A parrot ringing a bell and tapping another parrot’s portrait resembles a miniature version of human social media. The experiment asked a narrower question: could pet parrots learn a two-step request for a live call, and would they use it without food rewards?
The answer was encouraging, but more qualified than the viral retelling. Eighteen parrots entered the study and 15 continued into the open-calling phase. Those 15 initiated 147 calls. The work showed that carefully trained birds would voluntarily engage with a video-call system. It did not prove that video calls cure loneliness or can replace direct contact with other birds.
This is one exploratory study, not settled consensus about parrot welfare.
The experiment put a request before every call
Rébecca Kleinberger, Jennifer Cunha, Megha Vemuri and Ilyena Hirskyj-Douglas described the three-month project in the 2023 CHI Conference on Human Factors in Computing Systems proceedings. The peer-reviewed paper sits at the intersection of animal behavior and human-computer interaction: the goal was to evaluate an animal-controlled interface, not to run a clinical welfare trial.
The birds lived in family homes and represented several parrot species. Researchers recruited them through social media and the Parrot Kindergarten training network. Each had to be older than one, have no known behavioral problems and be comfortable with screens and objects. Caregivers needed animal-training experience.
During a two-week meet-and-greet phase, caregivers taught each bird an association: ring a dedicated bell, select a photograph of a bird and receive a video call with that bird. Food rewards helped establish the bell-and-picture sequence but were not given during the calls or the later open-call phase.
The second step was important. A bell ring alone might have been accidental or directed at the caregiver. Selecting a pictured bird provided corroboration that the parrot wanted the next event. Most selections used a tablet, while printed pictures or pointing could accommodate birds that were uncomfortable with direct screen contact.
The process was not an autonomous bird version of FaceTime. Human caregivers presented the choices, placed the calls through Facebook Messenger and ended a call if either bird disengaged. The Northeastern University account stresses that experienced handlers introduced the equipment slowly and watched for fear or aggression.
Eighteen started, but 15 made the 147 calls
Three of the 18 birds stayed far from the screen and did not engage despite extra support. The researchers excluded them after the first phase to avoid repeated frustration or distress. That attrition is not a footnote: the open-call results came from 15 parrots for whom the setup had already proved tolerable.
The continuing birds were reorganized into three trios and three pairs. Across seven or eight open-call sessions, each could initiate up to two outgoing calls. Calls were capped at five minutes, and the other household had to be ready.
The birds initiated 147 of 234 possible calls. Every participant that reached this phase made at least one. The range was wide, from three outgoing calls for one parrot to the protocol’s maximum 16 for two others. On average, a bird placed at least one call in 74 percent of its sessions.
Nearly 60 percent of calls ran for the full five minutes. Another 13.6 percent ended when the receiving bird disengaged, 21.7 percent when the caller disengaged and 4.7 percent because of technical problems. By the researchers’ definition, callers remained engaged in 86 percent of requested calls.
These figures support voluntary use and sustained attention under this protocol. They are not direct measurements of happiness, reduced loneliness or improved health. No hormone, stress, feather condition or long-term welfare outcome was measured.
Preferred partners is descriptive, not a formal diagnosis
The parrots were not choosing among all other study birds. A bird in a pair had only one possible partner. A bird in a trio had two choices, with the portrait positions shuffled between sessions to reduce the chance that it was simply tapping its favored side of the display.
Strong dyads nevertheless appeared. Particular birds interacted repeatedly, and birds that placed more calls also tended to receive more. The relationship between outgoing and incoming call counts was statistically significant in this small sample, with an R-squared value of 0.667 and a reported p value of 0.0072.
Caregivers described some pairs singing, dancing, preening and playing at the same time. Two elderly macaws that had rarely encountered other macaws became an especially visible pair. The project’s research summary calls the 147 interactions bird-triggered and reports high motivation for most continuing participants.
Calling these “preferred partners” captures the observed social pattern, but it should not be mistaken for a controlled test of friendship across a large network. Group composition reflected time zones, bird size, species and caregiver availability. In pairs, repeated contact could not demonstrate a choice between partners because no alternative was offered.
Mirroring was observed; learning was reported
During calls, the birds approached the screen, vocalized, touched it, preened, played with toys and sometimes appeared to coordinate singing or movement. In the first phase, two authors independently coded perception-related video and initially agreed on 89 percent of labels before the full team resolved differences.
Fourteen of the 18 initial birds reacted promptly when the other parrot appeared, five followed its position and 13 reacted to its vocalizations. Those behaviors indicate perception of the audiovisual event, but not precisely what each bird understood about remote presence.
The reports of new behavior had a different evidential basis. Caregivers said some birds began foraging, flying, waving or making new sounds after seeing a remote partner. One owner reported that a parrot foraged after a year of unsuccessful human attempts to teach the behavior. Another thought a bird became more willing to fly after observing its call partner.
Those accounts are interesting observations, not independent learning trials. There was no blinded observer, pre-registered test of each skill or control group receiving the same caregiver attention without live calls. Novelty, training, extra human attention and ordinary change over time cannot be separated from social learning in this design.
There is independent evidence that parrots and other birds can acquire behavior socially. ScienceBlog’s account of bin-opening cultures among Sydney cockatoos describes a much larger diffusion study. It supports the plausibility of learning from another bird, but it does not validate every learning report in the video-call project.
A live call may be different from bird video
The original study did not compare live calls with recordings. It therefore could not determine whether reciprocity mattered or whether moving images and parrot sounds would have produced similar engagement. The authors named that missing comparison as a limitation.
A 2024 follow-up study addressed the question. Over six months, participating parrots initiated significantly more live calls and stayed engaged longer with live partners than with recordings. The research program remains small and involves specially trained birds with committed caregivers.
Screens present an imperfect sensory world. Cameras lose birds when they move out of frame, displays are built for human vision and consumer speakers cannot reproduce every feature of parrot sound. The researchers proposed wider camera angles and species-appropriate audio.
A video window cannot reproduce flight, touch, shared space or the full sensory life of a flock. Nor did the study compare calling with in-person social contact. The system is better understood as possible enrichment for birds that cannot safely or practically meet, not as evidence that mediated contact is equivalent to living with compatible parrots.
The design lesson is about agency and restraint
The most original feature was not simply putting parrots on a call. It was giving them a repeatable way to request one, choose when possible and leave. Work on animal-computer interaction and welfare treats that control as potentially valuable while warning that human caregivers still have to protect long-term welfare.
The protocol’s safeguards were part of the result. Calls were short, adults remained nearby and researchers reviewed recordings for missed disengagement. They found eight occasions when caregivers should have ended a call earlier.
Caregiver enthusiasm itself was identified as a risk because an eager human could overlook a bird’s refusal. Sudden sounds or images could also cause fear, and a bird might become frustrated when a remote partner disappeared. Removing three uncomfortable participants showed that the system did not suit every parrot.
A recent systematic review of parrot-welfare research found many gaps in the evidence and wide variation in proposed welfare indicators. The 2023 video-call project belongs inside that uncertain literature. It offers a promising enrichment design, not a tested prescription for pet care.
The parrots did more than passively watch a screen. They requested calls, responded to remote birds and sometimes developed striking routines together. The careful conclusion is also the most useful one: under close supervision, some parrots treated a live video connection as a social opportunity worth asking for.