Look Out, Darwin

Mind over Matter (2)


Mind over Matter (Part 1)


Which tastes better, Coke or Pepsi? Must be easy to decide--have a sip of Coke and a sip of Pepsi, and then decide. Of course some would choose Coke and some would choose Pepsi. Right? Wrong!


Strangely, most people cannot tell the difference, unless if they drink both, successively. That is, give someone a glass of the dark fizzy stuff and ask what is it. Quite likely, they will guess wrong. Give someone a glass of water and a glass of water with a few spoons of sugar dissolved in it—everyone will correctly identify which one is which. Since many cannot identify whether a drink is Coke or Pepsi, it means there is not much of a difference.


If you serve both drinks, the taster tastes the difference (minor) but finds it hard to identify or to state a clear preference. Suppose your friend John is an ardent Pepsi drinker. You pour him a cup of Coke and a cup of Pepsi and ask him which tastes better (John should like the Pepsi). Very likely he will get it right. Now do it a bit differently—you pour the drinks, but get Mary to serve it to John. Chances are, this time John will get it wrong. Why?


When you serve the drinks, you are performing a “single-blind test”. That is, you know which cup contains which brand of cola, but John does not. When Mary serves the drinks it is a “double-blind test”, that is Mary and John have no idea of the contents. The difference between results of single and double blind tests have been shown to be statistically significant. The server transmits subconscious cues to the victim.


Why do people have strong preferences for similar fizzy drinks labeled Coke and Pepsi? Why are taste tests so complicated? The answer to both issues lies in the complexity of the human mind.


Preferences are often a cultivated phenomenon, deeply embedded in the psyche. If a child is told by his or her parents that Coke is better than Pepsi, he or she will believe it, internalize it and then adhere to it for ever. The reality may never override this perception. This phenomenon transcends taste and encompasses most of our perceptions, likes and dislikes, behaviors, emotions and choices.


Double blind testing tried to eliminate many subtle cues that science cannot pinpoint, but exists in real life. The smile, the gesture and the demeanor, all are things we pick up even if we think we do not. A simple thing like preference turns out to be quite immensely complex to measure. Even if during the taste test, John picks Coke to be better than Pepsi, he will not change his mind. Next time he knowingly drinks Coke, it will taste horrendous to him.


We all watch is amazement as Olympic figure skaters pirouette on the ice. They glide, they swoop, they spin, they jump. They seem to fly and sail, they seem to defy the laws of gravity. How do they do it? How does a fast moving woman jump up and land into the arms of a fast moving man without falling, all on a surface where friction is non-existent? It is an un-quantifiable coordination of body and mind. (On a personal note, I once observed a young lady do a jump on the ice skating rink. It looked so simple that I tried it. About a millisecond after my feet left the surface, my entire body made intimate contact with the ice. Every part (including parts I did not know existed) of my body hurt. For two weeks.)


While most people can ride a bike, and consider it to be easy; those who cannot find it terribly difficult. You are on a bike, riding down a winding path. Other people are walking, an occasional dog is scampering across. As you speed up and glide along, you see everything, yet things are blurred. You notice various moving objects in front of you, but you do not hit them. The objects move unpredictably, the dog crossing the road stops and then starts walking again. Yet you do not hit the dog. Is that difficult to do? Are you really thinking about the possible mishaps and are you computing the time and motion coordinations needed for collision avoidance? You must be, but not consciously.


The brain can be trained to predict and perceive and act without the person actively thinking about it. The brain often does it quite right. Throw a ball into the air and a dog can run up to the point where it will land and then jump and catch it in mid air. It is a phenomenal coordination of time, motion, muscle control and body control. Making a computer-controlled arm catch a ball has been a challenge for scientists and engineers. The results have been a mixed bag. In 1998 a team from MIT built a robotic arm that could catch a ball, and found that the computation and physical dexterity required was quite complex. The contraption worked, but the average dog can do it better.


Experiments to make computers act like dogs seem like fun, but machines to be more cognizant of the physical environment is important. A major challenge is to enable aircraft to land under automated control. Statistically the most accident-prone segment of a flight is the landing. Landings under human control can lead to pilot error. Taking the human out of the loop can improve aviation safety.


Prototype aircraft that have landed without human control. However reliability of such technology still cannot be trusted with human lives. The best reported landing automation (by NASA, 2001) actually uses a human—a camera mounted on the unmanned plane sends pictures to a screen in front of a trained pilot. Electrodes wired to his arms and feet pick up his muscle movements and transmit them to the controls of the drone. The pilot watches the video and his body impulses actually land the plane.


About a million humans are capable of landing airplanes. Commercial, private and recreational flying generates millions of safe landings every day, but every one of them is handled by a real live human and not machines. Computers and instruments aboard a modern jetliner may navigate the plane to the destination, but when it comes in for the touchdown, a human takes over total control.


Landing a plane involves the management of a few key parameters—airspeed, glide slope, drift and alignment. First, the airspeed has to be right. We want to land at the slowest possible airspeed, a bit higher than the aircraft’s stall speed (the stall speed is the point when the wind separates from the wings and the plane drops fast). The glide slope determines how fast the plane descends and must be a line that leads the plane from its current position to the touchdown point. Drift is evil, it is caused by winds that keep pushing the plane off of the intended path. Finally the plane not only has to be traveling in the right direction, it has to be pointed right—planes can fly slightly sideways (due to drift) but they have to land perfectly aligned.


You are behind the controls of an aircraft as we make the final approach to the landing strip. As a pilot, you have two feet on the rudders, one hand on the yoke, one hand on the throttle and one hand on the trim. One eye must be planted on the aim point, constantly judging the glide slope, drift and alignment. The second eye must be glancing down to judge your height, and correlate it with the glide-slope indicator and altimeter on the dash. If you have a third eye, it should watch the engine RPM, the manifold pressure, and the vertical speed indicator. The fourth and the most important eye must be glued to the airspeed indicator to ensure the beast does not stall. A stall, close to the ground is very irritating to your friends; they will be obliged to attend your funeral.


To cut a long story of wrestling with wind, physics and machine short, a good coordination of your three hands and four eyes coupled with intense electrical pingings flowing though the nerves, twitching many a muscle in harmony, brings you flying over the runway, a few feet off the ground at a little above the stall speed. The incessant chatter on the radio distracts you, while you pull the nose up to stop the descent (a descending plane hitting the ground will break off the wheels.) As the plane levels out, a cushion of air (called ground effect) bounces it up towards the sky. Some of your hands compensate for the bounce while others compensate for the irritating drift. You move the throttle and the trim to help in the compensation; you have no idea which way you moved them. Unknown to you, your feet are working overtime, managing alignment. You raise the nose up and the airspeed drops to the verge of a stall, just a few inches of the ground. The imminent stall reduces lift and the plane drops to the ground. Through this intense maneuver (about 2 seconds), you did not have time to think: the body was reacting to commands from the brain, with the mind turned off.


[To be continued]


Partha Dasgupta is on the faculty of the Computer Science and Engineering Department at Arizona State University in Tempe. His specializations are in the areas of Operating Systems, Cryptography and Networking. His homepage is at http://cactus.eas.asu.edu/partha




Partha Dasgupta