation until two specific brain regions are stimulated; inhibit those regions and they quit more easily. *C. elegans* can effectively give up too. Detecting that the world is arrayed against you, hunkering down, and waiting out the storm is an ancestral, conserved pattern. Humans run the same machinery inside a brain that tells stories, so the state acquires a plot: my future is grey.

Talk therapy is a large part of every case he sees, alongside medication and stimulation. Alone, it can help mild or moderate depression; in severe illness it is not enough by itself, but it is still required, because it reshapes complex activity patterns that drugs and stimulation do not reach. Cognitive behavioral therapy is structured, with manuals, homework, and ten or twelve sessions that can be enough to catch the triggers that turn an ordinary negative thought into a lasting negative state. Exercise and regular sleep, he says, are badly underestimated in mood maintenance and need no prescription. The cruel feature of depression is that hopelessness and low energy block people from seeking care. That is why loved ones matter so much: they help with motivation, safety, and insight, especially in the severe cases.

Psychoanalysis and modern talk therapy are neighbors, not the same street. Data for actual treatment, Deisseroth says, support cognitive behavioral therapy more than classical analysis, which has become niche and is used more for insight. Dreams are still useful as conversation starters. Compared with looser conversational methods such as interpersonal therapy, CBT looks cookbook-like — and randomized studies sometimes find the two comparable, which supports Fridman's hunch that artful conversation can be powerful if it is done well.

Fridman reads the park-bench speech from *Good Will Hunting* and asks whether Robin Williams's therapist is a caricature of the work. Deisseroth surprises him. The depth of connection is the problem. If a patient casts the relationship as friendship or as parent and child, they start selecting what to share to fit that story. He and many other therapists prefer not to let the bond get that deep. You want trust and a therapeutic alliance, and also enough of a blank slate that the patient is not constrained, consciously or not. You still have to leave some of yourself in the room. Interpersonal trouble is best seen when it replays in the office — transference — provided the therapist notices and does not let it run away. If a patient stirs frustration or even anger, the move is not to scold yourself for feeling it. Recognize it as information about other relationships in the patient's life, and use it.

Opening up has a cost. An early draft of *Projections* braided patients, science, and the history of the human family and still did not feel real. Deisseroth realized readers would not connect unless he put himself on the page. He did. It was not the plan going in; it was frightening; in retrospect he learned something. Fridman asks where the darkest places in his life have been. For years Deisseroth was effectively a single father during the hardest stretch of late medical school, internship, 3 a.m. call, surgery, and unforgiving rounds, with a personal life stripped to the bone. Then the wards hit him — connecting too deeply with a child who had a brain tumor. At the lowest point, with only a raw core left, he still had hope. Fridman asks if he has thought about killing himself. He has not, not even as a shape in the distance. There was, he says, no reason to feel hope in that moment. It was a light without reason. A psychiatrist does not need every demon. Having seen how low things can go is enough to confer empathy for struggle, even when the symptoms are not your own.

Fridman names Robin Williams and Anthony Bourdain: people who seemed to have everything, who made millions laugh, and who still died by suicide. Deisseroth separates suicidal thoughts from the act. Ideation can serve many purposes, some of them only in imagination, the way a sexual fantasy need not be a plan. Reasons people give include absence of joy, active psychic pain, and interpersonal or cultural pressure. The act of ending the self is different. As far as we know, no other animal we can study has the concept "this is myself, the situation is intolerable, therefore I will end the self." That is why the biology remains poorly understood and prediction remains poor.

Sergey, who has been pulling references throughout, finds that examination of Williams's brain tissue suggested diffuse Lewy body dementia. Deisseroth takes the finding seriously. Lewy body dementia involves frank neuron death, including dopamine neurons that help generate reward and pleasure. Williams may have reached a point where a long-standing balance between negativity and positivity was broken by progressive cellular loss. Even without degeneration, another path is available: decades of compensated psychic pain, held in check by comedy, social life, and work, until those props drop away late in life and the same symptom becomes intolerable.

In the emergency room the practical questions are grim and specific. Was it a gesture without intent to die? If death was the intent, are the reasons still present? What is the chance of repetition? Psychiatrists can use locked units and a stretch of care; then they release people. The cellular and circuit basis of the act is still a blank, which Deisseroth calls one of the most frustrating facts in the field. The hopeful move is to keep taking the puzzle apart. Anhedonia can already be dialed in animals. So can the motivation to overcome challenges. Internal negative states have an anatomical handle: the habenula and nearby structures, which become active in acute disappointment, unexpected pain, and outcomes that go worse than expected. Targeting that system to reduce psychic pain is, he says, a concrete hope.

Fridman jokes that his own habenula is very active, and that a harsh inner voice is useful for productivity if you keep it on the table and watch it. Deisseroth hears the half-joke as a sign the voice is being used rather than only suffered — and then reminds him that no outsider knows another person's internal state. His own negativity is the feeling of never having done enough. It works while there is runway. Later in life the same voice starts to have more of a point, which is exactly when it can become dangerous.

