The supernormal
stimulus
Her own egg — and the one she chose instead.
She is not confused, and she is not weak. She is doing exactly the right thing with the wrong object — because something was built to beat the right one.
One condition in the builder — Supernormal Stimulus Protection — covering every one of the 24 pathways this page sweeps. The science is all below, and the protective side of it is free; this button is only the narrow part a blend can reach. Educational only.
Before you read anything — look at these for one second.
Which one did your eye go to first? There is no wrong answer, and we are not storing it.
A supernormal stimulus is a counterfeit that beats the real thing. Build an egg bigger and more speckled than any egg a bird could lay, and she will roll it into the nest and sit on it while her own eggs go cold beside her. The instinct is not broken. It is working exactly as designed — and the design never anticipated a competitor.
Everything below stays exactly as it is — same words, same layout, same figures. Only the saturation goes. It is the one intervention on this page you can run on yourself right now, and the fastest way to understand why it is on the list.
Same page. Same argument. Notice what happened to the pull — and that nothing was taken away except one feature. That is the whole idea: you do not have to out-argue a craving if you can turn down the thing that produces it.
Dust, ochre, the brown of actual food. These are the colors of things that grow.
Scarlet, magenta, electric violet. No pigment makes these. Only emitted light does.
Where your eye went before you finished reading this. That is the whole mechanism, in one object.
Every ring is the same thing — a reward — and each one carries more of the feature than the one before it. A surface reflects light and can only return less than it receives. A screen emits it. That is not a metaphor; it is why a display can exceed any object in the world it is showing you. And it is why the switch above works: turn the color off and this diagram stops pulling, instantly, without a single word of it changing.
Tinbergen, De Levende Natuur 51: 66–69 (1948) — the Dutch original, not the English book that is usually cited for it. In his own words: “Wanneer we nu één reuzenei tegen drie normale gaven, viel de voorkeur toch op het reuzenei” — one giant egg against three normal ones, and the giant egg still won. She is not confused and she is not weak. She is doing the right thing with the wrong object, because something was built to beat the right one.
A ripe berry. Sweet because sweet was rare.
Sugar and fat combined in a ratio that occurs in nothing that grows.
A handful of new faces in a season.
A new one every time your thumb moves, without end.
Reward that cost a day of walking, and then stopped.
No travel, no season, no satiety point, no close.
The three dials that make a thing supernormal. The bars carry shape, not measurement — nobody has quantified how far the second exceeds the first, and a number here would be invented. What is measured is that these three features are what keep the dopamine bursts coming, which is why capping them, rather than capping the clock, is what the biology actually recommends.
Supernormal color
every one of these was run as an experimentColor is the oldest lever in this literature, and the pattern is the same every time: the animal is not fooled about WHAT the thing is. It is responding to a feature — red, contrast, size — and the counterfeit simply carries more of that feature than any real thing can. Modern packaging, thumbnails and screens are built on exactly this, and a backlit display can be more saturated than any surface in nature, which reflects light rather than emitting it.
A perfect white surface is the ceiling for every pigment in the natural world. It can return 100% of the light that falls on it and not one photon more. Daylight fluorescent pigments break that ceiling: they absorb ultraviolet and blue light and re-emit it as visible, so at their peak wavelength they return more light than falls on them.
“With fluorescence, the dye emits 77% more light between 500–550 nm than perfect white (100% reflectance). The dye is said to have a peak radiance factor of 177%… The peak radiance factors of some dye combinations exceed 300%.”
That is a supernormal stimulus written in physics, with a number attached. Not an analogy — a stimulus that exceeds the physical maximum of anything that merely reflects, which is precisely what a plaster egg did to an egg. It is why safety gear, highlighters and tennis balls are these colors, and why they read as glowing rather than as colored.
And here is the honest part: we cannot show them to you. No screen can display a radiance factor above 100% — sRGB has no coordinates for a color that emits more than it receives. Every swatch above is the closest approximation this display can make. It is already past what any reflecting pigment can do, and still short of the real thing. (If you are reading this on a recent Mac, iPhone or iPad, those swatches are being drawn in Display-P3 rather than sRGB — about a quarter more of visible color than the hex codes beneath them can express. Even that does not reach a fluorescent.) Which is a fair summary of this entire page: the counterfeit is not a lie about the world, it is an excess of one true feature of it, and the measuring instrument runs out before the stimulus does.
Source: “Fluorescence and the Color of Day-Glo Paints”, Optics & Photonics News, March 1992, Optical Society collection, Florida State University.
Tinbergen built an accurate three-dimensional model of the parent’s head and bill. The knitting needle beat it. The chick is not looking for its mother — it is looking for red against contrast, and the needle supplies more of that than a mother can.
Territorial males attacked the float more vigorously than they attacked actual invading males — provided the underside was redder than a real fish’s.
They preferred the dummy so large they slid off it, repeatedly, and kept climbing back on.
Males chose the monochrome paper over the living female. More orange, in a shape that could never mate.
Neither of these is moving. Both are flat, both are still, and nothing on this page is animating them around their own rings. Let your eyes wander across them rather than staring at one point, or blink quickly a few times, and they will boil.
As published, 2003 — sRGB · the four steps at the luminances the paper measured
The same figure, gap widened — Display-P3 · same four steps, same pairing, more span than sRGB can hold
Figure after Kitaoka & Ashida, VISION 15, 261–262 (2003), itself a variant of Fraser & Wilcox, Nature 281, 565–566 (1979). Luminance values, latency figures and the single-unit recordings: Conway, Kitaoka, Yazdanbakhsh, Pack & Livingstone, “Neural Basis for a Powerful Static Motion Illusion”, The Journal of Neuroscience 25(23):5651–5656 (2005).
