Find out how Berkeley professors and researchers are in the forefront of cutting-edge research and promising new discoveries. In ScienceMatters@Berkeley, we showcase scientific research taking place in the College of Letters & Science and the College of Chemistry.
Pulling a Voice out of a Crowd
You’re walking down a city street, and the din is deafening. Car engines roar, bus brakes shriek, and pedestrians are shouting just to be heard. Then, amid the racket, you hear it: someone speaking your name. Berkeley professor of physics Michael DeWeese is studying this remarkable ability to pay attention to certain sounds. His findings promise to improve both hearing aids and hands-free device interfaces in the future.

Michael DeWeese is also a member of the Helen Wills Neuroscience Institute. Photo credit: Vivek AyerThe next time you attend a packed party, take a moment to appreciate the brain’s remarkable listening powers. Its ability to focus in on a particular conversation, while tuning out the surrounding din, is something no artificial device can emulate.
“It’s like when you focus on one voice at a cocktail party,” says Michael DeWeese, a Berkeley professor of physics. “Your brain has top-down executive control that can direct your attention to sounds you want to focus on despite all the distracting sounds in your environment.” DeWeese is working out the neurological mechanisms behind selective auditory attention. His findings could someday generate hearing aids that filter out noise efficiently, but also inspire a whole new generation of voice activated devices.Trained in particle physics, DeWeese might seem an unlikely sort of brain researcher. He chose to study neuroscience, he says, because “I could do both theory and experiments in the same laboratory; it doesn’t require hundreds of people to do a single cutting edge experiment as it often does in particle physics. Also, the questions to me are the most compelling in all of science: discovering how the brain works.”
Auditory attention research, says DeWeese, offers an extraordinary window into the mind. “Attention is probably the best experimental handle we’ve got on consciousness. If I can get some behavioral readout that tells me an animal is paying attention to something, and I can see how its neurons are processing that signal, I can hope to figure out what the neural substrate is for our conscious experience,” DeWeese says.
In his laboratory, DeWeese trains rats to listen for certain sounds while disregarding others. Each subject rat is placed in a chamber punctuated by a row of three small portholes. Then it must respond to sounds such as a high tone versus a low one, or noise from the left hand speaker versus from the right.

Photo courtesy Wikipedia

DeWeese might reward the rat for poking its nose into the right hand porthole when it hears a high tone, and into a left hand porthole when it hears a low tone.  In the first set of trials, most of the sounds might come from a speaker placed over the rat’s head, with sounds coming from a speaker behind the animal on a few rare trials. Rats that don’t pay attention to the right location in space are more likely to err and get a “time out,” forcing them to wait before they can initiate the next trail. For the next set of trials, most sounds might come from the speaker placed behind the rat’s head, requiring the animal to shift its focus of attention to a different location in its environment. “That way we can see changes in the way the brain processes the same sounds coming from the same location in space depending on whether it’s paying attention to them or ignoring them,” DeWeese says.
The brain is thought to modulate attention by altering neural behavior. Just as aspirin can increase the amount of stimulus required to make a neuron pass along pain messages, neuromodulator molecules such as acetylcholine can make some neurons more or less likely to relay information about sound stimuli. “There is some change in the internal cell processing of signals,” DeWeese says. “In addition,  the transmission of sensory information is gated at the circuit level.” These changes likely occur within many of the neurons in a given circuit, and to different degrees in different brain regions.DeWeese records from neurons that are part of these circuits to determine the roles they play in propagating information. Eventually he plans to observe how such single neurons react during behaviors requiring shifts in attention.
The theoretical arm of DeWeese’s research program includes determining what aspects of sounds are most important to the brain, and how that information is encoded. For example, he is investigating how the mathematical structure of sounds recorded in nature differs from unstructured white noise. With this information, he seeks to predict what types of sounds should elicit the strongest responses from auditory cortical neurons, in a manner analogous to the way some neurons in the visual cortex respond most strongly to seeing edges. His group is also developing powerful machine learning algorithms to fit mathematical models of his neural data and testing whether they emulate actual brain behavior.
Encoding sound efficiently, and ignoring those deemed unimportant, offers strong evolutionary advantages. “It allows the brain to use those operations in a dynamical, smart way. You don’t want to waste your sensory processing resources on sounds that don’t matter,” DeWeese says.
Understanding how the brain normally focuses on sounds could help scientists identify anomalies in those who have difficulty focusing their attention, such as patients with schizophrenia and attention deficit hyperactivity disorder (ADHD). DeWeese’s findings could also contribute to the design of hearing aids and hands-free devices that will respond to nearby voices, and deemphasize background noise. In this way, his work could help even the hard of hearing appreciate a rousing party again.

Related Web Sites

• Mike DeWeese, Physics Department
• Michael DeWeese, Helen Wills Neuroscience Institute
• “Michael DeWeese: One Physicist’s View of the Cerebral Cortex,” lecture video, Physics Department

A Taste of Andean Culture
Of all of the advances people have developed over the millennia, food plants may be the most important. Humans domesticated wheat, rice, and potatoes; learned to cook and store them; and shipped these staples far and wide to enrich themselves and extend cultural influence. By examining the plant remains on early settlements, Berkeley professor of anthropology Christine Hastorf pieces together how ancient peoples worked, ate, traded and worshiped.
The Jekyll and Hyde Act of Oncogenes
Pull one strand of a spider’s web, and the rest of the silken construction will flex and shake with the strain. The same effect can be seen in the network of genes involved with cancer, says Kunxin Luo, a Berkeley professor of cell and developmental biology. Her work with the protein TGF-beta is laying bare pathways to some of cancer’s biggest culprits.