You’ve changed somehow. Is that a new turnip?

I spent about an hour yesterday at the National Gallery of Art, mesmerized by the Italian artist Giuseppe Arcimboldo’s paintings of human faces. They are not realistic depictions of faces, though some were meant as portraits of public figures. They are instead composites of familiar objects like flowers, fruit or fish and crustaceans—each rendered with scientific accuracy. Close examination reveals the intricate interplay of these objects, but at first glance they are unmistakably faces, with noses and ears and hair and chins. They are quite surreal, which is remarkable given that Arcimboldo created them in the 16^th century.

Arcimboldo clearly knew something about how we perceive faces. He had to know—intuitively, as an artist—that the features that make up a human face are less important than the whole. Even if the nose becomes a cucumber, we still see the whole face, automatically integrating the details into something meaningful.

Psychological scientists have since proven that we process faces more holistically than other objects in our world—presumably because rapid facial perception is beneficial. But is there a cost associated with seeing faces holistically? Iowa State University psychological scientists Miko Wilford and Gary Wells suspected that there might be. Specifically, they suspected that holistic perception might make us inattentive to particular features, so that we focus on the big picture at the expense of the details. They decided to explore the idea in the laboratory.

They showed people pictures of both faces and houses. Then later they showed them either the identical picture or one that had been slightly modified—a different nose or chin on a face, for example, and a different porch or roof on a house. They asked the volunteers to say whether or not the image had changed at all—and if so, what particular feature had changed.

Here’s what they found, and reported in the newest issue of Psychological Science: People were much better at spotting a changed face than a changed house—but they were much worse at knowing exactly what had changed in a face. They couldn’t put their finger on whether it was a longer chin or wider eyes, yet they knew automatically the face was not the same as before. With houses, it was the opposite. They didn’t readily detect any change at all, but if asked, they were more apt to spot a new window or veranda.

So it appears that people do see the whole face, not merely a collection of features. And this innate talent comes at a cost—inattention to detail. The scientists double-checked this finding by turning some of the changed pictures upside down. They figured that inverting a face would throw off the natural processing. And that’s exactly what happened. Inverted faces were no different than houses—assortments of individual features rather than integrated wholes.

Arcimboldo, by the way, intuitively knew this as well. Some of his most playful paintings are ordinary still lifes—a bowl of vegetables, for instance—which only form a human face when turned upside down.

Wray Herbert’s new book, On Second Thought: Outsmarting Your Mind’s Hard-Wired Habits, has just been published by Crown. Excerpts from his two blogs—“Full Frontal Psychology” and “We’re Only Human”—appear regularly in The Huffington Post and in Scientific American Mind.