Do cats understand the laws of physics?

Betteridge’s law of headlines would suggest that the answer is “no”, but the authors of a new paper in Animal Cognition beg to disagree. This short report (reference and free pdf below) tests the idea that cats can identify a rattling sound in a box as denoting an object in the box, and then, when the box is opened upside down, will get flummoxed if something doesn’t drop out of the rattling box. They will also get flummoxed if a toy drops out of a shaken box that didn’t make a rattle. In other words, cats can somehow sense the incongruity between an auditory stimulus (the rattling) and a visual stimulus (the expected object causing the rattle).

So, to be brief, here’s what Saho Takagi and her colleagues did.  They studied 30 domestic cats of both sexes, all tested in —yes—cat cafes: a delightful staple of Japanese culture.  Each cat was given four tests involving a box and a putative object. The box was designed with an electromagnet and all of them held three metal balls, with the magnet activated by pushing a button on the box. When the balls were in the box, shaking it would make a rattling sound—unless the balls were affixed to the electromagnet.

The kicker is that the electromagnet not only did the electromagnet allow a box to contain an object without making a rattling sound when shaken, but also enabled the investigator, when the bottom of the box was opened, to either release the ball to drop on a cushion, or keep the ball inside the box, stuck to the electromagnet.

Each cat was thus exposed to four conditions:

  1. Box rattles, cat hears it, then box opened and balls drop out.
  2. Box rattles, cat hears it, then box opened but NO balls drop out (electromagnet turned on).
  3. Box does not rattle (though it has balls in it), then box opened and balls drop out
  4. Box does not rattle (though it has balls in it), then box opened but NO balls drop out (electromagnet activated the whole time).

These four conditions, imposed in random order on each of the 30 cats, are shown like this: Screen Shot 2016-06-15 at 12.25.21 PM

As the caption above notes, conditions 1 and 4 are congruent with expectations from Cat Physical Law, but conditions 2 and 3 are “incongruous”: the cat either hears an object and perhaps expects it to fall out (but it doesn’t), or doesn’t hear an object and nevertheless sees it fall out. These two conditions should flummox the cat—IF cats can associate an auditory stimulus with the appearance of an object they haven’t yet seen.

The researchers made two predictions. The first one, which is not surprising, is that cats would spend a larger amount of time looking at a box that’s rattling before the object either does or doesn’t drop out. This was verified by the data below (“mean looking time” is the number of frames of the videotape that the cat looked at the box during the five-second shaking period; these frames were scored by observers who were blinded as to the condition). Whether or not an object was going to drop out of the box (“object” or “no object”), cats paid significantly more attention to the box that rattled. No surprise here:

Screen Shot 2016-06-15 at 12.35.11 PM

But then the cats got the chance to be flummoxed: the boxes were opened, and the ball either did or didn’t drop out. After 5 seconds of holding the box upside down, it was placed on the floor for 15 seconds, and the cats were released to look at the apparatus for a 15-second inspection period (they were held by their owners during the shaking phase and five-second post-opening phase). The authors predicted that cats would look at the box longer under the two “incongruent” conditions (2 and 3 above) than under the congruent conditions, because they’d be flummoxed by the lack of a visual stimulus matching the auditory one. And that, in general, is what they found (again, the length of inspection was judged by the number of frames of the videotape during which the cat was looking at the box:

Screen Shot 2016-06-15 at 12.41.15 PM

As the graph shows, cats looked the longest at the box when an object fell out but there was no sound, or when there was a sound but no object fell out, than under the other two conditions taken together. (Whether an object fell out also in general increased their inspection time). The comparisons above are statistically significantly different when connected by a bracket with an asterisk. In particular, the authors’ predicted differences in inspection (between sound/object vs. no sound/object, as well as between sound/no object vs. no sound/no object) were upheld. HOWEVER, there was no difference between inspection time for the sound/object vs the sound/no object condition.

