ASA 129th Meeting - Washington, DC - 1995 May 30 .. Jun 06

5aPP5. Cortical codes for sound location.

John C. Middlebrooks

Depts. of Neurosci. and Otolaryngol., Box 100244, Univ. of Florida, Gainesville, FL 32610

Ann Clock Eddins

Indiana Univ., Bloomington, IN 47405

Li Xu

David M. Green

Univ. of Florida, Gainesville, FL 32611

The responses of auditory cortical neurons carry information both in the number of spikes within a burst and in the timing of those spikes. The capacity of single neurons to encode sound locations was studied in areas A1 and AES of (alpha)-chloralose-anesthetized cats. The spike counts of about half of studied neurons were modulated by more than 50% as a noise source was varied in azimuth. Nevertheless, the ``best areas'' of most units, using a 50%-of-maximum-response criterion, were larger than 180(degrees). Temporal spike patterns of single neurons could be classified according to sound source azimuth, using an artificial neuron network. The classification of temporal patterns consistently surpassed a maximum-likelihood classification of spike counts. The spike patterns of single neurons could successfully encode locations throughout 360(degrees) of azimuth. The localization performance of any single neuron was substantially worse than that of an awake, behaving cat. When responses were averaged across multiple trials, however, performance improved monotonically with increases in the number of trials, suggesting that the information carried out by the temporal spike patterns of multiple neurons could account for a cat's localization behavior.