Several VLSI models of cochlear processing are available at present. Most are difficult to design accurately and have a fairly limited dynamic range, which may be a hindrance in many applications. To circumvent these problems, we plan to test new designs due to Rahul Sarpeshkar of Bell Labs that incorporate gain-control circuitry that yield scale-invariant discriminability at the input, independent of signal level [. Sarpeshkar.]. The auditory system operates with laws that approximate such behavior closely due to the effects of the outer hair cells and the olivocochlear efferent system. We will investigate different cochlear circuits and architectures with respect to their robustness in the presence of noise in the components and in the incoming signals. We will use such cochleas in higher-level systems to see what advantages intelligent front-ends can confer at higher levels of a system.