14 September 2006

BERKELEY – While it is widely accepted that the output of nerve cells carries information between regions of the brain, it's a big mystery how widely separated regions of the cortex involving billions of cells are linked together to coordinate complex activity.

A new study by neuroscientists at the University of California, Berkeley, and neurosurgeons and neurologists at UC San Francisco (UCSF) is beginning to answer that question.

UCSF neurosurgeons place 64-electrode grids on the surface of the brain's temporal and frontal lobes to locate regions where epileptic seizures originate. These grids allowed UC Berkeley neuroscientists to study the interaction of brain waves during simple tasks, such as word recognition or hand movements. (Images courtesy the Knight Lab)

"One of the most important questions in neuroscience is: How do areas of the brain communicate?" said Dr. Robert Knight, professor of psychology, Evan Rauch Professor of Neuroscience and director of the Helen Wills Neuroscience Institute at UC Berkeley. "A simple activity like responding to a question involves areas all over the brain that hear the sound, analyze it, extract the relevant information, formulate a response, and then coordinate your lips and mouth to speak. We have no idea how information moves between these areas."

By measuring electrical activity in the brains of pre-surgical epilepsy patients, the researchers have found the first evidence that slow brain oscillations, or theta waves, "tune in" the fast brain oscillations called high-gamma waves that signal the transmission of information between different areas of the brain. In this way, the researchers argue, areas like the auditory cortex and frontal cortex, separated by several inches in the cerebral cortex, can coordinate activity.

"If you are reading something, language areas oscillate in theta frequency allowing high-gamma-related neural activity in individual neurons to transmit information," said Knight. "When you stop reading and begin to type, theta rhythms oscillate in motor structures, allowing you to plan and execute your motor response by way of high gamma. Simple, but effective."

The findings are reported in the Sept. 15 issue of Science.

Tuning in high-frequency brain waves

The researchers found that when people are asked to do a simple task, such as listening to a list of words, the slow, theta oscillations in the hearing area of the brain become coupled with the fast, high-gamma oscillations in the same area. When two different brain areas then oscillate together at the same theta frequency and phase, it becomes much easier for these regions to tune in the high-gamma oscillations that transfer information between them.

Electrodes on the brain surface detect high-gamma activity (star) in only one area when the patient hears a tone (top), while more areas exhibit high-gamma activity in response to a recognizable phoneme (middle) or word (bottom). Theta waves in these active areas lock onto the same frequency and phase, tuning in high-frequency signals tranferring information.

"One theory about how the brain is organized says that there is a hierarchy of oscillations that can control how one neuron talks to another neuron, or how one brain area talks to another brain area," said lead author Ryan Canolty, a UC Berkeley graduate student in the Helen Wills Neuroscience Institute. "Our study was designed to test the idea that the high-frequency oscillations generated by the brain are coupled to the slower theta oscillations.

"This coupling is important because the two rhythms have different functions and operate on different spatial scales - a patch of high frequency activity is very localized, about the size of a dime or smaller on the brain, and is associated with bottom-up sensory or motor processing, while the theta rhythm is much more spatially widespread, the size of a silver dollar or larger on the surface of the cortex, and is tied to top-down executive processes like attention and memory. Coupling between these two rhythms could be what gives the brain a way to connect low-level perceptions and actions to high-level goals and intentions."

Brain waves - such as the slow alpha waves of the relaxed or idling brain or the fast, seemingly random pulses accompanying dream sleep - are generated by coordinated firing of neurons in the brain triggered by waves of excitability that wash over an area. Waves of excitability in the theta range of oscillations lower the threshold for neuron firing, making it more likely that input arriving at the critical time will make neurons in that area of the brain fire.

Typically measured with electrodes on the scalp (electroencelphalograms, or EEGs), brain waves are classified from the very slowest delta waves (1-3 oscillations per second) seen in very deep sleep, through theta (4-7 oscillations per second), alpha (8-13 oscillations per second) and beta (14-30 oscillations per second) to the most rapid firings in the human brain - gamma oscillations (30-60 oscillations per second). Because slow firings are detected when the brain is least active, while rapid firings accompany activity, neuroscientists think that information in the brain is carried by the high frequencies.

