The case against the dog in the bed was built with accelerometers, and it was built on small samples. In 2015, Mayo Clinic researchers put motion sensors on 40 healthy adults and their dogs for seven nights, with the people wearing a clinical actigraph and the dogs wearing consumer FitBark trackers. When the study appeared in Mayo Clinic Proceedings in 2017, the owners averaged about 81 percent sleep efficiency, the share of time in bed actually spent asleep, and that figure was significantly lower when the dog slept on the bed than when it slept in the bedroom but off it. Lois Krahn, the Mayo sleep specialist on the paper, drew the narrower lesson that having pets in the bedroom “per se does not affect their sleep in a negative way.” The bed itself stayed under suspicion.
Three years later an actigraphy study of 12 women in Animals pushed the suspicion further. Dog movement roughly tripled the odds that a woman lying still would start moving (an odds ratio of 2.98), and the authors read the influence as running from dog to human. The women themselves barely noticed. They reported being woken by the dog on only 22 of 124 sleep periods, which the researchers took as a sign that owners were unaware of what their dogs were doing to their nights. Meanwhile Christy Hoffman’s 2018 survey of nearly 1,000 American women found that they rated dogs as less disruptive bed partners than cats or human partners, and higher on comfort and security. So the field settled into an awkward split, in which people said the dog helped and the sensors said the dog moved first while the person followed.
A University of Toronto study published October 7 in PLOS ONE asks the question in its title, “Who is disrupting whom?”, and with three times as many dog-human pairs as the 12-woman study, it mostly gets the opposite answer.
Ten pairs to four
Between November 2 and December 9, 2021, anthropologist David Samson’s group monitored 25 adults in southern Ontario (21 women and four men, average age 38.9, per News-Medical’s summary) along with 38 dogs, every one a husky or malamute between two and 12 years old. Restricting the breeds held activity differences between breeds constant, and it fit a larger project comparing dogs with wolves; huskies and malamutes, the paper notes, descend from a Siberian dog lineage roughly 9,500 years old. People wore a MotionWatch 8 on the wrist, the dogs wore the same sensor on a collar, and both logged movement once a minute, which added up to 777 nights of human sleep and 1,093 dog-nights of activity. The authors frame the whole arrangement, two species sharing one bed, as “an extreme expression of multispecies entanglement.”
The method that answers the title question is simpler than it sounds. Take the person’s minute-by-minute movement trace and the dog’s trace from the same night, slide one against the other a minute at a time, and find the offset where the two match best. If the match peaks with the human trace running ahead, the person tends to move first; if the dog’s trace leads, the dog does.
Of 36 analyzed pairs, 22 of the 36 showed a clear peak. In 10 of those 22 the human moved first, in 8 the two peaked within the same minute, and in only 4 did the dog lead. Human-first was the largest group, not a majority, but it outnumbered dog-first by two and a half to one, which is close to the reverse of what the 2020 study described.
| Group | Value (pairs) |
|---|---|
| Human first | 10 |
| Same minute | 8 |
| Dog first | 4 |
The breakdown by sleeping arrangement sharpens it. All four dog-first pairs came from households where the dog never slept on the bed. Among the 11 synchronized pairs where the dog shared the bed always or sometimes, not one showed the dog leading; three always-bedsharing pairs split two human-first and one same-minute. Those are single-digit counts and should be held lightly. They also measure movement, not waking. A wrist twitching a minute before a collar does not prove anyone woke up, or that one sleeper roused the other, and the authors say the timing peaks describe correlation, not cause. Across whole nights, more dog movement went with more human movement, a small but statistically significant link that, in the authors’ words, “could not determine whether either individual’s movement caused the other’s activity.”
The dog on the bed, measured again
On the practical question, the Toronto numbers break with Mayo. The owners slept an average of 6.23 hours a night at 79.31 percent efficiency, in the same range as Mayo’s group, and neither figure was significantly associated with whether a dog slept on the bed, whether other dogs lived in the house, or whether a human partner shared the bed. In this sample, 16 percent of owners always shared a bed with their dogs, 40 percent sometimes did, and 44 percent never did. A study of 25 people can miss a modest effect, so the finding is that no bedsharing penalty showed up, not that one has been ruled out. But the penalty Mayo reported in 2017 did not reappear when someone looked again with the same class of instrument.
What owners reported was messier. According to News-Medical, 48 percent said their dogs interrupted their sleep, while 72 percent said the dogs made them feel safer. Owners of more than one dog scored worse on the Pittsburgh Sleep Quality Index, a standard questionnaire, than single-dog owners, though the authors take that apart themselves: the full model with household factors explained sleep quality no better than a model with none of them (p = 0.116). People who wake in the night go looking for a cause, and a 60-pound husky shifting on the duvet is an easy one to find. In this study the sensor on the person’s wrist usually got there first, or at the same time.
The limits are the ones you would expect from a small observational study, and the authors list them. Every dog was a husky or malamute, so a house with a restless terrier may look different. Monitoring ran under the COVID-19 restrictions of late 2021, when routines were anything but normal. There was no comparison group without dogs, and nobody watched what actually happened in the bedrooms. The MotionWatch has not been validated for telling sleep from wakefulness in dogs, which is why the team measured canine activity rather than canine sleep, and the human tri-axial mode used here had not been checked against polysomnography, the overnight lab test. The earlier studies share most of those weaknesses, and the cautionary reading of them (that the dog on the bed is a measurable drag on your sleep) always rested on a 40-person sample and a 12-woman one.
What comes next sits with the same Toronto group. Their husky nights were recorded as one half of a dog-versus-wolf comparison, and when Samson’s lab reports the wolf side, it will offer the first baseline for nighttime activity in a canid that has never slept on anyone’s bed. Until a sleep lab puts owners and dogs under polysomnography together, the 10-to-4 count from Ontario is the most recent word on who moves first, and it does not favor blaming the dog.
Sources
- Li et al., PLOS ONE (2026)
- News-Medical: “Who moves first at night, you or your dog?” (October 12, 2026)
- Hoffman, Browne & Smith, “Human-Animal Co-Sleeping,” Animals (2020)
- News-Medical: Mayo Clinic Proceedings study on dogs and human sleep in the home (2017)
- Canisius University: Hoffman survey of women’s sleep and pet bedsharing, Anthrozoös (2018)
- FitBark: Mayo Clinic studies using FitBark devices and data