Boston University Research Magazine 2007

The Conversation: Hearing and Speaking

Eavesdropping on the Conversations Between Neurons

Herbert Voigt, professor of biomedical engineering, also studies the ways in which our brain localizes sound. However, his interest lies in a part of the brain that interprets sound when our binaural hearing offers no clues, because the sound is equidistant from both ears in the medial plane either overhead or beneath us.

Here, “spectral notches” are the localizing information—the fact that thanks to the shape of our outer ear, energy is pulled out of a sound at certain frequencies, depending on its elevation. This spectral notch analysis is done in the dorsal cochlear nucleus, part of a brain-stem structure that is the first stop in the brain for information the auditory nerve has brought from the ear.

“When I was a graduate student,” Voigt points out, “the literature suggested that the dorsal cochlear nucleus was vestigial in humans, a meaningless leftover. Now we know it does spectral notch detection, and possibly much, much more. It may even be involved in the preliminary analysis of speech. Thanks to resonances of the vocal tract, there are spectral peaks and notches that characterize vowel sounds, too.”

Voigt's research uses gerbils to unravel the neural architecture of the dorsal cochlear nucleus, because gerbils, like humans, hear very well at low frequencies. Working with graduate student Han Zheng, Voigt has developed a computational model of the gerbil's dorsal cochlear nucleus that predicts the ways that key neurons will react when they are stimulated by particular sounds—as well as the ways different kinds of neurons converse with each other, either to excite or inhibit each other, depending on the qualities of the sound stimulus.

“At one time, the literature suggested that the dorsal cochlear nucleus was vestigial in humans. Now we know it does spectral notch detection, and possibly much, much more.”

Currently, Voigt is recording the ultrasonic vocalizations of gerbil pups calling for their mothers. These are then played back for adult gerbils, so Voigt and his team can record the neural activity caused by this behaviorally significant sound.

Working with graduate student Rapeechai “Pom” Navawongse, Voigt has developed a head stage implant that allows an electrode to be inserted into a single neuron in the dorsal cochlear nucleus without anesthesia or a decerebration that destroys the feedback from the upper brain. Voigt emphasizes that the neural responses of an awake and intact animal are likely to give a far more illuminating picture of the ways that the brain processes sound.

“Most studies of the central auditory system,” says Voigt, “have been in the anesthetized brain, so we really have no clue what's going on yet.”

For more information, see www.bu.edu/dbin/bme/faculty/?prof=hfv.