How it is done
the levers, namedNone of these require anyone to be weak, and none of them require a conspiracy. They are the features that reliably produce the response, found by iteration, kept because they work. The reason to name them is practical: these are the dials the evidence says to turn down, and they are easier to turn down than a craving is to argue with.
Identical payout. Identical brightness. Identical everything except when. Notice which one your eyes keep returning to, and notice that you already know what the other one is about to do — which is exactly why it stopped being interesting. That is the lever, and it is running on you right now, in a figure with nothing in it but two dots.
Find the one feature the instinct keys on — red, sweetness, contrast, a shape — and supply more of it than nature can. This is the original mechanism and it is still the main one.
Dopamine neurons fire to what is BETTER THAN EXPECTED. A fixed reward stops producing bursts once it is predicted; a variable one never does. Slot machines, feeds and endless video are all variable-ratio schedules, and that is not an accident of design.
Satiety requires an end. The lateral habenula — the anti-reward brake — fires when reward is omitted or the animal is sated. A stream with no last item gives it nothing to fire on.
The Coolidge effect: after satiety, a new partner restores dopamine. In five male rats accumbens dopamine peaked at 195% of baseline with the first female and renewed to only 34% with a novel one — smaller, not the same again. Novel frames mimic it before the refractory cycle can finish.
Sugar and fat occur together in almost nothing that grows. Combine them in a ratio no plant produces and the result is not a stronger food — it is a signal the system was never calibrated against.
Counts, streaks and scores convert an experience into a tracked quantity, which makes the next one a prediction-error event in its own right.
Sensitization means the same input stops producing the same response. The honest version of this is not moral failure; it is the system doing what the Pitchers and Wallace experiments show it does — and the pull is toward more, because more is what still registers.
Where you meet it
the same levers, in the wildNone of this requires anyone to be stupid, and none of it requires a villain. These are features that reliably produce a response, discovered by iteration and retained because they work. Naming them is useful for one practical reason: a feature can be turned down. A craving cannot be argued with.
A gesture that may or may not yield something new, on an unpredictable schedule. That is the slot-machine lever, and dopamine fires to the UNCERTAINTY, not the post.
There is no last item, so there is no moment of completion for the satiety machinery to act on. The brake never gets its cue.
Likes, views, streaks and followers convert an ordinary social act into a tracked quantity, which makes every check a prediction-error event with a number attached.
Not information. A scheduled interruption engineered to arrive when the estimate of your return has decayed.
Thumbnails selected and graded for contrast, on a backlit panel brighter than the reflected world. The gull-chick lever, in your hand.
The Coolidge effect is measured: in five male rats on microdialysis, accumbens dopamine peaked at 195% of baseline with the first female and fell back once mounting stopped. A novel female renewed it — but only to 34%, not to the original magnitude. A stream of novel frames supplies that rebound faster than the refractory cycle can finish it.
Selected, lit, edited and amplified past what any real situation presents — the same operation as the three-banded knitting needle.
Dopamine tracks better-than-expected. In an endless catalog the LOOKING generates more prediction error than the finding, which is why the hunt can outlast the interest.
Sensitization means the same input stops producing the same response. The pull is toward more, because more is what still registers.
ICD-11 classes compulsive sexual behavior under impulse control, not addiction, and DSM-5-TR has no diagnosis for this. Grubbs and colleagues found perceived addiction often tracks moral conflict more than hours of use. No human study has measured ΔFosB or spine density here. The distress people report is real; the mechanism is not settled, and anyone telling you it is has gone past the evidence.
The shell is perfectly bilateral — one half drawn, the other the same half mirrored, down to the fold. That symmetry is what your eye locks onto, and it is anatomically honest, because a brain really is built that way. Nothing inside it is mirrored. The storm, the short circuits and the blown core are all asymmetric, and one accumbens is burning while its twin is not — which is the only thing in the picture that could not happen to a brain at rest. Symmetry is why you cannot look away. Asymmetry is what you are looking at.
Follow the solid cyan line: dopamine neurons in the ventral tegmental area fire to what is better than expected and release into the nucleus accumbens. There, in D1 neurons, the cascade runs cAMP → PKA → CREB → ΔFosB, and the orange line is the handover — control moving from ventral striatum, which is goal-directed, to dorsal striatum, which is habit.
The two dashed lines are the ones that matter most and show least. The prefrontal brake is drawn dashed because imaging in problematic use repeatedly shows cue-heavy striatum with weaker prefrontal control. The lateral habenula is the anti-reward brake: it fires when a reward is omitted or the animal is sated — and a stream with no last item never gives it the cue.
Read this as a schematic. Positions are stylized for legibility, not anatomical. The circuit is the one the addiction literature establishes and the one the imaging studies above report; what has NOT been measured in a human being, for this stimulus, is ΔFosB, spine density, GABA, norepinephrine, orexin or habenula firing. The map is real. The arrows into it, for this particular behavior, are mostly inference from animals and from cue-reactivity imaging, and the page says so wherever they appear.
There is nothing in those tiles. No image, no person, no content of any kind — and your eye still went across them, and still registered each one as a separate event. That is the mechanism, running on nothing. Novelty is computed on the signal, not on the meaning, which is why a stream can keep producing the response long after it has stopped producing the experience.
This is how the original experiments worked too. Tinbergen never photographed a female butterfly — he held up a rectangle of orange paper, and the males chose it over a living female. He never photographed a gull; he used a knitting needle. The whole finding of this literature is that an instinct answers a feature, not a thing. So a page about it does not need the thing either, and showing one would depict a person when the mechanism was never about any person.
