People can tell the difference between abstract art and paintings by chimps

A study reported on in Psychology Today set out to determine whether people could distinguish abstract art from paintings done by monkeys (as well as children and other animals, like elephants). Turns out, most people can usually tell the stuff done by artists from the stuff done by your five-year-old.

For a paper in press at Psychological Science, Angelina Hawley-Dolan and Ellen Winner of Boston University collected 72 undergrads, 32 of which were studio-art majors, and showed them 30 paintings by abstract expressionists. Each painting was paired with a painting by a child, a monkey, a chimpanzee, a gorilla, or an elephant. The images were matched on superficial attributes such as color, line quality, and brushstroke, and subjects were asked which piece they personally liked more, and which they thought was a better work of art.
"My monkey could have painted that." Really?

Seeing the Mind Behind the Art People Can Distinguish Abstract Expressionist Paintings From Highly Similar Paintings by Children, Chimps, Monkeys, and Elephants (Thanks, MHutson, via Submitterator!)


  1. A chimpanzee is not a monkey. That’s like saying a dog is a bear. Monkeys are not self-aware but chimpanzees are. Basic taxonomy that indicates a set of common features: monkeys are not apes. Humans, on the other hand, are apes.

    1. A chimpanzee is not a monkey.

      Read the quoted excerpt: the professional paintings were compared with those of chimps AND monkeys (and gorillas and elephants and human children).

      Granted, if you only read Cory’s headline and first sentence it does kind of sound like he made that error.

  2. Maybe the elephant should serve some wine and light snacks or even hire some cute gallerinas for the elephant’s next show.

    Just sayin’

  3. I agree that no chimp can be an artist. A chimp could never conceive of creating art from the DNA found in William Burroughs’ shit, for example.

    1. They’ve probably been trying to tell us to do just that for decades whenever they fling poo, we just couldn’t figure out what they were getting at.

  4. I dunno… only 60-65% could guess correctly… or am I reading that table wrong? That doesn’t seem to imply to me that “people can tell.” That sounds only slightly better then guessing to me.

    The implication to me is something closer to “abstract artists are only 10-15% more coherent a child, monkey, chimp or gorilla.”

    1. The results show that people can tell human-drawn abstract art better than blind chance, which does show that it is possible to tell, and thus presumably possible to identify artefacts of humanity and start doing criticism about the message and other wankery that people who talk about art want.

      Me, I prefer the kids drawing on the left in the article, and thought it looked more like art. It’s more deliberate and structured to my eye, while the artist’s one on the right just looks boring. There are lots of little pointless splodges of colour that just make the whole thing look messy.

  5. Just looking at paint handling and brush technique alone would probably be enough to fairly accurately tell the difference between a painting done by an adult and a child or chimp. Also, asking art majors is stacking the deck a bit. They may have recognized some of the paintings.

    I wish they showed more than two paintings in the article. It would be fun to guess.

  6. If I’m reading that chart correctly….nonart students guessed the correct label more frequently than art students. (quite a bit of std dev. going on, though…)

  7. Without wishing to troll in any way, abstract art’s still a crock of shit. In my humble opinion.

  8. The December 2002 issue of Scientific American had an article the fractal quality of Jackson Pollock’s drip paintings.

    Link to a PDF of the article here:

    You can read the original paper here, it’s not very long:
    Taylor et al: Perceptual and Physiological Responses to the Visual Complexity of Pollock’s Dripped Fractal Patterns

    While not an explanation of why Pollock’s art is so valued, Taylor’s ideas can help in understanding perhaps why we can tell the difference between abstart paintings and random paintings.

  9. Giles: you’re not; I don’t think they were asked which they thought was by whom/what. Some of the pairs were unlabelled, some were labelled correctly with the type of “artist”, and on some the labels were reversed. So the non-art students (I’m sorry, nonart ISNOT AWORD) chose more correctly-labelled works as being “better” works of art.

    I think.

  10. So it seems evident that, most likely, your pet monkey could not have painted that.

    Yeah, but if you had two pet monkeys, you’d have a better than even chance one of them could have (65.56%). I am not sure if that’s great evidence for justification of a professional career in abstract art.