Until recently, scalp recordings could detect gamma waves only up to 70 firings per second, but in 1998 researchers at Johns Hopkins University discovered brain waves up to 100 oscillations per second by placing electrodes directly on the surface of the brain. Knight and his UC Berkeley group used the same technique to show last year that brain oscillations can occur up to 200 times per second - and perhaps as fast as 300 times per second. Waves with 80-200 oscillations per second are called high-gamma, though they likely play an entirely different role from the traditional lower frequency gamma waves, Knight said.

One theory is that brain oscillations organize neurons into cooperating groups: low-frequency waves synchronize the firing of large groups of neurons, while the higher frequencies synchronize smaller groups. Though neuroscientists don't know what underlying neural activity generates the waves recorded on the surface of the cortex, the oscillations may be generated spontaneously by neurons when grouped together in the hundreds of thousands.

"When you have to remember a new phone number or attend to moving cars as you cross an intersection, you'll have an increase in the strength of the theta rhythm in many different brain regions," Canolty said. "The idea is that this theta rhythm might be more of an executive control mechanism to tie different brain areas together, whereas high-gamma waves within a region tie groups of cells together and time when their output can be sent or when input from another area can be received."

Recording brain waves in epilepsy patients

To test these hypotheses, Knight and his colleagues teamed up with UCSF doctors Nick Barbaro, neurosurgeon, professor of neurosurgery and director of surgical epilepsy; Mitchel Berger, neurosurgeon, professor and chair of neurosurgery and director of the Brain Tumor Research Center; and Heidi Kirsch, epileptologist, assistant professor in residence of neurology, to record brain activity in brain tumor and epilepsy patients scheduled for surgery to remove a portion of their brains. The epilepsy patients typically have brain activity measured up to a week beforehand so that surgeons can localize important areas they need to avoid, such as centers of language, vision or motor activity.

The goal of the UC Berkeley-UCSF Intracranial Project is to use high-gamma waves to produce a finer map of the brain to guide neurosurgeons during brain surgery and potentially to use these same high frequency oscillations to control robotic devices in paralyzed patients.

"This represents a paradigm shift in how we map brain function," said Berger. "As a neurosurgeon, I work within these very complicated cortical and subcortical areas where regions talk to each other, and there is a level of connectivity between cortical regions not apparent by any current means of detection.

"By measuring high-gamma band activity, we will be able to see in real time, during surgery, how cortical regions are connected through subcortical systems, allowing us to understand how these regions process information. This hold the key to understanding diseases like autism, which clearly involves the subcortical pathways."

In these clinical procedures, Barbaro removed a portion of each patient's skull and placed a grid of 64 electrodes on the surface of the brain's frontal and temporal lobes to precisely localize the source of the seizure so it could be removed in a subsequent surgical procedure. Knight, Canolty and their UC Berkeley colleagues then recorded activity in response to sounds and visual stimulation.

Several hours of data from five different patients revealed that high-gamma activity was locked to the theta rhythm in many different areas of the brain. The stronger the theta wave, the stronger the coupling to high-gamma oscillations.

The pattern of coupling between theta and high-gamma also changed with the task. Patients listening to a list of words would show strong coupling in a particular set of brain regions, but when they then had to name pictures, a different set of brain areas would show strong coupling.

"We used to think that one little patch of cortex takes care of this function and another little patch takes care of that function, but now we see it's more about systems that are cooperating on one task, then they switch over and cooperate on another task," said UCSF's Kirsch. "On the fly you want to link these areas to do a task, and when the task is over, you want to decouple them and let them link up with someone else. Ryan has shown that the theta waves allow this coupling and uncoupling by locking into phase."

Knight and his colleagues continue to probe the connection between waves of different frequency in the brain and are building a more closely spaced grid of electrodes that can measure finer detail on the brain's surface. They also plan to combine brain grid recordings with recordings from individual neurons in the cortex to find out what really generates the brain waves that EEGs and ECoGs measure.

Coauthors of the Science paper include UC Berkeley graduate students Erik Edwards and Maryam Soltani of the psychology department and S. S. Dalal of the bioengineering department; and radiologist Sri S. Nagarajan of UC Berkeley's Department of Bioengineering and UCSF's Department of Radiology.

The work was supported by the Rauch family and by the National Institute of Neurological Disorders and Stroke and the National Institute on Deafness and Other Communication Disorders of the National Institutes of Health.