A dolphin invents its own whistle as a calf and answers to it forty years later — even when the voice is stripped out.
Dolphins are small toothed whales. Each develops its own whistle in its first year and keeps it for life, others recognise it, and some populations pass learned hunting techniques down the female line.
The killer whale is the largest dolphin, which is the first surprise, and the second is how much of what dolphins are known for turns out to be about learning rather than instinct. A dolphin invents a whistle for itself as a calf and uses it for forty years; play a synthetic copy with the voice stripped out and its relatives still respond, so the identity is in the pattern, not the sound of the animal making it. In one bay in Western Australia a handful of dolphins carry sponges on their beaks while probing the seafloor, and their daughters do too — a technique confined to one family line rather than to one habitat. Set against this is a long record of overstatement. Dolphins have been credited with language, with names in the human sense, and with a general benevolence towards people that the behavioural record does not support. NatureHQ keeps the demonstrated results and marks where the popular version overshoots them, which in this subject is often.
Developed coverage · 89% complete · reviewed 2026-08-09
What this page covers
The oceanic dolphins, Delphinidae — around 40 species. Most research concerns two: the common bottlenose dolphin and the Indo-Pacific bottlenose dolphin.
Often confused with: Phocoena phocoena; Platanista gangetica
Quick facts
Family
Delphinidae — around 40 species, including the killer whale
Signature whistle
Invented in the first year, retained for decades
Tool use
Sponge carrying, in one Shark Bay matriline
Lifespan
Bottlenose dolphins commonly reach 40–50 years in the wild
The strongest result in dolphin communication, and the one most often described slightly wrongly.
A bottlenose dolphin calf develops a distinctive frequency contour during its first year and then keeps it, largely unchanged, for the rest of its life. Other dolphins can tell whose it is. The obvious explanation is that they recognise the voice, in the way you recognise a friend on the telephone — and the experiment that matters is the one that removed that possibility.
Each dolphin invents its own whistle, and others recognise it
Well supported
Good evidence backs this, though some details remain open.
Bottlenose dolphins develop an individually distinctive frequency contour during the first year of life and retain it for decades. Conspecifics respond to a synthetic reproduction of that contour with all voice characteristics removed, showing that identity is carried by the learned pattern itself rather than by vocal timbre.
Who this applies to
bottlenose dolphins, principally the long-studied Sarasota Bay population
Studied in
Tursiops truncatus
You may have heard
“Dolphins have names for each other”
A signature whistle is self-produced — closer to a signature than to a name given by others — and while dolphins copy each other’s whistles when separated, using a whistle to talk *about* an absent dolphin has not been shown. The learned, arbitrary, individual-specific part of the analogy holds; the naming part does not.
Why we rate it this way, and what the caveats are
Well supportedHigh confidence
The voice-stripped playback design rules out the obvious alternative — that dolphins simply recognise each other’s voices — and the developmental work showing whistles are invented young and kept for life comes from an independently tracked wild population.
How far it can be extended
Signature-whistle-like individually distinctive calls have been reported in other delphinids, though the voice-stripped playback test has been run on bottlenose dolphins.
Caveats
Small samples of temporarily restrained wild animals in the decisive experiments.
Most evidence comes from one intensively studied population that may not be typical.
Copying a whistle when separated is consistent with addressing an individual and does not demonstrate reference to an absent third party.
Still unanswered
Do dolphins use another individual’s signature whistle to refer to it in that individual’s absence?
How much additional information beyond identity does a whistle carry?
Janik et al., 2006 · Proceedings of the National Academy of Sciences
Dolphins responded to synthetic copies of a relative’s whistle contour with voice features removed, isolating the learned pattern as the identity carrier.
Sayigh et al., 1990 · Behavioral Ecology and Sociobiology
Established that whistles are developed in the first year and retained for at least a decade — a label that changed would be useless for identifying an individual across a long life.
How we know
Playing a dolphin its relative’s whistle with the voice removed
Dolphins recognise each other’s signature whistles. Are they recognising the tune, or just the voice singing it?
This is the control that makes the whole "dolphins have names" question answerable. Wild bottlenose dolphins were temporarily restrained and played synthetic whistles that reproduced the frequency contour of a close relative’s signature whistle — the shape of the pitch over time — with every trace of the animal’s vocal characteristics computationally removed. Control contours of similar complexity, belonging to no relative, were played alongside. The measure was whether the dolphin turned towards the speaker.