Schematic, not plotted data. In male rats, accumbens dopamine falls toward satiety with a familiar partner and rebounds to a novel one — and each rebound is smaller than the one before. That shape is what the animal work establishes. We do not have a table of numbers for it, so we have not drawn axes that would pretend we do. No equivalent has been measured in a person.
Sugar, fat and salt tuned in combinations that occur in nothing that grows — formulated, tested and optimised for the maximum response rather than for nutrition.
Gloss, saturation and contrast past anything the food itself carries. The photograph on the box is the plaster egg.
An artificial deadline converts a steady offer into an unpredictable one, which is the same prediction-error trick a variable schedule plays.
Points, tiers and streaks attach a tracked number to ordinary purchasing.
Limited editions, drops and seasonal variants keep a familiar product producing the bursts a familiar product otherwise stops producing.
Proprietary blends that hide what is inside. Before-and-after photography. Scarcity timers. A hero molecule with one mouse study behind it. Packaging engineered to look like pharmacy rather than like a jar of powder.
Every claim on this site carries its study, every ingredient carries its grade, the limits sit in the open, and the things we map but cannot claim are listed as such. That is not modesty. It is the only posture from which a page like this one is worth anything — and it is checkable, which is the point.
Every technique on your screen was in use for a thousand years before there were screens. Scale, repetition, saturated color, rhythm, scent, the withholding of a sense and then the flooding of it. None of that is a modern discovery. What is modern is the frequency.
1,070 Papua New Guinean objects examined and 81 people surveyed. Eye designs turned up significantly more often on the protective objects — war shields, canoe prows — than on domestic ones, and objects carrying them drew more fear and anxiety from viewers while the others drew calm. But not for the reason everyone assumes. The best test of why eyespots work found the opposite of eye-resemblance: “conspicuousness, and not eye mimicry, is important” — circles survived no better than bars, and making the spots more eye-like did nothing at all (Stevens, Hardman & Stubbins, Behavioral Ecology 19(3): 525–531, 2008). The releaser is not an eye. It is contrast, where contrast does not belong.
Rhythm and dance; sensory deprivation; sensory overload; starvation and extreme diet; breathing maneuvers. Every one of them is a section heading in a review in one of the most conservative journals in psychology. The techniques are real and they are old. “Drumming and dancing have been practiced since ancient times to induce altered states and are still common today.”
The trance state itself is real and visible: 15 experienced practitioners in an fMRI scanner showed reorganized hub activity and reduced connectivity along the auditory pathway — they were decoupling from the very drum that was driving them. But when someone ran the controlled version, drumming at 4.2 Hz against instrumental music across 39 people, the instructions did more work than the drum, and the cortisol difference everyone expected did not appear at all. We are reporting that because it is what the measurement said.
We are not claiming to have noticed this first. There is a formal theory of ritual built around “sensory pageantry and emotional arousal”, which predicts that the rarer a rite is performed, the more sensory load it carries. There is measured work arguing that religious interactions restrict cognitive processing and raise receptiveness to authority (Schjoedt et al., Religion, Brain & Behavior 3(1), 2013) — and a published rebuttal saying the depletion mechanism is not well enough supported (Sousa & White, J. Cognition and Culture, 2012). What is ours is the synthesis and the vocabulary, not the insight.
Everything above has a political version, and it is used by every side. This page names no party and no person, because the mechanism does not care which one is using it — and because the useful thing to know is the shape of the technique, which lets you see it working on you regardless of whether you agree with the message.
Push the infant configuration up in a face and the nucleus accumbens — the same structure named everywhere else on this page — responds more strongly, along with anterior cingulate, precuneus and fusiform. That is the imaging companion to the behavioral study above. Every softened, grandfatherly, warmly-lit campaign portrait in the world is working that curve, and so is every photograph of a candidate holding a baby.
Eye designs cluster on the objects meant to guard — shields, prows — and viewers reported more fear and anxiety looking at them. The modern version is the opponent’s face in grainy monochrome, enlarged, lit from below, held on screen a beat too long. And the correction matters here too: the best test says the power is in conspicuousness rather than eye-resemblance, which is exactly why the trick survives being abstracted into a logo, a color, or a single looming shape.
Peak shift is the formal name for the political cartoon. Take the average face, subtract it from this face, and amplify what is left. Done gently it is portraiture; done hard it is caricature; done hard and repeated it becomes the only version of that person a lot of people will ever hold in mind. The measured exaggeration in ordinary portrait art is already over 2.5× on the eye. A caricature is simply further along the same line.
Where the idea comes from
three namesNothing was added from left to right except more of the same feature — bigger head, bigger eyes, set lower in the face. Most people find the right-hand one the most appealing, and it is the one that could not be born. When Glocker’s group did this properly with real infant photographs across 122 adults, pushing the schema up raised both rated cuteness and the measured motivation to care for the child (Ethology 115(3): 257–263, 2009). Every cartoon you loved as a child lives at the right of this row.
First described the effect, in Der Kumpan in der Umwelt des Vogels — the companion in the bird’s world.
Named it and ran the experiments, in The Study of Instinct. Oystercatchers abandoned their own clutch for an outsized plaster egg. Male butterflies courted cardboard dummies more vividly marked than any female of their species.
Carried it to people in Supernormal Stimuli: candy engineered past fruit, pornography past partners, feeds past conversation. That book is evolutionary psychology — it is an argument, not a measurement.
The idea did not start with a screen, and it did not start with Tinbergen either. It starts with a man watching a pheasant that had grown too ornamental to fly.