The authors interpret this as the cats looking longer at the apparatus when conditions 2 and 3 obtained: those conditions with physical incongruity. This is what they say:

This study may be viewed as evidence for cats’ having a rudimentary understanding of gravity. We have found no study specifically testing knowledge of this fundamental physical rule in cats. Some nonhuman animals have been shown to respond spontaneously in accordance with gravity (e.g., tamarins: Hood et al. 1999; dogs: Osthaus et al. 2003), which suggests that an innate tendency to react in accordance with the gravity rule may be common among mammals.

Well, forget the “understanding of gravity” part, and look at this as cats showing an association between a sound and the expected appearance of an object. The results support this to some extent, but the lack of difference in inspection time between conditions 1 and 2, in which 2 is incongruous and 1 is not, weaken this conclusion somewhat. (To be fair, one might argue that condition 1, in which a rattling box releases a ball, would really interest cats!) The similar results in studies with tamarins and dogs led the authors to suggest that an “understanding of gravity” may be “common among mammals.”

Finally, I’m pleased to see that the first author of the study owns a cat. Here’s a picture of Saho Takagi and her cat. Who else but a cat lover would even do such a study?


Source. Photographer: Mayu Takagi

h/t: Gethyn


Takagi, S. et al 2016. There’s no ball without noise: cats’ prediction of an object from noise. Animal Cognition, 2016; DOI: 10.1007/s10071-016-1001-6


  1. Ken Kukec
    Posted June 15, 2016 at 1:11 pm | Permalink

    … cats can identify a rattling sound in a box …

    Does that include Schrödinger’s feline death rattle?

    • rickflick
      Posted June 15, 2016 at 1:26 pm | Permalink

      Ya, that should be part of cat phizzicks.

      • gravelinspector-Aidan
        Posted June 15, 2016 at 3:06 pm | Permalink

        If you start to introduce Schrodinger to the system, that’ll (probably) let Heisenberg in though the window. Or the floor. And that’s going to bring the Singularity that much closer.

  2. Kevin
    Posted June 15, 2016 at 1:27 pm | Permalink

    Interesting study, but I think the reactions to the sounds are more complicated. My cats hear, daily, shaking noises in boxes and they think about what’s going to go into their bellies…gravity…not so much.

    An “understanding of gravity” is clear in the behavior of cats if not all mammals. I would content that they would be unable to move without an understanding of the asymmetry built into our terrestrial experience.

    Consider designing a robot that must walk up stairs. It is clear that it must be programmed to understand how to negotiate it’s own trajectory in the precise of gravity. Mammals do this effortlessly, because they have evolved to do it.

    • John Taylor
      Posted June 15, 2016 at 2:32 pm | Permalink

      I think understanding how gravity operates at the surface of the earth, stuff falls down, and understanding gravity in general are two very different things. I doubt that cats will be landing probes on other planets or predicting the reappearance of comets any time soon.

      • rickflick
        Posted June 15, 2016 at 3:09 pm | Permalink

        Cats do seem to have some difficulty extending their physics to unusual domains. Here we see some cats preparing for life in weightlessness. A skill they will need if they are to take to the spacelanes and survive the apocalypse.

        • gravelinspector-Aidan
          Posted June 15, 2016 at 3:15 pm | Permalink

          Seen that video before. It’s entertaining.
          One wonders if any of the cats suffered from freefall … well, they don’t call it the “Vomit Comet” for nothing.
          (The cats don’t seem particularly distressed, though definitely confused. It would be really interesting to see cats in the ISS in it’s latter days. Shortly before it is brought down. Unless someone has got some really good ideas for a cat-lat in space.

      • gravelinspector-Aidan
        Posted June 15, 2016 at 3:12 pm | Permalink

        Joke told – often – by David Levy (of Shoemakeer-Levy-9 fame) :
        Q. why are cats like comets?
        A. Because they both have tails, and do just what they want to.

        • Posted June 16, 2016 at 11:36 am | Permalink

          Also both involve a bit of eccentricity in what they do …

          • gravelinspector-Aidan
            Posted June 17, 2016 at 12:07 am | Permalink

            Both? Levy + cats, Levy + comets, or cats + comets?
            If you’ve ever seen Levy’s design for a fork-mounted telescope (it’s in Manly’s “unusual telescopes” book, and not on the web TTBOMK, so you’ll have to find dead tree version somewhere. It’s a couple of photos before the “18in f/3 shaving mirror”.), then you’d appreciate that he has non-zero eccentricity himself.
            While looking for the picture I wanted, I found this nightmare.

