A Complex Puzzle in Dynamic 3D

Neuroscientists are far from understanding how the brain works. Faulty assumptions are falling like trees in a timber-cutting contest -- but at least in neuroscience, scientists are generally free to do research without politicians-holding-pursestrings looking over their shoulders. And so we learn -- as science is supposed to do -- by testing and destroying the hard-won work of others who came before.

Researchers from the University of Maryland are unraveling more of the seemingly chaotic nature of the cerebral cortex:

All our knowledge of how the brain really works has been based on taking a small sampling of all available neurons and making inferences about how the other neurons respond, Dr. Kanold explains. "This is like showing someone who wants to know how America looks, 'Here is one person from New York City and one person from California.' You don't get a very good picture of what the country looks like from that sampling," says Kanold, originally from Germany.

In contrast, Kanold and colleagues were able to look at the activity of all the neurons in a large region of the auditory cortex simultaneously. To get the highest resolution picture to date of how auditory cortex neurons are organized, the researchers used a technique to fill neurons in living mice with a dye that glows brightly when calcium levels rise, a key signal that neurons are firing. They then selectively illuminated specific regions of the cortex with a laser and measured the neuronal activity of hundreds of neurons in response to stimulation by simple tones of different frequencies. _SD

Scientists from Germany, the UK, and the US worked together to unravel a small piece of the 3D dynamic brain function puzzle:

Tolias, who is also on the staff at with the Michael E. DeBakey Veterans Affairs Medical Center, said, "If you were to eavesdrop on the activity of a neuron in the visual part of the brain while a person is looking at a picture over and over again, the neuron will respond differently each time. In other words, a substantial part of the activity is unrelated to the picture itself. It is this activity that was believed to be common among many adjacent neurons because they are densely interconnected."

"Here is where problems begin to arise," Tolias said. "If the activity that is unrelated to the picture is common to many cells, it would build up from one stage of processing to the next, ultimately dominating brain activity and making information processing impossible -- a scenario called runaway synchrony."

To find an answer to this paradox, Tolias and his colleagues, including Alexander S. Ecker, the paper's first author who is a graduate student in Tolias' lab at BCM and the Max Planck Institute for Biological Cybernetics in Tübingen, Germany developed a new technology that allowed more precise measurement of action potentials. They found that the groups of neurons believed to be reacting in a related fashion actually had a weak relationship. They were reacting on their own, not dependent on each other.

"We measured correlations in awake, behaving primates, allowing us to have control of the experimental conditions. This gave us the chance to eliminate the possibility of a number of artifacts affecting our measurements," Ecker said. "For recording, we used chronically implanted multi-tetrode arrays -- a technique that offered us the chance to monitor many neurons at extremely high recording quality."

...The testing involved a variety of visual stimulation ranging from bars and grating to natural images. The groups of neurons tested were physically close to each other with highly overlapping receptive fields and all receiving strong common input.

One reason Tolias believes the neurons behave without correlation is to allow information to be sent through the brain in the most efficient way possible.

"Such a mechanism that allows the decorrelation might be a crucial prerequisite to prevent small correlations from accumulating and dominating network activity along the visual hierarchy," Ecker said.

The "decorrelated state" may also have other benefits, Tolias added. "Information processing in the brain is much easier if nerve cells' activity is uncorrelated. If one level of the hierarchy wants to know what the previous area is doing, it can simply forget about correlations in this case. Otherwise, it has to perform more complex computations to get to the same result." _ScienceDaily

Interesting looks at "uncorrelated activity" of neighboring neurons in both the auditory and visual cortices. It has been difficult to achieve this level of dynamic resolution in the past, when monitoring neuronal activity in the cortex.

The problem is one of scale: there are too many neurons in the cortex, too tightly packed together, to allow a complete monitoring of even a very small section of cortex. The finding would have been made long ago, if neuroscientists had only possessed a finer means of neuro-electrical monitoring.

Do we want to know how the brain works? If we want to build human-equivalent machine intelligences, we had better want to know. The various gung-ho projects to "reverse-engineer" the brain and to use "biomimetic" approaches to building machine brains, are all walking on foundations of air at this time. And if that is true, you know that the even more ad-hoc approaches being taken by most computer scientists and electrical engineers are even farther off the mark, and less likely to succeed in reaching human-equivalence.

Even the best of tools to study the brain's "connectome" are clumsily crude and inept. But how precise do we need to be? After all, human brains vary wildly in their knowledge, wisdom, intelligence, creativity, and competence. How can we know that we are "copying" a "better brain?" At this time we are not even close to being able to make that distinction.

As to the study described in the newsrelease above, Al Fin Neuroscientists can only say "I could have told them that!" And yet, science has to take these slow, plodding steps in order to make tomorrow's needful paths of study more clear.

If only climate science had been able to avoid a political takeover so early in its infancy. We might have avoided so much of the current popular and political insanity that has gripped Europe, North America, and Oceania. Better to be slow, plodding, and well-supported by observable data, than to be flamboyantly and Nobel Prize winningly wrong.

Modified 10Feb10 with additional content and editing.