Before anyone had a word for it, Darwin had the mechanism. On the Argus pheasant, whose display plumes are so large they stop the wings working: “the male Argus pheasant acquired his beauty gradually, through the preference of the females during many generations for the more highly ornamented males.” Not the ornamented ones. The more ornamented. A preference with no ceiling, running for long enough to build an animal that can no longer fly.
“The actions we call instinctive all conform to the general reflex type; they are called forth by determinate sensory stimuli.” Determinate. A particular signal, not a general understanding of the situation — which is exactly the opening a counterfeit signal needs.
Watching great crested grebes court, Julian Huxley coined ritualization for the way a signal becomes stereotyped, exaggerated and detached from the thing it started as. Half a century later he convened an entire Royal Society volume on it — A Discussion on Ritualization of Behavior in Animals and Man, Phil. Trans. R. Soc. B 251(772) (1966). The word for what a ritual does to a signal was coined for birds.
Teach a rat that a rectangle means food and it responds more to a rectangle that is longer and thinner than the one it was trained on. “What the rat is learning is not a prototype but a rule.” Push past the example and you get a stronger answer than the example ever produced. Ramachandran takes this straight into art — “All art is caricature” — and his worked example is a Chola bronze of Parvati, where the hips and breasts are, in his words, “grossly hypertrophied, the waist abnormally thin.” Correction worth having: the Venus of Willendorf is usually attached to this argument. It is not his example — the word Willendorf does not appear in that paper or in his Reith lecture on the subject.
776 artistic portraits covering the whole period of the history of art, against 289 photographs. Eye roundness, lip roundness, eye height, eye width and lip height all came out significantly larger in the paintings. Then the test that settles it: 42 art students drew themselves twice, once from memory and once with a mirror in front of them. The exaggeration was the same size either way. Looking at the real face did not correct it.
Measured against photographs, artistic profiles render the eye 2.51× taller and 2.76× wider than it is, and the mouth 1.94× wider. Eye width carried an effect size of η² = 0.42 — enormous for anything in psychology. When 13 students drew from a live model sitting in front of them, eye width came out 106% larger than the model’s. Not a style. A systematic distortion in one direction, made in the presence of the thing being copied.
Lorenz called the configuration that releases care the Kindchenschema — big head, big eyes, small body. Glocker took real infant photographs and pushed that configuration up and down. Across 122 adults, the exaggerated infants were rated cuter and drew a stronger motivation to care for them than the real, unmanipulated ones. Every cartoon character you have ever found appealing lives on that curve, and so does every doll, and so does the face on the packaging.
It is not looking for its mother. It is looking for a feature — and a painted stick can carry more of that feature than a mother can.
What the brain actually does
three steps, in orderDopamine is not reporting pleasure. It is reporting the difference between what happened and what was expected — which is why the first case, a reward you were certain of, produces nothing at all. A system built on this cannot be satisfied by a reward it can predict. It can only be satisfied by a surprise, and surprises have to keep escalating.
Dopamine neurons in the ventral tegmental area fire to what is better than expected — novelty, surprise, an unpredictable payoff. They release into the nucleus accumbens. This is not pleasure. Schultz showed it tracks PREDICTION ERROR; Berridge and Robinson showed it carries wanting, which can grow while liking stays flat or falls. A feed of unlimited novelty on a variable schedule is a prediction-error machine. That is why the search can matter more than the thing found.
In accumbens output neurons, dopamine raises cAMP, then PKA, then CREB — which, with SRF beside it at the fosB promoter, transcribes the Fos family. Repeated, a stable truncated variant accumulates: ΔFosB. The pool, not the molecule, lasts weeks. Nestler’s group calls it the molecular switch for sensitization. Downstream: more dendritic spines, altered AMPA trafficking, more dynorphin, stronger cue-triggered pursuit. Cocaine does this in D1 cells only; a natural reward in both.
Control migrates from ventral striatum, which is goal-directed, to dorsal striatum, which is habit. Prefrontal braking weakens. Koob and Volkow’s three-stage loop — binge, withdrawal, anticipation — closes: the extended amygdala raises CRF, norepinephrine and dynorphin, and the low after the high becomes its own reason for the next bout. Koob’s word for that low is hyperkatifeia.
ΔFosB, and the studies that found it
every one of these is a rodentBoth columns are driven by the same eight exposures. The difference is not how strongly either one responds — it is what is still there when the next exposure arrives. ΔFosB is unusually stable, so the stack is a running total, and you can count it. That is the entire reason this one protein keeps appearing in this literature, and it is why the thing being described is a change in the machine rather than a mood. Read the shape, not the height: the rows below are every study that found it, and every one of them is an animal.
Sucrose drinking and sexual behavior both raise ΔFosB in the accumbens. Overexpressing ΔFosB increases sucrose intake and promotes sexual behavior. Prior sex increases later sucrose intake.
Free full text ↗The cleanest causal paper. Sexual experience induces ΔFosB and dendritic spines and cross-sensitizes amphetamine reward. Blocking D1 during mating — not D2 — attenuated the first and prevented the other two. A dominant-negative construct blocks it too.
Free full text ↗The gate is double. Deleting CREB alone in adult accumbens left cocaine induction of ΔFosB intact; only CREB and SRF together blocked it. Both bind the fosB promoter after cocaine. Stress-driven ΔFosB needs SRF, not CREB.
Free full text ↗The cell-type wall. Cocaine induces ΔFosB in D1 cells only. Sucrose, enrichment and calorie restriction induce it in BOTH D1 and D2 cells, in all three striatal regions. A natural reward does not write the drug pattern.
