Physiological mechanisms of frequency discrimination

Abstract

I recall a conversation with Bob Bilger that occurred around 1971, when I was a graduate student in Bioacoustics at Pitt. He said (translated into English from the original Bilgerian), ‘‘Did it ever occur to you that you can apply principles of signal detection theory in physiological studies? That way, you could examine neural and psychophysical performance using comparable measures and possibly shed light on underlying mechanisms.’’ This idea, which predated its implementation in auditory research by at least 10 years, was initially lost on me. Much later, I recognized the possibilities and performed some pertinent experiments with various colleagues. We utilized 2AFC adaptive tracking procedures to investigate performance limits of single cat auditory nerve fibers and neural populations on pure-tone frequency discrimination tasks. For single-fiber responses, we found that: (1) decision strategies based on detecting differences in spike counts fail to account for perceptual findings by wide margins; and (2) strategies based on detecting differences in phase-locked activity produce data that match perceptual findings well at low frequencies but rapidly degrade at frequencies >1500 Hz. Examining neural population performance using a stochastic excitation pattern model, we found that a strategy which simply estimates the spatial location of the response peak accounts well for perceptual performance at most frequencies. See, Bob? Your idea works, and I WAS paying attention.