September 21, 2010


Sizing Up Consciousness by Its Bits (CARL ZIMMER, 9/20/10, NY Times)

“I love his ideas,” said Christof Koch, an expert on consciousness at Caltech. “It’s the only really promising fundamental theory of consciousness.”

Dr. Tononi’s obsession with consciousness started in his teens. He was initially interested in ethics, but he decided that questions of personal responsibility depended on our consciousness of our own actions. So he would have to figure out consciousness first. “I’ve been stuck with this thing for most of my life,” he said.

Eventually he decided to study consciousness by becoming a psychiatrist. An early encounter with a patient in a vegetative state convinced Dr. Tononi that understanding consciousness was not just a matter of philosophy.

“There are very practical things involved,” Dr. Tononi said. “Are these patients feeling pain or not? You look at science, and basically science is telling you nothing.” [...]

While in medical school, Dr. Tononi began to think of consciousness in a different way, as a particularly rich form of information. He took his inspiration from the American engineer Claude Shannon, who built a scientific theory of information in the mid-1900s. Mr. Shannon measured information in a signal by how much uncertainty it reduced. There is very little information in a photodiode that switches on when it detects light, because it reduces only a little uncertainty. It can distinguish between light and dark, but it cannot distinguish between different kinds of light. It cannot tell the differences between a television screen showing a Charlie Chaplin movie or an ad for potato chips. The question that the photodiode can answer, in other words, is about as simple as a question can get.

Our neurons are basically fancy photodiodes, producing electric bursts in response to incoming signals. But the conscious experiences they produce contain far more information than in a single diode. In other words, they reduce much more uncertainty. While a photodiode can be in one of two states, our brains can be in one of trillions of states. Not only can we tell the difference between a Chaplin movie and a potato chip, but our brains can go into a different state from one frame of the movie to the next.

“One out of two isn’t a lot of information, but if it’s one out of trillions, then there’s a lot,” Dr. Tononi said.

Consciousness is not simply about quantity of information, he says. Simply combining a lot of photodiodes is not enough to create human consciousness. In our brains, neurons talk to one another, merging information into a unified whole. A grid made up of a million photodiodes in a camera can take a picture, but the information in each diode is independent from all the others. You could cut the grid into two pieces and they would still take the same picture.

Consciousness, Dr. Tononi says, is nothing more than integrated information. Information theorists measure the amount of information in a computer file or a cellphone call in bits, and Dr. Tononi argues that we could, in theory, measure consciousness in bits as well. When we are wide awake, our consciousness contains more bits than when we are asleep.

For the past decade, Dr. Tononi and his colleagues have been expanding traditional information theory in order to analyze integrated information. It is possible, they have shown, to calculate how much integrated information there is in a network. Dr. Tononi has dubbed this quantity phi, and he has studied it in simple networks made up of just a few interconnected parts. How the parts of a network are wired together has a big effect on phi. If a network is made up of isolated parts, phi is low, because the parts cannot share information.

But simply linking all the parts in every possible way does not raise phi much. “It’s either all on, or all off,” Dr. Tononi said. In effect, the network becomes one giant photodiode.

Networks gain the highest phi possible if their parts are organized into separate clusters, which are then joined. “What you need are specialists who talk to each other, so they can behave as a whole,” Dr. Tononi said. He does not think it is a coincidence that the brain’s organization obeys this phi-raising principle.

Dr. Tononi argues that his Integrated Information Theory sidesteps a lot of the problems that previous models of consciousness have faced. It neatly explains, for example, why epileptic seizures cause unconsciousness. A seizure forces many neurons to turn on and off together. Their synchrony reduces the number of possible states the brain can be in, lowering its phi.

Dr. Koch considers Dr. Tononi’s theory to be still in its infancy. It is impossible, for example, to calculate phi for the human brain because its billions of neurons and trillions of connections can be arranged in so many ways. Dr. Koch and Dr. Tononi recently started a collaboration to determine phi for a much more modest nervous system, that of a worm known as Caenorhabditis elegans. Despite the fact that it has only 302 neurons in its entire body, Dr. Koch and Dr. Tononi will be able make only a rough approximation of phi, rather than a precise calculation.

“The lifetime of the universe isn’t long enough for that,” Dr. Koch said.

Enhanced by Zemanta

Posted by Orrin Judd at September 21, 2010 1:52 PM
blog comments powered by Disqus