    • gravelinspector-Aidan
      Posted June 15, 2016 at 3:09 pm | Permalink

      An “understanding of gravity” is clear in the behavior of cats if not all mammals.

      OK – counter test. Do cats understand gravity better than dolphins? (Or porpoises – for convenience of experimentation.)
      I’ve no idea of how to design and experiment – that’s what Sherlock would call a 3 pipe problem, and at least one of them would probably need to be opium.

      • Kevin
        Posted June 15, 2016 at 6:03 pm | Permalink

        You can teach a porpoise to jump through a hoop. They can do this with as much grace as Michael Jordan swishing a three pointer. Both have to have some idea of what gravity does and how to anticipate the forces needed. And both may very well have the same ‘mathematical knowledge’ about gravity…probably none. 😊

        • Gregory Kusnick
          Posted June 15, 2016 at 9:35 pm | Permalink

          You can teach cats to jump through hoops too. But training up the “muscle memory” of how hard to push off to reach the target and earn the treat is not the same as having an idea or understanding of what gravity is and how it works.

          It’s well established that even humans don’t solve physics equations when shooting free throws or catching fly balls. They apply muscle memory and simple heuristics; conceptualizing the effects of gravity doesn’t enter into it because that level of thinking is too slow to get the job done in real time.

          • Kevin
            Posted June 16, 2016 at 10:00 am | Permalink

            But one could argue the solution obtained is functionally equivalent to working out the physics. Though, working out the precise angular momentum will do Messe no good to bend his ball toward victory, or Ryan Locate, good grief, I dare say, may not be able to spell fluid mechanics, bless his massive heart.

            • Gregory Kusnick
              Posted June 16, 2016 at 11:45 am | Permalink

              Is it functionally equivalent? Let’s put it to the test with a thought experiment.

              Suppose we build a hoop-jumping robot that works by actually measuring the forces involved, solving the equations of motion, and applying the calculated amount of force to its leg actuators. Let’s suppose that the robot performs as well as our trained hoop-jumping cat.

              Now take them both to the Moon and repeat the contest there. My prediction: the robot will succeed on its first jump, because its methodology embodies a genuine understanding of jumping physics that works in any environment. The cat’s first jump will overshoot the mark, because its Earth-trained reflexes are inappropriate to the Moon’s lower gravity. It will need retraining to regain the jumping accuracy it had on Earth. So the cat’s “understanding” is not equivalent to the robot’s.

        • gravelinspector-Aidan
          Posted June 16, 2016 at 11:49 pm | Permalink

          You can teach a porpoise to jump through a hoop.

          Decent point. We may have to move up to the cuttlefish.

          They can do this with as much grace as Michael Jordan swishing a three pointer.

          на английском пожалуйста?

  3. Merilee
    Posted June 15, 2016 at 1:27 pm | Permalink


  4. mudskipper
    Posted June 15, 2016 at 1:38 pm | Permalink

    Of course, kitties understand gravity. They perform careful experiments in which they push fragile objects off of tables and observe them carefully as they fall to the ground. Being true scientists, they repeat these experiments frequently, using different objects, surfaces, and all hours of the day and night.

  5. Gregory Kusnick
    Posted June 15, 2016 at 1:56 pm | Permalink

    Given that cats’ natural prey includes creatures that routinely defy gravity by flying or clinging to vertical surfaces, I doubt that reasoning about gravity plays much role in their responses here. One way to test that might be to repeat the experiment with the objects emerging horizontally from the box and see if the results differ.

    My cat’s-eye-view reading of these results would be roughly as follows:

    No sound, no object: Nothing interesting here.

    Sound, with or without an object: Non-stealthy creatures in the box. Moderately interesting.

    No sound, followed by an object: Stealthy creatures in the box who don’t want me to know they’re in there. Very interesting.

    But the question I would really like answered is, why do they indicate the presence of sound with a lightning bolt?

