This fall, for the first time, medical students at the University of Washington are learning to "diagnose paintings." It's the subject of a new elective class. The term "diagnose," though it's the one they use, is actually the opposite of what they do. They look, they report the details they see, they become aware of how much they don't see by hearing what others see, and, above all, they try to resist the impulse to make a diagnosis.

The class alternates: On "art weeks" they stand in museums exhaustively describing artworks. Who made the artwork, when, and in what style—the usual art-historical details—don't matter; only what's in front of their faces does.

Last week was a "medical week." In a small conference room at the Henry Art Gallery, the lights are turned out and slides are glowing, just like in any art-history class. But these are medical slides. Dr. Andrea Kalus (the class is cotaught by Kalus, a dermatologist, and Tamara Moats, a longtime art-history educator) is trying to get the 19 students to respond to snapshots of patients with the same blank receptivity they bring to art. It is not really working.

"Is that cyanosis?" one student asks, faced with an image of a silvery-skinned man.

"I think that's Horner syndrome," is the response to a picture of a woman with two different-colored, differently shaped eyes.

For every slide Dr. Kalus shows, the students can manage about three minutes of observation—"one pupil is bigger than the other," etc.—before they succumb to their desire to name a diagnosis. Dr. Kalus stops them: "Just tell me what you see."

When they stick with observation for a prolonged period, the results are almost miraculous. From a single image of a dark-haired girl with a red, blistery pattern on her skin, they determine—by continuing to seek details against an onslaught of their mistaken theories—that the injury was caused by an external source (its borders are straight lines); not caused by liquid, chemicals, or radiation (given the patterning); and that it's a severe sunburn caused by the patient taking photosensitizing drugs (ibuprofen). They can even say where the sun hung. (I want to mention to them that the direction of light is a major preoccupation in art, especially painting.)

Typically, med students get specific instruction on how to listen to and how to touch (palpate) their patients—but not on how to look at them. "Physical diagnostic tools," as Dr. Kalus calls them, used to be emphasized before the rise of technology. Today's gap in visual training is made worse by the fact that medical students are extremely goal-oriented, the doc says: They want to get to the answer. The gift of art, for them, is that it can't be diagnosed. All they can learn from it is to keep looking.