Our hearing organ, the cochlea, evidently poises itself at a Hopf bifurcation to maximize tuning and amplification. We show that in this condition several effects are expected to be generic: compression of the dynamic range, infinitely sharp tuning at zero input, and generation of combination tones. These effects are "essentially" nonlinear in that they become more marked the smaller the forcing: there is no audible sound soft enough not to evoke them. All the well-documented nonlinear aspects of hearing therefore appear to be consequences of the same underlying mechanism. PACS numbers: 87.19.Dd, 43.66. + y, 87.17.Nn The classic Helmholtz theory [1] posits that our hearing organ, the cochlea, is arranged like a harp or the back plane of a piano, with a number of highly tuned elements arrayed along a frequency scale, performing Fourier analysis of the incoming sound. Although the notion that the inner ear works like a musical instrument offers a beautiful esthetic symmetry, it has serious flaws. In the 1940s, Gold [2] pointed out that the cochlea's narrow passageways are filled with fluid, which dampens any hope of simple mechanical tuning. He argued that the ear cannot operate as a passive sensor, but that additional energy must be put into the system. As in the operation of a regenerative receiver [3], active amplification of the signal can compensate for damping in order to provide highly tuned responses. von Békésy's classic measurements in the cochlea [4] demonstrated the mapping of sound frequencies to positions along the cochlea. He observed the tuning to be quite shallow and found cochlear responses to behave linearly over the range of physiologically relevant sound intensities. Gold's notions were largely set aside in favor of the hypothesis of coarse mechanical tuning followed by a "second filter," whose nature was surmised to be electrical.von Békésy conducted his measurements on cadavers, whose dead cochleas lacked power sources or amplifiers that might have provided positive feedback. Only fairly recently, laser-interferometric velocimetry performed on live and reasonably intact cochleas has led to a very different picture [5,6]. There is, in fact, sharp mechanical tuning, but it is essentially nonlinear: there is no audible sound soft enough that the cochlear response is linear. Although the response far from the resonance's center is linear, at the resonance's peak the response rises sublinearly, compressing almost 80 dB into about 20 dB (Fig. 1). The width of the resonance increases with increasing amplitude, being least for sounds near the threshold of hearing. Observation of the response's essential nonlinearity at the level of cochlear mechanics contradicts von Békésy's finding. Furthermore, this nonlinearity does not originate in the rigidity of membranes or in fluid-mechanical effects. Because it reversibly disappears if the cochlea's ionic gradient is temporarily disturbed, the nonlinearity depends on a biological power supply [7].Gold conjectured that a regenerative mechanism for hearing...