Read it ↗ΔFosB in the accumbens increases food-reinforced instrumental responding and progressive-ratio motivation — how hard an animal will work.
Free full text ↗Wheel running induces ΔFosB and escalates running. The switch is not exclusive to anything harmful.
Free full text ↗High-fat feeding raises accumbens ΔFosB, with D2 receptors and BDNF up and D1 down. The diet writes the same mark the behaviors do.
Read it ↗Not a ΔFosB measurement — a D2 one, kept here because it is the braking half. Chronic high-fat feeding downregulates striatal D2 and weakens D2-mediated braking. Human PET (Wang 2001) finds the same D2 pattern in obesity.
Dopamine is not pleasure. It is prediction error — which is why the search can outlast the interest.
The chemistry, one line each
what each molecule is forSorted by the role each molecule is given above, and nothing more. Notice what this figure does not show: the two sides are evenly matched, and a supernormal stimulus does not add a molecule to the left or remove one from the right. It changes how hard the left side is driven. That is why the useful question is never which chemical is involved — all of them are, in anyone — but how far the thing in front of you is pushing them.
Prediction error and incentive salience — wanting, not liking. Overactivation strengthens the habit that precedes the hit and downregulates receptors so that ordinary rewards stop registering. Volkow, Wise & Baler, Nat Rev Neurosci 2017.
The stable transcription factor that makes a few episodes into a disposition. Weeks, not permanent — it declines across abstinence in the animal work. Its induction needs CREB and SRF together, and we reach only CREB.
From prefrontal cortex, amygdala and hippocampus onto the accumbens, carrying cue-triggered seeking. Altered AMPA/NMDA ratios and spine density are the structural signature.
Local inhibition in accumbens and prefrontal cortex. A weakened excitation–inhibition balance is the usual account of poor braking — inferred from imaging hypofrontality, never measured directly after a feed or a meal.
5-HT2C in accumbens and VTA generally suppresses reward motivation and supports impulse control. Low tone tracks impulsivity broadly, not this stimulus specifically.
From locus coeruleus: arousal, attention, and the salience of intense cues. It is not the plasticity switch. It is loud, not formative.
Mu-opioid signalling in ventral striatum carries hedonic liking, which Berridge showed can come apart from wanting entirely.
Rises with chronic reward and acts on kappa receptors to damp dopamine and produce dysphoria. The within-system opponent process — the system’s own correction, felt as a low.
Rises in the extended amygdala during withdrawal, from drugs and from palatable food alike. The negative-reinforcement half of the cycle.
From lateral hypothalamus, driving VTA dopamine during seeking and elevated in withdrawal craving.
CB1 signalling modulates food and sexual reward at the same mesolimbic nodes.
Social contact mobilizes anandamide in the accumbens VIA oxytocin — block the receptor and it stops (Wei et al., PNAS 2015). This is the best transmitter-level account of why real affiliation competes with an endless stream.
Supports the synaptic change after reward — and exercise raises it in hippocampus and cortex, which is part of why exercise protects rather than merely inducing the same switch.
Fires when reward is omitted or the animal is sated, suppressing VTA dopamine. A stream that never satiates is, in this frame, a stimulus that keeps the brake off. Never tested directly with these stimuli.
What is not known
the most important section hereThis is the only figure here whose point is how much of it is empty. Everything hollow is real science — it is just science done in animals, or done in people for a different behavior. We draw it hollow rather than leaving it out, because a page that quietly omitted the gap would look stronger and be worth less. Seven of ten.
No human study, in any language, has measured ΔFosB, dendritic spine density, GABA, norepinephrine, orexin or habenula firing as a function of pornography, social media or ultra-processed food. Every molecular finding above is a rodent.
The human work is imaging. Gola 2017: in 28 men treated for problematic use against 24 who were not, ventral striatum differed on cues predicting erotic pictures — not on the pictures themselves. Kühn & Gallinat, JAMA Psychiatry 2014: in 64 men, more weekly hours correlated with smaller right caudate — cross-sectional, and it cannot show which came first.
The dissent belongs here. Steele, Staley, Fong & Prause (2013) found EEG P300 to sexual images tracked libido, not problem severity. Grubbs et al. (2018) found perceived addiction often tracks moral incongruence more than hours of use — the distress is real; sensitization is not always the cause.
The formal status is modest. ICD-11 places compulsive sexual behavior under IMPULSE CONTROL, not addiction — while describing gambling disorder alongside the substance use disorders. DSM-5-TR has no pornography diagnosis. Food addiction remains a research construct.
And the consequence people attribute to it does not track the use. Across four independent samples, in different countries and from groups on opposing sides, the hours a man watches barely predict erectile difficulty or sexual satisfaction — while his own distress about watching predicts both. Those instruments measure distress, not exposure. No randomized trial exists, the one prospective test was null, and none of it controls for moral conflict.
And the claims you have heard are mostly wrong. Pornography does not "deplete dopamine." A weekend "dopamine detox" is not what the biology describes. Two numbers, not one: a single ΔFosB molecule has a half-life near ten hours in culture; the pool that repeated episodes build is what lasts WEEKS in the animal work. What changes it is removing the features that generate prediction error: unbounded novelty, variable timing, escalating intensity.
Everything above about ΔFosB was measured in rodents that mated, not in people who watched. Four steps stand between that evidence and the word in this section’s title, and only one of them has been measured in a person. They are set out in order below, because a reader who finds the seam for himself has been handled badly.
Gola 2017 compared 28 men in treatment for problematic use with 24 who were not. They differed in ventral-striatum response to cues that predicted erotic pictures — and not in their response to the pictures themselves, nor to cues predicting money. The difference sits in the anticipation.
