Processing of Natural Sounds in Auditory Cortex

Thursday, March 11, 1999 - 10:30am - 11:15am
Keller 3-180
Israel Nelken (Hadassah Medical School)
In this talk, I will show that responses of neurons in primary auditory cortex of cats are modified strongly by weak acoustic components, provided the weak components appear in what I call, for lack of better characterization, natural settings. In one set of experiments, bird chirps extracted from natural recordings were used. These sounds contain, in addition to the main chirp, background components such as echoes. The main chirp was extracted from the natural recording using signal processing techniques. Subtracting the main chirp from the natural sound extracted the noise component of the sound. It is shown that the responses of the neurons to the full natural sound are much more similar to their responses to the noise component, presented alone, than to the main chirp, although the main chirp may be 20 dB stronger within the tuning curve of the neuron. Conversely, the responses to gaussian noise bands multiplied by a temporal envelope, mimicking the structure of a large class of natural backgrounds, are strongly affected by the addition of a weak tone at the neuron's BF, enabling the neuronal responses to signal the presence of the tone at lower levels than when the tone is added to non-modulated gaussian noise bands. This effect may underlie the psychophysical phenomenon of comodulation masking release.

These somewhat counterintuitive properties of neurons in primary auditory cortex are absent in a study of a secondary auditory field, Field AES, which has been implicated in sound localization. Neurons in FAES show spatial sensitivity to virtual space stimuli, and modifications of these stimuli cause predictable changes in the neuronal responses based on their spatial response profile. Thus, in this respect at least, primary auditory cortex is complex, whereas the secondary auditory field is simple.