Extending signal detection theory approaches up the central pathway: Auditory midbrain models

Abstract

The application of signal detection theory (SDT) to peripheral encoding of stimulus parameters provided a quantitative basis for understanding auditory-nerve (AN) responses. Siebert’s development of this strategy, based on relatively simple analytical models, laid the groundwork for Heinz’s extension that took advantage of computational AN models. These studies applied the Cramer-Rao lower bound (CRLB) to estimate just-noticeable differences of stimulus parameters, assuming that underlying neural responses could be described as non-homogenous Poisson processes (NHHP) and that the brain acts as an optimal processor. Krips and Furst carried this approach into the central nervous system, first showing that coincidence-detector neurons that receive excitatory and/or inhibitory NHHP inputs produce outputs that are also NHHP. These models were tested using the CRLB for sensitivity to interaural differences in time and level. Here, we carry the approach a step further along the auditory pathway to the auditory m...