Sexual experience raises accumbens ΔFosB, adds spines and cross-sensitizes amphetamine reward; D1 blockade during mating attenuated the first and prevented the other two. Every one of those animals copulated. Carrying it across assumes a species boundary, an image doing what an act does, and one mark.
ΔFosB is a transcription factor inside the cell, and reading it takes brain tissue. Every human figure in this literature is post-mortem. This is not an experiment somebody skipped; it is one that cannot be run in the people the question is about. A living human number would be an invention.
Tinbergen’s test was a choice: the bird rolled the plaster egg in and let her own clutch go cold — that is what earns the word supernormal. No such choice has been run. Unlimited novelty, no refractory limit, variable timing and escalation are features of the delivery, not a measure of the effect.
What the biology predicts, if the analogy holds, is more of the natural-reward pattern — and that pattern, where it was measured, is not the drug pattern: cocaine marks D1 cells only, while sucrose, enrichment and calorie restriction mark both. The honest inference is therefore narrower than the one told elsewhere, and it points the other way — an appetite being outbid, not a drug being taken. None of the protective side rests on any of it.
Every molecular finding on this page is a rodent. Every human finding is an image. Both of those are worth knowing, and neither is the other.
What is protective
and which of it we can honestly reachThe right-hand tile is dark for most of this loop, so it is “less screen time” by any measure you could put on a dashboard. Look at it in the moment it comes back on. It is exactly as saturated, exactly as irregular and exactly as novel as it ever was — because nothing about the stimulus changed, only its schedule. The left-hand tile is on the whole time and pulls at almost nothing. That is what this section means by cap the features, not the clock, and it is why the mechanisms listed below are all aimed at the gain rather than at the hours.
Unlimited novelty, variable-ratio timing and escalating intensity are what keep the bursts coming. Structured limits on those match the biology; a one-day fast does not.
The striking one: exercise induces ΔFosB too (Werme 2002), yet in a separate experiment wheel access during cocaine self-administration cut intake — significantly in females only (Cosgrove 2002). Sex and high-fat diet cross-sensitize instead. Human PET after eight weeks in methamphetamine users: striatal D2/D3 up about 15%, against 4% in controls.
Dopamine-receptor trafficking and prefrontal function are sleep-dependent. Short sleep raises impulsivity and the pull of high-calorie food.
Oxytocin → anandamide → CB1 in the accumbens. Social interaction biases VTA dopamine toward social cues and away from drug cues.
Withdrawal craving is stress-amplified; CRF and cortisol are the machinery. This is the single mechanism where our own evidence is deepest.
Environmental enrichment induces plasticity that lowers drug-seeking — Olsen’s 2011 review is the best map of which natural rewards push which way.
Brewer’s work: reduced posterior cingulate activity during craving, weakening the link from urge to action. The effect is on reactivity, not on "raising dopamine". The posterior cingulate is the main hub of the default mode network, which is the pathway named here.
This is the color lever, pointed at your own screen. Every case above works by supplying more saturation and contrast than anything in nature — and a backlit display emits light where a real surface only reflects it, so it can exceed any natural object. Setting a phone or a monitor to grayscale, or simply lowering vibrancy, removes the feature rather than asking you to resist it.
THE EVIDENCE, HONESTLY: one direct experimental test exists. Wickord and Quaiser-Pohl (Computers in Human Behavior Reports, vol 10, 2023, p.100294) switched participants’ phones to grayscale for at least 24 hours and concluded it decreased usage time. 240 people started; 92 completed the follow-up. It is self-report, it is short, and more than half dropped out. That is a real finding and a thin one, and the mechanism it rests on — that exaggerated color is a releaser — is far better established than the intervention is.
On iPhone: Settings → Accessibility → Display & Text Size → Color Filters → Grayscale. On Android: Settings → Digital Wellbeing, or Developer options → Simulate color space → Monochromacy. On a Mac: System Settings → Accessibility → Display → Color Filters. Most people find they keep the shortcut and toggle it at night.
The study ↗What this catalog actually acts on
the sweep, measuredThis is the honest half. Below is what the catalog actually acts on — every pathway on this site that touches the map above, with its grade and the blends that carry it. None of it treats a disorder, and none of it reverses sensitization. These are the mechanisms underneath the protective side: stress and cortisol, sleep, BDNF and plasticity, GABA, serotonin, dopamine handling, and the social, self-regulatory and plasticity mechanisms the protective rows name.
This figure is counted from the same list printed below it, every time the page loads, so the picture and the rows cannot drift apart. The third bar is not a failure. It is the part we have mapped and are not claiming — and the next movement draws it.
