Pulling an all-nighter doesn't work because learning requires a good night's sleep
Over the next month high school and college students across the nation will face that scary rite of spring, final examinations. As a teacher I get to give exams, rather than take them, but I can remember with crystal clarity when the shoe was on the other foot. I didn't like exams a bit as a student -- what student does? But in my case I was often practically paralyzed with fear. What scared me about exams was the possibility of unanticipated questions. No matter how much I learned, there was always something I didn't know, some direction from which my teacher could lob a question I had no chance of answering. I lived and died by the all-nighter. Black coffee was my closest friend in final exam week, and sleep seemed a luxury I couldn't afford. My parents urged me to sleep more, but I was trying to cram enough in to meet any possible question, and couldn't waste time sleeping.
Now, forty years later, I find I did it all wrong. In work published last month, Robert Stickgold, a psychiatrist at Harvard Medical School, demonstrates that our memory of newly learned stuff improves only after sleeping at least six hours. If I wanted to do well on final exams, I could not have chosen a poorer way to prepare. God must look after the ignorant, as I usually passed.
Learning is, in its most basic sense, a matter of forming memories. Dr. Stickgold's experiments show that a person trying to learn something does not improve his or her knowledge until after they have had more than six hours of sleep (preferably eight). It seems the brain needs time to file new information and skills away in the proper slots so that it can be retrieved later. Without enough sleep to do all this filing, the new information does not get properly encoded into the brain's memory circuits.
To sort out the role of sleep in learning, Stickgold trained Harvard undergraduates to look for particular visual targets on a computer screen, and to push a button as soon as they were sure they had seen one. At first, responses were relatively sluggish -- it typically took 400 milliseconds for a target to reach a student's conscious awareness. With an hour's training, however, many students were hitting the button correctly in 75 milliseconds.
How well had they learned? When retested from 3 to 12 hours later on the same day, there was no further improvement past a student's best time in the training session. If Stickgold let a student get a little sleep, but less than six hours, then retested the next day, the student still showed no improvement in performing the target identification.
For students who slept more than six hours, the story was very different. Sleep greatly improved performance. Students who achieved 75 milliseconds in the training
session would reliably perform the target identification in 62 milliseconds after a good night's sleep! After several nights ample sleep, they often got even more proficient.
Why six or eight hours, and not four or five? The sort of sleeping you do at the beginning of a night's sleep, and the sort you do at the end are different, and both, it appears, are required for efficient learning.
The first two hours of sleeping are spent in deep sleep, what psychiarists call slow wave sleep. During this time, certain brain chemicals become used up, which allows information that has been gathered during the day to flow out of the memory center of the brain, the hippocampus, and into the cortex, the outer covering of the brain where long-term memories are stored. Like moving information in a computer from active memory to the hard drive, this process preserves experience for future reference. Without it, long term learning cannot occur.
Over the next hours the cortex sorts through the information it has received, distributing it to various locations and networks. Particular connections between nerve cells become strengthened as memories are preserved, a process that is though to require the manufacture of new proteins, a slow process. If you halt this process before it is complete, the day's memories do not get fully "transcribed" and you don't remember all that you would have, had you allowed the process to continue to completion. A few hours are just not enough to get the job done. Four hours, Stickgold estimates, is a minimum requirement.
The last two hours of a night's uninterrupted sleep are spent in rapid-eye-movement (rem) sleep, when dreams occur. The brain shuts down the connection to the hippocampus and runs through the data it has stored over the previous hours. This process is also important to learning, as it reinforces and strengthens the many connections between nerve cells that make up the new memory. Like a child repeating a refrain to memorize it, the brain goes over what it has learned, till practice makes perfect.
That's why my college system of getting by on three or four hours of sleep during the week and crashing for twelve hours on weekends didn't work. After a few days, all the facts I had memorized during one of my "all-nighters" faded away. Of course they did. I had never given them a chance to integrate properly into my memory circuits. If I could look back, how well I did on my exams probably had far less to do with how hard I studied than with how much I slept. It doesn't seem fair that after all these years I have to admit that my parents were right all along.