Brain-controlled games and other devices should soon be on sale

yujian623

Mar 15th 2007 From The Economist print edition


How would you like to rearrange the famous sarsens of Stonehenge just by thinking about it? Or improve your virtual golf by focusing your attention on the ball for a few moments before taking your next putt on the green-on-the-screen? Those are the promises of, respectively, Emotiv Systems and NeuroSky, two young companies based in California, that plan to transport the measurement of brain waves from the medical sphere into the realm of computer games. If all goes well, their first products should be on the market next year. People will then be able to tell a computer what they want it to do just by thinking about it. Tedious fiddling about with mice and joysticks will become irritants of the past.



Controlling things by mere thought is a staple of science fiction. That fiction, though, is often based on a real technique known as electroencephalography (EEG). This works by deploying an array of electrodes over a person's scalp and recording surface manifestations of the electrical activity going on under his skull.



At the moment, EEG's uses are mostly medical. Though the output of the electrodes is a set of crude brain waves, enough is now known about the healthy patterns of these waves for changes in them to be used to diagnose unhealthy abnormalities. Yet, because parts of a person's grey matter exhibit increased electric activity when they respond to stimuli or prepare for movements, there has always been the lingering hope that EEG might also manifest someone's thoughts in a machine-readable form that could be used for everyday purposes.



To realize that hope means solving two problems—one of hardware and one of software. The hardware problem is that existing EEG requires a helmet with as many as 120 electrodes in it, and that these electrodes have to be affixed to the scalp with a gel. The software problem is that many different types of brain waves have to be interpreted simultaneously and instantly. That is no mean computing task.



The power of thought



Both Emotiv, which is based in San Francisco, and NeuroSky, of San Jose, think they have cracked these problems. Emotiv recently unveiled a prototype headset that has a mere 18 electrodes. Moreover, no gel is needed for these electrodes to make a good contact with the headset-wearer's scalp. Emotiv claims that its system can detect brain signals associated with facial expressions such as smiles and winks, different emotional states such as excitement and calmness, and even conscious thoughts such as the desire to move a particular object. It will not say precisely how this trick is done, but it seems to work well enough to make a virtual character in a game mimic a player's own facial expression, as well as permitting that player to move things around just by thinking about it.



To encourage others to add to this bag of tricks, Emotiv has also created a special kit for software developers. This should enable programmers to rewrite their code in a way that will allow Emotiv headsets to be plugged into existing applications.



According to Nam Do, Emotiv's boss, those applications are most likely to be single-player computer games running on machines such as Microsoft's Xbox 360 and Sony's PlayStation 3. In the longer term, though, he thinks the system will be ideal for controlling avatars (the visual representations of players) in multiplayer virtual worlds such as Second Life.



For Stanley Yang, the boss of NeuroSky, even 18 sensors seems too clunky. His firm's technology has reduced the brainwave pickup to the minimum specification imaginable—a single electrode. Existing versions of this electrode are small enough to fit into a mobile phone and Mr Yang claims they will soon be shrunk to the size of a thumbnail, enabling people to wear them without noticing.



Reducing the mind-reader to this bare minimum makes it cheap—about $20, compared with several hundred for Emotiv's headset—though it is not as precise. But that lack of precision may not matter. According to Klaus-Robert Müller, a computer scientist at the Fraunhofer Institute in Berlin who has been studying the problem for years, a single well-placed electrode is sufficient to gather meaningful information from brain waves. On the other hand, Dr Müller and his team have been unable, as yet, to produce a device that works well outside the cosseted environment of a laboratory.



There are several reasons for this. First, although human brains are similar to one another in general, they are different in detail, so a mass-produced headset with the electrodes in standard locations may not work for everyone. Second, about one-third of the population is considered “illiterate”, meaning in this context that not even a full-fledged medical EEG can convert their brain activities into actions. Third, electrical signals generated by muscular activity such as blinking are easily confused with actual brain-wave readings. Wink at a fellow player at the wrong moment, then, and you might end up dropping that sarsen you have lifted so triumphantly from the fields of Salisbury Plain on the toes of your avatar's foot.