    I think a better way to phrase the results is by saying that professional abstract art has only a 2:1 chance of being judged as better art than the psuedo-random smearings of a not-fully developed mind.

    Unless you’re being judged by amatuers.

    (Also, there’s a slightly lower chance your picture will actually be liked more).

  11. Anyone else find it telling that this study acknowledges no middle ground between “paintings by animals & infants” and “works by successful professional artists”?

  12. Also, it doesn’t seem like you can claim that people can “tell the difference” as it doesn’t sound like anyone was actually asked to explicitly do so.

    Without paying for the login to see the full study, the Psych Today article reads like people were just told to make judgment calls about which painting in a pair was subjectively better. Psych Today doesn’t make it clear that the subjects were instructed to specifically choose the one painted by a professional. The setup explained implies that the subjects are simply *going* to pick the painting they think was done by the professional. What if one of the subjects purposefully thought one of the two paintings actually was painted by a child and actually thought that one was better?

    I think the experimental setup would call for presenting people with pairs of paintings and asking them which they thought was painted by a professional and which they thought was painted by one of the other four. You could still use the same “No Label” “Correct Label” and “Incorrect Label” schema.

  13. Wait, really?

    Note that the numbers in parentheses are standard deviations; three quarters of those numbers include 50% within a single standard deviation, and I don’t think there’s a single value on that chart that doesn’t fall inside a 95% confidence interval.

    So basically what this study says is that in a sample of about thirty people, they found a fifty-fifty chance of guessing right.

    So it says nothing. Why is this newsworthy?

    1. And yes I know they had 72 people in the study, but those groups in that table are split into groups of 30/32 non-majors/majors.

    2. The different averages seem to be clustered near 65, with standard deviations near 15. That’s a bit too close to say any one of them is clearly better than chance. But when you have a dozen averages and standard deviations that all fall into a similar range- 2/3 choosing the artist’s work, regardless of labeling or academic art training- that is very suggestive. I’d expect more sophisticated statistical tests could make that explicit.

      A average of averages of normally distributed values varies less than the averages themselves (standard deviation decreases as sqrt(n)) ; this is why scientists repeat experiments. These are not repeats, but they’re clearly not independent either.

      1. Yes, it’s definitely suggestive, but I strongly object to citing a merely suggestive study without any sort of caveat to that effect.

        Citing a scientific study is, on the internet, like bringing out the big guns. It lends a strong weight of credibility to your argument. I think it’s dishonest to do so when the evidence you’re citing is less concrete than the argument you’re making.

        The original post by Cory said nothing about doubts or uncertainties. It didn’t even gloss over them, it just left them to the conscientious observer to notice in the table.

    3. So basically what this study says is that in a sample of about thirty people, they found a fifty-fifty chance of guessing right. So it says nothing. Why is this newsworthy?

      Um, what? That would be extremely newsworthy. Don’t make the mistake that failing to reject the null hypothesis makes something non-interesting.

      On the contrary, I assume that the vast majority of art professors and critics would roundly scoff at the notion that “real” art could be confused with children’s scribbles at an almost equal chance. This study, if the results were solid and valid, could theoretically dismantle huge implicit assumptions in the art world. It’s the other result — that the difference were statistically significant — that would be yawn-o-rama. Indeed, I rolled my eyes at the headline of this post, thinking “of course.” And then I read how close the results actually were.

      That said, this result does not prove anything one way or the other. A tiny sample size, almost guaranteed bias in choosing which children’s and animal pieces to use, and a weird methodology means that this study proves just about nothing right now. But, if these weak results did hold up, then it would be pretty bad for the abstract art world.

      1. Hmm. Yes, you’re right, I wasn’t considering the implication of the failure to prove that people can tell the difference.

        And you’re definitely right, that would be a very newsworthy study; I’d be interested in seeing a study designed to test that hypothesis rather than testing whether people can tell the difference. Hopefully with slightly bigger groups.

  14. And when you take the standard deviations into account it looks like they performed more or less at chance.

  15. This bit amuses me: ” (as well as children and other animals, like elephants)” as though it’s lumping in children with animals and keeping teenagers and adults as a discrete category of being.

  16. I also preferred the child’s painting. I thought that it had more coherence of colour, and a more interesting structure than the professional one.