What happened
Dolphins turned towards the speaker significantly more often for the voice-stripped version of a relative’s whistle than for control contours.
What it shows
Identity is carried by an arbitrary learned pattern, independent of who is producing the sound. That is a property a name has and a voice does not — which is what makes the analogy defensible at all.
What it does not show
It does not show dolphins using a whistle to refer to an absent third party, which is the other half of what a name does and remains undemonstrated. Fourteen restrained animals from one intensively studied population is also a narrow base, and restraint is itself a strong context.
The controls — what makes this evidence rather than a story
Voice removal is the point: any response to the synthetic version cannot be voice recognition, which every social mammal does.
Control contours matched the test whistles in complexity but not identity.
Playback order was varied so responses could not track habituation or sequence.
Researchers synthesised the contour — the shape of the whistle in frequency and time — with every trace of the animal’s voice removed, and played it to its relatives. They responded anyway. What identifies the individual is an arbitrary learned pattern, which is a property a name has and a voice does not.
Dolphins copy the signature whistles of close associates, and do so most when the two are separated — the nearest thing yet observed to calling someone by name.
What has not been shown is a dolphin using another’s whistle to talk about it to a third dolphin. That is the difference between a signature and a name, and it is the whole of the gap between the evidence and the headline.
Each bottlenose dolphin develops its own whistle, others learn to copy it, and a dolphin answers when its own whistle is played back.
Established
Specialists would state this without hedging. Multiple independent lines of evidence agree.
Bottlenose dolphins develop individually distinctive signature whistles within the first months of life and retain them for decades. Playback of a synthesised copy of an animal’s own signature whistle, with voice characteristics removed, elicits a vocal response, while playbacks of familiar and unfamiliar animals’ whistles do not.
Who this applies to
Demonstrated in bottlenose dolphins. Signature-whistle-like signals are reported in a few other delphinids but are far less well characterised.
Studied in
Tursiops truncatus
Why we rate it this way, and what the caveats are
EstablishedHigh confidence
Fifty years of observational work established the whistles; the playback experiment established that the contour alone is enough for an animal to recognise it is being addressed. Both halves have been independently replicated.
How far it can be extended
The experimental work is almost entirely on one species and on a small number of intensively studied populations.
Caveats
Copying happens mostly between close associates — mothers and calves, allied males — rather than across a population.
A dolphin responding when addressed is not the same as a dolphin talking about a third animal that is not there.
Still unanswered
Whether a dolphin can use another’s whistle to refer to it in its absence.
What, if anything, is added to the whistle beyond identity.
Janik et al., 2006 · Proceedings of the National Academy of Sciences
Established that identity is carried by the whistle contour rather than by voice quality, which is what makes the signal a learned label rather than a recognisable voice.
How we know
Calling a dolphin by its own name, in somebody else’s voice
When a dolphin hears its own signature whistle, is it recognising the whistle, or recognising the voice of whoever produced it?
Each animal’s signature whistle was resynthesised — the contour preserved, the voice characteristics that identify the producer stripped out. That synthetic version was played back to the animal, alongside two controls: the synthesised signature whistle of a familiar animal from the same group, and that of an animal it had never met. Whether the subject whistled back was recorded.
What happened
Dolphins answered playbacks of their own whistle and did not answer either control.
What it shows
The contour by itself is enough for a dolphin to recognise that it is the one being addressed. That makes the whistle a learned, arbitrary, individually specific label with a receiver who responds to it — the strongest evidence for a language-like property in any animal signal.
What it does not show
It does not show that dolphins use each other’s labels the way people use names. Nothing here involves one dolphin referring to a third animal that is not present, which is the property that would make it displacement. Copying another animal’s whistle happens mostly between mothers and calves and between allied males rather than across a population, so this is not a general addressing system. The sample is small and the animals were temporarily restrained.
The controls — what makes this evidence rather than a story
Voice features removed, so a response cannot be recognition of who is speaking.
A familiar animal’s whistle as a control, which rules out a general response to anything familiar.
An unfamiliar animal’s whistle as a second control, which rules out a general response to any whistle.
