Tim Haresign
James A. Simmons
Michael Ferragamo
Hunter Lab., Brown Univ., Providence, RI 02912
Traditional views of neural processing of auditory signals in the nervous system have been shaped by conceptual models based on Fourier analysis---in the frequency domain for spectral selectivity or in the time domain for periodicity analysis. However, in recent years wavelet analysis has been proven to be a powerful new technique for signal processing. Evidence has been found for a wavelet-like representation of ultrasonic auditory signals in the inferior colliculus of the big brown bat, Eptesicus fuscus. The inferior colliculus is a laminar structure, with the neurons in each layer being sensitive to a specific, narrow frequency band. In response to a broadband stimulus the averaged activity of small groups of neurons exhibits a structured temporal pattern in its response. The temporal response pattern is dependent on the spectral-temporal structure of the input sound, and varies as a function of the frequency lamina from which the measurements are taken. Based on these results, it seems likely that information about ultrasonic signals is being segregated according to frequency and then transformed onto a different time scale within each frequency lamina as a wavelet-like burst, so that the time and phase relations within ultrasonic signals can be displayed by time and phase relations on a corresponding expanded physiological scale. [Work supported by ONR Grant No. N00014-89-J-3055, NIMH Grant Nos. MH-00521 and MH-19118, and NIH Grant No. DC-00511.]