Human beings sit on a spectrum of how social they can be. At one end are hypersocial states, including some chromosomal deletions, with instant affinity, fluent language, and apparently deep connection. At the other is autism spectrum disorder: increasing difficulty keeping up with social interaction. Milder forms leave language intact but make the next move in a social situation opaque. Further along, eye contact itself becomes overwhelming. At the far end, language and social communication break down: no reciprocity, no shared enjoyment, an absence of social cognition. Autism is among the most genetically determined psychiatric conditions, which is part of why those slight positive correlations with intelligence and education are interesting rather than paradoxical.

Deisseroth's account turns on information, not on a missing social instinct. A person who is gifted at social life is gifted at unpredictable, high-bit-rate information: each word, glance, and shift of posture updates a model of the other person, of you, and of their model of you. You cannot pause to decode. You have to run predictions. Another brain state is built for things that do not move — a proof, a geometry, a large set of static parts. That state is slow in conversation and unusually good at holding a system still long enough to see it from every angle. He thinks that way of being has contributed a great deal to human design. Social difficulty shows up first because social life is the most unpredictable stream we have. Unexpected sounds, lights, and touches are problems too. The core, in his view, is unpredictable information.

Fridman describes closing his eyes to think: not shyness, but too much incoming information, plus the social expectation that eye contact means you are keeping up. Deisseroth says that is an accurate description of what many people on the spectrum report. There is no medication for autism itself. Early behavioral intervention is what works, so a child who is not interacting does not fall further behind. Anxiety, which often comes along because the world is hard to predict, can be treated. Deisseroth sees high-functioning patients who make a good living in places like Silicon Valley and are still miserable, never sure what to say or how anyone else knows. He can treat the anxiety. It does not touch the autism. And if someone has autistic traits but is socially and occupationally successful, he does not call it a disorder. Symptoms without functional harm do not get that name — he has had patients who were pleasantly hallucinating, frankly psychotic, and undiagnosed for the same reason.

Schizophrenia, Deisseroth says, is a break from reality. Positive symptoms include hallucinations (usually auditory), paranoia, and delusions — fixed false beliefs that will not yield. Negative symptoms are progressive: flattening of emotion toward a neutral state, thought disorder, poor working memory, the inability to keep a plan or a place in a conversation. Why they travel together is unknown. The illness is extremely genetic, perhaps 80 percent, and about one percent of people worldwide, across cultures. The typical story is a first break: a young man, often 18 or 19, who has been academically, athletically, and socially typical, arrives at college and is hit by terrifying paranoia or a voice screaming at him. In women it often comes later, in the twenties. Untreated, it progresses. Antipsychotics that block the serotonin 2A receptor and modulate dopamine can return some people to a fully normal state and keep them there; more often they land in an intermediate condition, with a lasting drop in function and with side effects, including movement problems. One working idea is a communication failure inside the brain: the inner monologue is not tagged as one's own, so it is heard as coming from outside, or from another room in the same building.

Around 1810, James Tilly Matthews drew his experience for his doctor: a cowering figure controlled by a loom sending long threads — projections — into his arms and body. He called it the air loom, using the most complex machine of the early industrial revolution to explain a body that moved without his will. Today the same experience is attributed to satellites, aliens, or a government agency beaming radio. Fridman has found people with schizophrenia, some of them unhoused, unusually well-read and attuned to the world's complexity. Deisseroth agrees: they generate unlikely explanations, which is part of why they cannot function, and the impulse to make sense of the world is still, in its way, refreshing.

Why do we cry? The lacrimal gland evolved to wash grit from the eye, and it is controlled from the pons, an ancient structure that also runs breathing. Forebrain fear-and-anxiety regions send long-range projections down to those breathing cells so that threat can raise heart rate and respiration. The tear-gland neurons sit next door. Crying, Deisseroth argues, is almost certainly a slightly misdirected projection — a missed signpost, a few steps too far. It landed in extremely valuable social real estate. In controlled tests, tears on a face stir a desire to help more powerfully than any other facial feature. Because the signal is involuntary, it is hard to fake. It reports, at once, frailty, need, and hope that the community might answer.

Then he runs a thought experiment on the hard problem of consciousness. His lab can already play a pattern into about 25 visual-cortex cells and make a mouse behave as if it saw something that is not there. Suppose you could do that for every neuron in a human brain: show a rich red, then strip the stimulus and replay the exact activity. Most people say the subjective feeling would remain. If every cell's activity is supplied from outside, the cells no longer need to sit in one skull; you could spread them across North America and play the same pattern. That a "Lex" would still feel red is absurd — and it is the same situation as the previous step, where people said yes. The paradox, for Deisseroth, means we are missing something large about inner subjectivity. In late 2020 his lab used optogenetics to drive a ketamine-like rhythm in retrosplenial cortex and knocked brain regions out of sync, so they could not form a joint representation: a toehold on how the self unbinds from the body, not an answer. He has hope for a breakthrough in his lifetime and does not know. Asked the meaning of life, he reaches for an Aristotelian point: an organism is happy when it performs to its design, and we are clearly designed to ask why. Fridman closes with Jung: knowing your own darkness is the best method for dealing with the darkness of other people.