Some dolphins wear sponges to forage, and daughters learn it from their mothers
Well supported
Good evidence backs this, though some details remain open.
A minority of Indo-Pacific bottlenose dolphins in Shark Bay carry marine sponges on the rostrum while probing the seafloor. The behaviour is confined almost entirely to a single matriline, is not explained by nuclear relatedness or habitat, and is acquired by calves during a prolonged association with the mother.
Who this applies to
one population in Shark Bay, Western AustraliaDo not extend this beyond the taxa listed — the popular version over-reaches.
Studied in
Tursiops aduncus
Why we rate it this way, and what the caveats are
Well supportedHigh confidence
Two independent lines converge: genetic analysis excluding ecological and nuclear-genetic explanations, and decades of direct observation of individually known mothers and calves.
How far it can be extended
Sponging is essentially confined to this population and to a subset within it. It is not something dolphins in general do.
Caveats
The cultural conclusion rests partly on excluding alternatives rather than on observing transmission directly.
Overwhelmingly a female behaviour; why sons rarely take it up is unresolved.
A tiny fraction of a single population.
Still unanswered
Why is sponging almost absent in males?
Does the sponge protect the rostrum, flush prey, or both?
Long-term observation of the same population showing calves of sponging mothers acquiring the behaviour.
The genetic work is the interesting part, because it was used to rule genetics out. If sponging were inherited, spongers should be related through their nuclear genes; if it were about habitat, spongers should be found wherever the right seafloor is. Neither holds. What predicts sponging is descent through the mother’s line, which is what social learning looks like when written into a pedigree.
In the 1980s two dolphins were taught artificial languages — one of gestures, one of sounds — with a vocabulary of objects, actions and modifiers and a fixed word order. They were then given sentences they had never seen, including reversible ones where the same words in a different order mean something different. They got them right.
This is a real and impressive result about what a dolphin brain can parse. It is not evidence that dolphins have a language, and the distinction is comprehension versus production: these animals understood structured instructions and never produced a structured utterance. Nothing in natural dolphin communication has been shown to have this property.
Dolphins investigate marks on their own bodies using a mirror
Emerging evidence
Real findings exist, but too few or too recent to be settled.
Bottlenose dolphins marked with non-toxic ink on body regions they cannot see directly spend significantly more time at reflective surfaces and orient to bring the marked region into view, relative to sham-marked controls.
Who this applies to
two captive bottlenose dolphins in the original test
Studied in
Tursiops truncatus
Why we rate it this way, and what the caveats are
Emerging evidenceModerate confidence
The sham-mark control is the right design and the result has been replicated in later work, but the original sample was two animals and the mark test is asymmetric — passing is informative, failing is not.
Caveats
Two individuals in the original study, both captive and long habituated to humans.
A dolphin cannot touch a mark, so the classic self-directed behaviour is replaced by positioning.
What the mark test measures is itself disputed; passing it is not equivalent to self-awareness.
Still unanswered
What exactly does mirror mark-directed behaviour indicate about self-representation?
Why do some species with complex social lives fail the test?
Not vision with sound — a sense that reports what things are made of.
A dolphin hears through its jaw, and its sonar reports what a thing is made of, not just where it is
Established
Specialists would state this without hedging. Multiple independent lines of evidence agree.
Dolphins generate broadband clicks in the nasal passages, focus them through the fatty melon into a directional forward beam, and receive returning echoes through fat channels in the lower jaw conducting to the middle ear. Trained animals discriminate targets differing in wall thickness, internal contents or material at ranges of tens of metres.
Who this applies to
bottlenose dolphins, principally trained animals in controlled discrimination tasks
Studied in
Tursiops truncatus
Why we rate it this way, and what the caveats are
EstablishedHigh confidence
Decades of controlled discrimination testing with measured acoustics, supported by anatomical and acoustic modelling of the sound path.
How far it can be extended
The anatomical apparatus — melon, fat-filled jaw, nasal click generation — is shared across odontocetes, though click structure varies considerably by species.
Caveats
Performance figures come from trained captive animals on artificial tasks.
What wild dolphins actually attend to is far less characterised.
Echolocation is not vision with sound — the information returned is about material and structure, not colour or fine shape at distance.