HPA Axis / CortisolA22 blends
The HPA axis governs cortisol, the stress hormone. Adaptogens help normalize an over- or under-active axis, improving resilience and sleep. Key botanicals that drive it
Carried by — 22 blends: Anti-Fatigue (Mitochondrial Endurance) 63 ingredients → · Appetite Suppressant (Satiety Signaling) 32 ingredients → · Autonomous (Cognitive Performance) 72 ingredients → · Autophagy Activator (Cellular Self-Renewal) 59 ingredients → · and 18 more
Mood (serotonin)A20 blends
Serotonin underlies mood and calm. Some botanicals modulate its synthesis and reuptake, with clinical data in mild-to-moderate low mood. Key botanicals that drive it
Carried by — 20 blends: Anti-Fatigue (Mitochondrial Endurance) 63 ingredients → · Appetite Suppressant (Satiety Signaling) 32 ingredients → · Autonomous (Cognitive Performance) 72 ingredients → · Autophagy Activator (Cellular Self-Renewal) 59 ingredients → · and 16 more
AnxietyB+17 blends
Carried by — 17 blends: Anti-Fatigue (Mitochondrial Endurance) 63 ingredients → · Apigenin (Calming Senolytic) 1 ingredient → · Appetite Suppressant (Satiety Signaling) 32 ingredients → · Ascendance (Analgesic & Anti-Inflammatory) 33 ingredients → · and 13 more
BDNFA15 blends
BDNF is a growth factor that builds new neurons and synapses — central to learning, memory and mood. Exercise and specific botanicals raise it. Key botanicals that drive it
Carried by — 15 blends: ALZ (Cognitive Preservation) 43 ingredients → · Autonomous (Cognitive Performance) 72 ingredients → · Ecstatic (Nervous System Stand-Down) 36 ingredients → · Endurance Eternal (Pigment Preservation) 33 ingredients → · and 11 more
SleepA13 blends
A full eight hours in the wrong shape leaves you unrestored. Explore the research
Carried by — 13 blends: Anti-Fatigue (Mitochondrial Endurance) 63 ingredients → · Appetite Suppressant (Satiety Signaling) 32 ingredients → · Ecstatic (Nervous System Stand-Down) 36 ingredients → · Endurance Eternal (Pigment Preservation) 33 ingredients → · and 9 more
GABAA6 blends
GABA is the brain's main calming neurotransmitter. GABAergic botanicals promote relaxation and sleep without sedative dependence. Key botanicals that drive it
Carried by — 6 blends: Apigenin (Calming Senolytic) 1 ingredient → · Ecstatic (Nervous System Stand-Down) 36 ingredients → · Endurance Eternal (Pigment Preservation) 33 ingredients → · Hypnotic Sleep (Slow-Wave Architecture) 23 ingredients → · and 2 more
Dopamine clearanceA3 blends
Catechol-O-methyltransferase — catecholamine (dopamine/norepinephrine) breakdown; prefrontal signaling tone. Explore the research
Carried by — 3 blends: Apigenin (Calming Senolytic) 1 ingredient → · Lipolytic (Stored-Fat Mobilization) 35 ingredients → · Luteolin (The Histamine Flavone) 1 ingredient →
Sleep hormoneA2 blends
Melatonin signals darkness to set sleep onset and is itself a potent antioxidant; some botanicals raise endogenous levels. Key botanicals that drive it
Carried by — 2 blends: Hypnotic Sleep (Slow-Wave Architecture) 23 ingredients → · Peel (Flavonoid Powerhouse) 54 ingredients →
Social cognition2 blends
The thinnest domain here — no study yet links an upstream pathway to a social endpoint. Explore the research
Carried by — 2 blends: Autonomous (Cognitive Performance) 72 ingredients → · Interstellar Matcha (Ceremonial Grade) 1 ingredient →
Stress resilience1 blend
Reactivity, not resting level, is what predicts: cortisol reactivity tracked incident hypertension, adjusted for baseline bp and resting cortisol. Only ~40% of people mount that response at all. Explore the research
Carried by — 1 blend: Trinity (Mind Body Spirit Harmonizer) 75 ingredients →
Endocannabinoid toneB+1 blend
The endocannabinoid system is the body's balance network for mood, pain, appetite and sleep. Supporting its tone helps the system return to equilibrium. Explore the research
Carried by — 1 blend: Trinity (Mind Body Spirit Harmonizer) 75 ingredients →
Dopamine breakdown1 blend
It has a high affinity for dopamine and phenethylamine, and its activity in the brain rises with age. Explore the research
Carried by — 1 blend: Rewire (Total Brain Reset) 32 ingredients →
Learning & memory1 blend
Separated from attention deliberately — cholinergic imaging in Parkinson's found denervation tracked attention, executive and visuospatial function but not memory. Explore the research
Carried by — 1 blend: ALZ (Cognitive Preservation) 43 ingredients →
Recognition memory1 blend
The weakest-evidenced pathway here — its only upstream support is preclinical and unpublished. Explore the research
Carried by — 1 blend: ALZ (Cognitive Preservation) 43 ingredients →
Sleep pressureA1 blend
Adenosine builds 'sleep pressure' across the day; supporting deep, restorative sleep clears it. Key botanicals that drive it
Carried by — 1 blend: Hypnotic Sleep (Slow-Wave Architecture) 23 ingredients →
OrexinA1 blend
Orexin is the wakefulness signal; dialing it down at night — the target of newer sleep drugs — eases sleep onset. Key botanicals that drive it
Carried by — 1 blend: Hypnotic Sleep (Slow-Wave Architecture) 23 ingredients →
Object recognition1 blend
Lost in agnosia after ventral-stream damage while reaching and grasping stay intact. Explore the research
Carried by — 1 blend: Autonomous (Cognitive Performance) 72 ingredients →
Brain adaptability4 blends
The brain's capacity to rewire in response to use. Explore the research
Carried by — 4 blends: Event Horizon (Post-Ethanol Restoration) 40 ingredients → · Liver Regenerator (Hepatic Restoration) 23 ingredients → · Plush (Collagen, Firmness & Glow) 40 ingredients → · Rewire (Total Brain Reset) 32 ingredients →
Executive function2 blends
One of the six DSM-5 cognitive domains: planning, decision-making, mental flexibility and self-regulation. Explore the research
Carried by — 2 blends: Autonomous (Cognitive Performance) 72 ingredients → · Nebula (Warrior Mode Rocket Fuel) 29 ingredients →
Social bonding hormone1 blend
The hormone and brain messenger involved in bonding, trust and calming the brain's threat response. Explore the research
Carried by — 1 blend: Trinity (Mind Body Spirit Harmonizer) 75 ingredients →
Social engagement1 blend
Social connection as a physiological input. Explore the research
Carried by — 1 blend: Trinity (Mind Body Spirit Harmonizer) 75 ingredients →
Impulse control1 blend
A subdomain of executive function: withholding a response you have already started to make. Explore the research
Carried by — 1 blend: Autonomous (Cognitive Performance) 72 ingredients →
Stress & cortisol1 blend
The body's mobilization response, built to be brief. Explore the research
Carried by — 1 blend: Trinity (Mind Body Spirit Harmonizer) 75 ingredients →
Default mode network1 blend
The brain's inward-facing default network, and how well it stands down when a task begins. Explore the research
Carried by — 1 blend: Autonomous (Cognitive Performance) 72 ingredients →
You do not have to out-argue a craving if you can turn down the thing that produces it.
