48 Examples Of Ordinary Objects Looking Like Something Else Entirely

If you have ever started seeing faces in clouds or were thinking that your teapot might be smiling at you, don't worry, you're not crazy. Pareidolia is common for a lot of us. Still, it's interesting why our brains are seemingly wired to see faces everywhere: is it because we like to give anthropomorphic qualities to everything?

Researcher Mark Hamilton and his team completed their research on pareidolia in 2024 and have some theories about why we tend to see eyes, noses, and mouths everywhere. The researchers think that pareidolia is just our way of protecting ourselves.

In a way, pareidolia is the result of our survival instinct. "Our research points to the fact that it might be because of the need to quickly parse and identify animal faces," Hamilton told Bored Panda. "Possibly for things like avoiding predators and catching prey."

Interestingly, humans aren't alone in seeing faces in places where there are none. "Other research that we reference in our paper shows even rhesus monkeys experience pareidolia," Hamilton added. "Pareidolia happens very quickly in the brain, a sign that it's a deep-rooted evolutionary artifact."

In their research, Hamilton and his team first fed a state-of-the-art face detector images of human faces. They asked it to recognize faces in pareidolic images. The algorithm didn't do so well, with only a 9% average precision. When the algorithm was trained on pareidolic images, on the other hand, it was way more precise and had a 36% AP. 

"Our research specifically showed that algorithms trained on human faces alone didn't see nearly as much pareidolia as algorithms trained on human and animal faces," Hamilton says. Training the algorithm on animal faces made it recognize pareidolic images even better. Hamilton believes it's because the algorithm generalized beyond just human facial features.

When the algorithm is trained on animal faces it starts thinking about faces more abstractly. "If you think about it, animal faces have a ton of different variants and colors, which might explain why we can see pareidolic faces that look nothing like human faces," Hamilton explains.

If the AI recognizes pareidolia better when it's trained on a more diverse set of faces, does that mean that we, humans, start seeing more faces in things as we age and have seen more living faces of all kinds? Not really. According to Hamilton, other research shows that pareidolia in humans emerges very young. One study even found that fetuses respond to face-like patterns by shining lights into the womb!

Another interesting fact is that there are also gender differences in who sees more pareidolia. "Women see more pareidolia than men, and pareidolic faces tend to be perceived more often as male, both in our dataset and in the broader literature," Mark T. Hamilton tells Bored Panda.

The thing that surprised Hamilton and his team most during their research was that pareidolia and detecting animals are linked in modern algorithms. "That's not often connected," he told us. "This helps make sense of a lot of the different pareidolia observations out there. Like its observation in monkeys, its fast response time, even perhaps its skew towards male faces (which one could argue are more correlated with predator traits like strong jaws, harder noses, etc)."

Hamilton's team also had some fun during their research. Apparently, the algorithms are quite good at finding pareidolic doppelgängers for animals. It turns out that lions can look like pancakes, some seals look eerily similar to pears, and mice resemble melons! You can see the hilarious examples right here!

The team also wrote down some equations that roughly predict when and where people might see pareidolia. Then, they verified them with human psychophysics studies. "The equation and studies show that there's a "goldilocks" zone of "complexity" in a scene where faces are more likely to pop out," Hamilton explains. You can see a graphic or what's too much pareidolia and what's too little here."

You might also like: 38 seriously confusing photos that require a second look