Skip to content
NatureHQ

species group

Crows

Corvus Linnaeus, 1758

Crows are medium-sized black birds in the genus Corvus, found on every continent except Antarctica and South America. They are among the most intelligent birds studied, though the headline abilities belong to particular species rather than to crows as a whole.

Crows sit at the point where "clever animal" stops being a figure of speech. Wild American crows recognise individual human faces and remember a threatening one for years. Carrion crows have neurons that fire for specific numbers. New Caledonian crows manufacture hooked tools to a standard design. Ravens, their close relatives, appear to plan for events hours ahead. But these are four different birds, and treating them as one animal is how a remarkable island species' tool-making becomes a claim about the bird on your fence. This page keeps them apart, says which species each finding comes from, and shows the experiments underneath.

In-depth record · 92% complete · reviewed 2026-08-09

What this page covers

Around 45 species in the genus Corvus. Most research on "crow intelligence" comes from four of them, and they are not interchangeable.

Often confused with: Corvus corax (common raven — larger, wedge-tailed, deeper call); Corvus frugilegus (rook — pale bare face, sociable); Corvus monedula (jackdaw — smaller, grey nape, pale eye)

Quick facts

Group
Genus Corvus, family Corvidae — around 45 species
Where
Worldwide except Antarctica and South America
Brain
Large relative to body size, with a dense forebrain region called the nidopallium caudolaterale
Face memory
At least five years, in wild American crows
Tool makers
Two Corvus species out of roughly 45
Diet
Omnivorous — carrion, invertebrates, grain, eggs, fruit and human refuse

Which crow are we actually talking about?

The single most useful thing to know before reading anything about crow intelligence.

The word "crow" covers around 45 species. Research on crow cognition is concentrated in four of them, and they differ enormously. Attaching a New Caledonian crow's tool-making to the birds in a British field is a bit like reporting a chimpanzee result as a fact about mammals.

The four species behind most crow research
SpeciesWhereWhat it is known for
American crow (Corvus brachyrhynchos)North AmericaFace recognition, long memory for threatening people, gathering around the dead
Carrion crow (Corvus corone)Europe and AsiaNumber-selective neurons; work on perception and reporting
New Caledonian crow (Corvus moneduloides)New CaledoniaManufacture of hooked twig and Pandanus-leaf tools
Hawaiian crow (Corvus hawaiiensis)Hawaiʻi — extinct in the wildSpecies-wide tool use, tested in almost the entire surviving population

Ravens (Corvus corax) are in the same genus but are a separate species with their own research literature, and the two are constantly confused in popular writing. Rooks and jackdaws are corvids too, but not crows.

Related

  • Common raven

    Bigger, wedge-tailed, and a different set of experiments

Words used here
Corvid
A member of the crow family, Corvidae — crows, ravens, rooks, jackdaws, magpies and jays.
Genus
A group of closely related species. Corvus is the genus containing crows and ravens.

Crows and people

The face research, and what it does and does not show.

The best-known work on crows concerns their relationship with us. Researchers in Seattle wore a rubber mask while catching and ringing wild American crows, then had different people walk the same routes wearing either that mask or a neutral one. The birds went after the mask that had caught them — not the person wearing it.

Crows can recognise individual human faces

Well supported

Good evidence backs this, though some details remain open.

Wild American crows (Corvus brachyrhynchos) discriminate between human faces and associate a specific face with threat, responding to that face independently of the body wearing it.

Who this applies to
American crows, in the wild
Studied in
Corvus brachyrhynchos

You may have heard

“Crows never forget a face, and will hold a grudge against you personally.”

The research is about a face linked to a genuinely threatening event — being caught and handled — and about a species studied in North America. Nothing suggests a crow forms an opinion of a person who has simply walked past, and "grudge" describes a human motive the experiments cannot detect.

Why we rate it this way, and what the caveats are
Well supportedHigh confidence

The mask design rules out the obvious alternatives — different people wore the same face — and the effect was replicated across sites, persisted for years, and has a matching brain-imaging result.

How far it can be extended

Comparable face and person discrimination has been reported in other corvids and in several unrelated urban birds, and the ability fits the ecology of any long-lived bird living alongside people. It has not been demonstrated experimentally in most Corvus species, so it remains an expectation rather than a finding.