What we map but do not claim
the gap, shownThis map carries mechanisms our formulas act on and mechanisms they do not. We map the second kind because they are how the thing works, not because we reach them. Which is which is not a list we type — the figure below reads it off the catalog every time the page loads, so the gap is shown rather than hidden and cannot quietly go stale.
A hollow circle is not an empty one. It is a pathway that exists in the literature, that this catalog can point at, and that we have chosen not to make a claim about because the evidence does not reach the bar. Showing them is the point: a page that printed only the filled circles would look twice as strong and be worth half as much.
Read the papers yourself
free, in full, and legalEverything above rests on published work, and most of the load-bearing papers happen to be open access. These are the full texts, free and legal, on PubMed Central.
No color was pushed, no motion was added, nothing below is arranged to keep you here. It is a list of papers with links on them. Everything else on this page used the mechanisms it was describing, on purpose, so that you would feel them rather than take our word for it — and the fair thing to do at the end is take all of it back off and let the sources stand by themselves. If the argument only worked with the lights on, it was not an argument.
9 papers, free in full. One mark each. Every one is a complete peer-reviewed text you can open and read without paying, without an account and without asking us to summarize it for you. If this page has told you something, the thing that makes it checkable is underneath.
ΔFosB and natural reward — sucrose and sex raise it; overexpressing it raises intake.
Open the full text ↗Sex, ΔFosB, dendritic spines, and the D1 dependence. The cleanest causal paper in the set.
Open the full text ↗The transcriptional mechanisms behind ΔFosB, from the group that characterised it.
Open the full text ↗Natural rewards, neuroplasticity and non-drug addictions — the best single map of which natural rewards push which way.
Open the full text ↗The opponent-process review: why the low after the high becomes its own reason for the next one.
Open the full text ↗Pornography considered as a supranormal stimulus, in the neuroplasticity frame — the paper that makes the analogy explicitly.
Open the full text ↗The internet-pornography review. Read with care: its ΔFosB passages assert the mechanism without citing a human study.
Open the full text ↗Oxytocin → anandamide in the accumbens: the transmitter-level account of why real social contact competes.
Open the full text ↗The Coolidge-effect measurement itself: accumbens dopamine across satiety and a novel partner.
Open the full text ↗Named here and NOT linked, because they are not free to read in full and we will not send you to a paywall as though it were a source: Tinbergen 1951 and Barrett 2010 are books; Volkow, Wise & Baler 2017 (Nature Reviews Neuroscience), Kühn & Gallinat 2014 (JAMA Psychiatry — abstract free, full text not), Voon 2014, Johnson & Kenny 2010 and Grubbs 2018 are behind publisher access. A library, an interlibrary loan or an author’s posted manuscript will reach them.
Not one of these eight can be bought. Five run on pathways we carry.
Everything in What is protective above is real, most of it is free, and not one of the eight comes in a jar — every row of it is behavior you do, and a page that offered a capsule instead would be doing to you exactly what it just spent twelve sections describing. What a blend reaches is narrower, and it is already printed upstairs: five of those eight rows name pathways this catalog carries, at the grades the next section gives them.
- The stress arm — HPA axis, stress and cortisol, anxiety. The page’s own line: “the single mechanism where our evidence is deepest.”
- Sleep and circadian regularity — sleep, sleep hormone, sleep pressure — on which dopamine-receptor trafficking and prefrontal control both depend, and the impulse control and executive function that rest on them.
- The plasticity enrichment runs through: BDNF, and brain adaptability itself. The substrate, not the enriching.
- Endocannabinoid tone — with the social bonding hormone and social engagement itself, which real social contact runs on. The thinnest of the five, and the next section prints how thin.
- The default mode network, which the mindfulness row is about: the posterior cingulate is its hub.
- Cap the features · do the exercise · have the conversation
- Learn the new thing · sit the practice · turn the color down
- Nor the sleeping itself. A blend can act on sleep pathways; only you can go to bed.
This hands one condition — Supernormal Stimulus Protection — to the builder, and the promise ladder picks the blends. It is the combo this page’s own sweep asks for: every pathway in the section below is in it, including the four rows above that name pathways we carry. We do not name the blends here, because the answer is computed from the pathways behind that condition and it should change when the evidence does. Educational only. If a behavior is costing you your work, your sleep or your relationships, that is worth taking to a clinician — not to a supplement.
Educational only. These statements have not been evaluated by the Food & Drug Administration, and nothing here is intended to diagnose, treat, cure or prevent any disease. If a behavior is costing you your work, your sleep or your relationships, that is worth taking to a clinician — not to a supplement.