Sigmund Freud thought dreams were a window into our unfulfilled sexual desires. But the dreams of video game players suggest they have a more practical role: helping us to learn new skills.
"It really looks like if you're not dreaming about it, you're not getting better," says Robert Stickgold of Harvard Medical School, who carried out one of the video game studies.
That sleep can help with learning and memory is well established. What's more, the more people dream during the light sleep characterised by rapid eye movements (REM), the better they recall memories. But whether the specific content of dreams plays a role in this sleep-learning process wasn't clear.
To find out, Sidarta Ribeiro and André Pantoja of the Edmond and Lily Safra International Institute of Neuroscience of Natal in Brazil turned to the visceral, monster-filled, first-person shoot-'em-up game Doom.
Monsters and Chainsaws
Their team persuaded 22 volunteers – some Doom experts, some casual players, and mostly male – to spend two nights in their laboratory hooked up to an electroencephalograph (EEG), which measures the brain's electrical activity.
On the first night, the volunteers did not play any video games, but researchers woke them during REM sleep – the stage of sleep most associated with dreaming – and asked them to recall their dreams.
On the second night, volunteers played Doom for an hour just before bedtime. The next morning, Ribeiro's team roused the volunteers out of slumber and again catalogued their dreams. They also asked them to play Doom again.
To estimate the fraction of each person's dreams that were devoted to the video game, the researchers also asked volunteers to list things they associated with Doom and things they associated with the sleep lab in general.
Words like "blood", "monster", "chainsaw" and "shotgun" topped the Doom list, while the lab environment brought beds, pillows and electrodes to mind. Ribeiro's team then scoured the dream reports for these words to estimate the extent that the video game was intruding into the dreams.
On the second night, nearly everyone dreamed about Doom and about half dreamed about the laboratory. However, when the researchers compared players' performance before and after sleeping, only Doom-related dreaming was correlated with improvement at the game.
The researchers scored performance by analysing video recordings of the game and noting the number of player deaths, enemies killed, secret passages discovered, shooting accuracy and five other measures.
They found that the more Doom-dreaming a player did, the more likely he or she was to improve on these measures – but only up to a point. Volunteers who dreamed most about Doom didn't see the biggest gains.
"If you're too obsessed about something, you're dreaming about blood and monsters, you can't do well," says Ribeiro, who presented the work at the Society for Neuroscience meeting in Chicago earlier this month.
He says this doesn't invalidate the results. In fact, the same pattern has been seen in other measures of how a stimulus such as stress or caffeine affects learning: a little can help, but too much can hinder.
What's more, the EEG measurements suggest that volunteers seemed to be replaying their personal experience of the game during their REM sleep, a further indication that it might be involved in learning. Those with little experience of Doom activated brain areas that control their hands both while playing the game and while sleeping – and presumably dreaming. Meanwhile, more adept Doom players tended to activate frontal regions of the brain, associated more with higher thought and decision-making, during both play and sleep.
Further insights into dreaming and learning came from a presentation at the same conference from Stickgold and his colleague Erin Wamsley.
Instead of Doom, they asked volunteers to play a virtual maze game. After several practice sessions, volunteers took a nap, which was too short to allow REM sleep, described their dreams, and then played the game again.
The work is still preliminary, but Wamsley and Stickgold found that four volunteers who dreamed about the game solved the maze dramatically faster after their siesta, while those whose dreams didn't incorporate the maze improved little or not at all, just like control volunteers not allowed a nap.
The researchers say their study indicates that non-REM dreams also seem to accompany learning.
"I think Stickgold and Ribiero's reports do a very nice job of correlating dream reports with learning itself. I think that's the most direct evidence we've got," says Gina Poe, a neuroscientist at the University of Michigan in Ann Arbor, who caught both presentations in Chicago.