My latest column in the Wall Street Journal is on the purpose of dreams:
Chancing last week on a study about the calming effect of dreams on people with post-traumatic stress disorder, I decided to read recent research on dreams. When I looked at this topic about 20 years ago, it was clear that our ignorance of the purpose of dreaming was almost total, notwithstanding the efforts of Sigmund Freud, Francis Crick and other fine minds. Is that still true?
To my delight, the answer seems to be no. Some ingenious experiments have replaced general ignorance with specific and intriguing ignorance (as is science's habit). We now know enough to know what it is we do not know about dreams.
In the past, people often had one explanation for sleep and another for dreams. That now seems wrong. One of the chief functions of sleep seems to be achieved during dreaming: the consolidation of memory. Sleep certainly improves memory performance of several different kinds: emotional, spatial, procedural and verbal.
But the new thinking is that, during sleep, the brain reprocesses or transforms fragile new memories into more permanent forms, sets them in mental context and extracts their meaning. And dreaming is a symptom that this is going on.
In one experiment, subjects were asked to remember a novel word that sounds like a real word: "cathedruke," for example. Much later they were asked to recall the real word "cathedral," which it resembles. People who had spent the intervening time awake recalled the word more quickly than those who had spent the time asleep. To cut a long experimental story short, it seems that the sleepers, probably by dreaming, had refiled cathedruke in "lexical" memory in the higher brain region known as the cortex rather than in the hippocampus, where short-term memory resides. There it interferes briefly with recall of cathedral.
In another experiment, people were asked to remember a series of words, such as nurse, ill, patient, etc. Some hours later, they were asked to recall if certain words were in the list. They correctly rejected most of the wrong words-except the word "doctor," which was not in the list but sounds as if it should have been. Here's the surprise: The people who made this mistake most were the ones who had been to sleep in the meantime. Dreaming sleep had extracted the "gist" of the list.
According to this theory, dreaming is a symptom of such memory processing. Contrary to popular belief, dreaming occurs throughout sleep, not just in rapid-eye-movement sleep. But the dreams reported by people woken from non-REM sleep tend to be literal and straightforward recitations of recent experiences stored in the hippocampus. The later dreams of REM sleep incorporate more distant memories, becoming more fantastic and more emotional as the new memories get mixed with old ones in the cortex.
Even the neurophysiology is becoming clearer. The same brain circuits active when a rat learns a maze are reactivated during sleep, though in faster and more fragmented bursts, and seem indirectly linked with bursts of synchronized activity in neurons in the cortex known as sleep spindles. People who experience more spindles retain memories better.
In the laboratory of Erin Wamsley and Robert Stickgold at Harvard Medical School, people were asked to play a game that required navigating a virtual 3-D maze. During the next 90 minutes, some napped while others stayed awake. Over that period, they were asked several times what they were thinking (if awake) or dreaming (if asleep). Those who were dreaming about the game were 10 times better at it the next time they played. Those who were just thinking about the game improved hardly at all.
When we replay a recent experience in a dream, we enhance our memory.