At 8:30 a.m., a bleary-eyed young woman wearing a bright orange bandanna and fuzzy, gray slippers is staring at a computer screen. When a number appears in a small box, she hits the space bar. She does it again and again, sometimes quickly and other times much more slowly.

“She’s very sleepy. She has a lot of lapses,” whispers Namni Goel, a sleep researcher at the University of Pennsylvania Perelman School of Medicine.

The young woman, Amanda Burton, is taking a test of her attention and reaction speed. As a participant in a scientific study at the Sleep and Chronobiology Laboratory here, she has had just four hours of sleep each night for the past five nights. The Penn lab is one of the busiest human sleep research labs in the world. It has deprived more than 1,500 people of sleep over the decades with the goal of figuring out what happens in the bodies and brains of people when they skimp on shut-eye.

It is an urgent question. Experts generally recommend that healthy adults sleep between seven and nine hours a night. But about 35% of adults say they typically get less than seven hours, according to a survey of 75,000 people conducted by the Centers for Disease Control and Prevention in 2008. Chronic insufficient sleep has been linked to everything from diabetes and obesity to high blood pressure and an increased risk of death.

“We have to add sleep to this crazy world,” says David F. Dinges, a professor in the department of psychiatry and the 66-year-old director of the Penn lab. “And if we can’t stop short sleeping, the question is: How do we optimize the ability to tolerate [sleep loss], what are the effects of this and how do we reverse it, stop it and alter it?”

On a recent Monday, Ms. Burton was one of three subjects living in the Penn sleep lab, a small, windowless space with linoleum floors and dim fluorescent lighting. The participants—who are paid about $2,000—won’t leave at all during the 14-day study. They can’t exercise, use their phones or access the Internet. (Light and physical activity can affect sleep, mood and test performance. So can an argument with a significant other.)

The study subjects are hooked up to wires that monitor their brain activity. They sleep—the few hours they’re allowed to—in hospital beds in double rooms. Meals come from the hospital cafeteria. There’s no caffeine, chocolate, nicotine or alcohol allowed. But there are unlimited snacks—like Kind granola bars and spicy sweet chili Doritos—available in the small kitchen.

They are watched by technicians and “minders” (usually undergraduate employees) at all times—except for bathroom breaks and showers. The monitors make sure study subjects are completing the various study tasks and staying awake when they’re supposed to.

“Between 2 and 4 [a.m.] I was dozing off the entire time,” says Ms. Burton, a 33-year-old administrative assistant from Plainfield, N.J., who is taking time off to participate in her third sleep study at Penn. The monitors “had to constantly yell my name. ‘Amanda, Amanda, hey, are you awake?’ ” she says.

This particular study of about 100 subjects, which Dr. Goel is leading, aims to see if people respond equally to two different types of sleep deprivation. It compares acute, total sleep loss (going 36 hours without sleep) with chronic sleep deprivation (in this case, just four hours of sleep between 4 a.m. and 8 a.m. for five nights). Preliminary results indicate that the answer is yes, Dr. Goel says.

“There are huge individual differences. [But] if you are resilient to total sleep loss, you are resilient to being chronically sleep-deprived or if you’re vulnerable to one, you are vulnerable to the other,” she says.

The study is also looking for genes and other so-called biomarkers—substances in the blood—that can predict which people are more vulnerable to the effects of sleep loss. Dr. Goel says the researchers expect to finish the study later this year.

About every two hours, the study subjects complete a battery of tests that measure things like attention and reaction time, working memory and how quickly they can process information. They complete questionnaires that ask about mood and how sleepy they feel. Researchers calculate the number of calories they eat.

In earlier studies, the Penn lab and others have found that attention, reaction time and cognitive speed tend to be particularly affected by sleep loss. This is a big problem for driving. Higher cognitive functions, like reasoning, tend to be less affected. Mood plummets: People interpret even neutral facial expressions as more negative. Sleep-deprived subjects also eat more and gain weight: In Penn studies, they eat about 500 extra calories a day, veer toward fat-laden foods and gain about 2 pounds in a week.

Most Penn sleep-deprivation studies include a control group that lives in the same environment and does the same testing protocol, but doesn’t lose any sleep.

But not everyone tolerates sleep loss the same: In general about one-third of people are resistant to the effects of sleep loss, one-third are vulnerable and one-third are somewhere in the middle, Dr. Dinges says.

Ms. Burton seems to fall into the vulnerable group, Dr. Goel says. On the test where she had to hit the space bar when the number appeared, dubbed the psychomotor vigilance test or PVT, it took Ms. Burton sometimes as long as 2 seconds to react accurately. By contrast, another subject, Daniel Dougherty, usually took less than 300 milliseconds to respond. A typical, well-rested person would need about 250 milliseconds.

Other studies have shown that most people aren't good judges of how sleepy—and how impaired—they are.

“I was like, that will be a breeze,” says Mr. Dougherty, a 32-year-old from Goldsboro, Md., of what he expected the study to be like. “But it is actually kind of hard. Just feeling tired and wanting to go to sleep but you can’t.” Mr. Doughtery is changing careers and plans to use the sleep study money to repair his 15 year-old car.

Between tests the study subjects read, watch TV and play board games. There’s an entire bookcase filled with games like Monopoly, Sorry and Pictionary. They binge-watch shows on Netflix. Ms. Burton got through five seasons of “Portlandia.” Mr. Dougherty was plowing through “Narcos” episodes. Researchers decided that TV-watching would be acceptable because the subjects sit far enough from the televisions that the light emitted wouldn't interfere with the results.

The study is being funded by the Office of Naval Research. Dr. Dinges has also just completed a study looking at how long it takes to recover from sleep deprivation. In earlier studies, it seemed to only take a few days of regular shut-eye for people to recover from sleep loss and perform like they were well-rested.

But this new study seems to refute that. People might seem fine, but if they lose sleep again, their performance plummets more than expected. “There’s a memory in this biology,” Dr. Dinges says. “They are carrying a vulnerability.”

If the results hold up and are replicated, he says, the days of human sleep-deprivation studies might be numbered. “If I can’t reverse what I’ve induced, then these experiments are no longer ethical,” he says. “I’ve got to stop doing this.”

Source: The Wall Street Journal. Long Nights in a Sleep Lab. http://www.wsj.com/articles/long-nights-in-a-sleep-lab-1445272717