Dana Anderson and CU engineers have developed a nonlinear optical dynamical system to process microwave communication signals in a novel and powerful way. The system solves what is popularly referred to as the "cocktail party problem," which refers to the human ability to pay attention to one voice among many despite the presence of noise and other interfering signals. Also known as "blind signal separation," this problem calls for the descrambling of mixed signals with little or no prior knowledge about the signals or how they are mixed. The human brain seems to do this effortlessly but when done by machine, it involves a computationally intensive statistical analysis of the incoming signals. Using a creative synthesis of holography, electronics, and nonlinear dynamics, Anderson has developed a system whose dynamics are mediated by input signal statistics in such a way that the system automatically unscrambles incoming mixed signals. Because of the high speed and parallel nature of the optics, the system is capable of handling the signal bandwidth demands of communications networks.