Think back to when you slipped on the ice or in the shower: the ground rushing up, your feet shooting out, terror building even as your mind is working a mile a second to plot a soft landing.

This is what Emily Keshner studies, in a lab designed to mimic all the above. Balance is something most people don't think about unless they're learning to snowboard or walking up the aisle when their plane hits turbulence.

Or recovering from a stroke, as more people will be doing in an aging America. More than a quarter of people ages 65 to 74 experience dizziness or trouble with balance at some point each year, according to federal statistics, and nearly 16,000 elderly people die from fall-related injuries.

Keshner is one of a handful of scientists who pioneered the use of virtual reality - the technology of games and flight simulators - to study why we fall.

"I think her emphasis on having people navigate in a normal world situation is nearly unique," said Christopher Platt, balance and hearing program director for the National Institute on Deafness and Other Communication Disorders, which contributes part of the $300,000 that funds the lab's current research.

Balance is the result of complex interrelationships among eyes, ears, brain, and our awareness of other body parts. Lots of things can go wrong.

"The complaint that people often go to the hospital with is 'I'm dizzy,' " Platt said.

Just narrowing it down to the correct sensory system - or any system as opposed to, say, blood pressure - can be a challenge for doctors.

For researchers, studying one system historically has not illuminated how the body works as a whole. When the systems are examined in combination, however, isolating cause and effect can be hard. And painful for those being studied: People with balance disorders fall down.

Over the last decade, however, faster computers and sharper graphics have opened a whole new way to study balance - in a virtual environment.

"It has allowed us to look at a person as they would be in the real world," says Keshner, who brought the lab with her from Chicago when she became chair of the physical therapy department in Temple University's College of Health Professions two years ago. The first research subjects came through last fall.

On a small, dark stage in the basement laboratory a few weeks ago, the room began to turn around Joe Turman. The three-sided "room" - an artificial, dreamlike scene that looks out over straight-edged rugs and pillars to a mountainous horizon - appeared in rich detail through Turman's 3-D glasses.

At a computer 20 feet away, a post-doctoral fellow named Jill Slaboda swiftly built a stick figure from dots that marked the locations of sensors affixed to Turman's head and key joints.

Turman, 73, suffered his third stroke last year, and he struggled to remain upright in the lab as his surroundings appeared to spin. Suddenly, the floor lurched forward. Turman lost his balance, falling into a safety harness. (Another researcher was poised to catch him, just in case.)

People fall more after a stroke, and Keshner, who is interested in how the central nervous system coordinates movements from the feet up through the head, is investigating what those patients do differently.

She places them in the lab's virtual environment, and then disrupts it. Receptors track the pressure of feet on the floor during an attempt to balance, and electrical activity in key muscles, allowing her to measure reactions with precision.

Last month, she presented at a conference findings that stroke patients rely more heavily than others on visual stimuli and less on what's coming up through their feet.

So while a healthy person who steps on a rock during an afternoon stroll will put a priority on that new information, she explained, a stroke patient is more likely to continue relying on the flat path he sees ahead, and trip.

The data behind that finding, like many of her other discoveries, will likely shape rehabilitation programs, some of which she hopes to develop in her lab. Perhaps, she said, stroke patients can be trained to pay more attention to nonvisual cues.

Keshner spent years studying the vestibular system, the hair-lined organs of the inner ear that signal the brain when the head moves, allowing the body to maintain balance. (The brain's response to that signal keeps your eyes steady on this line even if you shift in the chair.)

When the vestibular and visual systems relay conflicting information - as when the inner ear detects motion on a boat but the eyes are fooled by the stationary wall inside the cabin - you get dizzy. (Going up on deck to look at the rolling horizon can put them in sync.)

The third system involved with balance, called proprioceptive, consists of sensory information from muscles and joints that indicate their relative locations. (It's how you can touch your nose with your eyes closed.)

Defects in the brain's processing ability or in any of the systems - caused by infection, trauma, or slowed response time - can result in symptoms ranging from vertigo to nausea.

While the Temple lab does mainly basic science that may shape future rehab, some other labs are using virtual reality for rehabilitation. Iraq war veterans with post-traumatic stress disorder, for example, have been desensitized through gradual exposure to virtual war. The University of Southern California is creating virtual environments that require specific motor movements within a game for rehabilitation.

Researchers at the University of Pittsburgh study and treat people with balance-related anxiety disorders in a virtual grocery store - it makes them sweaty and nauseous to move down the aisle in combination with all the head movements required to scan products in a busy environment.

Slaboda spent her first year as a Temple post-doc studying how children and adults stand up. Each subject was seated on a stool surrounded by the artificial scene, which she made to pitch (nose up or down, as in an airplane) or roll (wing up or down).

Under one condition, they were told to stand up as soon as the room began to move. Under another condition, they sat and watched for 10 seconds, and then were told to stand.

With all the data creating stick figures in her computer, Slaboda analyzed how they stood up - calculating the angles of head to trunk, trunk to leg, and plotting graphs of flexion and extension and time between movements.

Bottom line: "If you sit and watch first, you'll stand up slower," she said, because you had more time to become disoriented. She submitted her findings to a journal for publication last month and will soon begin the second half of her study: how stroke patients stand up.

The handful of researchers affiliated with the Temple lab come from a variety of disciplines. Slaboda has a doctorate in biomedical engineering. Keshner has expertise in dance, engineering, special education and movement science, in addition to physical therapy.

Jay Barton, the lab manager, is researching the brain-machine interface for the creation of prosthetic limbs controlled by thought. His training as an electrical engineer took him only so far. The biomechanics of real people offers surprises. Such as:

"How an engineer would design the body is not how the body operates," he said.