  6. Joseph Stans
    Posted June 15, 2016 at 2:12 pm | Permalink

    Shaking the box with balls and then preventing the balls from falling out is cruel.

  7. Michael Hart
    Posted June 15, 2016 at 2:26 pm | Permalink

    Is it interesting that in general the cats spent much less time looking in the later turning-over phase (20-80 frames, Fig. 3) than in the initial shaking phase (70-100 frames, Fig. 2) — compare the scales of the two figures. Only in the ‘object – no sound’ treatment do cats spend about the same amount of time looking in the two phases (about 75 frames, or about 2 seconds?). In the other three treatments, cats spend much less time looking. In the ‘no object – no sound’ treatment the time spent looking goes down from about 80 to about 20 frames.

    This might suggest a slightly different interpretation: that cats are not flummoxed, they’re simply less bored when nothing surprising occurs in the turning-over phase.

  8. Monika
    Posted June 15, 2016 at 2:29 pm | Permalink

    But did it confuse the cat?

  9. Posted June 15, 2016 at 3:54 pm | Permalink

    Early experiments of this kind were conducted by Cleese, Idle et al in the 1970s in Britan. See the documentation of their experiments here: Confuse A Cat

    Posted June 15, 2016 at 4:37 pm | Permalink

    I had a cat whose understanding of gravity was amazing–if I were lying on the couch, he would jump up and land on my chest, calculating the jump so he would achieve zero velocity at the level of my body so it felt like he had just stepped off equivalent height with no impact velocity

    • W.Benson
      Posted June 15, 2016 at 7:45 pm | Permalink

      All life, intelligent or not, evolves to obey the law of minimum expenditure.

  11. Steve Brooks
    Posted June 15, 2016 at 4:53 pm | Permalink

    I think cats recognize that objects can fall. Does this demonstrate an understanding of gravity as a force? I doubt it.

  12. Shawna718
    Posted June 15, 2016 at 9:08 pm | Permalink

    Reblogged this on Wishes On Whiskers and commented:
    But of course!

  13. Posted June 15, 2016 at 9:30 pm | Permalink

    An interesting experiment, but I think the interpretation in terms of the laws of physics is a bit misdirected; it’s more about feline cognition/epistemology. As noted here at WEIT before, cats have what is known as “object permanence”: they continue to ‘believe’ in the existence of objects, even when they can no longer see them. Thus, if you act is if you are throwing a cat toy, but take it out of the cat’s view and then don’t actually throw it, the cat will move to where the trajectory of the toy would have taken it, if you had let it go.

    In this experiment, what I think we are seeing is that the cat will come to apprehend that an object is present, even if its only clue to the object’s existence is the sound that it makes. It can ‘predict’ where the object should appear, based on the noise (just as it can ‘predict’ the trajectory of the unthrown object). When the ‘prediction’ doesn’t hold, it still thinks there should be an object, so it ‘investigates’.

    That a noise can generate a ‘belief’ in the cat’s cognition is interesting, but I think to be expected, given that in nature they live off the capture of small prey. To associate rustling in the leaves with the presence of (potentially delicious) objects, and to realize the mouse is still there trying to get away, even if you lose sight of it, is clearly adaptive for a cat.

  14. DiscoveredJoys
    Posted June 16, 2016 at 3:22 am | Permalink

    Let’s try an easier question first. Do people understand the laws of physics?

    I’d argue that most people don’t ‘understand’ physics, even a lot of trained physicists don’t ‘understand’ physical laws.

    Unless you take ‘understand’ to mean ‘form expectations based on previous observations of cause and effect’. In this sense both cats and people (and even d*gs) understand the laws of physics.

  15. Scote
    Posted June 16, 2016 at 7:56 am | Permalink

    So… Not so much an understanding of physics, but rather that cats have an understanding that prey noise in a box mean there should be prey to attack.

  16. Nat
    Posted June 16, 2016 at 10:53 am | Permalink

    Ah, but you most read the short story by Fritz Leiber, “Spacetime for Springers.”

One Trackback/Pingback

  1. […] Do cats understand the laws of physics? […]

%d bloggers like this: