From dogs recognising other dogs on screen to cats tracking moving images and parrots engaging through live video calls, what experiments reveal about how pets perceive digital content.
The Appearance of Understanding
Looking at a screen, tracking a moving figure or reacting to a sound can resemble understanding. The studies reviewed here, on dogs, cats and parrots, do not directly show that any of these animals understands an onscreen image as a representation of something in the world.
In a 2021 experiment, researchers showed 32 pet dogs videos of dogs, cows or horses, paired with either a dog bark or a frog’s croak. After the dog-video and bark combination, the dogs looked less towards the centre of the display area and more towards the entrance side, where the onscreen animal had entered, than after the other combinations. The researchers interpreted this pattern as recognition of another dog in the video. It is evidence from a controlled audiovisual task, and it does not establish that the dogs believed the animal was physically present.
These behaviours are not equivalent. Attention shows that a stimulus draws an animal in, and a cross-modal experiment can test whether an image and a sound are treated as belonging together. Representation poses a harder question: what does the animal expect the image to refer to after it disappears or changes? The studies below test parts of that problem, not one general capacity called screen comprehension.
The Canine Evidence: From Recognition to Reported Behaviour
The laboratory result is narrower than the behaviour owners report at home. Experiments can test whether a dog responds differently to particular combinations of image and sound, while surveys describe what people notice during everyday viewing.
On the experimental side, a 2023 experiment involving 64 dogs tested whether the recognition reported in 2021 extended to cats. Each dog heard a dog bark or cat meow, followed by a video of a dog or cat. The researchers found no expected surprise response to the mismatched combinations: a cat vocalisation followed by a dog video, or a dog vocalisation followed by a cat video. The experiment also provided only limited evidence of recognition of another dog under its own conditions. Read together with the 2021 result, this suggests that recognition in such tasks depends on the particular images, sounds and sequence presented.
Surveys can only report what owners notice. In an online survey, 1,077 of 1,246 dog owners reported that their dogs watched screen content. Among dogs reported to watch screens, owners commonly described approaching the screen (78%) and vocalising (76%). Most respondents, 89%, lived in the United States. These are owner-reported observations from a selected sample, not direct measurements of comprehension or a prevalence estimate for dogs generally.
In a 2025 study published in Scientific Reports, researchers analysed owner responses for 453 dogs whose owners reported that they watched television. Higher owner-rated excitement, labelled PANAS Excitement in the paper’s results, was associated with higher scores on the study’s “Following” measure. Higher PANAS Negative Activation scores, which the authors define as reflecting fearful reactions to stimuli and changes in the environment, were associated with higher scores for non-animal content, including humans and inanimate objects. These are associations in owner-reported ratings, not evidence that temperament caused the behaviour or that dogs interpreted the images as real objects. Owners were recruited through social media, and dogs reported not to watch television were excluded, so the findings describe a selected group of viewers.
Cats and Moving Images
The cat studies ask something more basic than the dog experiments: what holds a cat’s gaze. The evidence primarily concerns attention to screen-based movement. A 2008 study in Applied Animal Behaviour Science compared a control condition with no visual stimulation against four experimental conditions for 125 shelter cats: a blank television screen and televised images depicting humans, inanimate movement or animate movement. The cats were observed for three hours daily over three days. Across the observation period, the cats spent 6.1% of their time looking at the television monitors. Cats exposed to moving animals and inanimate movement looked at the screens more than those shown humans or a blank screen. Their attention declined significantly across each three-hour presentation, a pattern the researchers interpreted as suggesting habituation.
A 2022 controlled experiment published in Animal Cognition showed domestic cats and dogs abstract geometric shapes moving either as if one chased the other or independently. Cats first looked longer at the independent movement and later shifted towards the chasing pattern, so they discriminated between the two. The authors cautioned that they could not conclude the cats perceived the chasing as animate, which leaves open whether a cat would treat a bird or mouse on television as a living creature.
A 2025 analysis of online videos found little evidence that cats treated mirrors or augmented-reality reflections as images of themselves. It did not test television viewing.
These studies show that movement attracts cats, not how cats interpret a natural animal image. The parrot study asks whether a screen can do more when another animal answers back.
Parrots: When Screens Become Social
A 2024 study by Hirskyj-Douglas, Cunha and Kleinberger, Call of the Wild Web, presented at the CHI Conference on Human Factors in Computing Systems, examined nine parrots using a bell-and-tablet system. A parrot could signal that it wanted a call and choose another bird on the tablet, and its caregiver then connected the call. Over six months, the birds initiated 65 of 108 possible calls in the live phase and 40 in the later prerecorded phase. Total call time, excluding training, was 561 minutes live and 142 minutes prerecorded. Call initiation was significantly higher in the live phase (p = 0.00424), as was engagement duration (p = 0.0278). But every parrot experienced live calls first, and only live calls allowed another bird to respond, so the study cannot isolate responsiveness as the cause. It measured use of the system and engagement, not what the parrots believed the screen represented.
Biological and Technical Barriers to Understanding
A dog’s screen experience may differ from a human’s, although the implications for ordinary television remain incompletely tested. In behavioural discrimination experiments with three dogs, Neitz and colleagues found evidence consistent with dichromatic colour vision (Neitz et al. (1989), Color vision in the dog). Typical human colour vision relies on three classes of cone photopigment, so a programme designed to look vivid to a person may present different contrasts to a dog. A 2026 review of touchscreen-based dog cognition research recommends treating colour composition and contrast as variables when designing experiments. That guidance concerns stimulus design, and touchscreen tasks say little directly about ordinary television.
Motion creates a second design problem. Screens present movement as rapid sequences of images, so researchers must check that the display is perceptible as intended. Coile, Pollitz and Smith measured critical flicker fusion, the rate at which a flickering light looks steady, behaviourally in four beagles (1989), and the 2026 review identifies temporal resolution as a consideration in canine experiments. Neither source shows that television looks unnatural to every dog. The point is narrower: before reading a dog’s gaze as recognition or understanding, researchers need to confirm that the relevant visual cues are presented in a form the animal can perceive.
Even a well-designed display lacks the smell and tactile feedback of a physical encounter, so an onscreen animal supplies only part of the information available in a real one.
Welfare, Enrichment, and Future Directions
The welfare question requires different evidence from a viewing test. The shelter-cat study found that attention declined within each presentation and recorded behavioural differences between conditions, but it did not establish durable welfare benefits. The dog surveys relied on owner reports, and the parrot experiment measured calling and engagement in a structured video-call system. None shows that leaving a television on reduces loneliness or separation anxiety, or that screens can replace companionship, exploration or play. A more decisive study would measure both positive and adverse behaviour over time, with and without screen access, rather than treating attention as a proxy for wellbeing.
The three species were tested on different things, so their results do not add up to one finding. Dogs matched a dog’s bark with a video of a dog in one controlled test, cats changed their attention between two geometric motion patterns, and parrots used a live-call system more than prerecorded playback. The next scientific question is whether a response to a screen changes what an animal expects to happen beyond it.
Researchers can test that by varying whether an image disappears, reappears somewhere else or is followed by the same event in real life. A dog that looked for an onscreen animal where it should reappear would show something that attention alone cannot.