Caveats

  • Demonstrated with masks, which are exaggerated and unmoving compared with a real face.
  • The measure is alarm behaviour, so it shows the birds told faces apart — not which features they used.
  • Shown in American crows; carrion crows, hooded crows and rooks have not been tested the same way.

Still unanswered

  • Which facial features do crows actually use?
  • Do crows recognise faces they associate with something positive as reliably as threatening ones?

Last reviewed 2026-08-09

The evidence (2 studies)

Crows remember a threatening person for years

Well supported

Good evidence backs this, though some details remain open.

American crows continued to respond defensively to a mask associated with capture for at least five years after the original event, with no reinforcement in between.

Who this applies to
American crows
Studied in
Corvus brachyrhynchos
Why we rate it this way, and what the caveats are
Well supportedHigh confidence

Tracked across multiple sites over several years by the same team using a consistent protocol, with the response persisting rather than fading.

Caveats

  • Because the response spread socially, some later scolding came from birds reacting to other birds rather than from personal memory.
  • Individual birds were mostly unmarked, so the duration is a population-level observation.

Still unanswered

  • How long does an individual crow retain the association without seeing others react?

Last reviewed 2026-08-09

The evidence (2 studies)

The part that turns a memory result into something closer to culture.

Crows tell each other about dangerous people

Well supported

Good evidence backs this, though some details remain open.

Defensive responses to a specific human face spread through American crow populations to individuals with no direct experience of the threatening event, including juveniles born afterwards.

Who this applies to
American crows
Studied in
Corvus brachyrhynchos
Why we rate it this way, and what the caveats are
Well supportedModerate confidence

The spread is well documented across five years and several sites. Whether it happens through deliberate signalling or by simply observing other birds mob is not resolved, and "tell" is a convenient shorthand rather than a description.

Caveats

  • Transmission was inferred from how the response spread, not observed between named individuals.
  • Learning by watching a mobbing crowd is a simpler explanation than intentional communication, and fits the data.

Still unanswered

  • Is there a specific signal, or do birds simply learn by joining a mob?
  • How accurately is the identity of the dangerous face preserved as it spreads?

Last reviewed 2026-08-09

The evidence (2 studies)

How we know

The caveman mask experiment

Can a wild crow tell one human face from another, and remember which one is dangerous?

Researchers put on a rubber mask — memorably, a caveman — and used it while catching and ringing wild American crows. The birds were released unharmed, but they had now seen that face doing something unpleasant. Afterwards, volunteers walked fixed routes through the same areas wearing either the caveman mask or a completely different, neutral one, and noted how the crows behaved. Crucially, the masks were passed between different people, of different heights and builds, and worn in different clothes.

What happened

Crows scolded, dived at and mobbed the caveman mask far more than the neutral one. They kept doing it for years, and birds that had never been caught joined in. In later work the reaction spread through the population and to birds too young to have been present.

What it shows

Wild crows can identify a particular human face, attach a meaning to it, and hold onto that association for years. Because the face moved between bodies, the face itself must be what they were tracking.

What it does not show

It does not show that crows see faces the way people do, or that they could pick a face out of a crowd with no threat attached. It measures alarm behaviour, so it tells us the birds discriminated — not which features they used to do it.

The controls — what makes this evidence rather than a story
  • A neutral mask, used at the same sites by the same people, gave a baseline for ordinary human presence.
  • Swapping masks between wearers meant the birds could not be responding to a particular person's body, gait or clothing.
  • Routes and timings were fixed in advance, so the birds' reaction — not the walker's behaviour — was what varied.

From Lasting recognition of threatening people by wild American crows

How clever is a crow?

Number neurons, perception, and the limits of what any of it establishes.

Bird brains are organised nothing like ours — there is no cortex — yet corvids solve problems that were once assumed to need one. Two lines of work in carrion crows go further than behaviour and record what individual neurons are doing.

Individual neurons in a crow's brain respond to specific numbers

Well supported

Good evidence backs this, though some details remain open.

Neurons recorded in the nidopallium caudolaterale of trained carrion crows (Corvus corone) show tuning to specific numerosities, with response profiles comparable to number neurons in the primate prefrontal cortex.

Who this applies to
carrion crows, in a trained laboratory task
Studied in
Corvus corone
Why we rate it this way, and what the caveats are
Well supportedHigh confidence

Direct single-neuron recording is the strongest available evidence for a neural code, and the tuning curves match a well-characterised primate result. The constraint is two animals, which is normal for the method.

Caveats

  • Two heavily trained birds.
  • Small quantities only; this is a sense of number, not counting.
  • Bird and primate forebrains are organised differently, so the similarity is functional rather than structural.

Still unanswered

  • Do untrained crows use these neurons in natural foraging decisions?

Last reviewed 2026-08-09

The evidence (1 study)

Crows show a brain signature associated with conscious perception

Emerging evidence

Real findings exist, but too few or too recent to be settled.

In carrion crows performing a threshold detection task, neuronal activity in the nidopallium caudolaterale tracked the bird's reported percept rather than the physical stimulus — the signature used to study conscious perception in primates.

Who this applies to
carrion crows, in a trained detection task
Studied in
Corvus corone

You may have heard

“Scientists have shown that crows are conscious.”

What was shown is that crows' brain activity tracked what they reported seeing rather than what was actually shown — the same marker used in human and monkey research. Whether that marker means the same thing in a bird is precisely the open question, and no experiment of this design can close it.

Why we rate it this way, and what the caveats are
Emerging evidenceModerate confidence

The neural result is solid and elegantly designed. What it means is contested: a correlate of a report is not a demonstration of experience, and that gap is philosophical as much as empirical.

Caveats

  • Two animals.
  • Reported as "crows are conscious" far more confidently than the study claims.
  • A neural correlate of a report is not the same as subjective experience.

Still unanswered

  • Does the same signature appear in untrained birds and in other species?
  • What would count as evidence of experience rather than of reporting?

Last reviewed 2026-08-09

The evidence (1 study)

Both results come from two heavily trained birds each, which is normal for single-neuron recording and still a real constraint. The consciousness study in particular is reported far more confidently in the press than in the paper.

Compare across species

Words used here
Nidopallium caudolaterale
A region at the back of a bird's forebrain that does much of the work a mammal's prefrontal cortex does, despite being built differently.
Numerosity
How many items are in a set — the quantity itself, before any counting or naming.

Do crows understand why things work?

The failures are the informative part, and they are rarely reported.

A crow that has learned to avoid a hole in a tube and a crow that has grasped that objects fall through holes behave identically — until you change the apparatus. That is the whole design of this literature: keep the causal structure and change how it looks, then keep the look and change the structure, and see which the bird follows.

New Caledonian crows grasp causes they can see, and fail when the mechanism is hidden

Well supported

Good evidence backs this, though some details remain open.

New Caledonian crows transfer learned solutions to physical tasks across changes in appearance that preserve causal structure, and fail across changes in causal structure that preserve appearance. On water-displacement tasks they select sinking over floating and solid over hollow objects, but fail when success requires inferring a connection between two tubes hidden beneath the apparatus.

Who this applies to
New Caledonian crows, a tool-using species unusual even among corvidsDo not extend this beyond the taxa listed — the popular version over-reaches.
Studied in
Corvus moneduloides

You may have heard

“Crows understand physics”

The failures are the informative part and they are rarely reported. These crows handle causes they can observe and do not infer a mechanism they cannot see — a specific and interesting profile that "understands physics" flattens into something both grander and less useful.

Why we rate it this way, and what the caveats are
Well supportedModerate confidence

The transfer design is the right test and the pattern of success and failure is consistent across two independent task families. Confidence is moderate because both studies used six wild-caught birds with variable individual performance.

How far it can be extended

New Caledonian crows are exceptional among corvids in habitual tool manufacture, and these results should not be read as describing crows in general.

Caveats

  • Six birds per study, with meaningful variation between individuals.
  • Trap-tube and displacement tasks both have histories of lower-level explanations.
  • The comparison to children’s performance draws on different laboratories and procedures.

Still unanswered

  • Is the hidden-connection failure a limit on inference, or on attention to the relevant part of the apparatus?
  • Do non-tool-using corvids show the same profile?

Last reviewed 2026-08-09

The evidence (3 studies)

How we know

Six water-displacement tasks, and the one the crows failed

New Caledonian crows drop stones into water to raise a floating reward. Do they understand why it works?

A floating treat sat out of reach in a tube of water, with objects available to drop in. Rather than one task, the crows faced a series designed to pull apart what they might have learned: sinking objects versus floating ones, solid versus hollow, a water-filled tube versus a sand-filled one, a narrow tube versus a wide one — and, finally, two visibly separate tubes that were connected out of sight beneath the table, so that dropping objects into one raised the level in the other.

What happened

The crows chose sinking over floating objects, solid over hollow, and water over sand. They failed the connected-tubes condition.

What it shows

The crows track causal properties they can observe. The profile is specific: visible causes yes, unobservable mechanism no — roughly where a five-to-seven-year-old child sits on the same task.

What it does not show

It does not show that crows understand physics, and the failure is the part usually left out of the coverage. Six birds, some tested across several tasks in sequence with possible carry-over, and object preferences could partly reflect how things feel to handle rather than what the bird predicts they will do.

The controls — what makes this evidence rather than a story
  • Each variant holds the reward and the general procedure constant and changes one causal property, so a bird following a surface rule and a bird following a causal one diverge.
  • The hidden-connection condition is the deliberate limit case: everything visible is unchanged, and only the unobservable causal structure differs.
  • Object preferences were measured against what was available, not merely counted.

From Using the Aesop’s Fable paradigm to investigate causal understanding of water displacement by New Caledonian crows

On water-displacement tasks the same profile appears. New Caledonian crows chose sinking objects over floating ones, solid over hollow, water over sand — and failed the condition where success required working out that two visibly separate tubes were connected out of sight beneath the table.

Visible causes yes; unobservable mechanism no. That is a specific and interesting profile, and roughly where a five-to-seven-year-old child sits on the same task. It is also considerably more useful than "crows understand physics", which flattens a result into a slogan.

Words used here
Transfer test
Changing the problem after training to find out what the animal actually learned — the rule, or the appearance.

The tool-making crows

Two species out of roughly forty-five, both on islands.

Habitual tool manufacture is the ability most often attached to crows in general, and it is the one that generalises least. It is known in two species, both island birds with unusual ecology.

New Caledonian crows make hooked tools to a standard design

Established

Specialists would state this without hedging. Multiple independent lines of evidence agree.

Corvus moneduloides manufactures hooked twig tools and stepped Pandanus-leaf tools to consistent designs, and uses them to extract invertebrate prey from crevices.

Who this applies to
New Caledonian crows onlyDo not extend this beyond the taxa listed — the popular version over-reaches.
Studied in
Corvus moneduloides

You may have heard

“Crows use tools.”

Two crow species out of roughly forty-five do this habitually, and both live on islands. Applying it to "crows" attaches an extraordinary island adaptation to the ordinary bird on a lamp post.

Why we rate it this way, and what the caveats are
EstablishedHigh confidence

Documented in the wild since 1996, replicated in the laboratory many times, and studied in enough detail that regional design variation has been mapped.

How far it can be extended

Habitual tool manufacture is known in exactly two of the roughly forty-five Corvus species — the New Caledonian crow and the Hawaiian crow — and both are island species with unusual ecology. The crows most people see do not make tools.

Caveats

  • This is one island species. It says nothing about the crows in a European or North American garden.
  • The famous wire-bending demonstration is now understood as an expression of a species-typical skill.

Still unanswered

  • How much of the design is learned from other birds and how much develops without a model?
  • Why has habitual tool use evolved in island corvids in particular?

Last reviewed 2026-08-09

The evidence (3 studies)

The Hawaiian crow uses tools too — and almost the whole species was tested

Well supported

Good evidence backs this, though some details remain open.

The great majority of surviving Corvus hawaiiensis use sticks to extract food from crevices, and juveniles develop the behaviour without demonstration, indicating a species-wide predisposition.

Who this applies to
Hawaiian crows, or ʻalalā
Studied in
Corvus hawaiiensis
Why we rate it this way, and what the caveats are
Well supportedHigh confidence

Essentially the entire surviving population was tested, which is an unusually complete sample — though only because the species is extinct in the wild.

Caveats

  • Every bird tested was in a conservation breeding programme; no wild population remains for comparison.
  • Tool use in the wild was never systematically documented before the species disappeared from it.

Still unanswered

  • What did tool use look like in the wild, and can it be restored with reintroduced birds?

Last reviewed 2026-08-09

The evidence (1 study)

The famous "Betty the crow" experiment, in which a bird bent wire into a hook, was reinterpreted after fieldwork showed that hook-making is normal behaviour for wild New Caledonian crows. The result stands; the explanation changed.

Based on New Caledonian crows make hooked tools to a standard design

Compare across species

Gathering around the dead

Crows will mob and shout around the body of another crow, a behaviour widely described as a funeral. It has been tested directly, with controls, and the answer is more practical than the name suggests.

Crows gather around a dead crow — but it is danger learning, not a funeral

Well supported

Good evidence backs this, though some details remain open.

American crows presented with a dead conspecific mob the person holding it, avoid the location afterwards, and continue to scold that person's face for weeks. The behaviour is consistent with learning about danger; no measure in the study addresses grief.

Who this applies to
American crows
Studied in
Corvus brachyrhynchos

You may have heard

“Crows hold funerals for their dead.”

The gathering is real and well documented. What the experiment measured was the birds learning that a place and a person were dangerous — they avoided the spot and scolded the face for weeks afterwards. Whether anything like grief is also present is untested, and the study was never designed to find out.

Why we rate it this way, and what the caveats are
Well supportedHigh confidence

Directly tested with appropriate controls, including a dead pigeon condition and the mask alone. The evidence for the behaviour is strong; the funeral interpretation is not something the design could support either way.

Caveats

  • Avoidance and alarm are what was measured. Nothing in the design could detect an emotional state.
  • The dead crow was always presented by a human, so the corpse and the person are confounded.

Still unanswered

  • Do crows respond differently to a dead crow they knew individually?

Last reviewed 2026-08-09

The evidence (1 study)

How we know

The experiments behind the claims on this page, explained from scratch.

How we know

The caveman mask experiment

Can a wild crow tell one human face from another, and remember which one is dangerous?

Researchers put on a rubber mask — memorably, a caveman — and used it while catching and ringing wild American crows. The birds were released unharmed, but they had now seen that face doing something unpleasant. Afterwards, volunteers walked fixed routes through the same areas wearing either the caveman mask or a completely different, neutral one, and noted how the crows behaved. Crucially, the masks were passed between different people, of different heights and builds, and worn in different clothes.

What happened

Crows scolded, dived at and mobbed the caveman mask far more than the neutral one. They kept doing it for years, and birds that had never been caught joined in. In later work the reaction spread through the population and to birds too young to have been present.

What it shows

Wild crows can identify a particular human face, attach a meaning to it, and hold onto that association for years. Because the face moved between bodies, the face itself must be what they were tracking.

What it does not show

It does not show that crows see faces the way people do, or that they could pick a face out of a crowd with no threat attached. It measures alarm behaviour, so it tells us the birds discriminated — not which features they used to do it.

The controls — what makes this evidence rather than a story
  • A neutral mask, used at the same sites by the same people, gave a baseline for ordinary human presence.
  • Swapping masks between wearers meant the birds could not be responding to a particular person's body, gait or clothing.
  • Routes and timings were fixed in advance, so the birds' reaction — not the walker's behaviour — was what varied.

From Lasting recognition of threatening people by wild American crows

How the picture changed

  1. 1996

    First observation

    Tool manufacture found in the wild

    New Caledonian crows are documented making two standardised tool types, opening the field of corvid tool research.

    Manufacture and use of hook-tools by New Caledonian crows

  2. 2002

    Landmark experiment

    A captive crow bends wire into a hook

    Betty's wire-bending becomes the headline example of animal insight for the next two decades.

    Shaping of hooks in New Caledonian crows

  3. 2005

    Landmark experiment

    Ravens are shown to track what a competitor has seen

    Caching experiments suggest corvids adjust behaviour to what another bird was in a position to observe — opening two decades of argument about whether that requires representing another mind.

    Ravens, Corvus corax, differentiate between knowledgeable and ignorant competitors

  4. 2010

    Landmark experiment

    Wild crows recognise individual human faces

    The caveman mask experiment shows American crows tracking a face independently of who wears it.

    Lasting recognition of threatening people by wild American crows

  5. 2012

    Modern discovery

    Brain imaging finds the circuitry

    Threatening and caring faces engage different regions of the crow brain.

    Brain imaging reveals neuronal circuitry underlying the crow's perception of human faces

  6. 2015

    Modern discovery

    Number-selective neurons recorded

    Individual neurons in carrion crows are shown to be tuned to specific quantities.

    Neurons selective to the number of visual items in the corvid songbird endbrain

  7. 2016

    First observation

    A second tool-using crow species

    Almost the entire surviving Hawaiian crow population is tested, and most use tools.

    Discovery of species-wide tool use in the Hawaiian crow

  8. 2016

    Reinterpretation

    The wire-bending result is reframed

    Wild New Caledonian crows are shown to bend plant material routinely, so hook-making is species-typical rather than spontaneous invention.

    Changes how the 2002 result reads

    Betty was not deleted from the record and not quietly rewritten. What changed is the explanation: bending is species-typical behaviour, so the 2002 result demonstrates a capacity rather than an invention. The original finding stands; the word "insight" does not.

    Tool bending in New Caledonian crows

  9. 2016

    Replication

    The knowledge-attribution question is sharpened, not settled

    Ravens adjust caching when a peephole suggests they could be watched, with no bird visible. It narrows the behaviour-reading alternative without eliminating it.

    Changes how the 2005 result reads

    The 2005 result could be explained by ravens reacting to a visible competitor’s behaviour. Removing the competitor and leaving only a peephole tests the same claim without that escape route — which strengthens the interpretation rather than replacing it.

    Ravens attribute visual access to unseen competitors

  10. 2020

    Modern discovery

    A neural correlate of perception

    Crow neural activity is shown to track what the bird reports seeing rather than what was presented — widely over-reported as proof of consciousness.

    A neural correlate of sensory consciousness in a corvid bird

Worth knowing

A crow that has never been near a trap can learn to scold a particular human face by watching other crows do it. The knowledge outlives the birds that acquired it first-hand.

Based on Crows tell each other about dangerous people

Bird brains have no cortex. Corvids arrived at comparable abilities using a differently organised forebrain, which is why they matter so much to researchers studying how intelligence evolves.

Almost the entire Hawaiian crow species — 104 birds — was tested for tool use in a single study. That was possible only because the species is extinct in the wild.

Based on The Hawaiian crow uses tools too — and almost the whole species was tested

Crows in your garden

Animal welfare

Before you feed crows

Crows are quick to associate a person and a place with food, and the association is durable. Regular feeding can concentrate birds, increase conflict with neighbours and, in some places, breach local rules on feeding wildlife. Where crows are protected, disturbing nests or removing birds may also be an offence. Rules differ enormously between countries and even between local authorities.

Where this applies: Varies by country and locality — check your national wildlife authority

When to get help: For an injured or grounded bird, contact a licensed wildlife rescue or veterinary practice rather than attempting to care for it yourself.

If a crow is dive-bombing you, it is almost certainly defending a nest or a recently fledged chick nearby. It usually stops within a couple of weeks. Taking a different route for a fortnight solves it more reliably than anything else.

What crows eat — and what you should not give them

Crows are omnivores and generalists, which is most of the explanation for their success alongside people. Wild diets across the genus include insects and other invertebrates, grain and seed, fruit, carrion, small vertebrates, eggs and nestlings, and — in towns — a great deal of human food waste. The proportions shift with season and location rather than being fixed.

The eggs-and-nestlings part is worth stating plainly, because it is the usual reason people dislike crows. Predation on other birds’ nests is real. Whether it depresses songbird populations is a separate question, and the evidence for that is much weaker than the folk belief — nest predation is one pressure among many, and studies removing corvids have generally not produced the songbird recoveries predicted.

Animal welfare

If you do feed crows, what is safe

The safest offerings are unsalted nuts in shell, whole peanuts, eggs, and small amounts of pet food. Avoid anything salted — birds handle salt poorly — and avoid bread, which fills a bird without feeding it. Chocolate, caffeine, avocado, onion and alcohol are toxic to birds and must never be offered. Put food out in small quantities, in the open where the bird can watch for danger, and not in the same place every day: predictable feeding sites attract rats, concentrate disease, and can create dependence.

Where this applies: Feeding wild birds is restricted or prohibited in some localities. Check your national or local wildlife authority before establishing a regular feeding routine.

When to get help: For a bird that appears sick — fluffed, lethargic, or with growths around the beak — stop feeding, clean the area, and contact a wildlife rescue. Concentrated feeding spreads avian disease efficiently.

NatureHQ’s position on feeding is not that it is wrong, but that it is an intervention rather than an observation. A crow that learns to associate a person with food is a crow whose behaviour you have changed, and the face-recognition research is a good reason to expect that change to be durable and to spread to birds you never fed.

Related

A crow year

Crows pair for the long term — often for life, though "divorce" and replacement after a death both occur — and hold a territory year-round in many populations. A pair typically raises one brood a year of four or five young, and in several species the young from previous years stay on and help feed the next brood rather than dispersing immediately.

  • Nest: a bulky stick structure high in a tree, built each year rather than reused.
  • Eggs incubated around 18 days; young leave the nest at four to five weeks and are fed for months afterwards.
  • Cooperative breeding: helpers at the nest are usually the pair’s own offspring from earlier years.
  • Wild lifespan commonly seven to eight years, with ringed individuals recorded beyond fifteen.
  • Winter roosts of hundreds or thousands form outside the breeding season, then break up again in spring.

The long fledging period is worth connecting to everything else on this page. A young crow spends months being fed while it is capable of flight, which is time available for learning — and the face-recognition and tool results all depend on a bird that learns slowly and remembers for years.

Feeding young is done by regurgitation and by carrying, and the food is whatever the parents are eating, which is close to everything. That generality is why crows do well alongside people and why the winter roosts can be so large: a bird with no dietary specialism can concentrate wherever food happens to be.

Most crows do not migrate, but some northern populations move south for winter while their southern neighbours stay put — so "do crows migrate?" depends on which population is being asked about.

Words used here
Cooperative breeding
Young from earlier broods staying to help raise the next one instead of leaving to breed themselves.
Fledging
Leaving the nest. In crows it is followed by months of continued feeding, which is unusually long.

Cawing, mobbing, and the business with shiny things

The calling is not one signal. Crows produce a large repertoire — the familiar caw in several distinguishable patterns, rattles, clicks and soft sounds used close to a partner — and the same bird uses different structures for assembling a group, for alarm, and for contact. Playback work shows crows responding differently to the calls of familiar and unfamiliar individuals.

Mobbing is the behaviour most people notice: several crows harassing a hawk, owl or fox, diving at it and calling continuously. It is defence rather than aggression for its own sake, it recruits other crows, and — as the mask experiments showed — the memory of what was mobbed can persist for years and pass to birds that were not there.

The shiny-object story deserves handling carefully, because it is mostly folklore. Corvids are neophilic — interested in novelty — and will pick up and examine unfamiliar objects, sometimes caching them. Experimental tests of a preference for shiny things specifically have found the attraction weak or absent, and captive birds often avoid novel objects rather than approaching them. What is real is that crows investigate, and that a bird investigating a bright object is more memorable than one investigating a leaf.

Gifts to people are the other much-repeated story. There are well-documented cases of crows leaving objects near people who feed them regularly. Whether this is reciprocity, cache displacement or coincidence is not established, and the honest position is that the observations are real and the interpretation is not.

Crows can be taught to produce sounds resembling human speech in captivity, as ravens can. It is imitation rather than language, and it is not something wild crows do.

Words used here
Mobbing
Group harassment of a predator by prey animals. Noisy, coordinated, and effective at driving it away.
Neophilia
Attraction to novelty. Common in corvids, and frequently mistaken for a preference for shiny objects.

What is still unknown

  • Which facial features do crows actually use to tell people apart?

    Why it matters: The mask experiments show crows discriminate faces, but not what they are looking at. Without that, we cannot say how similar their face processing is to ours.

    What would settle it: Systematic testing with faces varied one feature at a time.

  • Why has habitual tool use evolved in island crows in particular?

    Why it matters: Both tool-using Corvus species are island birds. If isolation, or the absence of woodpeckers, is the driver, that tells us something general about how tool use arises.

  • Do the number-selective neurons in trained crows do anything in an untrained bird?

    Why it matters: The recordings come from birds trained for months on a numerical task. Whether wild crows use the same machinery in foraging decisions is unknown.

Claims about this, checked

Things people have heard, and what the evidence actually supports.

The research behind this page

17 studies, newest first. Each one has a page explaining what it found and what it could not show.

You may also find this interesting

What this page is still missing

NatureHQ publishes its own gaps. This record is at 92% completeness against what we would call a finished subject.

  • 6 high-priority search intent(s) not yet covered
  • Diet, breeding and lifecycle sections are not yet written — this record leads with cognition because that is where the search demand and the research depth are.
  • European carrion and hooded crow research is under-represented relative to North American work.
  • Study identifiers have not yet been resolved against Crossref.

Last reviewed 2026-08-09 · 11 claims · 76 search questions answered on this page