Still unanswered
How much of a wild dolphin’s foraging depends on echolocation versus passive listening?
How do dolphins avoid being deafened by their own clicks?
Griffin and Galambos, 1941 · Journal of Experimental Zoology
The independent evolution of echolocation in bats, for comparison of what the two systems can and cannot do.
How we know
Flying bats through a room strung with wires
Spallanzani showed in 1794 that blinded bats navigate and deafened ones cannot. For 150 years nobody could say how. What is the missing sense?
A room was strung with a grid of vertical wires, and bats were flown through it under four conditions: intact, with vision blocked, with the ears plugged, and with the mouth covered. Collisions were counted in each. The genuinely new ingredient was not the flight room but the instrument — a newly available detector that could hear above 20 kHz, so the researchers could finally listen to what the bats were emitting while they flew.
What happened
Blindfolded bats flew the grid as well as intact ones. Ear-plugged and mouth-covered bats collided repeatedly. Ultrasonic pulses were detected throughout flight, rising in rate as obstacles approached.
What it shows
Bats orient by emitting high-frequency sound and listening for the echoes. The hypothesis was 150 years old; what changed was that a microphone existed which could hear the evidence.
What it does not show
Avoiding wires in a room is a far simpler task than catching an insect in flight, and it says nothing about the range or resolution of echolocation in the field. It also covers two North American species — Old World fruit bats mostly do not echolocate at all, so "bats navigate by echo" needs its exceptions.
The controls — what makes this evidence rather than a story
Removing one channel at a time is what stops the result being attributed to general impairment from being handled or covered.
Intact bats flying the same grid gave the baseline collision rate.
Covering the mouth tests emission specifically, separately from hearing — a bat that can listen but not call should still fail.
The anatomy is strange enough to be worth stating plainly. A dolphin makes its clicks in the nasal passages, not the larynx. It focuses them through the melon, a fatty lens in the forehead, into a narrow beam pointed forward. And it does not hear the returning echoes through its ears in any ordinary sense: sound enters through fat-filled channels in the lower jaw and is conducted to the middle ear.
What comes back is not a picture. A trained dolphin can tell a cylinder with slightly thicker walls from one with thinner walls, and a water-filled target from an oil-filled one, at tens of metres. Sound passes into objects and returns altered by what is inside them, so echolocation reports material and internal structure — information vision simply does not carry.
A dolphin can adjust its beam width and click rate continuously, effectively zooming as it closes on a target. Click rate rises so high in the final approach that it becomes an audible buzz.
Male dolphins form alliances — and their alliances form alliances
Well supported
Good evidence backs this, though some details remain open.
Male Indo-Pacific bottlenose dolphins in Shark Bay form stable first-order alliances of two or three individuals that cooperate to herd single females, and those alliances combine into larger second-order alliances that contest access to females with other such groups.
Who this applies to
male dolphins in the Shark Bay population, Western Australia
Studied in
Tursiops aduncus
Why we rate it this way, and what the caveats are
Well supportedModerate confidence
Based on years of observation of individually recognised animals, which is the only way to see structure at this level. Confidence is held at moderate because association is inferred from co-occurrence, and the pattern is documented in one exceptionally well-studied population.
Caveats
Co-occurrence can overstate social bonds; not every animal seen together is allied.
One population, with an unusually long observation history and unusually clear water.
The alliances function largely in coercive herding of females, which the popular picture of dolphin sociality omits.
Still unanswered
Do dolphins track third-party relationships — who is allied with whom — as the structure implies they might?
How common are nested alliances in other dolphin populations?
Krützen et al., 2005 · Proceedings of the National Academy of Sciences
The same population, and evidence that behaviour there passes along social lines.
In Shark Bay, male dolphins form stable pairs and trios that work together to herd individual females. Those small alliances then combine into larger groups that take females from other alliances and defend against the same. Keeping track of that requires holding two social maps at once: who your partners are, and which other groups your group works with.
It is worth being honest about what the alliances are for. This is coercive herding of females by cooperating males, and it sits awkwardly with the benign social intelligence dolphins are usually credited with. The cognitive achievement is real; the behaviour it supports is not gentle.