  17. To me, the child’s was obvious. Definitely preferable, since it had more freedom to it, though.

    The “real” artist seems, consciously or unconsciously, to have been constrained by his training to a constant brush pressure, and so all the lines are confident, assured, and constant-pressure. Changes in brush width are mostly due to the movement of the brush rather than changes in pressure. The child used varied brush (well, crayon) pressure, which meant that the texszture of the paper could come out; and also used stuff that wasn’t even a brush.

    Something else I’ve picked up from watching kids paint: younger artists use black as a focus in their art quite often, as the topmost layer of ink or paint. It’s the “finishing touch”, to them. Older artists tend to use it at a backdrop, or as the inked line-art between the details, instead. To them, it’s the “basis of the work”.

    The kid’s way of doing it seemed wrong to me at firest, and it was all I could do not to say “No! You’re obscuring all the lovely detail with this dark funk you’re slathering over it!”

    Although that’s another explanation – they’ve worked their way through all the interesting colours in the limited palette they’re given, and by the time they get to the boringest, black, they’re bored of the picture, so just slather it around boredly. And stamping with that round circley thing must’ve been definitely fun enough to get carried away with :D

  18. I preferred the child’s painting, but the other one clearly demonstrated more technical skill. Another clue is that the cross symbol is not something an ape would have drawn.

    Up to now I’ve always thought it was hyperbole when people say that a chimp could draw better, but this study shows that for an unbiased viewer it’s true a third of the time.

    I think an interesting way to redo the test would be to take three paintings by professional artists and three by chimps and have people put them in order from best to worst.

  19. “The images were matched on superficial attributes such as color, line quality, and brushstroke…”

    Am I reading it wrong, or does this study basically prove that, if you deliberately choose paintings that look like monkey drawings, it’s difficult to tell them apart?

  20. “monkeys (as well as children and other animals”

    Awesome. I hope I can see baby children next time I go to the zoo.

  21. I’ve never understood the deep-seated negativity toward modern, abstract art, especially in the US. It can be quite visceral at times.

    1. My guess is that its a reaction against the primitive. Euro cultures, at least, tend to denigrate ‘savages’ and things ‘savage’ (see e.g. 19th century books … before we learned to mask our true feelings). Maybe it’s a hardwired remnant of the ancient battle against Neaderthals and/or the wildness and threatening arbitrariness of the world. Or maybe it’s just plain reality tunnel.

  22. It was pretty obvious to me that the picture on the left was done by a kid and the one on the right was done by an adult artist. But then, I’ve looked at a lot of art by kids and abstract artists, so perhaps the clues were more obvious to me.

    I agree that if the study was asking the subjects to choose the “better” painting, it was ill-designed. I personally preferred the adult painting over the kid painting, but my reasons for preferring it were distinct from the clues that I observed to its provenance.

  23. That’s the thing, though. This study DOESN’T show for an unbiased viewer that it’s true a third of the time. The statistics they include show that there’s no reasonable way to assume these results are different than a 0.5, or a pure guess.

  24. As an abstract artist I must say, artists like that make bad name for abstract art. when you need art majors to tell the difference between “profecianal art” done by adults and something done by a 3 year old or someone of a different species and even with that the results were just a little better then guessing… Why can’t we stop justifying bad art and start supporting artists who have some depth to their work and spend more then 15 minutes on a painting? There is plenty of good quality abstract art out there after all.

  25. The study was out of Boston College, not Boston University. Please see the linked article.

  26. LOL, “People can tell artists from chimps/children/elephants about 2/3 of the time”. This is supposed to be a defence of art?

    For the non-label comparison, artists and non-artists preferred the ‘real’ art 13% and 6% (respectively) more than would be expected if the two pieces were equally good.

    Simply looking at the techniques should easily show who the ‘true’ artist is. Elephants aren’t good at delicate brushwork.

    Wow, those artists must be really good…

    I hope laymen would prefer my work rather more than 6% more often than that done by a child/chimp/elephant.

    If it wasn’t for those pesky child labour laws, I’d employ a staff of 4-year olds and get them churning out art tomorrow. If only elephants had opposable thumbs – we’ve clearly been missing out on their hidden talents. Maybe those cave drawings of mammoths were actually self-portraits.

Comments are closed.