Almost every page that answers this question answers it about a bird, and the answer is almost always yes. This one is organized the other way around, by experiment, because that is where the difference between the species actually lives.
Five designs have produced real results. Below is what each one did, which birds it was run on, and what it is fair to conclude. Where a bird is missing, it is missing because the design has not been run on it.
The short answer
- Birds shown to tell one person from another
- American Crow, Common Raven, Northern Mockingbird, and the feral pigeon. Each by a different method, each with a control
- Birds shown to read what a person is doing
- The herring gull, in two experiments. That is a real and separate finding
- Birds with a famous result that answers a different question
- The European magpie and the Graylag Goose. Both are good studies. Neither is about you
- The honest gap
- For most backyard birds, none of these designs has been run. That is a fact about the literature, not a verdict on the bird
What counts as an answer here
A bird that comes to the window when you pick up the seed scoop has learned something. It has not necessarily learned you. It may have learned the scoop, the window, the hour, or the shape of a person on that path at that time.
Every result below has the same shape. Two things are made as similar as possible, one is attached to an experience, and the bird is later given both. If the bird splits them, it is tracking the thing that differed. That is the whole trick, and it is why a study with a control beats a hundred anecdotes.
Design one, the mask: crows and ravens
The most famous version. Wear an arbitrary rubber mask while doing something the birds dislike, then come back later wearing that mask and a different one. The mask is the point. It removes clothing, height, gait and voice from the equation.
- The crow work
- John Marzluff at the University of Washington. In 2006 the team netted and banded crows around campus and Seattle while wearing rubber masks, then returned in various masks later
- What the crows did
- They scolded and mobbed only the banding mask, and it did not matter who was underneath it or what they wore
- How long it has lasted
- Nearly seven years on, the birds were still harassing that mask, while seeing it only twice a year for a few hours
- The part that is stranger
- The share of birds joining in roughly doubled over that period, and many of the scolders were young birds hatched after the banding. They never saw it happen
- The brain scan version
- A 2012 study put 12 adult male crows in front of one mask or the other for about fifteen minutes, then imaged them. The two masks lit up differently
The full account is in the crow article. The reason it anchors this whole subject is that no other result has run for seven years.
The raven version compressed the same idea and got a sharper number. At the Haidlhof station in Austria, Blum, Fitch and Bugnyar showed 16 captive ravens two masks. One mask then appeared held by a person carrying a dead raven, four times. The other appeared empty handed. The test phase ran from November 2011 to October 2015, 28 trials, no dead bird ever shown again.
- What was measured
- Scolding duration, meaning how much of each presentation the bird spent giving alarm calls
- How fast they learned
- The authors report discrimination often on the first trial
- How long it held
- Four years
- The caveat the study carries
- Captive birds, and rubber masks rather than faces
Detail and figures in the raven article.
Design two, the stranger on day five: mockingbirds
This one needs no mask, and it is the cleanest study in the set for a backyard bird.
Levey and colleagues at the University of Florida used 24 mockingbird nests and 10 student volunteers over five days. On the first four days a volunteer walked up to a nest, gently touched the edge of it through the foliage, and left. On the fifth day a volunteer the birds had never seen did exactly the same thing.
- What was measured
- How soon the bird flushed, how many alarm calls it gave, and whether it attacked
- Across the repeat visits
- The response escalated. Birds left the nest sooner and called more with each visit, and some began diving at the intruder head
- On day five, new person, same act
- The birds barely reacted. Fewer calls, far less attacking
- Why that fifth day is the study
- It rules out the boring explanation. If the birds were just getting jumpier about nest visits in general, the new volunteer would have taken the worst of it
- The exposure required
- The release describes roughly sixty seconds of total contact across brief encounters
Published in the Proceedings of the National Academy of Sciences in 2009. More in the mockingbird article.
Design three, the coat swap and the photographs: pigeons
Pigeons have been done twice, once in a park and once in a lab, and the two halves fit together unusually well.
- The park
- Dalila Bovet and colleagues at the University of Paris Ouest Nanterre La Defense worked in a city center park with feral pigeons that had never been caught or handled
- The setup
- Two women of similar build and skin tone, in different colored lab coats. One fed the birds, one chased them off
- The swap
- The two later exchanged coats. The pigeons kept avoiding the same person
- What that isolates
- Bovet put it plainly. The birds used the relevant features of the individual, probably facial, rather than the coats covering most of the body
- The lab half
- Ed Wasserman and Fabian Soto at the University of Iowa trained pigeons on photographs of human faces that varied by identity and by expression
- The result
- Pigeons showed the same asymmetry humans show. Easier to ignore the emotion when judging identity than to ignore identity when judging emotion
Both are in the pigeon article. The coat swap is the part worth borrowing. It is the cheapest control in the whole field.
What gulls read instead
The herring gull work out of the University of Exeter is often filed under recognition. It is not, and the distinction is useful.
- Experiment one, 2019
- Approach a gull with food while looking at it, or while looking away. Gulls took on average 21 seconds longer to approach when they were being watched
- Experiment two, 2020
- Offer two identical wrapped items, one of which a person had held for twenty seconds in view of the bird
- The split
- 19 of 24 gulls, which is 79 percent, went for the handled one
- What that shows
- Gulls are reading human behavior in real time, closely, and adjusting
- What it does not show
- That a gull can tell you from the person next to you. Nobody has run a mask, a swap or a matched pair on a gull
Both experiments and the practical version are in the gull article.
Two famous results that answer a different question
- The magpie mirror
- Prior, Schwarz and Guentuerkuen put a colored mark on the throat of five hand raised European magpies and gave them a mirror, with a black sham mark and a nonreflective plate as controls. Two of the five showed mark directed behavior
- Why it is not this question
- That is self recognition. It is a serious result and it is about the bird and the mirror, not the bird and you. The detail is in the magpie article
- The goose faces
- Kleindorfer and colleagues identified individual Graylag Geese from bill shape alone at 98.4 percent within a year and 97.1 percent across a year, then put life sized photos in front of 111 free flying birds
- What the geese did
- They behaved differently in front of a photo of their own partner than in front of a flock mate or a blank board. The recognition being measured is goose to goose, and that is covered in the goose article
Hummingbirds land in the same category for a different reason. Their memory is genuinely remarkable and it is aimed at flowers, down to which one has already been emptied and roughly how long it takes to refill. None of that is a memory for a person, which is the point of the hummingbird article.
The scoreboard
- Strongest evidence
- Crow and raven, because the mask removes almost everything except the face, and because the follow up ran for years rather than weeks
- Cleanest backyard evidence
- Mockingbird, because the fifth day control is simple and the birds were wild and unhandled
- Best isolation of the cue
- Pigeon, because the coat swap took clothing off the table in one move
- Real but different
- Gull, magpie, goose and hummingbird
- Everything else
- Untested by these designs as far as we could find
What this means at a bird bath
Very little of this changes what you should do, and that is worth saying instead of pretending otherwise. None of the five designs found a bird that behaves better toward a person it knows. What they found is birds that behave worse toward a person who did something bad. The crows, the ravens and the mockingbirds were all measuring a grudge.
The practical version is short. Be boring near the water. A bird bath that gets topped up by someone who never lunges, never chases and never reaches into it is a bird bath that stops generating alarm calls. Where you put it matters more than any of this, and that is the placement guide.
- Approach the same way every time. Same path, same speed. You are trying to be a low information event.
- Do the maintenance when the birds are not there. Scrubbing a basin while a bird watches is the closest a normal person gets to running one of these experiments on themselves.
- Do not chase anything, ever. The pigeon study needed one person doing exactly that to produce an effect that outlasted a coat swap.
- Assume the nest area is different. The mockingbird result came from nests. Near one, an ordinary person becomes an intruder without doing anything wrong.
Frequently asked questions
Which bird recognizes humans best
Of the birds that have been tested with a control, the crow has the longest running evidence by a wide margin. Nearly seven years, still going, and spreading to birds that were not there.
Do birds remember people who feed them
The tested results are all about people who did something unpleasant. That is not an accident. A bad experience produces a measurable, obvious behavior, and a good one mostly produces a bird showing up, which is much harder to attribute to you rather than to the food.
My bird is not on the list. Does that mean it cannot
No. It means nobody has run a mask, a swap, a matched pair or a stranger control on it. Absence of a study is not a result, and this article is trying hard not to treat it as one.
Why do so many pages say every bird recognizes people
Because the crow result is spectacular and it travels. It gets carried onto other species by resemblance, which is how a corvid finding ends up attached to a bird in a different family on a different continent.
Is a bird approaching me evidence
On its own, no. It becomes evidence the moment a second person does the identical thing and the bird treats them differently. That comparison is free and you can run it yourself.
What we could not check
- What we read in full
- The University of Washington release and the National Wildlife account of the crow work, the 2020 raven paper, the release carrying the mockingbird study, both pigeon releases, both gull releases, the 2008 magpie paper and the 2024 goose paper
- What we could not open
- Several of the underlying journal articles sit behind checks we do not get past. Where that happened we have leaned on the institutional release and said so
- The number we are not giving
- How many crows were trapped in the original banding. Neither account we could open states it
- What we are not claiming
- That the five designs cover the field. They cover what we could read and verify
References
- Reference: Crows react to threats in human-like way | University of Washington news, on the mask work and the 2012 imaging study
- Reference: Crows Can Distinguish Faces In a Crowd | National Wildlife, National Wildlife Federation, on the seven year follow up
- Reference: Rapid Learning and Long-Term Memory for Dangerous Humans in Ravens (Corvus corax) | Blum, Fitch and Bugnyar, Frontiers in Psychology, 2020
- Reference: Mockingbirds, no bird brains, can recognize a face in a crowd | ScienceDaily, carrying the University of Florida release on the PNAS study, 2009
- Reference: Pigeons never forget a face | ScienceDaily, on the Paris park study led by Dalila Bovet, 2011
- Reference: Some features of human face perception are not uniquely human, pigeon study shows | ScienceDaily, on the University of Iowa study by Wasserman and Soto, 2011
- Reference: Staring at seagulls could save your chips | ScienceDaily, carrying the University of Exeter release on the gaze study, 2019
- Reference: Seagulls favor food humans have handled | ScienceDaily, carrying the University of Exeter release on the handling study, 2020
- Reference: Mirror-Induced Behavior in the Magpie (Pica pica): Evidence of Self-Recognition | Prior, Schwarz and Guentuerkuen, PLOS Biology, 2008
- Reference: Cues to individuality in Greylag Goose faces: algorithmic discrimination and behavioral field tests | Kleindorfer and colleagues, Journal of Ornithology, 2024