The wider society is fission–fusion: animals associate, separate and reassociate constantly, so a dolphin meets the same individuals repeatedly across decades in changing combinations. That is the condition under which a stable individual label — a signature whistle held unchanged for a lifetime — earns its keep.
Words used here
Fission–fusion society
A society where group membership changes constantly while the wider community persists. Chimpanzees, elephants and dolphins all live this way.
A dolphin sleeps half its brain at a time, with the opposite eye shut
Established
Specialists would state this without hedging. Multiple independent lines of evidence agree.
Cetaceans exhibit unihemispheric slow-wave sleep: one cerebral hemisphere shows sleep-pattern EEG while the other remains awake, with the eye contralateral to the sleeping hemisphere closed. The hemispheres alternate. REM sleep is minimal or undetectable by standard criteria, and newborn calves with their mothers show almost no conventional sleep for the first weeks of life.
Who this applies to
cetaceans in which sleep has been recorded, principally in captivity
They sleep constantly — just never all at once. Breathing is voluntary in a dolphin rather than automatic, so a fully unconscious dolphin would drown. Half the brain rests while the other half swims, watches and surfaces, and then they swap.
Why we rate it this way, and what the caveats are
EstablishedHigh confidence
Directly recorded by EEG in multiple species across several laboratories, and consistent with the behavioural requirement to surface and breathe voluntarily.
How far it can be extended
Unihemispheric sleep has been recorded across several cetacean species independently, and follows from obligate voluntary breathing shared across the group.
Caveats
Almost all recordings are from captive animals; wild sleep is very hard to measure.
Absence of detectable REM may reflect criteria developed for land mammals rather than its true absence.
The near-absence of sleep in newborn calves and their mothers is remarkable and not well explained.
Still unanswered
Do cetaceans have REM sleep in some form that standard criteria miss?
How do mothers and calves function for weeks with almost no conventional sleep?
Lyamin et al., 2008 · Neuroscience & Biobehavioral Reviews
Synthesis of EEG and behavioural sleep recordings across cetaceans.
Breathing in a dolphin is voluntary rather than automatic, so a fully unconscious dolphin would drown. The solution is to sleep one cerebral hemisphere at a time, with the eye opposite the sleeping half closed, and swap over — so the animal is never entirely asleep and never entirely awake.
Two things about this remain genuinely puzzling. REM sleep, near-universal in mammals and often tied to memory, is minimal or undetectable by the usual criteria in cetaceans. And newborn calves, along with their mothers, show almost no conventional sleep for the first weeks of life — an animal that swims continuously from birth, apparently without the rest every other mammal requires.
Words used here
Unihemispheric slow-wave sleep
Sleeping with one brain hemisphere at a time. Found in cetaceans, some seals and many birds.
Melba and David Caldwell noticed that individual captive dolphins each produced a distinctive, repeated whistle contour, and proposed that it functioned as an individual identifier.
1984
Landmark experiment
Dolphins shown to parse taught sentences
Two dolphins trained on artificial gestural and acoustic languages responded correctly to novel sentences, including reversible ones where word order determines meaning.
Recording individually known wild dolphins repeatedly over years established that signature whistle contours persist for at least a decade, and that calves develop their own rather than copying the mother.
Long-term tracking of individually recognised males revealed alliances that themselves form alliances — a two-tier social structure otherwise known mainly from humans.
The voice is stripped out, and recognition survives
Wild dolphins responded to synthetic reproductions of a relative’s whistle contour with all voice characteristics removed, showing identity is carried by the learned pattern rather than by vocal timbre.
Changes how the 1965 result reads
The Caldwells’ proposal had an obvious alternative explanation that stood for forty years: dolphins might simply be recognising each other’s voices, as most social mammals do. Removing the voice entirely and finding recognition intact is what converted a plausible interpretation into a demonstrated one.
Dolphins were shown to reproduce the signature whistles of close associates, and to do so most when separated from them — the nearest thing yet observed to addressing an individual by name.
A dolphin breathes air through a blowhole on top of its head — a nostril that migrated backwards over evolutionary time — and it breathes voluntarily, one conscious decision at a time. That single fact explains a surprising amount, including why a dolphin cannot sleep the way a land mammal does and why anaesthesia is dangerous for one.
No gills, and no ability to extract oxygen from water. A dolphin held underwater drowns.
