Can computers tell us what cats do all day?

It’s an often-asked question – what DO our cats do all day? Of course, now that many more of us work from home, we probably have a greater awareness of our cats’ activity patterns and hobbies, but it never hurts to have a little science to back us up!

Activity budgets

It can be challenging to get detailed information about animal’s behavior. Often researchers are trying to assess an animal’s “activity budget” – meaning, determining how much time an animal spends on different, presumably important behaviors. Depending on the species, the behaviors of interest could include feeding, grooming, resting, walking or other active behaviors. For captive animals, we are often interested in abnormal behaviors that might indicate stress such as pacing or repetitive grooming.

To put together activity budgets, researchers have relied on techniques such as live observation and video-recording; live observation can be challenging for logging details or duration of behaviors. Video-recordings allow you to use more care, but require a very time-consuming process of coding. Both techniques can be subject to attention bias (you see what you pay attention to).

Getting data from cats

More recently, studies have used technology to both make data collection easier, and to hopefully also make it less biased. A 2019 study showed how cameras worn by cats could reliably capture many of the adorable things that cats do. Researchers have also placed accelerometers on cats to determine when they are active and inactive, or to compare activity levels of cats receiving a treatment vs placebo, or assess how activity might be related to other behaviors.

A new way to study cat activity

A new study, “How Lazy Are Pet Cats Really? Using Machine Learning and Accelerometry to Get a Glimpse into the Behaviour of Privately Owned Cats in Different Households” has taken technology to a new level, using accelerometers PLUS machine learning! Machine learning is essentially training computers to learn a relationship based on a set of data, then seeing how well it can make predictions based on new data. In a previous study, the same authors trained an algorithm using videos that had been coded by a human with matching accelerometer data. They determined that the computers could do a pretty darn good job of identifying specific feline behaviors.

What did the researchers do?

In the new study, the researchers placed accelerometers (via a collar) on 28 owned cats (17 female, 11 male, all spayed/neutered) living in homes. They were interested of the effects of cat-related factors (sex, age), environmental factors (season, housing, presence of other pets or children in the home, diet) on cats’ activity budgets. The eight behaviors of interest that the computers could identify accurately were: active, lying down, sitting, standing, grooming, eliminating (that’s peeing and pooping), eating, and scratching. Seven days of accelerometer data was collected for each cat in summer and winter.

So what did the computer say the cats did all day?

Cats were only active for about 3% of their time (that’s less than 45 minutes a day). Younger cats were generally more active, as were cats with outdoor access. However, cats with outdoor access were less active in winter compared to summer. Cats who lived indoors only had no differences in activity level based on season.

Cats spent about 5.5% of their time eating, with rural cats spending more time eating than urban kitties. Cats also spent about the same amount of time each day grooming, although grooming decreased with age.

Although going to the bathroom is very important, cats are efficient. They spent less than six minutes a day taking care of business. Cats spent around 17 minutes per day scratching or grooming themselves.

Where cats excelled was at lying and sitting. Cats averaged about 37% of their time on each, for almost three-quarters of their day loafing. Females spent more time lying down than male cats, and single cats spent more time lying than cats in multi-cat households.

Cats averaged almost 13% (around 3 hours) of their day standing – meaning they might be more alert than when lying or sitting, but still not active.

Speaking of activity, overall, there were clear activity patterns among the cats. In the summer, cats were most active between 5 AM and 9 AM, and 8 PM to 11 PM. In winter, cats activity peaked between 6 and 8 AM and 4 and 7 PM, demonstrating some clear effects of light cycles on our cats’ behavior! This pattern was less prominent, although still present, for indoor only cats.

Implications for cat research

I get pretty excited about how technology can help us do MORE and BETTER science, so I am very interested in where these methods can take us next. This research was done in the cats’ natural living environments, and allowed for the efficient collection of data giving us many insights into just what it is your cat is doing all day! Perhaps before long this technology will allow those of us at home to better understand our own cat’s activity patterns and what might be impacting them.

In this study, we can see some clear effects of being indoors only, sex, season, living in a rural or urban environment, and living with other cats. For example, why do single cats spend more time lying down than cats in multi-cat households? Is this because of more play or tension between the cats? Or something else altogether? The effects of other factors were more complicated, but hopefully future studies can explore further how things like diet and living with other children affects cats’ activity budgets!

References

Smit, M., Ikurior, S. J., Corner-Thomas, R. A., Andrews, C. J., Draganova, I., & Thomas, D. G. (2023). The Use of Triaxial Accelerometers and Machine Learning Algorithms for Behavioural Identification in Domestic Cats (Felis catus): A Validation Study. Sensors, 23(16), 7165.

Smit, M., Corner-Thomas, R. A., Draganova, I., Andrews, C. J., & Thomas, D. G. (2024). How Lazy Are Pet Cats Really? Using Machine Learning and Accelerometry to Get a Glimpse into the Behaviour of Privately Owned Cats in Different Households. Sensors, 24(8), 2623.