Brain can be trained to process sound in alternate way, study shows

UCSF scientists have found that the brains of rats can be trained to learn an alternate way of processing changes in the loudness of sound. The discovery, they say, has potential for the treatment of hearing loss, autism, and other sensory disabilities in humans. It also gives clues, they say, about the process of learning and the way we perceive the world.

“We addressed a very fundamental question,” says Daniel B. Polley, PhD, lead author of the study.  “When we notice a sound getting louder, what happens in our brain so that we know it’s getting louder?”

Polley is a postdoctoral research fellow in the laboratory of senior author Michael M. Merzenich, PhD, co-director of the Coleman Memorial Laboratory in the UCSF Keck Center for Integrative Neuroscience and UCSF professor of otolaryngology.

The study was published recently in Proceedings of the National Academy of Sciences (November 16, 2004).

“This is a very old idea,” Polley notes. “How to relate the bigness of a stimulus to the bigness of its internal representation in the brain.” Over the centuries, philosophers and scientists have put together a picture of how our brains model the world through the mechanism of our senses. Physical stimuli such as light, sound, and touch are converted by our sensory organs—eyes, ears, and skin—into electrical signals, which are processed by neurons in different areas of the brain. As those neurons fire, we see, hear, and feel. When the light or sound changes in intensity, our neurons fire faster or slower in direct ratio to the change. That ratio varies depending on the sense involved, but is constant for each sense: the louder a sound, the faster the neurons in the auditory cortex fire.

But now that picture has changed. Polley trained two groups of rats to become ” experts” at discriminating between very small differences in loudness—an ability that untrained rats do not have. He then looked at how the expert rats processed changes in loudness compared to two groups of untrained rats, and found that the auditory cortex in the expert rats contained groups of neurons that had become selective for specific volume levels—they fired only at those levels and were quiet otherwise. This physiological change in the brain, called “plasticity,” has been widely observed in humans and animals who have learned new skills.

Then came the breakthrough discovery: the expert rats were processing volume changes in a new and different way. In the brains of the untrained rats, the overall neural response rate increased as the sound got louder and louder, as the classical model would predict. In the expert rats, however, the overall response rate of the selective neurons increased until the sound reached a loudness threshold of 40 decibels—and then leveled off while the loudness increased 100-fold, from 40 to 80 decibels. “At first glance, this was not good,” observes Polley: If their neurons were not increasing their firing rate, how were the expert rats registering the increase in volume? David T. Blake, PhD, UCSF assistant research physiologist and a co-author of the study, cracked the puzzle. Instead of looking for a simple increase in firing rate, Blake measured the rate at which the firing changed, either up or down. This rate turned out to be in exact proportion to the increase in volume—and at the same ratio as the firing rate increase. Tests confirmed that the untrained rats’ brains were not registering volume increases in this new way; it had been learned by the expert rats as they became better at discriminating changes in volume.

Polley concludes, “There is still proportionality between response strength in the brain and the stimulus. But now neurons are much more selective, and can represent sound intensity with decreasing firing rates as well as increasing firing rates.” This system is “optimal” for representing subtle changes in loudness, reasons Polley, because “it gives you two directions to change through,” making it many times more responsive than a simple firing rate increase. “And it becomes optimized through learning.”

The discovery has several implications. From a practical viewpoint, “I think it has quite a bit to offer,” says Christoph E. Schreiner, UCSF professor of otolaryngology and a co-author of the study. In particular, it might present a technique for retraining people with partial hearing loss, who often cannot hear very soft sounds but have normal hearing at higher volume levels.

“There’s a very steep volume curve that goes from soft to very loud right away, and people have a hard time with that,” Schreiner explains, “especially for hearing-aid users.” However, they—or their auditory cortexes—might be trained to be more sensitive to minor volume changes at the lower threshold of hearing, “so this steep transition doesn’t bother them anymore.” Similarly, such training might be of value to profoundly hearing-impaired people with cochlear implants, which replace the function of the inner ear but are not as sensitive to small volume changes.

Another group that might be helped is children with sensory-modulation disorders, including children with autism. These children are “overwhelmingly sensitive” to changes in their environment, explains Polley. “So when presented with a moderate stimulus—a sound, a touch, a flash of light—they respond as if their entire sensory systems have become overwhelmed. What might be needed in their brains is greater selectivity.” Potentially, they could be trained to distinguish smaller degrees of change in their environments. Being perceived as gradual, these changes would be less overwhelming.

From a psychological viewpoint, the study says something about how we acquire and refine new skills. When we speak of training a musician’s ear or a painter’ s eye, speculates Polley, we may be referring to the alternate sensory processing system employed by the expert rats. “This is implicit learning,” he says. “How do we learn the skills that distinguish one tradesman from another tradesman? These processes are undoubtedly operating in these types of learning behaviors, and they most likely are responsible for expertise. We are looking at the neural substrate for these lifelong learning processes.”

Another co-author of the study was Marc A. Heiser, a student in the MD/PhD Neuroscience Graduate Program at UCSF.

The research was funded by the Coleman Fund, the Sooy Fund, grants from the National Institutes of Health, and a National Research Service Award Fellowship.