Teeth, in a single set that is never replaced — a dolphin that loses one does not grow another.
Hair, briefly: calves are born with a few whiskers on the rostrum which are lost within weeks.
No drinking. Fresh water comes from food and from metabolising fat, since drinking seawater would cost more water than it gained.
Cannot survive on land: without the water’s support, body weight compresses the lungs and the animal overheats.
The jumping has more than one explanation and no single answer. Leaping is cheaper than swimming at speed — air offers far less resistance than water — so a travelling dolphin porpoising is saving energy. It also dislodges parasites, gives a look above the surface, and appears in social and play contexts where no practical function is obvious.
Riding the bow wave of a boat is the same economics. A vessel pushes a wall of water ahead of it, and a dolphin in that pressure field is carried along for very little effort — the same trick they use on the wave in front of a large whale, which long predates boats.
A few dolphin species do live in fresh water, including the Amazon and Ganges river dolphins. They are not the same species as the bottlenose dolphins of the sea, and they are among the most threatened cetaceans.
Words used here
Blowhole
The nostril on top of a cetacean’s head. Breathing through it is a voluntary act, not an automatic one.
Porpoising
Leaping clear of the water while travelling at speed. Cheaper than swimming, because air resists less than water.
A bottlenose dolphin is pregnant for about twelve months and gives birth to a single calf, tail first — an arrangement that matters, because a calf emerging head first would have to hold its breath through the whole delivery. The mother, sometimes helped by another female, guides it to the surface for its first breath within seconds.
Gestation around 12 months; a single calf, born tail first.
Nursing for one to two years, on milk delivered by muscular contraction rather than suckling — a calf cannot form a seal underwater.
Calves stay with their mothers for three to six years, learning foraging techniques that differ between families.
Females typically calve every three to six years, which makes population recovery slow.
Wild bottlenose dolphins commonly live 40 to 50 years, with females often outliving males.
The learning period is the part that connects to everything else on this page. Sponging — carrying a sponge as a tool while foraging — passes from mother to daughter over those years, which is why it runs in matrilines rather than spreading across a population. A dolphin calf is not simply growing; it is being taught a local way of making a living.
Captive lifespans have historically been shorter than wild ones, and comparisons remain contested because captive populations are small, unevenly aged and change with husbandry practice. NatureHQ records the wild figures and treats the captive comparison as an open question rather than a settled number.
Words used here
Matriline
A line of descent through females. Dolphin foraging traditions frequently follow one.
Dolphins are whales. All dolphins are toothed whales, the killer whale is the largest dolphin, and the animals people call porpoises are a separate family with blunter heads, no beak and spade-shaped rather than conical teeth.
Delphinidae — oceanic dolphins, around 40 species, including the killer whale and the bottlenose.
Phocoenidae — porpoises. Smaller, beakless, spade-shaped teeth. Not dolphins.
River dolphins — several unrelated families that arrived at a similar body separately; the Ganges and Amazon river dolphins are not close relatives of each other.
The confusion is worth clearing up because research findings do not transfer across these groups. A result about a bottlenose dolphin says nothing reliable about a harbour porpoise.
The reputation is built on a facial structure that cannot change expression.
A dolphin appears to be smiling and cannot do otherwise. The upturned line of the mouth is fixed anatomy — the jaw and the melon give the face that shape whatever the animal is doing, and there are no facial muscles to move it. A dolphin in pain, in distress or dying looks exactly the same as one that is not. Almost everything people believe about dolphin temperament traces back to that accident of skull shape, and the belief has real consequences: captive animals in obvious distress have been read as content by visitors for decades.
The behaviour underneath is more interesting than either the friendly version or the corrective version that has replaced it in some quarters. Wild dolphins are curious, socially complex, and entirely capable of aggression. They kill harbour porpoises in some populations — repeatedly, with severe injuries, and not for food, which remains unexplained. Males form coalitions that coerce females. Infanticide has been documented. None of that makes them malicious; it makes them a large wild predator with a rich social life, which is exactly what they are.
