The Brown Ghost electric fish detects prey signals using electric fields generated around its body. Prey detection performance is dictated by the standard trade-off between sensitivity versus accuracy, commonly computed through a receiver operator characteristic (ROC) curve. Performance is both enabled and limited by intrinsic biophysical noise that must be accounted for in the statistical decision-making process involved in prey detection. In this talk I will present modeling and experimental analyses of the biophysical features in the Brown Ghost's sensory neural network that enable the best possible sensitivity and accuracy for prey detection. The main take-away is that standard physiological processes within sensory neurons can, in theory and empirically, compute the optimal decision rules given a ROC analysis for prey detection at the lowest possible levels of stimulus intensity.