Deisseroth, Karl

professor of bioengineering and psychiatry, Stanford University; practicing psychiatrist; pioneer of optogenetics; author of Projections: A Story of Human Emotions

Karl Deisseroth: Depression, Schizophrenia, and Psychiatry

Lex Fridman Podcast · Apr 7, 2022

• Karl Deisseroth
• Projections (book)
• optogenetics origins and technique
• depression and anhedonia biology
• suicidality and the habenula
• autism as unpredictability processing
• schizophrenia and thought disorder
• crying as an evolved social signal
• the hard problem of consciousness
• psychoanalysis vs. cognitive behavioral therapy

Core ideas

Deisseroth frames his book Projections around the biologist's principle that dysfunction reveals the true purpose of a normal function, using patient stories (with identifying details changed but symptom descriptions drawn from real cases) to illuminate what the "unbroken" human mind is for. He resists a hard line between disorder and order, noting psychiatry only labels something a disorder when it disrupts social or occupational functioning, and cites population studies suggesting roughly 25% lifetime prevalence of some psychiatric disorder — while cautioning that self-report surveys (versus rigorous structured clinical interviews) tend to inflate such numbers further. He highlights a genuinely puzzling finding: severe, highly heritable disorders like autism, bipolar disorder, and anorexia are each positively correlated at the population level with measures of intelligence, educational attainment, and income, suggesting these conditions sit on a spectrum with adaptive as well as destructive expressions rather than being purely negative aberrations.

On psychoanalysis, Deisseroth credits Freud with legitimizing serious attention to the unconscious mind but considers Freud's near-total focus on libidinal drives too narrow, crediting Jung with recognizing a broader range of unconscious motivations; he suggests, however, that modern neuroscience finds even the classic id/ego/superego framework somewhat artificial, since brain activity looks more like competing drives in a "town hall" than cleanly separable structures — a caution he extends to other tidy dichotomies like Kahneman's System 1/System 2. He describes a central, ongoing puzzle in his own recording-based neuroscience: even with rich, brain-wide, single-cell-resolution activity data, researchers cannot identify a single origin point for a decision or choice, because activity representing an impending action appears to arise nearly simultaneously across many brain regions (cortex, striatum, thalamus) rather than in one privileged location — pushing his lab toward causal (rather than purely observational) methods.

He explains optogenetics — a technique he helped pioneer, using light-sensitive proteins borrowed from algae and other microbes to make specific neurons respond to light pulses, enabling precise activation or suppression of defined cell populations in behaving animals — tracing its lineage back to an 1866 observation by St. Petersburg botanist Andrei Famintsyn on light-responsive algae, and describing its rapid technical progress (from cultured neurons in 2005 to controlling up to hundreds of individually specified single cells across a mouse's visual cortex by 2022, sufficient to make a mouse behave as if it saw a stimulus that was never presented). He cites a real-world therapeutic application — Swiss researcher Botond Roska's use of an optogenetic tool to partially restore light sensitivity and functional vision in a blind human patient's retina, published in Nature Medicine — while stating his own primary interest is optogenetics as a discovery tool for uncovering causal principles (e.g., identifying specific brain circuits that causally control motivation, anhedonia, or sociability in animals) rather than directly running clinical trials himself.

Deisseroth discusses depression as a syndrome with strong genetic/heritable roots but no known objective biomarker — no blood test, imaging finding, or activity signature — despite intensive current research; he distinguishes ancestral, plausibly adaptive roots of low mood and motivational withdrawal (observed even in simple animals like C. elegans and zebrafish, which "give up" under sustained adverse conditions unless specific brain regions are stimulated) from the debilitating, often reason-independent suffering seen in severe clinical cases. He emphasizes that talk therapy — particularly structured cognitive behavioral therapy — remains a core, evidence-supported treatment even alongside medication and brain stimulation, and offers concrete guidance for mild-to-moderate cases (exercise, sleep regularity, identifying and interrupting catastrophizing thought patterns) while stressing that severe, hopelessness-driven depression especially requires support from loved ones to overcome the illness's own barriers to seeking treatment.

On suicide, Deisseroth distinguishes suicidal ideation (thoughts, which can serve a psychological function without leading to action, analogous to fantasy) from the act itself, which remains poorly understood biologically and essentially unstudiable in animal models since no other known species appears to conceptualize ending its own existence; he points to the habenula and connected structures as an identified brain region generating acute internal psychic pain/disappointment as one promising, mechanistically tractable target. Discussing Robin Williams specifically, he notes that postmortem examination reportedly found diffuse Lewy body dementia — a progressive neurodegenerative condition involving loss of dopamine neurons among others — suggesting a concrete, biological disruption of whatever internal balance the actor had long maintained between negativity and positivity, rather than a purely psychological explanation.

On autism, Deisseroth proposes reframing the core deficit not as social dysfunction specifically but as difficulty processing unpredictable, rapidly changing information generally (citing aversive reactions to unexpected sounds, lights, or touch, not just social cues) — with social interaction simply being an especially unpredictable, high-bandwidth domain — while autistic individuals may be comparatively strong with static, complex-but-predictable systems (mathematical or geometric reasoning). He describes schizophrenia as a roughly 80%-heritable, globally uniform (~1% prevalence) disorder combining "positive" symptoms (hallucinations, delusions, paranoia — a break from consensus reality) with progressive "negative" symptoms (flattened affect, disorganized thought, impaired working memory), often beginning with a dramatic "first break" in previously high-functioning young adults; he frames psychotic hallucinations as arising from failed internal communication — an inner monologue that should be recognized as self-generated instead being misattributed to an external agent — and notes historically that patients' delusional explanations (a Victorian-era patient's "air loom" device, versus a modern patient's satellite or radio-frequency control narrative) track the era's most sophisticated available technology.

Deisseroth explains crying as an evolutionary repurposing: the lacrimal (tear) gland is controlled by brainstem circuits near respiratory control centers, and long-range projections from fear/anxiety regions that evolved to regulate breathing and heart rate during threat appear to have "slightly misdirected" onto tear-duct neurons, incidentally producing an involuntary, hard-to-fake emotional signal that experiments show is uniquely powerful (compared to other facial expressions) at eliciting help from onlookers. On consciousness, he presents a thought experiment: if every neuron in a person's brain could be driven with the exact activity pattern that occurred during a genuine subjective experience, most people intuit the same subjective feeling would occur — but if those neurons were then spatially spread across a continent while receiving identical inputs, the same logic seemingly still implies the feeling would occur, which he calls a genuine paradox pointing to a deep unsolved question in neuroscience: what allows spatially and causally connected neural activity patterns to bind into a single, unified conscious representation. He connects this to his own optogenetic experiments inducing dissociative, ketamine/PCP-like states by desynchronizing brain regions from forming a joint representation, published in Nature.

Excerpts