杏吧原创

GRADUATE student Sam Hincks sticks a wet electrode to my forehead, then another, tucking them in place under a black Tufts University sweatband.

鈥淎re you nervous?鈥� he asks.

I am, but I don鈥檛 want to lose my cool. 鈥淎 little,鈥� I say.

Hincks flips the switch. It takes a moment, then I feel a slight, sharp tingle, crashing in waves somewhere just out of my line of sight. One milliamp of current is flowing between the electrodes 鈥搕hrough my brain.

The little zap is called transcranial direct current stimulation (tDCS) and I am in lab, up on the fourth floor of a Tufts University research building in Medford, Massachusetts. The researchers are exploring the possibilities for computers and wearable devices to read what鈥檚 going on in the brain and stimulate it in specific ways.

Jacob鈥檚 lab is dedicated to improving the relationship between humans and machines, finding a way for one to communicate more easily with the other. He imagines the fluctuating stress and thoughts of the brain as a dial: if you want to let a computer know how you are feeling, you could manually turn a knob up or down, or you could find a way for the computer to tune into those changing states automatically. Perhaps the computer might even start turning the dial itself.

鈥淚 think of the human and the computer as two powerful information processors connected by a narrow channel,鈥� says Jacob. 鈥淥ur goal is to improve the bandwidth between the two.鈥�

鈥淥ur goal is to improve the bandwidth between two powerful processors: the human and the computer鈥�

To get information out of the brain, Jacob鈥檚 lab relies on a technique called functional near infrared spectroscopy (fNIRS). Tack two sensors onto the forehead and shine harmless red light through a few centimetres of skull and skin. The light is absorbed and scattered by blood in vessels at the brain鈥檚 surface. The amount that bounces back to the sensors is a proxy for the oxygen levels in the brain. High oxygen means high activity, a sign that you鈥檙e thinking hard.

The team has already used this system to enable a computer to track and adjust to a person鈥檚 cognitive state. One recent device sends the oxygen levels to a Google Glass. If it judges the user鈥檚 brain to be busy, it holds off sending any notifications until activity levels die down. Another system follows the progress of novice piano players as they plunk their way through a new piece. It ramps up the difficulty of the song when the players鈥� workload dips below a certain threshold, indicating they鈥檝e mastered a section.

Such systems mean the world can start adapting to the brain鈥檚 ability to cope with it. When someone tries one of these devices for the first time, a machine-learning algorithm steps in to calibrate the sensors for their brain. This takes a while, and is one of the barriers to consumer adoption. When I try fNIRS, I spend 5 minutes doing simple mental arithmetic while Hincks gets calibrating. By the end, he says the computer has learned enough to predict my cognitive workload with 75 per cent accuracy.

The device can tell if a person is working hard or cruising. But as computers move onto our foreheads and arms (see 鈥�Arm hacking鈥�), what if they took the next step, giving us a little zap when we seem to be struggling? That鈥檚 where tDCS comes in. Jacob wants to use it to tune the brain for the task at hand.

It is simple and cheap to set up: aspiring biohackers could make their own tDCS devices for about $20 using instructions off the internet. Just place spongy electrodes, wet with salt water, on the head, then run current from a 9-volt battery through them. The idea is that the electricity will change the excitability of some neurons, making them more or less likely to fire. The technique has already been studied as a treatment for depression, strokes, and even tinnitus. Jacob鈥檚 lab wants to use it to interact with our devices.

His team鈥檚 first goal is to understand how different people respond to tDCS. 鈥淲e think that people who have more of a response as measured by fNIRS would be more helped by stimulation,鈥� says Hincks.

After that, the first test for tDCS might involve flying virtual drones. Jacob鈥檚 lab works with a simulation which puts the player in control of a number of imaginary UAVs, each of which needs to be steered around obstacles to its target. In early tests, players were fitted with the fNIRS sensors, and the computer added or removed drones from their control according to their cognitive workload.

鈥淭he computer adds or removes drones from the player鈥檚 control according to cognitive workload鈥�

With tDCS, the computer could give the user a zap when it senses a dip in their abilities, adapting the user鈥檚 brain to their task, rather than the other way round.

鈥淲e want to just crank it up for a minute or two and then crank it down. We鈥檙e looking for this very fine-grained control,鈥� Jacob says. 鈥淲e鈥檙e looking to measure you with fNIRS and, based on what we measure, slowly tweak this. It鈥檚 a sort of two-way communication with the brain.鈥�

Roi Cohen Kadosh, a cognitive neuroscientist at the University of Oxford, cautions that tDCS may not offer a boost to everyone. In a study , he and his colleagues stimulated the brains of people who had high levels of anxiety about mathematics. For them, the stimulation seemed beneficial: their reaction times on simple arithmetical problems improved and they had less cortisol in their saliva, a sign of lower stress. But when a group with low anxiety about mathematics tried to solve the same kinds of problems after tDCS, their performance actually got worse. Similarly, he says, other groups of people may not get an edge from tDCS.

Baby aspirin stage

鈥淭hose with high cognitive abilities might not benefit from stimulation. They might even show impairments,鈥� Cohen Kadosh says.

And not everyone is convinced of the current鈥檚 power. Jared Horvath and colleagues at the University of Melbourne in Australia have reviewed the results of hundreds of studies involving tDCS, and found that its reported benefits 鈥� increased speed when completing tasks, higher accuracy, better memory 鈥� were inconsistent.

, a biomedical engineer at the City College of New York, who studies electricity鈥檚 effect on the body, says there are some essential questions scientists must answer before tDCS becomes widespread: what brain region should be stimulated and at what strength; and is stimulation better before, during or after an activity?

鈥淲e鈥檙e in the 鈥榖aby aspirin鈥� stages of tDCS,鈥� says Bikson. 鈥淲e have a tremendous amount to learn about how to optimise it.鈥�

So tDCS won鈥檛 be out in the real world just yet. But if it can get there, it may usher in an era where not only do our devices adapt to us, we adapt to them.

(Image: John Lund/Superstock)

This article appeared in print under the headline 鈥淢ind adapts to machine鈥