Attacks on people are rare and are almost entirely associated with feeding or with lone habituated animals. A dolphin that has learned to associate people with food loses its wariness, and injuries follow — which is why feeding wild dolphins is illegal in many places. The stories of dolphins protecting swimmers from sharks are a different matter: some incidents are well attested, and there is no way to distinguish protection from a dolphin group mobbing a shark for its own reasons, with a person incidentally inside the circle. The more careful reading is that dolphins harass sharks, which they do, and that a swimmer sometimes benefits.
Bow-riding — the behaviour that puts dolphins in front of a moving ship — is the one that looks most like friendliness and is best explained without it. A moving hull pushes a pressure wave ahead of it, and a dolphin positioned in that wave is carried along with almost no effort. They did it with whales before there were ships. The dolphins riding a bow wave are getting a free ride, and whether they also enjoy it is not something a boat can tell you.
Safety
Around wild dolphins
Do not feed them, do not swim towards them, and let any approach be theirs. Feeding wild dolphins is illegal in many jurisdictions because it produces habituated animals that later injure people, and swim-with operations are regulated or prohibited in many places for the same reason.
Where this applies: Approach distances, feeding and swim-with activity are regulated nationally and vary widely.
When to get help: Consult the local marine wildlife authority before any organised interaction.
Words used here
Bow-riding
Swimming in the pressure wave pushed ahead of a moving vessel, which carries the animal along for free.
Habituation
Losing wariness of people through repeated contact, usually because of food. The main route to a dolphin injuring someone.
The pufferfish story is one of the most widely repeated claims about animal behaviour, and it rests on considerably less than its circulation suggests. Footage from a 2013 documentary showed young dolphins passing a pufferfish between them, mouthing it gently rather than eating it. The narration proposed that they were deliberately taking small doses of tetrodotoxin to get high, and that framing is what travelled.
What can be said: the behaviour was filmed, the fish was passed around, and it was not eaten. Tetrodotoxin is genuinely one of the most potent toxins known, and it is a nerve blocker rather than anything resembling a recreational drug — at low doses it produces numbness and paralysis, not euphoria, in every animal it has been studied in. No measurement of dose, of uptake, or of any physiological effect on the dolphins was taken. Dolphins also handle, toss and pass all sorts of objects, which is the ordinary explanation and the one the footage on its own supports.
NatureHQ records this as an observation of object play with an untested interpretation attached. It could be right. Nothing published establishes it, and a documentary hypothesis repeated for a decade does not become evidence by repetition.
The shark question is better grounded. Dolphins do attack sharks, ramming them in the gills and flanks, and groups have been recorded driving sharks away and occasionally killing smaller ones. They are also eaten by large sharks, frequently — bite scars are common on wild dolphins, and shark predation is a leading cause of calf mortality in some populations. The two facts sit together: a dolphin group can beat a shark and an individual dolphin usually cannot, which is why the behaviour is collective.
Dolphins do not eat people and there is no reason they would. The diet is fish and squid, taken whole; the teeth are conical and built for gripping rather than cutting, and a dolphin cannot chew.
Words used here
Tetrodotoxin
The pufferfish toxin. A sodium-channel blocker that causes numbness and paralysis, with no known intoxicating effect.
Do dolphins use a signature whistle to refer to an absent individual — the last missing piece of the "name" claim?
Why it matters: This is the single result that would make the word “name” defensible rather than an analogy, and popular coverage of dolphin communication already assumes it has been shown.
How much information beyond identity does a whistle carry?
Why it matters: If whistles carry emotional state, location or intent as well as identity, the signalling system is far richer than a set of individual labels.
Why is sponging almost exclusively female?
Why it matters: The sex bias is the strongest clue to why the tradition persists, and nothing currently separates foraging need from opportunity to learn.
What does passing the mirror mark test actually indicate about a dolphin’s self-representation?
Why it matters: The test is used across species as a proxy for self-awareness while what it measures remains disputed, so its results are routinely over-read.
How complete this page is, and what it is still missing
NatureHQ publishes its own gaps. This page is at 89% completeness against what we would call a finished subject, and was last reviewed on 2026-08-09. It carries 10 claims and answers 66 mapped search questions.
2 high-priority search intent(s) not yet covered
no research from the last few years is attached — check for newer work
Echolocation in dolphins is mentioned only in passing and deserves its own section.
Social structure — fission–fusion societies and male alliances — is not yet covered.
Captivity, and the welfare literature around it, is absent and should not stay absent.